Genetic complementation screening and molecular docking give new insight on phosphorylation-dependent Mastl kinase activation

Abstract Mastl is a mitotic kinase that is essential for error-free chromosome segregation. It is an atypical member of AGC kinase family, possessing a unique non-conserved middle region. The mechanism of Mastl activation has been studied extensively in vitro. Phosphorylation of several residues were identified to be crucial for activation. These sites correspond to T193 and T206 in the activation loop and S861 in the C-terminal tail of mouse Mastl. To date, the significance of these phosphosites was not confirmed in intact mammalian cells. Here, we utilize a genetic complementation approach to determine the essentials of mammalian Mastl kinase activation. We used tamoxifen-inducible conditional knockout mouse embryonic fibroblasts to delete endogenous Mastl and screened various mutants for their ability to complement its loss. S861A mutant was able to complement endogenous Mastl loss. In parallel, we performed computational molecular docking studies to evaluate the significance of this residue for kinase activation. Our in-depth sequence and structure analysis revealed that Mastl pS861 does not belong to a conformational state, where the phosphoresidue contributes to C-tail docking. C-tail of Mastl is relatively short and it lacks a hydrophobic (HF) motif that would otherwise help its anchoring over N-lobe, required for the final steps of kinase activation. Our results show that phosphorylation of Mastl C-tail turn motif (S861) is dispensable for kinase function in cellulo. Communicated by Ramaswamy H. Sarma GRAPHICAL ABSTRACT


Function of Mastl in mitosis
Post-translational regulation by phosphorylation is the major regulator of mitosis. Initiation of mitosis is governed by mitotic protein kinases (Kim et al., 2016). Cdk1 is present at the top of the mitotic signaling cascade (Diril et al., 2012;Santamar� ıa et al., 2007). Cdk1 and its downstream protein kinases activate their mitotic substrates, initiating a signaling cascade. Most of these phosphorylations result in activation of factors that promote mitosis and inhibition of factors that promote mitotic exit (Gharbi-Ayachi et al., 2010). PP2A (protein phosphatase 2A) is the primary antagonist of Cdk1 during mitosis. This phosphatase is indirectly inhibited by Mastl (Microtubule-associated serine/threonine kinase like). Mastl is the human orthologue of Greatwall kinase (Gwl), first identified in Drosophila (Yu et al., 2004). Mastl kinase phosphorylates two highly homologous, small, thermostable proteins: Arpp19 and ENSA. These two phosphoproteins specifically bind and inhibit B55o-bound PP2A. Inhibition of PP2A-B55o results in rapid accumulation of phosphorylations on mitotic substrates, supporting the autoamplification loop (Kim et al., 2012;Kishimoto, 2015;Mochida & Hunt, 2012;Mochida et al., 2010;Slupe et al., 2011).
Due to this crucial function in mitosis, Mastl is an essential protein kinase for cell division and proliferation. Loss-of-function studies have shown that, in its absence, SAC (Spindle Assembly Checkpoint) signaling is weakened, resulting in premature onset of anaphase. Spindle-kinetochore attachment errors cannot be corrected, leading to anaphase bridges and DNA breaks during chromosome segregation (Bisteau et al., 2020;Diril et al., 2016). Anaphase bridges prevent closure of the cleavage furrow, causing mitotic collapse at the end of mitosis (Alvarez-Fernandez et al., 2013).

Activation mechanism of AGC family protein kinases
The common kinase fold is constituted of a b-sheet-rich smaller N-lobe and an a-helix-rich larger C-lobe, connected through a loop segment. This loop encompasses a 20-30 residue-long activation loop that contains the catalytic aspartate within its conserved DFG (Asp-Phe-Gly) motif (Modi & Dunbrack, 2019). The hinge region, glycine-rich loop's backbone, chelated cofactors and a conserved lysine together coordinate ATP through electrostatic interactions (Endicott et al., 2012) (Figure 1(A)).
According to the general kinase activation model, the active DFG-in conformation is induced by activation loop phosphorylation (Adams, 2003;Vijayan et al., 2015). In its active state, activation loop allows the binding of ATP and peptide substrate to their respective docking sites. Simultaneously, catalytic aspartate approaches these binding sites (Supporting Information Figure S1) (Huse & Kuriyan, 2002;Zhou et al., 2010). In AGC family of protein kinases, phosphorylation of activation loop can only partially activate the enzyme. A unique feature of AGC kinases is that the core catalytic elements need to be stabilized through intramolecular allosteric interactions. A series of conformational changes are induced by C-tail/N-lobe interactions. C-tail has two important regions that directly interact with N-lobe. These are the turn motif and the HF motif (hydrophobic motif, FXXF). According to the currently accepted model, conserved turn motif phosphosite aids docking of C-tail to the basic patch of N-lobe. Extending from this initial docking point, HF motif reaches the hydrophobic groove of N-lobe. HF motif enhances hydrophobic packing by inserting the benzene rings of its conserved phenylalanines into this hydrophobic groove. This hydrophobic packing allows C-tail to reposition the N-lobe aC-helix. Ultimately, the conserved glutamate within aC-helix and the conserved lysine located in ATP binding pocket form a salt bridge. As explained previously at the beginning of this section, this conserved lysine functions in coordination of ATP. In other words, C-tail indirectly aids the active site for coordination of ATP through a multistep intramolecular allosteric regulation (Figure 1(B)). Using the crystal structure of an atypical AGC kinase (Arencibia et al., 2017), human PKC iota (PDB entry 3A8W) (Takimura et al., 2010), the complete molecular machinery is visualized in Supporting Information Figure S2. Of note, PKA possesses a C-tail phosphate that does not contribute to intramolecular interactions (Supporting Information Figure S3), making the generalizability of current model of AGC kinase activation questionable.

Activation of Mastl kinase
Mastl is a unique serine/threonine protein kinase among AGC family kinases. It has an unusually long (�500 residues) linker region between its N-terminal and C-terminal lobes. Studies on Xenopus egg extracts have revealed three phosphosites that are critical for activation of Mastl Vigneron et al., 2011). Positions of these residues in mouse Mastl are T193, T206 and S861 (T194, T207, S875 in human; T193, T206, S883 in Xenopus). T193 and T206 are located in activation loop within Cdk1 consensus motifs (proline-directed) and are phosphorylated by Cdk1 in vitro. These two phosphosites are necessary for nuclear export of Mastl before NEBD (nuclear envelope breakdown) (Alvarez-Fernandez et al., 2013). Upon phosphorylation, activation loop shifts from DFG-out to DFG-in conformation. Then, Mastl undergoes cis-autophosphorylation by phosphorylating its turn motif (tail/linker) phosphosite S861 . As explained in the previous section, in AGC kinases, this turn motif phosphoresidue is docked to a basic patch on N-lobe, serving as a means to guide HF motif towards N-lobe hydrophobic groove. However, in the case of Mastl, C-tail of the enzyme is shorter and it does not possess a known HF motif. Therefore, it cannot reach the hydrophobic groove (Figure 1(C)).
Results of Vigneron et al. (2011) demonstrated that activity of constitutively active Mastl increases up to 156% in the presence of a synthetic peptide (PIFtide) that mimics hydrophobic motif. Their finding suggests that Mastl utilizes the HF motif of other AGC kinases in a heteromeric complex to complete its activation. However, it is still not clear what purpose would C-tail/N-lobe intramolecular docking would serve, when the C-tail is devoid of HF motif. This raises the question whether the C-tail phosphorylation (S861) is essential for the cellular functions of Mastl.
In the present study, we reevaluated the biological significance of these three phosphosites and two additional phosphosites (S212 and T727) that were proposed to be contributor to Mastl activation Hermida et al., 2020). Other than these, we tested the hyperactive Drosophila Scant mutation K71M (Archambault et al., 2007), thrombocytopenia-associated mutation E166D (Hurtado et al., 2018), and two different non-conserved middle region (NCMR) deletions (D194-725 and D305-620). We obtained viable Mastl knockout mouse embryonic fibroblasts (MEF) clones that express the ectopic E166D, K71M, S861A or S861D mutants. Given the vital role of turn motif phosphosite S861 in AGC kinase activation, we focused our study on functional characterization of this particular phosphosite.

Software
CLC Main Workbench 7.9.1 software was used for sequence visualization and analysis (https://digitalinsights.qiagen.com). The gel images, immunoblot images and micrographs were processed using Adobe Photoshop and GIMP (version 2.10.12) (The GIMP Development Team, 2019). The plots were drawn in R (v3.6.1) (R Core Team, 2019), invoked by RStudio (v1.2.5001) (RStudio Team, 2015), using the ggplot2 (Kassambara, 2019) R package.  (Zheng et al., 1993). The hinge region and glycine-rich loop backbones, catalytic aspartate, conserved lysine and ATP are shown in stick form. The manganese ions are depicted as transparent spheres. The observable ion-ion (salt bridge), ion-dipole and dipole-dipole (hydrogen bonds only) interactions between the enzyme and ATP are depicted as red, gray and blue colored dashed lines, respectively. (B) The crystal structure of catalytic subunit of ATP-bound c-AMP-dependent protein kinase (PDB entry 3X2W) (Das et al., 2015). The conserved lysine and glutamate, conserved phenylalanine residues of HF motif and ATP are shown in stick form. aC-helix (violet), HF motif (yellow) and the proximal b-strand (transparent gray) together form the hydrophobic groove. Hydrophobic groove is indicated by the transparent yellow area. The magnesium ions are depicted as transparent spheres. The observable ion-ion (salt bridge) and ion-dipole interactions between the enzyme and ATP are depicted as red and gray colored dashed lines, respectively. The blue, orange and red colored atoms are nitrogen, phosphorus and oxygen, respectively. (C) Comparison of the general AGC kinase and Mastl kinase activation models. The bilobal protein kinase structure is depicted as light-brown bubbles. The white indents at top-left and top-right positions of N-lobe indicate the basic patch and the hydrophobic groove, respectively. Activation loop and C-tail are depicted as blue and red lines, respectively. The cartoon illustrations show the difference between activation of Mastl (left) and general AGC kinase activation model (right). For simplicity, extended NCMR is not represented in the illustrated Mastl kinase structures.
Structural refinements, binding energy and buried surface area calculations were performed using the Guru interface of HADDOCK2.2 web server (https://milou.science.uu.nl/services/ HADDOCK2.2/haddockserver-guru.html) (van Zundert et al., 2016). When the docking stages are skipped, HADDOCK2.2 is capable of refining the interaction of separate molecules. To refine the C-tail and N-lobe interactions, the protein kinase chain was separated with the N-terminal starting position of C-tail. C-tail and the rest of the protein were submitted as if they were separate chains.
Following changes were applied to the default docking protocol; � 'Remove non-polar hydrogens?' checkbox was unchecked under the 'Distance restraints' tab. � 'Define center of mass restraints to enforce contact between the molecules' checkbox was checked under the 'Distance restraints' tab. � 'Randomize starting orientations', 'Perform initial rigid body minimization' and 'Allow translation in rigid body minimization' checkboxes were unchecked under the 'Advanced sampling parameters' tab. � 'number of MD steps for rigid body high temperature TAD', 'number of MD steps during first rigid body cooling stage', 'number of MD steps during second cooling stage with flexible side-chains at interface' and 'number of MD steps during third cooling stage with fully flexible interface' parameters were set to 0 under the 'Advanced sampling parameters' tab.
HADDOCK outputs residue-based energy contributions of each interfacial amino acid (expressed in electrostatics, vdW (van der Waals) and electrostatics þ vdW energy terms), deposited in ene-residue.disp file (located under HADDOCK output folder: structures/it1/water/analysis). For each interface residue, various types of scores were extracted from the HADDOCK runs for its 200 different states. These scores were presented in boxplot format in which the energy distribution of residues can be compared to one another ( Figure 6 and Supporting Information Figure S7). The codes and computational methods used in the present study were adapted from our previous study (https:// github.com/CSB-KaracaLab/BINDKIN) .

Molecular cloning
Retroviral expression constructs (pBABE-Puro) for N-terminally HA-tagged mouse Mastl WT, D155N (kinase-dead, KD), E166D (thrombocytopenia-associated mutant) and K71M (Drosophila Scant mutation) were generated in Kaldis Laboratory (Institute of Molecular and Cell Biology, Singapore). The other mutations were created by site-directed mutagenesis. The primers used for site-directed mutagenesis are given in Supporting Information Table S2.

PCR genotyping
A three-primer strategy was utilized for genotyping of Mastl locus, to distinguish FLOX and KO alleles ( Figure 2(A)). The primers used for genotyping are given in Supporting Information Table S2. For PCR genotyping, the samples were prepared according to the adapted HotSHOT protocol (Truett et al., 2000). DNA sample of 0.5 mL and 0.25 units of Taq DNA polymerase (A111103; Ampliqon) was used for each 10 mL of reaction volume. Final primer concentrations were 1 mM for 'FOR' and 'REV1' and 0.15 mM for 'REV2' primers. Thirty-five PCR cycles were performed using 67 � C annealing temperature. Under these optimized conditions, the PCR reaction consistently yields similar intensity bands for Mastl FLOX (300 bp) and Mastl KO (500 bp) alleles from a heterozygous Mastl FLOX/KO cell clone ( Figure 2).

Cell culture, retroviral transduction and stable cell line engineering
PLAT-E cells (Cell Biolabs) and Mastl conditional knockout (CKO) immortalized MEFs (Diril et al., 2016) were grown in DMEM supplemented with 10% FBS and antibiotics. Cells were cultured in a 95% humidified incubator with 5% CO 2 at 37 � C. Lipofectamine 3000 transfection reagent (L3000015; Thermo Scientific) was used for transfection of PLAT-E cells.
PLAT-E culture and virus production were carried out according to the manufacturer's instructions. The viral supernatant was harvested on second and third days post-transfection and stored at 4 � C up to a week until use. Polybrene was mixed with retroviral medium to 8 mg/mL final concentration, immediately before infection of MEFs. MEFs were transduced at 40% confluence. Viral media collected at the second day of transfection were used in the first round of transduction. One day post-transduction, the used viral media were removed and a second round of transduction was performed by using the viral media that was harvested at the third day of transfection. Eight hours after the second round of transduction, the viral media were removed. Cells were cultured further in normal growth medium overnight for recovery, prior to antibiotic selection. The stable cell lines were generated under 2-4 mg/mL puromycin selection. Cells were selected for a week and negative control cells died within the first two days of selection. Mastl knockout was induced by adding 4-OHT to the culture medium at 20 ng/mL final concentration. The control cells were treated with an equal volume of DMSO. The knockout-induced stable cell lines were subjected to limited dilution approximately 6 days post-induction.

Cell viability and proliferation assays
The alamarBlue proliferation assay was carried out in 96-well plate format in three replicates. The daily measurements were initiated one day after seeding the cells. Cells were incubated in 150 mL of assay medium for 4 h. The assay medium was prepared by diluting 1 volume of alamarBlue dye reagent (BUF012A; Bio-Rad) in 9 volumes of growth medium. The metabolic activity was quantified fluorometrically by using 560 nm excitation wavelength and recording the emission at 590 nm. The assay was performed for six successive days.
The proliferation rate was also measured by a modified 3T3 assay (Diril et al., 2012) in 6-well plates. 20,000 cells were seeded per well and cells were counted after 3 days and 20,000 cells were plated again. Counts were repeated for six successive passages.

A robust rescue-assay for evaluation of the Mastl mutants
We used the following workflow in order to assess biological significance of the previously described mutations and NCMR deletions Vigneron et al., 2011). Briefly, (i) CKO Mastl FLOX/FLOX cell pools stably expressing Mastl mutants were generated, (ii) endogenous Mastl gene was deleted by Cre-mediated recombination, (iii) clonal cell lines were isolated and genotyped and (iv) the clones whose endogenous Mastl genes were knocked out were further characterized.
MEFs were infected with retroviral expression constructs (pBABE-Puro) and selected with puromycin, in order to establish stable cell pools expressing ectopic Mastl mutants. To quickly assess complementation capacity of the ectopic constructs, the stable pools were treated with 4-OHT and the cells were collected before and after KO induction (3 and 14 days) were genotyped (Figure 2(A,B)). Due to the nature of the inducible Cre systems, CKO MEFs include a basal level of knockout cells/loci as a result of spontaneous Cre leakage into the nucleus, even without 4-OHT induction (Zhong et al., 2015). Three days post-induction, a complete knockout of Mastl gene was observed. However, within fourteen days, initially undetectable levels of Mastl FLOX/FLOX or Mastl FLOX/KO cells, where endogenous Mastl could not be knocked out, proliferate. Cells escaping knockout is a known disadvantage of the inducible Cre systems (Diril et al. 2012). Cells expressing endogenous Mastl account for majority of the cell population when the ectopic constructs cannot complement endogenous Mastl loss (see day 14 in Figure 2(B)).
Although, nonfunctional Mastl-expressing (i.e., Mastl KD ) cell pools eventually proliferate after 4-OHT treatment, there is a lag compared to the functional Mastl-expressing (i.e., Mastl WT ) cells. Analysis of different Mastl mutants six days after 4-OHT treatment suggests that S861A, S861D, K71M and E166D mutations retain partial or full kinase activity compared to WT (Figure 2(E) and Supporting Information Figure S5). These results are consistent with the genotyping analysis (Figure 2(C,D) and Supporting Information Figure  S4A) which confirms that cells expressing these mutants are viable when their endogenous Mastl is knocked out.
Clonal cell lines were generated after 4-OHT treatment of cell pools expressing various Mastl mutants. PCR genotyping of WT and KD expressing clones confirmed that our experimental system works. When cells had been infected with WT Mastl, in almost all clones both copies of endogenous Mastl were found to be deleted. Whereas, KD-Mastl-infected clones kept at least one copy of their endogenous Mastl ( Figure  2(C)). K71M and E166D Mastl mutants can also complement endogenous Mastl loss (Supporting Information Figures S4  and S5). This is an expected result as these mutations do not have a significant effect on cellular proliferation (Archambault et al., 2007;Hurtado et al., 2018). None of the NCMR deletion clones had homozygous KO genotype, suggesting that the deletions cover catalytic or structural elements that are essential for Mastl function in vivo. This is in line with the previous observations for NCMR deletion mutants (Hermida et al., 2020;Vigneron et al., 2011). Conversely, the number of clones we genotyped for our deletion mutants were limited. Therefore, we cannot rule out the fact that with an increased sample size, a homozygous KO clone may be obtained.
After establishing the complementation screening strategy and successfully testing it on known viable mutations (K71M and E166D), we next focused on Mastl phosphorylation sites with putative roles on kinase activation. We created constructs with point mutations that have non-phosphorylatable (T193V, T206V, S212A, T727V and S861A) or phosphomimetic (T193E, T206E and S861D) residues. We were able to isolate clonal cell lines for S861A and S861D, in which, the endogenous Mastl gene was knocked out (Figure 2(D)). Other mutations did not yield viable knockout cell lines in our screen, as all the clones expressed one or both copies of the endogenous Mastl (Supporting Information Figure S4).

The turn motif phosphosite mutant Mastl is viable
Our complementation screening showed that the turn motif phosphosite mutant-expressing clones (S861A and S861D) were viable when endogenous Mastl gene was deleted. (Figure 2(D)). As previously explained, the turn motif phosphorylation is known to be crucial for AGC kinase activation. Therefore, we focused our study on S861 mutants for functional characterization of this particular phosphosite. We selected one clone from each of the WT, S861A and S861D clonal cell lines for further experiments.
Analysis of the selected clones by immunofluorescence microscopy and western blotting confirmed stable expression of the ectopic constructs ( Figure 3). Although the expression levels of the ectopic proteins were lower compared to endogenous Mastl (Figure 3(B)), this was sufficient for cell division. S861A and S861D clones could still proliferate albeit at a slightly reduced rate (Figure 4). The band upshift of the phosphomimetic 861D may indicate a constitutive activity, resulting in hyperphosphorylation of Mastl. Subcellular localization of the ectopic Mastl variants was analyzed by immunocytochemistry staining using anti-Mastl antibodies. As expected, endogenous and ectopic Mastl WT proteins localized to the nucleus. Mastl S861A and S861D mutants also localized to the nucleus (Figure 3(A)). Therefore, S861 phosphorylation has no effect on Mastl localization.

The Mastl S861 phosphosite mutations do not impair cell proliferation
We have demonstrated that deletion of endogenous Mastl can be rescued when Mastl S861A or Mastl S861D mutants are expressed ectopically. To compare the proliferation rates of these clones to that of control MEFs or Mastl WT-expressing knockout clones, we decided to undertake quantifiable assays.
First, the proliferation rates of these clones were measured indirectly by alamarBlue proliferation assay which measures the overall metabolic activity over time. The results suggest that the mutant Mastl-expressing clonal cell lines have proliferation rates comparable to control cells or Mastl WT clonal cell line ( Figure  4(A)). Next, we measured the proliferation rates directly by a modified 3T3 assay (see Methods). The mutant clones had slightly reduced proliferation rates compared to the control cells or WT clone (Figure 4(B)). We observed that 861A and 861D clones had slightly larger cells compared to parental cells or the WT clone (Figure 4(C)). This finding could be best explained by an increased length of the cell cycle allowing a longer period for cell growth.
These results suggest that, S861 phosphorylation site mutations do not have a lethal negative effect on cell proliferation rate. In fact, the slight reduction in proliferation rates  . Proliferation rates of control, WT, 861A and 861D clones (A) alamarBlue proliferation assay of the control and clonal cell lines. The metabolic activity was fluorometrically measured as a function of resazurin reduction capacity. The cells were monitored for six days. (AFU: Arbitrary Fluorescence Unit) (B) A modified 3T3 assay was performed to directly measure the increase in cell number over the course of six passages. Number of cells at the beginning of the assay were normalized to one. (C) The cells were seeded at low density in order to show their unique morphologies in the absence of cell-cell contact restrictions (upper row). In a 6-well plate, the semi-confluent asynchronous cells were treated with DMSO or 20 ng/mL 4-OHT. The next day, 20,000 cells/well were seeded. The bright field images were acquired two days later (three days in the case of 861A due to its slower growth rate) (lower row). Scale bar is 100 mm.
of the mutant clones is possibly due to the lower expression level of ectopic Mastl proteins (See Figure 3). However, reduced kinase activity, especially for the 861A variant cannot be ruled out.

Mastl kinase C-tail non-phosphorylatable mutant (S861A) is capable of hyperphosphorylation
Xenopus Mastl kinase undergoes hyperphosphorylation during its activation . In mitotically arrested primary MEFs, a slower running phosphorylated isoform appears in immunoblots. The apparent high molecular weight band is due to the decreased electrophoretic mobility of the hyperphosphorylated Mastl protein (Diril et al., 2016).
To investigate whether the hyperphosphorylation can take place in case of C-tail phosphosite-deficient Mastl, we treated the control cells and the ectopic WT Mastl-expressing homozygous knockout clones with nocodazole for different time intervals to arrest cells in mitosis. The Western blots for Mastl protein show that both cell groups display the band upshift ( Figure 5(A)). This means that the ectopically expressed wild-type Mastl can complete its phosphorylationdependent activation.
Next, we asked whether Mastl S861A mutant can become hyperphosphorylated. To that end, we repeated the mitotic arrest and phosphorylation analysis, this time including the 861A clone ( Figure 5(B)). It is observed that 861A clone is also capable of phosphorylation. This finding could be corroborated in HEK293T cells transfected with HA-tagged WT or 861A mutant constructs ( Figure 5(C)). All in all, these results show that 861A is capable of completing its activation steps, suggested by the mitotic arrest experiments ( Figure 5). Which explains why the mutant is able to rescue the cell cycle in Figure 4.

Conformation of the phosphosite defines the C-tail/ N-lobe interactions
Our experimental results suggest that pS861 phosphoresidue is auxiliary for Mastl kinase activation. To mechanistically explain this result, we analyzed the C-tail intramolecular interactions of Mastl and other AGC kinases. Currently, there is only one crystal structure of Mastl (Homo sapiens, 5LOH) available in the PDB. Due to solubility issues, the construct was truncated at its NCMR to obtain the minimal kinase domains (Ocasio et al., 2016). Despite the construct contained C-tail of human Mastl, C-tail was not present in the final structure. Because of this, we decided to find suitable templates for homology modeling of Mastl. To this end, we have collected all available AGC kinase crystal structures that possess a phosphorylated C-tail. According to this criterion, 17 structures of two different AGC kinases are available in PDB, which are PKC iota and PKA (Supporting Information Table S1). The structures fall into two distinct groups based on the conformation of their phosphorylated C-tail residue. The phosphoresidue is either buried in N-lobe, or it is solvent-exposed (Supporting Information Figure S3). Upon observing that there are two distinct C-tail phosphoresidue conformations, we first decided to elucidate which conformation would Mastl pS861 possess (buried or solvent-exposed). For this, we performed pairwise sequence alignments between the kinase domains of mouse Mastl and human PKC iota or mouse PKA (Supporting Information Figure S6). The sequence alignments revealed that mouse Mastl C-tail phosphosite does not correspond to a phosphoresidue in PKA and PKC iota, but it corresponds to an aspartate in PKC iota. To get a detailed insight, we obtained two separate homology models for Mastl, by using PKC iota (3A8W) or PKA (3FJQ) as the templates. Expectedly, the C-tail phosphosite of the model structures possesses the distinct conformation (i.e., solvent-exposed or buried) of their respective templates ( Figure 6). Using HADDOCK2.2 web server, the binding score of C-tail to N-lobe was predicted by using van der Waals score, electrostatics score, desolvation energy and the combined HADDOCK scoring function (Supporting Information Figure S7). The calculations were performed on nonphosphorylatable, wild-type, phosphomimetic and phosphorylated forms of PDB 3A8W (PKC iota), PBD 3FJQ (PKA), 3A8W-based Mastl model and 3FJQ-based Mastl model. Due to the charged interactions between the C-tail phosphoresidue and the N-lobe basic patch residues, we have turned our attention to the electrostatics scores of interfacial residues. As expectedly, when C-tail phosphoresidue is buried  inwards N-lobe, its contribution to the binding energies dramatically increases ( Figure 6).
The energetics analyses expectedly verified the notion that when the phosphoresidue is buried, it greatly contributes to the intramolecular interactions and when it is solvent-exposed, it does not. However, intriguingly, the unique patterns of polar contact distribution unveiled that in PKC iota and PKC-iota-based Mastl model, the contacts are clustered at the center, whereas in case of PKA and PKA-based Mastl model, the contacts are clustered more at the edges of the interface (Figure 6). Given that the PKA C-tail phoephoresidue does not contribute to the C-tail docking, this twosided stitching of its C-tail might be a structural feature that has evolved as a means to promote phosphorylation-independent docking. We then evaluated the structural impacts of nonphosphorylatable or phosphomimetic mutations of Ctail phosphosite. To this end, we selected the HADDOCKrefined structures with the lowest HADDOCK score for the said mutants of PKC iota (based on PDB entry 3A8W) and PKA (based on PDB entry 3FJQ) in order to use for APBS (Adaptive Poisson-Boltzmann Solver) calculations. The surface electrostatics potentials were calculated and mapped onto three-dimensional structures (Figure 7). The electrostatics potential maps show that phosphomimetic glutamate mutation in PKC iota is interacting with basic patch (the blue cavity). When this C-tail phosphosite is mutated into nonphosphorylatable valine, basic patch is expanded. In other words, distance between the edges of basic patch increases. This is because of the fact that, in the absence of a negative charge where there should be the phosphoresidue, basic patch residues cannot engage in a polar contact network. Conversely, in case of PKA, the phosphomimetic or nonphosphorylatable mutations of C-tail phosphosite do not cause a remarkable structural difference. Because, C-tail phosphosite is already tilted away from N-lobe (Figure 7). These results together reinforce our hypothesis that Mastl pS861 phosphoresidue does not contribute to intramolecular Ctail docking. . Phosphomimetic (T564E) and nonphosphorylatable (T564V) mutants of PKC iota and phosphomimetic (S339D) and nonphosphorylatable (S339A) mutants of PKA were generated using HADDOCK2.2. For each four mutant, post-simulation structure with the lowest HADDOCK score was subjected to APBS calculations. The surface electrostatics potentials were calculated and mapped onto three-dimensional structures. Red indicates negative potential and blue indicates positive potential. In each structure, C-tail is represented as white cartoon. The 90 � rotated state of C-tails are given on top of their respective structures in order to show the surface electrostatics potentials of each interface in detail. The C-tail binding electrostatics scores that were obtained from HADDOCK2.2 are given on top of each structure. Lower electrostatics score corresponds to stronger binding.

Conclusion
In AGC kinases, activation loop and C-tail phosphorylations are crucial for kinase activation. Activating phosphorylations govern the conformational transition from inactive to active state through a chain of molecular interactions. One feature that makes Mastl a unique AGC kinase, other than its long NCMR domain, is that, it has an N-lobe hydrophobic groove, yet it is devoid of a hydrophobic motif (HF) (FXXF) to bind it. In other words, Mastl's C-tail is partially missing. Another protein kinase that shares this unique feature is PDK1. PDK1 circumvents this structural shortfall by binding the HF motifs of other AGC kinases to achieve complete activation (Biondi, 2001;Biondi et al., 2000). Consistently, Vigneron et al. (2011) demonstrated that a partially active mutant form of human Mastl (K72M) could achieve 156% of its basal activity in the presence of a synthetic peptide (PIFtide) that contains HF motif. Given the lack of HF motif in Mastl and given that Mastl possibly compensates for this by utilizing the HF motif of other kinases, one cannot directly assign a biological function to its C-tail phosphosite.
Although AGC kinases have been studied for decades, the role of C-tail phosphosite in the final activation step is still a mystery for atypical AGC kinases. To date, several studies have sought to characterize the activation steps of Mastl kinase using different experimental approaches Hermida et al., 2020;Vigneron et al., 2011). Ultimately, C-tail phosphorylation was proposed to be one of the essential steps for Mastl activation. Vigneron et al. (2011) used mitotic CSF egg extracts and depleted endogenous Xenopus Greatwall kinase which induces mitotic exit . They supplemented these extracts with in vitro-transcribed human Mastl mRNAs coding for various mutants. The mutants were evaluated for their potential to restore the mitotic state by Western blot analysis of Cdk phosphorylations on mitotic substrates. Additionally, they performed kinase assays to measure kinase activity of recombinant Mastl variants using MBP (myelin basic protein) as substrate. Results from both assays showed that, the activity of 875A mutant (861A in mouse) was significantly reduced (31% of WT activity on MBP). Nonetheless, there was a residual activity compared to the kinase-dead mutant G44S (5% of WT).
Blake-Hodek et al. (2012) essentially used the same experimental methodology. However, they produced the different Mastl mutants (Xenopus) by baculovirus infection of Sf9 insect cells. They quantified the kinase activity by measuring Mastl autophosphorylation or using MBP as substrate in kinase assays. They also observed a significant decrease in the activity of 883A mutant (861A in mouse). Nevertheless, the activity of Mastl 883A mutant was more than the kinasedead mutant G41S (Xenopus).
Finally, in their elaborate phosphoproteomics study aiming to analyze the mechanism of MASTL activation and regulation, Hermida et al. (2020) used recombinant human Mastl kinase (WT or kinase-dead mutant) and mitotic HEK293 cells extracts. They identified the differentially phosphorylated sites between the variants, verifying that S875 is a cis-autophosphorylated residue. Although their observations show that S875 phosphorylation occurs on human Mastl, whether it is essential for activity was not investigated.
In this work, we have used Mastl CKO MEFs to remove the endogenous Mastl kinase and sought to rescue its loss by various mutants. The primary readouts of our system, cellular proliferation and autophosphorylation, show that S861 phosphorylation is not essential for activation or mitotic division. The apparent discrepancy between our results and the aforementioned papers could be due to two reasons: (i) The impaired activity of Mastl 861A mutants may be sufficient to drive mammalian cells through mitosis. (ii) Xenopus CSF egg extracts constitute a simplified model system and do not fully reflect the elaborate regulatory network responsible for mitotic entry in mammalian cells. Therefore, our model system allows us to functionally characterize the Mastl variants at the level of cell fate and our observations are directly relevant to mitosis.
Taken together, our present results render the generalizability of final step of current AGC kinase activation model (Arencibia et al., 2013;Hauge et al., 2007;Kannan et al., 2007) questionable for AGC kinases that are devoid of an HF motif. We hope that our results encourage researchers in the field to engage in deeper structural and enzymological studies for atypical AGC protein kinases. Conversely, we have demonstrated the complementation cloning strategy as a robust assay that allows cell culture-based mutagenesis screening of essential proteins. ME and MKD wrote the paper. All authors contributed to careful revising of the manuscript.