A first generation inhibitor of human Greatwall kinase, enabled by structural and functional characterisation of a minimal kinase domain construct

MASTL (microtubule-associated serine/threonine kinase-like), more commonly known as Greatwall (GWL), has been proposed as a novel cancer therapy target. GWL plays a crucial role in mitotic progression, via its known substrates ENSA/ARPP19, which when phosphorylated inactivate PP2A/B55 phosphatase. When over-expressed in breast cancer, GWL induces oncogenic properties such as transformation and invasiveness. Conversely, down-regulation of GWL selectively sensitises tumour cells to chemotherapy. Here we describe the first structure of the GWL minimal kinase domain and development of a small-molecule inhibitor GKI-1 (Greatwall Kinase Inhibitor-1). In vitro, GKI-1 inhibits full-length human GWL, and shows cellular efficacy. Treatment of HeLa cells with GKI-1 reduces ENSA/ARPP19 phosphorylation levels, such that they are comparable to those obtained by siRNA depletion of GWL; resulting in a decrease in mitotic events, mitotic arrest/cell death and cytokinesis failure. Furthermore, GKI-1 will be a useful starting point for the development of more potent and selective GWL inhibitors.


Supplementary Videos -HeLa cells treated with:
SV1: scrambled siRNA control; SV2: 25 µM GKI-1, SV4: 50 µM GKI-1, SV4: siGWL The supernatant arising from this step was applied to a batch / gravity column containing 10 ml of Amintra Glutathione Resin [Expedeon Ltd, Harston, UK] pre-equilibrated in Buffer A. The column containing the cell extract and resin was rotated / rolled at 4 °C for a period of 1-2 hours to facilitate protein binding, and then allowed to pack under gravity flow. The resin was then washed with the addition of 10 column volumes (CV) of Buffer A to remove any unbound material, then re-suspended in an equal volume of Buffer B: 50 mM HEPES.NaOH pH 7.5, 500 mM NaCl, 150 mM KCl, 10 mM MgCl 2 , 10 mM ATP, 0.5 mM TCEP, and rotated / rolled for a further period of 2 hours at 4 °C. The resin was once again allowed to pack under gravity flow, followed by an additional wash with a further 5 CV of Buffer A. Rhinovirus 3C-protease was then added to the resin slurry (200 µl of a 2 mg/ml stock solution) and the cleavage reaction allowed to proceed overnight, with rotation / rolling at 4 °C as before.
The following day, the resin was again allowed to pack under gravity flow, but this time collecting the flowthrough. Fractions containing hGWL-KinDom were identified by SDS-PAGE, pooled, and then concentrated using centrifugal ultra-filtration [

Phasing, model building and refinement
All diffraction data were collected at 100K. Data were integrated using the software package XDS (1), and then processed using the Pointless / Aimless / Ctruncate pipeline of the CCP4 software suite (2)(3)(4)(5). Phases were obtained by molecular replacement using Phaser (6) with a polyalanine model based on PDB: 1H1W (PDK1; (7)) as a search model. An iterative combination of manual building in Coot (8) and refinement with Phenix.refine (9) produced the final model.

Thermal shift assay
For thermal denaturation, samples containing protein at 1.5 M and 5 x SYPRO Orange (diluted from a 5000 x stock supplied in DMSO; Sigma-Aldrich, Gillingham, UK) were prepared in sample buffer: 20 mM HEPES.NaOH pH 7.5, 300 mM NaCl, 0.5 mM TCEP. Denaturation curves were monitored in 96-well PCR plates using a Roche LightCycler 480 II, with 465 and 580 nm filters for excitation and emission wavelengths, respectively. The program was as follows: 1 min at 20°C, followed by a continuous increment of 0.03°C/s to a final temperature of 85°C. Temperature midpoints (Tm) for each folded to unfolded transition were determined by non-linear regression fitting of a modified Boltzmann model to normalized data in Prism 6.0 (version 6.0h, GraphPad Software, La Jolla, CA USA).
Where: a n and a d are the slopes, and b n and b d the y-intercepts, of the native and denatured baselines respectively. T m is the temperature midpoint of the transition from native to denatured states, and m represents a generic slope factor.

Kinase Assays
Non-radioactive kinase assays were set-up by combining hGWL-KinDom (305 M in 20 mM HEPES pH 7.5, 500 mM NaCl, and 0.5 mM TCEP) diluted to desired concentration or immunoprecipitated hGWL FL

Radioactive Kinase Assay
To detect ENSA phosphorylated at Ser67 (p-ENSA), the standard kinase reaction was spiked with 0.075 MBq -32 P ATP (PerkinElmer, Seer Green, UK) per 20 l reaction. After the reaction was stopped with 5 l 5X SDS loading buffer and boiled for 5 min at 95 °C the mixture was resolved via SDS-PAGE (4-15% Criterion pre-cast gels; Bio-Rad Laboratories, Hemel Hempstead, UK, or 13% SDS-polyacrylamide gels). Staining with coomassie-blue revealed an ~17 kDa band that was imaged by autoradiography. All kinase assays were repeated thrice per assay condition and each point on the concentration-dependent response curves represents the mean ± 1 S.D.

Kinase-Glo Max Assay
After stopping the above non-radioactive kinase assay with 1 volume of Kinase-Glo Max reagent the mixture was allowed to incubate for a further 15 min. To detect consumption of ATP, samples were transferred to white, flat-bottom 96-well plates [Thermo Fisher Scientific, Loughborough, UK] and luminescence was detected using a POLARstar Omega micro-plate reader [BMG Labtech GmbH, Ortenberg, Germany]. As this assay measures ATP, activity or ATP turnover was measured as the -ΔRLU. Non-linear regression using Prism 6.0 and the built-in enzyme kinetics module were used to determine Michaelis-Menten kinetic parameters, k cat and K m , for both ATP and ENSA. All kinases assays were conducted thrice in triplicate (n = 9) and IC 50 values are reported as the mean ± 1 S.D.

Immunoprecipitation Kinase Assay
HEK293T cells were transfected with 10 g of plasmid (Flag-tagged hGWL FL or GFP; pcDNA3-EGFP, Addgene, Cambridge MA, USA) using CaCl 2 . 48 hours later 100 ng/ml nocodazole was added to the media and incubated for a further 16-18 h. Cells were collected and washed in PBS-A followed by lysis on ice for 20 min in 1mL IP buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 10% v/v glycerol, 0.5% v/v NP-40, 1 mM EDTA supplemented with protease and phosphatase inhibitors (Complete and PhosStop; Roche Diagnostics)). Samples were then sonicated and clarified by centrifugation at 13,000 rpm for 30 min at 4°C. To capture Flag-tagged hGWL FL , 100 l of lysate was diluted 5-fold with IP buffer and combined with 3 l anti-Flag M2 magnetic Dynabeads [Sigma-Aldrich] overnight at 4 °C. After sequential washes with IP buffer, 1:1 mix of IP and kinase assay buffers and a final wash in kinase assay buffer, the beads were then used in the kinase assay. The reaction was stopped by adding SDS loading buffer and analysed by western blot: p-ENSA and GWL levels were detected using the anti-phospho(Ser67)-ENSA and anti-MASTL(GWL, Prestige) antibodies. Activity was measured as the p-ENSA/GWL ratio and % kinase activity was normalised to the DMSO control. All kinases assays were repeated thrice per assay condition and each point on the concentration-dependent response curves represents the mean ± 1 S.D.
PKA, ROCK1 and CDK2 kinase assays PKA and ROCK1 assays (Promega) were carried out according to the manufacturer's protocols and CDK2 assays were carried out with Histone H1 as substrate (1 g per 20 l reaction) according to our Kinase-Glo Max assay. CDK2 was expressed and purified from E. coli as previously described (10).

Inhibitor Screens with hGWL-KinDom
Two small-molecule compound libraries, enriched with pharmacophores known to interact with the ATPbinding site of protein kinases, were kindly provided by GlaxoSmithKline and Roche. We used this pooled library (582 compounds, 10 M screening concentrations) in a Kinase-Glo Max luminescence kinase assay (11,12) as part of a two-point screen against hGWL-KinDom kinase activity, using ATP at fixed concentrations of 15 M and 45 M; ~1-and 3-fold the Km value for ATP respectively. With STU as positive control at a concentration of 200 M, Z-factor values for screening plates ranged from 0.41 to 0.69 (Supplementary Figure  4A,B). A threshold value of 3 was used to identify 34 hits, all with inhibitory activity greater than 30% (hit rate = 5.8%; Supplementary Figure 4A).

Molecular Docking -AutoDock
AT13148 and GKI-1/2 were docked into the ATP-binding site of the hGWL-KinDom crystal structure (where the disordered C-helix was homology-modelled) using AutoDock 4.2.6. PDB format files for the ligand and kinase domain were pre-processed using AutoDock Tools 1.5.6 as described in Mohamed, et al. (13). (Figure 4 and Supplementary Figure 6A,B).
Immunofluorescence and Time-lapse video microscopy 2.5 X 10 5 HeLa cells were plated onto glass coverslips in a 6-well plate, or 4-well -slides [Ibidi GmbH] were seeded with 5 X 10 4 cells and then transfected with Qiagen AllStars negative control or Hs_MASTL_6 siRNAs (14)  images per cover slip. Data were normalised against the AllStars negative control siRNA treated cells and values obtained from three biological replicates were reported as the mean ± 1 S.D. To assist with selecting mitotic cells, ScanR imaging software was utilised to gate for mitotic cells using the circularity factor and total DAPI parameters.
To produce time-lapse videos, after 48 hours and cell treatment with DMSO or compound (cells were treated with GKI-1 for 4-h before image acquisition) cells plated in 4-well -slides were imaged every 5 minutes for 8.5 hours on an Olympus IX73 microscope, using a 40 x 0.95NA air objective, and recorded on an Orca-Flash 4.0 CMOS camera [Hamatsu]. The computer program Fiji (http://www.fiji.sc/Fiji) was used to generate videos from image sequence TIF files. Data were normalised against the AllStars negative control siRNA treated cells and values obtained from 3 -5 biological replicates were reported as the mean ± 1 S.D.     69 (B). To identify hits, a threshold of 3 standard deviations of the DMSO controls was set, which led to identification of 34 'hits', all with an inhibitory activity greater than 30%; hit rate of 5.8%. Figure S6. Confirmatory screen of GR1-1-4 using immunoprecipitated hGWL FL Immunoprecipitated Flag-tagged hGWL FL was treated with increasing concentrations of (A) GRI-1, (B) GRI-2, (C,E) GRI-3 and (D,F) GRI-4 and kinase activity was measured using the immunoprecipitation kinase assay. Non-linear regression with Prism 6.0 was used to determine IC 50 values.

Figure S7. Inhibition of hGWL-KinDom by GRI-3 and GRI-4
hGWL-KinDom was treated with increasing concentrations of GRI-3and GRI-4 and kinase activity was measured using the Kinase-Glo Max assay. Non-linear regression with Prism 6.0 was used to determine IC 50 values. Immunoprecipitated hGWL FL was treated with increasing concentrations of GKI-2 and kinase activity was measured using the immunoprecipitation kinase assay. (D) % kinase activity was normalised to the DMSO control and plotted against GKI-2 concentration using Prism 6.0.