In 2018, we reported the discovery of the first WWP2 inhibitors (Watt et al., 2018). An ELISA based autoubiquitination assay was used to screen 1593 compounds from the NCI Diversity Set V and 450 compounds from the NCI Oncology Set V. The assay set up works by anchoring GST tagged WWP2 FL to glutathione coated wells. UbcH7 charged with flag ubiquitin is then added to the well and WWP2 undergoes autoubiquitination. The wells are then treated with an anti- flag HRP antibody conjugate. HRP has the ability to mediate the hydrogen peroxide dependent oxidation of TMB, resulting in a colour change which can be read at 450 nm. It is this read out which is used to measure autoubiquitination activity in the presence of potential inhibitors (Fig 5.1). The initial screen yielded 10 hits and subsequent concentration dependant assays calculated IC50s ranging from 10.28-0.38 µM.
The next step required confirmation of binding to the catalytic HECT domain via an orthogonal biophysical technique. STD and CPMG (Carr-Purcell-Meiboom-Gill) NMR experiments confirmed NSC2805, NSC650438, NSC369066, NSC3064, NSC194308 and NSC288387 to be binding to the HECT domain. The additional compounds which did not show binding by ligand NMR does not necessarily mean they are not binding to WWP2, as only the HECT domain is used for these experiments. Whilst it may be interesting to look at these compounds in the future, the main focus was to target the HECT domain and prevent its catalytic activity. The absence of unambiguous crystallographic data to elucidate binding mechanisms and inform analogue design of these compounds meant alternate techniques were required. The in silico docking of compounds represents the obvious alternative in the generation of structural data. The results from in silico docking can be strengthened by ligand NMR techniques.
Fig. 5.1 -Schematic of the WWP2 autoubiquitination assay
5.1.2
Ligand NMR
5.1.2.1 STD Epitope maps
STD NMR (descibed in Section 3.1.1.3) can be used to produce ligand epitope maps (Mayer and Meyer, 2001), assisting in the validation of a computational derived protein ligand model. Ligand epitope maps reveal the binding mode of the bound ligand in the binding pocket. The ligand protons which are closer to the protein protons receive more saturation and the relative STD intensity can be analysed to build the epitope map. Epitopes are derived from the initial rate of STD build up (STD0), which is calculated by fitting an exponential curve of saturation time against STD intensity. Once the curve has been fitted, the STD0for a selected proton can be calculated using the equation B shown in Figure 5.2. The STD0value is normalised
by dividing each value by the largest one. Mapping these values onto the ligand produces the ligand binding epitope map (Walpole et al., 2019).
Fig. 5.2 -Equations used for calculating ligand epitope maps
5.1.2.2 DEEP STD NMR
Monaco et al. (2017) reported the development of a novel STD NMR technique named DiffErential EPitope mapping (DEEP), utilising the nature of surrounding amino acids to validate ligand binding poses, aiding further validation of an in silico model. Orientation of aliphatic, aromatic and polar amino acids around the binding sites can be determined. DEEP STD works on the basis of selectively saturating a particular type of proton environment and if that type of proton is present in the binding site, ligand protons close to those particular residues will receive more saturation relative to protons close to other residues. STD data sets are acquired at two different on-resonance frequencies. A comparison of aliphatic against aromatic residues is done by running an on-resonance experiment at 0.6 ppm (aliphatic) and a second on-resonance experiment at 6.55 ppm (aromatic). The DEEP STD factor for each proton is calculated by the STD intensity ratio of the two experiments, minus the average ratio. A positive factor indicates the ligand protons are close to an aliphatic region, whilst a negative factor indicates the protons are close to aromatics. Proximity to polar residues are calculated by use of differential solvents as opposed to differential saturation frequency. The first DEEP STD experiment is performed in D2O and the second in H2O. The ∆STD is calculated as previously mentioned. Polar protons have the ability to exchange with the solvent, therefore the polar residues will contain deuterium in the D2O experiment, which is
inactive in NMR and protons in the H2O experiment, which are active in NMR. Interpretation of calculated ∆STD from these experiments means negative value indicates ligand protons are close to polar residues as the STD signal in the H2O experiment would be stronger (Monaco et al., 2017)(Walpole et al., 2019).
Fig. 5.3 -Equation used to calculate the DEEP STD factor
5.1.3
Experimental aims
Early hit- to- lead development of compounds in drug discovery is when initial hits are improved to have higher affinity and greater selectivity to the target. By using the techniques described, a computational derived NMR validated protein ligand model can be established. NSC288387 was picked out from the 10 initial WWP2 inhibitor hits for further analysis as it gave the clearest STD signals, suggesting a clear binding epitope map could be built and used to inform analogue development. NSC288387 is a flavin compound consisting of a tricyclic isoalloxazine core with a N-substituted phenyl ring.
The NMR validated in silico model can then to be used to drive analogue design and synthesis, with the aim of bringing the ligand activity down to the sub micromolar range. IC50s of the synthesised analogues are to be initially generated for WWP2. This information can be used to investigate SAR for a putative binding site. Activity is then to be tested against Nedd4 to test for specificity.
N N N N O O O Fig. 5.4 -Structure of NSC288387