Validation of a small molecule inhibitor of PDE6D-RAS interaction with favorable anti-leukemic effects

RAS mutations prevalent in high-risk leukemia have been linked to relapse and chemotherapy resistance. Efforts to directly target RAS proteins have been largely unsuccessful. However, since RAS-mediated transformation is dependent on signaling through the RAS-related C3 botulinum toxin substrate (RAC) small GTPase, we hypothesized that targeting RAC may be an effective therapeutic approach in RAS mutated tumors. Here we describe multiple small molecules capable of inhibiting RAC activation in acute lymphoblastic leukemia cell lines. One of these, DW0254, also demonstrates promising anti-leukemic activity in RAS-mutated cells. Using chemical proteomics and biophysical methods, we identified the hydrophobic pocket of phosphodiester 6 subunit delta (PDE6D), a known RAS chaperone, as a target for this compound. Inhibition of RAS localization to the plasma membrane upon DW0254 treatment is associated with RAC inhibition through a phosphatidylinositol-3-kinase/AKT-dependent mechanism. Our findings provide new insights into the importance of PDE6D-mediated transport for RAS-dependent RAC activation and leukemic cell survival.


LC-MS/MS Analysis of Peptide Mixtures
The peptides were analyzed by nanoLC-MS/MS using an Ultimate 3500 RSLC System (Dionex) coupled to a Q-Exactive Plus mass spectrometer (Thermo

In-vitro Photolabeling of PDE6D with PAL Probe and LC-MS/MS Analysis
In vitro competitive photoaffinity labelling Proteins were then precipitated using the chloroform-methanol method described by Wessel and Flügge (Wessel and Flügge, 1984) and the precipitated protein pellets were air-dried for 10 min at room temperature.

Preparation of Labelled PDE6D for MS-Analysis
Recombinant human His-TEV-PDE6D-Avitag protein (50 µM, 2.5 nmol) in 100 µL PBS was preincubated with 50 µM of DW0254 or DMSO for 5 min and then isolation window 2 m/z; 10 MS2 scans per full scan. Statistical differential analysis was performed using Limma ("Linear Models for Microarray Data") t-test (Smyth, 2004) and Benjamini-Hochberg false discovery rate correction was applied for multiple testing correction. Proteins with a FDR < 5% and a fold change of at least 2 were selected to be differentially modulated. A protein was considered as a target hit of DW0254 when identified with at least two peptides in at least 2 out of 3 replicates, FC > 2 and adjusted p-values < 0.05 in the two comparisons, PAL/DMSO and PAL/PAL+DW0254.

MS Data Processing for In-vitro Photolabeling of PDE6D with PAL Probe
The raw files were processed with the MaxQuant software for peptide and protein identification and quantification. MS/MS raw files of the digests were searched using the Andromeda search engine against a database containing only the recombinant human His-TEV-PDE6D-Avitag sequence using the following parameters: carbamidomethylation of cysteine was set as fixed modification whereas N-terminal acetylation and methionine oxidation were set as variable modifications.
All peptides were required to have a minimum peptide length of seven amino acids and a maximum of two miss cleavages. Specificity for Glu-C cleavage was required allowing cleavage after glutamate and aspartate. The false discovery rate (FDR) for protein and peptide identifications was set to a maximum of 1%. To validate and transfer identifications across different runs, the "match between runs" option in The algorithm performs an unbiased search for modified peptides that are derived from an already identified peptide. If an unidentified spectrum matches an identified spectrum, the mass shift (corresponding to the modification of the peptide) between the theoretical and the observed precursor mass and the matched sequence will be reported. Modified peptides will be only identified if they are derived from an already identified unmodified peptide with an FDR of 1% and a mass tolerance of 6.5 mDa.
Modified peptides were extracted from allPeptides.txt along with the ΔM mass shift between the unmodified "base peptide" and the modified peptide. All amino acids were considered as possible residues for modification. The mass of the modification used to search for probe-modified peptides was +581.3002 m/z for PAL which is the mass for the corresponding probe minus two nitrogen atoms. This modification was set as a variable modification in all MaxQuant searches. In brief, for "dependent peptides" analysis, the "all.peptides.txt" file was loaded and filtered for DP Proteins = "sp|DPE6D|", DP Mass Difference = 581.3002 +/-10 ppm and DP Score ">60".
Selected peptides with a DP mass shift corresponding to the photoadduct with a tolerance of 10 ppm and which are only present in the two conditions "PAL" and "PAL+DW0254" and absent in the control "DMSO" were considered as positive hits.
Remaining hits were further validated in a manual fashion. MS spectra were visualized with the Xcalibur software to validate the presence of the unmodified and modified peptides. Ideally, the unmodified peptide should be detected in all three conditions whereas the peptide modified with a photoadduct should be detected in the condition "PAL" and to a lesser extent in the condition "PAL+DW0254" but not in the control "DMS0". MS2 spectra were visualized using the viewer program of MaxQuant to annotate y and b ions of the unmodified peptide. MS2 spectra of the unmodified and modified peptides of interest were analyzed using XCalibur to determine the position of the photo adduct in the sequence. A mass shift corresponding to the photo adduct on a y and/or a b ion is expected.

Computational Analysis of PDE6D mutants and PAL docking
All computational analysis was carried out using MOE. Homology models of the PDE6D R48delV49del reported in Figure 5E were built by using the apo structure of PDE6D as template. Default homology modelling settings in MOE were used. Ten homology models per mutants were built and the best scoring model according to the GB/VI function was selected. Induced fit docking of DW0254 and Deltarasin were carried out. The bioactive conformation of the two ligands was extracted from the respective co-crystallized structure with PDE6D, while positioning in terms of translation and rotation was randomized. The binding site was identified by the superposed 3D coordinates of DW0254. The triangle matcher placement method was used to generate 1000 initial poses disabling the conformational search of the ligand as the bioactive conformation was already available. The generated poses were scored with the London dG function, the 30 best scoring poses were submitted to the induced fit refinement stage allowing residues within 6 Å of the ligand to move.
The refined complexes were scored with the GBVI/WSA dg function, and the best scoring one was finally selected. Usually, an RMSD of the docked ligands to crystallographic coordinates lower than 2 Å is considered a successful docking An initial docking pose for PAL was obtained by manually extending DW0254 with the linker attached to D37. The same energy relaxation protocol described above including minimizations and MD simulations was applied to refine the complex.

Bioluminescent imaging for DW0254 ex vivo efficacy studies
To generate a cell line with luciferase expression, P12-ICHIKAWA cells were infected with Lenti-FUW-Luc-mCh-puro virus and selected in liquid culture with puromycin (Sigma-Aldrich) 2.5µg/mL for 7 days following mCherry + cell sorting.
All animal studies were approved by the Boston Children's Hospital or Dana-Farber Cancer Institute Animal Care and Use Committee. 8-to 10-week-old NODscid IL2Rgamma null (NSG) mice (Jackson laboratories, Bar Harbor, ME) were sublethally irradiated with 300 cGy and injected with 1x10 6 luciferase expressing P12-ICHIKAWA cells treated for 12 hours with DMSO or 3µM DW0254. Disease burden was assessed using bioluminescence imaging starting six days after injections. Prior to imaging, each mouse was given an intra-peritoneal (i.p.) injection of luciferin (PerkinElmer, Part Number #122799) at a dose of 150mg/kg body weight. General anesthesia was then induced with 5% isoflurane and the mouse was placed in the light-tight heated chamber; anesthesia was continued during the procedure with 2% isoflurane introduced via nose cone. Both prone and supine images were recorded.
Optical images were displayed and analyzed with the Igor (WaveMetrics, Lake Oswego, OR) and IVIS Living Image (Xenogen) software packages. Regions were manually drawn around the bodies of the mice to assess signal intensity emitted. Optical signal was expressed as photon flux, in units of photons/s/cm 2 /steradian. The total photon flux for each mouse was calculated as the sum of prone and supine photon flux.