Dyr726, a brain-penetrant inhibitor of PI3Kα, Type III receptor tyrosine kinases, and WNT signaling

Abstract

The vast majority of clinical small molecule multi-kinase inhibitors (mKI) report abject failures in targeting cancers with high stem cell contents like high-grade glioma and colorectal cancers. The FDA-approved mKIs to date ablate receptor tyrosine kinase signaling but do not target the paradoxical WNT signaling which is a key survival driver for the self-renewing cancer stem cells. The WNT pathway enhances cancer plasticity and triggers relapse of highly heterogenous tumours. Using de novo synthesis and structure-activity-relationship (SAR) studies with blood-brain-barrier (BBB) penetrant mKI scaffolds, we designed a highly potent and selective small molecule inhibitor of PI3Kα, PDGFR/KIT, and the WNT pathway denoted Dyr726. Dyr726 is superior to clinical mKIs and inhibits PI3K-AKT-mTOR and WNT-pathway signaling at multiple nodes thereby impeding proliferation, invasion, and tumour growth. Phospho-proteomic, structural, and target engagement analyses, combined with in vitro, in vivo efficacy, and pharmacokinetic studies reveal that Dyr726 is a brain-penetrant small molecule which effectively reduces tumour volume and extends survival of murine orthotopic models. Our current work establishes a first-in-class brain penetrant small molecule mKI which simultaneously antagonize the PI3K-AKT-mTOR and WNT pathways in preclinical cancer stem cell cultures, adult and pediatric primary organoids, and orthotopic murine models with positive efficacy in combination with clinical standard of care.

Figure 1:

Dyr726 is an efficacious multi-kinase inhibitor (A) Kinase profiling of at 1 μM was carried out at the International Centre for Protein Kinase Profiling (http://www.kinase-screen.mrc.ac.uk/). Please also refer to Supplementary Table T1 and Fig S1 for extended data on kinase selectivity using DiscoverX platform. (B) Dyr726 potently kills a panel of genetically diverse primary patient-derived adult and pediatric glioma cells between 0.3-3 μM EC₅₀. (C) 2,000 U87-MG or GBM102 cells were seeded per well with their respective cell culture media in a U-Shaped-Bottom Microplate for 2-3 days to allow spheroids to form. Once spheroids had formed, they were isolated and embedded into Geltrex^™. Complete medium was placed on top with either DMSO or 500nM Dyr726. Images were taken after 3 days on the EVOS XL core microscope. The areas of invasion were quantified using FIJI. **** P < 0.0001 2-way ANOVA, mean +/− SD from n = biological replicates. (D) GBM12 and GBM76 neurospheres were allowed to form for 14d and 7d respectively. Once formed neurospheres were treated with Dyr726, crenolanib, avapritinib, alpelisib, capivasertib, abemaciclib, or DMSO control for 14d with the indicated concentrations. After further 14d, neurosphere diameters were measured using FIJI. (E) Colony formation assay was carried out with MGMT methylated GBM22 cells with or without the treatment with 1 μM Dyr726, 50 μM temozolomide (TMZ), and combination (Combi) of both over 7 d. Cells were fixed with 100% EtOH and stained using crystal violet solution. 24h later, the crystal violet solution was reconstituted using 10% acetic acid. Absorbance was measured at 590nm using the TECAN plate reader. (F) G7 cells were pretreated with Dyr726 or DMSO control for 1hr prior to irradiation at the indicated doses. 14d post irradiation at the indicated Gy doses, colonies were fixed and stained and manually counted using FIJI.

Figure 2:

Dyr726 specifically inhibits PI3Kα and gain-of-function mutants. (A) PI3Kα, PI3Kβ, PI3Kγ, PI3Kα E545K, PI3Kα E542K, PI3Kα H1047R, and AKT1 were assayed using the indicated concentrations of Dyr726. The IC₅₀ graph was plotted using GraphPad Prism software. (B) GBM6 cells were treated with increasing concentrations of Dyr726 for 1hr. Cells were lysed, and immunoblotting was performed using the indicated antibodies. (C) GBM215 cells were plated at 10,000 cells per well and were incubated for 7d to form neurospheres. Once neurospheres were formed, they were treated with indicated concentrations of Dyr726 for 14d. FIJI was used to measure the diameter of the neurospheres. * P < 0.05; **** P < 0.0001 one-way ANOVA, mean +/− SD from n = biological replicates. (D) HSJD-DIPG-007 cells were plated at 10,000 cells per well and were incubated for 1wk to form neurospheres. Once neurospheres were formed, they were treated with indicated concentrations of Dyr726 for 2wks. FIJI was used to measure the area of the neurospheres. **** P < 0.0001 one-way ANOVA, mean +/− SD from n = biological replicates. (E) EGFR amplified GBM143 cells were treated with hEGF and the indicated concentration of Dyr726 or DMSO control for 2hr. Cells were then lysed and immunoblotting was performed using the indicated antibodies. (F) Volcano plot depicting the log2 of fold phosphorylation change versus −log10(p-value) for phosphosites on secretory pathway proteins. The phospho-peptides marked in blue displayed a statistically significant increase while those in red displayed decrease in GBM6, GBM22, and GIN28 cells treated with 10 μM Dyr726 for 0.5hr. TMT-6-plex phospho-proteomics was carried out using three biologically distinct primary glioblastoma cultures.

Figure 3:

Dyr726 inhibits type III RTK signaling (A) EGFR, PDGFRα, PDGFRβ, KIT, FLT3 were assayed using the indicated concentrations of Dyr726. The IC₅₀ graph was plotted using GraphPad Prism software. (B) GBM6 cells overexpressing c-KIT WT were serum starved for 2hr, then treated with SCF indicated inhibitors. 2hr later cells were lysed and immunoblotting was performed using the indicated antibodies. (C) GBM22 cells overexpressing PDGFR WT were serum starved for 2hr, then treated with PDGF-BB and indicated inhibitors. 2hr later cells were lysed and immunoblotting was performed using the indicated antibodies. (D) HSJD-DIPG-007 cells were treated with PDGF-BB and the indicated concentration of Dyr726 or DMSO control for 2hr. Cells were then lysed and immunoblotting was performed using the indicated antibodies. (E) SU-DIPG-VI cells were treated with PDGF-BB and the indicated concentration of Dyr726 or DMSO control for 2hr. Cells were then lysed and immunoblotting was performed using the indicated antibodies. (F) GBM102 cells were treated with the indicated concentrations of either Dyr726, crenolanib, avapritinib, nilotinib, imatinib, or DMSO control for 7d. After 7d, cells were fixed with 100% EtOH and stained with crystal violet solution. 24hr later, the crystal violet solution was reconstituted using 10% acetic acid. Absorbance was measured at 590nm using the TECAN plate reader. Viability was measured as a fold change compared to DMSO treated cells. (G) ARPE-19 mitoQC cells were treated with 300nM Dyr726, Crenolanib, or DMSO for 24hr. Cells were then stained with Hoechst and mitolysosomes were imaged. Images were analyzed using the macro mito-QC Counter in FIJI. Mitolysosome size was quantified, ** P > 0.01 *** P < 0.001 (one-way ANOVA, mean +/− SD from n = 3 biological replicates).

Figure 4:

Dyr726 targets PDGFRA and PI3K activating mutations (A) GBM22 WT or mutant overexpressing PDGFRA cells were treated with either 3μM Dyr726, avapritinib, crenolanib, or DMSO control for 7d. After 7d, cells were fixed with 100% EtOH and stained with crystal violet solution. 24h later, the crystal violet solution was reconstituted using 10% acetic acid. Absorbance was measured at 590nm using the TECAN plate reader. Viability was measured as a fold change compared to DMSO treated cells. (B) GBM22 cells overexpressing PDGFRA WT, D842V, Y288C, T674I, P345S, V561 were serum starved for 2hr, then treated with PDGF-BB and inhibitor. 2h later cells were lysed and immunoblotting was performed using the indicated antibodies. (C) Indicated GBM22 WT or mutant cells were plated at 10,000 cells/well. 24hr later they were treated with 3μM of Dyr726, alpelisib, Osimertinib, or DMSO control. 1 week later, cells were fixed with 100% EtOH and stained using crystal violet solution. 24h later, the crystal violet solution was reconstituted using 10% acetic acid. Absorbance was measured at 590nm using the TECAN plate reader. Viability was measured as a fold change compared to DMSO treated cells. (D) MCF10A WT, H1047R, or E545K mutant cells were treated with either DMSO, or Dyr726 or alpelisib, or lapatinib. After 7d, cells were fixed with freshly made 4% paraformaldehyde (PFA). Cells were stained with crystal violet solution and then left to air dry overnight. Plates were scanned and then FIJI was used to analyze colonies. This was performed by cropping to an individual plate and converting to a binary image. The fill holes, watershed and analyze particles functions were then used to count colonies. (E) Representative images from (D) shown. (F) MCF10A WT, E545K, or H1047R mutant cells were treated with indicated concentrations of Dyr726, alpelisib, or lapatinib for 1hr. Cells were then lysed and immunoblotting was performed using the indicated antibodies.

Figure 5:

Dyr726 ablates the WNT pathway through DYRK/CLK inhibition. (A) DYRK1A, DYRK1B, DYRK2, DYRK3, CLK1, CLK2, CLK3, and GSK3β were assayed using the indicated concentrations of Dyr726. The IC50 graph was plotted using GraphPad Prism software. (B) Proteasome activity in total cell lysates from GBM22 cells with or without 10 μM 726 treatment for 2 h was measured with Suc-LLVY-AMC or Ac-RLR-AMC or Ac-GPLD-AMC. *P < 0.05, **P < 0.01 (compared to control treated, 2-way ANOVA, mean ± SD from n = 3 biological replicates). (C) HEK cells transiently expressing GFP-SF3B1-NT were treated with the indicated concentrations of 726 or 1 μM AnnH31 for 48 h. The phosphorylation state of SF3B1 was determined by immunoblotting with pThr434 antibody, and total amount of GFP-SF3B1 was utilized as loading control. (D) Comparison of Dyr726 treatment on TopGFP HCEC WNT reporter assay with or without either CHIR99021 (+GSK3i) or WNT3A treatment. The EC₅₀ graphs were plotted as mean±SEM using GraphPad Prism software. (E) TopGFP WNT reporter assay was utilized to benchmark inhibitory activity of 726 against known WNT inhibitors that has undergone clinical trials. (F) Relative expression of WNT target genes were carried out in a WNT qPCR array in GBM22 cells upon 10 μM Dyr726 treatment for 24 h. (G) Relative expressions of indicated transcripts were carried out in the indicated cells upon 10 μM 726 treatment for 24 h. (H) GBM22 cells treated with or without indicated concentrations of Dyr726 for 18 h. Later cells were lysed and immunoblotting was performed using the indicated antibodies.

Figure 6:

Dyr726 reduces ex vivo organoid and in vivo tumour volume (A) Indicated WNT-dependent patient-derived CRC organoids were treated with indicated concentrations of Dyr726 and cell viability were measured after specific time points using CellTiter-Glo^® 3D reagent. Data was generated as fold of untreated form 3-6 biological replicates. (B) Representative images of specific organoids from (A) shown (C) Pharmacokinetic profile of Dyr726. Unbound brain exposure (ng/mg) with unbound blood exposure (ng/mL) are plotted at the indicated times (x-axis) following a single dose of 25 mg/kg (n = 3 mice per time point). The biochemical IC₅₀ for each of the key targets is overlayed with the PK plot. Also see Supplementary Fig S2. (D) 2,000 GL261 cells were surgically injected intracranially into C57BL/6 mice of either gender. 5d post-surgery, mice were randomized and treated with either vehicle (n=14), Dyr726 25mg/kg (n=8), or abemaciclib 25 mg/kg (n=8) 3x weekly. At designated endpoint of 10% weight-loss, mice were culled, and Kaplan-Meier curve was derived (P value derived from survival curve comparison using Mantel-Cox Log-rank test). (E) 12,000 GBM12 cells were surgically injected intracranially into J:NU mice of either gender. 14d post-surgery, mice were randomized and treated with either vehicle (n=5), or Dyr726 25 mg/kg (n=6) 3x weekly. At designated endpoint of 10% weight-loss, mice were culled, and Kaplan-Meier curve was derived (P value derived from survival curve comparison using Mantel-Cox Log-rank test).