Reaching a pocket in FZD4 that was thought to be closed
Triple-negative breast cancer has no ER, no PR and no HER2 amplification, so hormone and HER2 drugs do not apply. It is 10 to 20% of breast cancers and the most aggressive of them.
Question
Wnt/β-catenin signaling keeps these tumors proliferating and helps them resist chemotherapy, and Frizzled receptors start the cascade. Blocking Frizzleds broadly has already been tried in patients. The antibody vantictumab hits five of them at once and showed real responses in a phase Ib trial, but it caused bone loss, because the same pathway maintains healthy bone.
That argues for a small molecule that binds one receptor at its own core. The problem is where that core sits. FZD4 is a class F GPCR, its orthosteric pocket is buried inside the seven-helix transmembrane bundle, and the bulky cysteine-rich domain on the outside was assumed to shield it. FZD4 was called undruggable for that reason. We asked whether the pocket can be reached anyway.
Process
Our chemistry collaborators in Chieti and Rome built 31 compounds on indole, pyrazole and indazole scaffolds, starting from lonidamine and indomethacin. My job was to find out what they do in cells.
I cultured HEK293 and 4T1 cells and transfected them with PEI. The first readout was DVL1 recruitment. Without FZD4, DVL1-GFP sits in bright dots inside the cytoplasm. When FZD4 is there, it moves to the plasma membrane, so a compound that breaks the receptor-effector contact leaves the dots in place, and the difference can be measured. Next was a TCF/LEF-GFP reporter read on a plate reader, turned on with WNT5A-conditioned medium, to see whether transcription follows. Then Western blots for β-catenin and MTT assays for viability in 4T1 cells.
Result
One compound of the 31 carried the series. It blocks DVL1 recruitment with an EC50 of 0.10 µM and shuts down TCF/LEF transcription with an IC50 of 0.35 µM, so the block at the receptor reaches all the way to the nucleus. In 4T1 cells β-catenin drops, the Wnt target genes c-myc, tcf, lef-1 and cyclin D1 go down with it, and viability falls at 30 µM over 48 hours.
Docking and point mutants made by the collaborating groups placed the compound inside the transmembrane bundle, held by an aromatic cage of three tyrosines. Mutating them costs the compound its activity but leaves the receptor answering normally to its natural ligand. The orthosteric core of FZD4 is reachable, which makes it a target rather than a dead end.
