Cytoskeleton microtubules have a critical role in mitosis, offering an established
target for antiparasitic drug development. Despite toxicity, the natural product (NP)
colchicine from Colchicum autumnale, is considered a classical antimitotic. Moreover, structural knowledge of the β-tubulin
colchicine-binding site led to safer and successful therapeutics. Such information
provides opportunity to establish in silico-in vitro workflows searching for new lead
compounds. To discover potential tubulin inhibitors, a pharmacophore-based virtual
screening approach was utilised in LigandScout and Discovery Studio programmes. Structure-based
models were built on the PDB crystal information of β-tubulin/colchicine-binding domain.
Additionally, ligand-based models were designed from known inhibitor NPs scaffolds.
Model optimization was performed with datasets of known actives -including NPs- and
corresponding decoy set. Virtual screening of vendor databank (SPECS) led to predicted
hits prioritised for in vitro testing based on consensus overlap, fit scores, and
drug-likeness. To validate virtual hits, 41 commercially available compounds were
tested in a cell-free fluorescence-based tubulin polymerisation assay (Cytoskeleton).
Test compounds (30 or 10 µM) were initially discriminated as active/inactive. Inhibition
observed across varying scaffolds led to a hit-rate of 26.83%. Dose-response experiments
showed a benzimidazole derivative as most active (IC50=2.9 µM) with potency in range of colchicine for other structural classes. Next, inhibitors
could be screened in cytotoxicity or antiparasitic assays for biological activity.
Validated models recognising NPs scaffolds can be used cross- functionally to identify
bioactives in extracts. Results demonstrate this coupled in silico-in vitro approach
offers an efficient tool to identify NPs and synthetic tubulin inhibitors [1]
[2]
[3]
[4].
Funding EUREGIO (Interregional Project Networks) (IPN 119) ‘HERBAL’.
