Objective/Rationale:
Parkinson’s treatment on the molecular level involves the removal of the insoluble protein molecules called the ‘Lewy bodies’ formed by the misfolding of these proteins.This can be done by using correcting proteins called ‘chaperones’ like ‘heat shock protein70’. In our project we would like to find compounds, which will initiate the production of correcting proteins with minimal side effects.
Project Description:
In this project, an Artificial Neural Network (ANN) will be trained using the molecular orbital energies and dipole moments as input and the IC50 (concentration of a drug that is required for 50% inhibition of the enzyme) values for Geldanamycin (GD) and itsderivatives as the output layer. Once an ANN has been trained, any molecule having a similar structure as GD (but not necessarily already synthesized) can be used as a potential candidate for testing and its IC50 value can be predicted using the ANN. Thiswill reduce the number of molecules, which will need to be synthesized, and therefore the experimental costs incurred to carry out assays to identify potential targets like GD.
Relevance to Diagnosis/Treatment of Parkinson’s Disease:
Current therapies for Parkinson’s involve mostly a Symptoms & Side Effects treatment by replenishing a compound called ‘dopamine’ which is lost due to the death of nerve cells in the brain. Our project will try to find out new compounds, which will help prevent the death of these nerve cells.
Anticipated Outcome:
We hypothesize that we can predict the structure of new molecules as potential medications for the treatment of Parkinson’s disease which will have the ability to mediate the neurodegenerative effects of the disease with minimal negative side effects, by using the procedures developed in our lab.
Final Outcome
Parkinson’s treatment on the molecular level involves the removal of the insoluble protein lumps called “Lewy bodies” formed by the misfold of certain proteins. Geldanamycin (GD) binds to and inhibits the molecular protein chaperone Hsp90, inducing the production of another chaperone molecule, Hsp70. The ability of Hsp70 to prevent neurotoxicity is due to its activity of refolding misfolded proteins and its interaction with co-chaperones and the parkin-associated ubiquitinylation pathway.
An Artificial Neural Network (ANN) was trained using orbital energies and dipole moments as inputs and the GD IC50 values (concentration of a drug that is required for 50% inhibition) and its derivatives as the outputs. Once the ANN was trained, compounds having a common fragment as GD were identified from the National Cancer Institute database and were used as test compounds. The IC50 values of these test compounds were predicted using the ANN. This will give us an idea about their inhibitory activity, and will reduce the number of molecules that need to be synthesized, reducing the experimental costs incurred to carry out high throughput screening assays to identify potential inhibitors of Hsp90.