BREED: Generating Novel Inhibitors through Hybridization of Known Ligands(A. C. Pierce, G. Rao, and G. W. Bemis)

Richard S. L. Stein

CS 379a

February 21, 2006

Examining Protein-Ligand Interactions

• Comparison of more than a small number of protein-inhibitor complex structures is impossible manually

• Hundreds of complex structures are available and more are being discovered constantly

• Combining fragments of known ligands is a promising approach for finding new compounds

• Manual recombination of fragments is tedious and can take prohibitively long

The Role of BREED

• Automates the recombination of fragments of known ligands to produce new inhibitors

• Generates a large number of new compounds from a small number of initial ligands

• Applied to HIV-1 aspartyl protease and protein kinases

Methodology• All ligands are saved in a

database with implicit hydrogen atoms

• Ligands are examined pairwise to find matching bonds

• Starting molecules are divided into two parts on either side of any matching bond

• New compounds formed by mixing & matching parts of initial molecules

• No bond contained in a ring may be used as a matching bond for this process

• Above steps can be repeated to form further hybrid compounds

Initial and Generated Compounds from HIV-1 Protease Inhibitors

Activity of Newly Generated Ligands

• Compound 5 was synthesized and found to have HIV-1 protease inhibitory activity (IC50 = 160 nM)

• Compound 6 was synthesized with a modification (aniline benzoxadiazole) and found to have high HIV-1 protease affinity (Ki = 0.1 nM)

• A variation of compound 7 was synthesized (t-butyl sulfamide N,N-dimethyl sulfamide) and found to have affinity Ki = 42 nM

Novel Structural Features of Newly Generated Ligands

• Compound 8 exhibits a hydroxy-ketone backbone, a new functional group for HIV-1 protease inhibitors, which has been used in inhibitors of renin (a related aspartyl protease)

• Compounds 10 & 11 have the backbones of initial structures 2 & 3 but have exchanged side chains, suggesting substituent transferability

• Compound 12 is a “second-generation” result produced by taking initial structure 3 and replacing each end with groups from initial structure 1

Initial and Generated Compounds from Protein Kinase Inhibitors

• New compound 24 produced from original structures 15 & 16 has been found to possess inhibitory activity against protein kinase p38 (IC50 = 160 nM)

• Fewer compounds generated by BREED overall (abundance of rings limits the possible number of matching bonds)

Advantages of BREED

• Efficiently automates recombination of molecule fragments to produce novel and effective inhibitors, avoiding tedious manual analysis

• Relatively small number of initial structures required to generate large number of possibilities for compounds that may have enzyme inhibitory action

• Use of experimentally determined ligands for initial structures eliminates need to search for docking conformation

Drawbacks to BREED

• Structures of experimentally known compounds (ligands) still required for the process to get started

• Few new structures generated for ring-rich inhibitors (ex. macrocycles/steroids) because of non-ring matching bond requirement

• Some hybrids formed by conformational alterations that are missed by the current BREED algorithm (ex. the HIV-1 protease inhibitor below)

Sample Discussion Questions• (i) What changes could be made to the BREED algorithm

to find ligands such as the protease inhibitor that given in example, without generating many inactive compounds?

• (ii) What should be the major criteria for deciding which of the generated structures to first synthesize and test? Ex.:– Binding affinity– Novelty– “Druglikeness”– Solubility– Absorption into the circulatory system– Metabolic stability– Toxicity considerations