Results and Discussion

Structural Analysis of Single-Point Mutations

Structural Analysis of Homo-tetramers

DNA Docking with CDKN1A

Quality check of WT structure:

Docking Results:

Discussion of Results

The prediction of monomer structures for each single-point mutation and calculation of RMSD values compared to wildtype (WT) allowed for initial analysis of these mutations’ impacts on tertiary structure. ESMFold was used as a prediction tool over AlphaFold because it handles point mutations slightly better due to its training on a larger protein database and their evolutionary relationships. While global RMSD calculations provide minimal information on local structures surrounding each mutant site, general trends could still be extrapolated. For example, the largest RMSD was found for the C176Y mutant which matches expectations. The original cysteine residue is involved in zinc coordination, so mutagenesis at this site to tyrosine interrupts zinc binding which we predict would destabilize the associated region, affecting activity. The predicted structures were qualitatively analyzed by zooming into mutated regions of the structure to see smaller changes in amino acid bond angles and location within surrounding environments. 

Calculations of RMSD for homotetrameric structures of each p53 mutant showed similar trends to their monomer structures. The largest structural differences were observed in the C176Y, H193Y, and R273C mutants. The inflated RMSD values were attributed to structural differences in the external loop region, which were predicted with extremely low confidence. To account for this, residues outside of the DNA binding domain were removed while maintaining the remaining predicted structure. RMSD values were recalculated to be over tenfold smaller, better reflecting the conformational change in the regions of interest. Because the tetramers were predicted using AlphaFold-Multimer, which struggles to account for structural changes from single point mutations, it’s likely that the differences between the wild type and mutants were understated. It should also be noted that any error caused by this circumstance could have influenced the results of the docking studies. 

DNA docking studies completed with PyDockDNA show how each tetramer was predicted to bind the CDKN1A DNA sequence. PyDockDNA ranks protein-DNA conformations based on electrostatics, Van Der Waals energy, and desolvation energy. To test for the quality of these computations, the WT structure generated by PyDockDNA was first compared to a previously solved crystal structure10. The conformations were conserved between the two, so we continued to use the WT structure as our baseline. While the structures visually appeared similar, slight changes were observed by the RMSD values. In designing our experimental methods, we intended to find a program that could use our monomer PDBs to predict the tetramer conformations. However, with AlphaFold-Multimer being the only accessible program, there’s a potential disconnect between the tertiary and quaternary structures.

Overall, accounting for conformational changes in higher-order protein structures resulting from single point mutations proved difficult. AlphaFold-multimer in tandem with PyDockDNA demonstrated to be useful tools in predicting protein-DNA interactions, as shown by the similarity in experimentally-determined crystal structure and predicted structure of the WTp53-CDKN1A complex. In future experiments, it would be ideal to employ predictive tools that better account for slight sequence changes. Models like AlphaFill have already shown potential in transplanting ligands and metal cofactors into AlphaFold outputs, providing an extra layer of molecular context. ESMFold’s implementation of evolutionary relationships is crucial for model accuracy at a monomer level, and would prove necessary for predictions similar to ours. In conclusion, discrepancies in our results emphasize the necessity for precise structure prediction tools to accurately predict complex, higher-order protein-ligand interactions.