(482f) Azide and Alkyne Charmm Parameterization: Resolving Dihedral Interactions That Contain Linear Moieties

Drug-like small molecules that contain azido and akynyl groups are structurally unique because they contain linear angles useful in therapeutics, protein functionalization and biomarking, and in bioconjugation reactions like "click" reactions. The click reaction has proven to be integral in drug design and in research with unnatural amino acids (uAA); however, these technologies rely heavily on molecular modeling. Improving the the accuracy of these models will lead to more accurate simulations and better rational design.

The CHARMM force field (FF) is a standard when it comes to all-atom protein and DNA modeling. CGenFF is the leading program used to generate CHARMM parameters for "drug-like" molecules that have not been explicitly modeled. However, both these FF lack appropriate parameterization to simulate any molecules that contain linear angle moieties, such as azides and alkynes, and current literature does not contain any solution for modeling these structures.

We propose a method to address this issue by (1) developing CHARMM parameters for four small molecules that contain terminal azido and alkynyl groups using the Force Field tool kit (ffTK), (2) addressing linear structure issues seen during ffTK analysis by modifying the procedure so that all CHARMM parameters can be solved, and (3) validating those results via in silico molecular dynamic (MD) simulation. We then combine these parameters with CGenFF to generate parameters for all-atom protein simulation with uAA mutations.

In this presentation we describe the process used to obtain CHARMM parameters for our selected small molecules, with a particular emphasis on determining and validating the dihedral parameters for linear moieties since these have proven very difficult to parameterize in the past. During validation, we prove that the physics of the linear terminal dihedral angles are well-represented by a large number of parameter sets without compromising molecular structure. Next, we prove the reliability of our parameter set by running MD simulations to prove our modeled structures match those found in literature and quantum theory. Protein MD next compares all-atom protein simulation with crystallographic data using a protein that contains an azide uAA mutation within its structure.

The final portion of the presentation extrapolates the all-atom parameters to generate Go-like coarse-grain uAA parameters useful for functionalized protein design. We concluded by showing uAA mutation effects on protein structure, stability and activity.

The results of this presentation resolve issues with linear moiety parameterization using the CHARMM force field, and provide an unprecedented view into how unnatural amino acids influence protein structure.