Physics-Based Modeling of Engineered Biopolymers for Monitoring Gene Expression in Disease

Jeffrey Michael Sanders, Thomas Jefferson University


One of the longstanding goals of biology is to explain physiological processes at the molecular level in living organisms. Knowledge of the chemical and biological cues that drive both normal and abnormal cellular functions in their native environments would greatly increase our understanding of nature. The highly interdisciplinary field of molecular imaging has become an important area of scientific research as the techniques can be applied to monitor biological changes in living organisms. Molecular imaging researchers draw upon the fields of engineering, chemistry, biology and physics to develop molecular probes to study diseases in vivo. Probes can be designed to target specific physiological or genetic signatures thought to be involved in disease progression. Genetic imaging seeks to identify changes in the genetic landscape as a result of disease and identify key players. This knowledge can then be used to guide drug development or monitor response to therapy. In cancer, genes that drive tumor development, also called oncogenes, can be monitored in vivo using oligonucleotide-based imaging probes. The time and cost to identify a suitable imaging probe can be prohibitive. While many cancers share common genetic and proteomic features, some are unique. In the past, computer aided drug design (CADD) has been utilized to discover small molecules for cancer treatment. CADD uses first principles of physics to model how drugs interact with biological targets. Millions of small molecules can be virtually screened in a fraction of the time required to test them experimentally. These computational methods are just now beginning to be applied for molecular imaging probes. In this study, molecular modeling techniques were applied to identify biopolymers capable of targeting genetic changes in disease. In order to computationally design imaging probe molecules, molecular docking and dynamics simulations were utilized to identify molecules capable of binding to cellular receptors with known roles in disease. Using molecular dynamics simulations, the binding process of epidermal growth factor receptor (EGFR) ligands was studied. The EGFR receptor is important in numerous cellular processes and is commonly misregulated, in some cases through ligand activation, in cancer. Molecular Mechanics Poisson Boltzmann Surface Area (MM-PBSA) methods were utilized to determine the relative binding affinities for the seven EGFR ligands from simulation. Relative affinities calculated were in qualitative agreement with experimental binding affinities: EGF>HB-EGF>TGF-α>BTC>EPR>EPG>AR. We next sought to use the computational approaches taken with EGFR and ligands for a clinically relevant target. The neuropilin (Nrp) co-receptor is overexpressed in disease, and peptides targeting Nrp receptors have been previously identified. Flexible docking methods were applied to peptide sequences taken from endogenous Nrp ligands. Docking results could not rank the five peptide sequences studied. MD simulations of docked peptides with Nrp1 were performed to generate more binding poses. To calculate the binding energies, steered molecular dynamics (SMD) simulations were incorporated. SMD simulations are used to calculate the nonequilibrium work required to dissociate a ligand from its receptor. To validate SMD, binding affinities were measured using isothermal titration calorimetry (ITC). A strong correlation was observed between SMD binding energies and those from ITC. The peptide sequence identified in this study with the highest affinity for Nrp1 was RPARPAR. This sequence can potentially be used as a scaffold for our dual specific imaging probes. To target genetic changes, synthetic oligonucleotides called peptide nucleic acids (PNAs) were used. PNAs are oligonucleotide derivatives with a peptide-like backbone and nucleobases as side chains. Due to their favorable pharmacokinetic properties, PNAs can be exploited as genetic imagine agents. Using quantum mechanical simulations, we developed force fields to simulate PNAs capable of targeting KRAS2 oncogenic mRNA sequences. PNAs with a promiscuous nucleobase capable of base pairing with adenine, guanine, cytosine and uracil were modeled and characterized for the PNA molecule's ability to target multiple oncogenic KRAS2 sequences. Using accelerated MD simulations, thermal stabilities based on MM-PBSA values were computed and compare experimentally to those obtained using circular dichroism. Hypoxanthine-containing PNAs were predicted to have a higher thermal stability when paired with mutant KRAS2 sequences while the wild type sequence melting temperature were predicted to be on the order of mismatch PNA-RNA duplexes. This suggests that in the case of genetic imaging probe development, computational modeling can potentially screen many biological targets and accurately predict high affinity molecules without the need for large scale experimental screening.

Subject Area

Condensed matter physics|Biophysics|Computer science

Recommended Citation

Sanders, Jeffrey Michael, "Physics-Based Modeling of Engineered Biopolymers for Monitoring Gene Expression in Disease" (2013). ETD Collection for Thomas Jefferson University. AAI3604995.