Graduation Year

2015

Document Type

Dissertation

Degree

Ph.D.

Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Medical Sciences

Major Professor

Yu Chen, Ph.D.

Committee Member

Burt Anderson, Ph.D.

Committee Member

David Kang, Ph.D.

Committee Member

Lindsey Shaw, Ph.D.

Committee Member

Edward Turos, Ph.D.

Keywords

beta-lactamase, molecular docking, virtual screening, penicillin binding proteins, X-ray crystallography, slingshot phosphatase

Abstract

Structure based drug design is a rapidly advancing discipline that examines how protein targets structurally interact with small molecules, or known inhibitors, and then uses this information to lead inhibitor optimization efforts. In the case of novel inhibitors, protein structural information is first obtained via X-ray crystallography, NMR studies, or a combination of both approaches. Then, computational molecular docking is often used to screen, in silico, millions of small molecules and calculate the potential interactions they may have with the target protein’s binding pocket, in hopes of identifying novel low affinity inhibitors. By examining the interactions these small, low affinity, inhibitors have with the binding pocket, optimization efforts can be focused on maximizing interactions with “hot spots” within the pocket, thus leading to larger, high affinity inhibitors. A similar optimization technique can also be applied to known inhibitors. By examining the interactions of a known inhibitor with the binding site, new compounds can be designed to target “hot spots” in the binding pocket using the known inhibitors core structure as a starting point. The affinity of the newly designed compounds can then be compared to the affinity of the original inhibitor, and further rounds of optimization can be carried out. While simple in design, there are many challenges associated with structure based drug design studies, and there is no guarantee novel inhibitors will be found, but ultimately, it is an extremely powerful methodology that results in a much higher hit rate than other, similar, techniques. The work herein describes the use of structure based drug design to target several different proteins involved in bacterial antibiotic resistance, and a protein that has been implicated in the development of Alzheimer’s disease.

The goal of the first project was to design a new PBP inhibitor based upon an existing scaffold, and to better understand the binding mechanism and molecular interactions between penicillin binding proteins and their inhibitors. PBPs are a group of proteins that catalyze the last steps of bacterial cell wall formation, and are the targets of the β-lactam antibiotics. Two compounds were designed which conjugated a ferrocene or ruthenocene group to 6-aminopenicillinic acid, and their antibiotic properties were tested against a range of bacterial strains. To get a better understanding of how the 6-APA organometallic compounds interacted with the PBP active site, a CTX-M-14 β-lactamase model system was used for X-ray crystallographic studies. CTX-M-14 was chosen as its active site shares many key catalytic features with PBPs, and it easily, and reproducibly, yields crystals capable of diffracting to sub-atomic (< 1.0 Å) resolution.

I determined a 1.18 Å structure of 6-APA-Ru in complex with CTX-M-14 E166A β-lactamase and was able to gain unprecedented details of the interactions of the ruthenocene group with the CTX-M active site. This structure also revealed that the compound bound in the CTX-M active site was actually the decarboxylated and hydrolyzed product, which was the first time a decarboxylated product had been captured in the CTX-M active site. A second, 0.85 Å, structure of CTX-M in complex with 6-APA-Ru was determined and shed light on how the hydrogen bonding network in the CTX-M active site changes in response to the 6-APA-Ru product binding. A final, 1.30 Å, structure captured the carboxylated and hydrolyzed 6-APA-Ru product in complex with CTX-M, which was the first time the carboxylated product had been captured in the CTX-M active with the catalytic Ser70 residue intact. The results show the potential of the ruthenocene group in improving antibiotic potency, and help to better elucidate the changes that occur in the CTX-M active site upon inhibitor binding, while at the same time, telling us what changes could occur in the active site of PBPs.

The next project was focused on novel inhibitor discovery against several different PBPs. PBPs have been successfully inhibited by β-lactam antibiotics for decades, but the alarming rise of bacteria resistant to these antibiotics has placed increased urgency on the discovery of novel PBP inhibitors. A fragment based molecular docking approach was employed to virtually screen millions of small compounds for interactions with the targeted active sites, and then high scoring compounds were selected for visual inspection and inhibitory testing. Virtual screening was first done against Staphylococcus aureus monofunctional transglycosylase, a type of PBP. MTG provided a good binding pocket for virtual screening, but proved challenging to purify and crystallize. However, through great effort MTG crystals were eventually obtained. After repeated rounds of virtual screening against MTG, multiple compounds were selected for inhibition testing, and testing is currently ongoing. Virtual screening was also done against Pseudomonas aeruginosa PBP5 and PBP1a. Purification and crystallization of these proteins proved to be easier than MTG, and both yielded diffraction quality crystals.

The final project focused on virtual screening against a protein implicated in the development of Alzheimer’s disease, Slingshot Phosphatase 1. The brains of AD patients have been found to contain elevated levels of active Cofilin, and these elevated levels of active Cofilin may lead to the overproduction of amyloid β. Aβ overproduction, and its resulting accumulation, is believed to be one of the pathways that lead to AD symptoms. Cofilin is activated when it is dephosphorylated by SSH1, and inhibiting this activation may decrease the production of Aβ and the development of AD symptoms. There is no known structure of SSH1, so to perform virtual screening a SSH1 homology model was constructed using the homolog SSH2 as a starting point. Virtual screening was then performed using the SSH1 homology model and many compounds were selected for inhibition testing. Initial testing found several compounds that could prevent Cofilin dephosphorylation at levels > 10μM. However, three compounds were found to be exceptionally active, and could prevent Cofilin dephosphorylation at both 1 and 10 μM. One of these three compounds was tested directly against purified SSH1 and found to inhibit its activity, and reduce Aβ production. Crystallization of purified SSH1, and SSH2, was attempted in order to get complex structures with the three best compounds. SSH2 crystals were obtained which diffracted to 1.91 Å, and several initial hits were found for SSH1. Optimization of crystals for both proteins is currently ongoing. The SSH1 inhibitor, along with the two other highly active compounds, provides an excellent starting point for the development of highly potent SSH1 inhibitors.

Download

Share

COinS