DOI
https://doi.org/10.25772/755R-FN70
Author ORCID Identifier
0000-0002-6075-853X
Defense Date
2024
Document Type
Dissertation
Degree Name
Doctor of Philosophy
Department
Chemistry
First Advisor
Brian Fuglestad
Abstract
Membrane proteins act as gateways to cells and are responsible not only for signaling, lipid transport, and cellular homeostasis, but they are also integral in disease pathologies. One class of membrane proteins, peripheral membrane proteins (PMPs), reside in aqueous compartments of the cell until they are recruited to the membrane to perform their function. Glutathione peroxidase 4 (GPx4) is a PMP enzyme responsible for engaging the membrane and reducing lipid hydroperoxides, thereby preventing cellular death via ferroptosis. Due to its central function surrounding ferroptosis, GPx4 is an interesting drug target that has recently gained interest for inflammatory diseases as well as drug-resistant cancers. Despite their importance, nearly all target-based ligands screens are conducted against PMPs, including GPx4, in their water-soluble state. However, the aqueous state of PMPs is generally an inactive state. Major roadblocks for PMP inhibitor discovery originate from technical challenges with screening against proteins bound to membranes, an issue that needs to be circumvented to target these proteins in their active state. The previous screens for GPx4 have resulted in the discovery of covalent inhibitors, which have poor selectivity and drug-like qualities, but they circumvent issues with GPx4 targeting due to its shallow active site surface. However, non-covalent inhibitors need to be pursued to eventually lead to a drug that specifically targets GPx4.
One important step to being able to effectively design activators or inhibitors for GPx4 is a more thorough understanding of its function at high-resolution. While its cellular functions are well characterized, molecular details at high resolution are currently lacking, particularly in its functional membrane-bound state. The first NMR assignments of GPx4 are reported and an experimental based model of GPx4 engaging with a membrane are presented utilizing a wide variety of membrane models, including those incorporating native substrates for GPx4. Tryptophan fluorimetry was used to determine the affinity of GPx4 to liposomes, as well as to experimentally determine that this interaction is primarily driven by electrostatics. Mutational analysis of the enzyme confirms the importance of a cationic patch for membrane interaction and lipid substrate engagement. Additionally, a moonlighting function of GPx4 was investigated by the first direct observation of an interaction between the enzyme and DNA, highlighting its nuclear function to crosslink DNA-bound protamines. The results presented here offer the highest resolution yet of intermolecular GPx4 interactions. However, despite the insights gleaned with the use of these membrane models, drawbacks arose throughout the studies. Micelle and bicelle models showed reduced protein stability and extended NMR experiments due to slower tumbling of the large assembly, leaving room for the development and optimization of better membrane models for protein NMR investigations.
To circumvent the issues seen with micelle and bicelle models, phosphocholine-based surfactants were used to formulate a novel DLPC:DPC reverse micelle (RM) system that can act as a membrane model for future PMP studies, confirmed by mapping membrane embedment of GPx4 and other proteins. The novel DLPC:DPC formulation is the first step in developing RMs as more biologically accurate membrane models that can be tuned depending on membrane chemistry. The new system has promising applications for membrane interfacial studies with proteins as well as the possibility to screen for small molecule binders for membrane associated proteins.
To test if the DLPC:DPC membrane-mimicking reverse micelles (mmRMs) are capable of housing a membrane-associated protein while screening for drug building blocks that target the membrane-bound state, a novel screening method was developed using GPx4 and the mmRM system against a small fragment library. This screening method allowed for the successful identification of nine confirmed hits for the protein within its membrane interface. Apparent Kd values were extracted using NMR titrations and revealed high-quality fragment binding. Partitioning experiments reveal the utility of the screening method in detecting a wide range of fragment types. Fragments were advanced, demonstrating an ability to build toward inhibitors from these initial hits. Moving forward, these fragment hits can be used as building blocks of inhibitors or activators of GPx4 and may potentially be developed into drug leads. Broadly, this method can be applied to fragment screening of other membrane interacting proteins to enable inhibitor design for this challenging class of proteins.
Rights
© The Author
Is Part Of
VCU University Archives
Is Part Of
VCU Theses and Dissertations
Date of Submission
5-9-2024