Mycobacterium tuberculosis (Mtb), the causative agent of tuberculosis (TB) is notoriously difficult to treat due to its impervious cell wall. However, the biosynthetic pathways responsible for its protective barrier have proven to be an Achilles heel, as many TB drugs target the enzymes within these unique pathways. Based on this proven approach, we too have decided to target enzymes responsible for the construction and maintenance of the outermost lipid membrane, the mycomembrane, for drug development. The focus of this dissertation is on the structural, enzymatic, and inhibitory study of two Mtb mycomembrane lipid esterases. Both enzymes are secreted, essential, have similar protein folds and utilize classical catalytic triads, yet perform different types of chemistry.
The Antigen 85 complex (Ag85) of Mtb catalyzes the acyltransfer of the mycolic acids (MA) to produce trehalose-dimycolate (TDM) or the mycolylarabinogalactan (mAG), two hallmark lipids of the unique mycomembrane. Three homologous mycolytransferases comprise the Ag85 complex (Ag85A, B, and C) and despite the term complex, act independently through a ping-pong catalytic mechanism. Based on early structural studies, an interfacial mechanism model was proposed that detailed substrate arrangement within Ag85s; however, numerous problems exist with this model. To address those concerns, we sought a crystal structure of the acyl-enzyme intermediate form of Ag85C; however MAs are highly insoluble. Therefore, Ag85C was successfully co-crystalized with tetrahydrolipstatin (THL), an FDA approved lipase inhibitor, which mimics core structural attributes of MAs. The Ag85-THL structure served as a basis for the modeling of the acyl-enzyme intermediate. Based on structural similarities to previously solved Ag85 structures, mutagenic studies, and computational simulations, Ag85s were shown to undergo structural changes upon acylation that limit substrate hydrolysis and promote substrate transfer. Based on these findings, a new model of substrate binding was proposed that satisfies issues with the previous interfacial model and results in the chemical requirement for the ß-hydroxyl of MAs for Ag85 catalysis. Outside the TB field, the Ag85C-THL structure is the second protein-THL structure to ever be solved and provides a new molecular perspective on lipase inhibition by THL.
Ebselen and thiopene compounds have previously been reported to inhibit Ag85C covalently and non-covalently, respectively. A library of ebselen and thiophene derivatives has been synthesized and tested for in vitro and in vivo properties. As a result of no clear structure activity relationship, two chemically dissimilar ebselen compounds were selected for further characterization. Based on the crystal structures of Ag85C in covalent complex with azido-ebselen and adamantyl-ebselen, key protein-drug interactions were assessed that influence in vitro inhibition properties. Furthermore, using differential scanning fluorimetry, ebselen modification was shown to significantly influence protein stability. Unfortunately, all thiophene derivatives displayed no inhibition towards Ag85C due to the loss of specificity to the sugar binding site. Finally, screening Ag85C against a small set of lactone containing compounds and two drug libraries comprising over 1500 TB active compounds resulted in the identification of two scaffolds that can be utilized for further development targeting Ag85s.
The second lipid esterase of interest is encoded by the rv3802c gene, which resides in gene clusters responsible for cell wall biosynthesis. Rv3802c encodes an enzyme with thioesterase/phospholipase A activity, is retained in the cell wall, and has been shown to modulate lipid content of the mycomembrane as a result of cellular stress and overexpression. To further characterize Rv3802, we set out to obtain the first crystal structure of the enzyme. The X-ray crystal structure of Rv3802 was solved with two molecules of polyethylene glycol (PEG) bound within the enzyme active site. On the basis of PEG binding, the lipid binding site of Rv3802 was structurally identified and characterized. Comparison of the Mtb Rv3802-PEG structure with an apo Mycobacterium smegmatis (Msmeg) ortholog structure resulted in the identification of dynamic regions required for lipid binding. The Rv3802-PEG structure provides a molecular basis for the binding of phosphatidylinositol-based substrates, further suggesting the enzyme responsible for the decomposition of glycerophospholipids of the mycomembrane.
In efforts to develop inhibitors of Rv3802, two fluorescence based assays were developed and compared. Using these assays, two drug libraries were screened against Rv3802, which resulted in the identification of multiple compounds with low micromolar affinity. Unfortunately, identified compounds failed to produce morphological differences in Msmeg cells and displayed modest in vivo activity against two strains of Mtb.
THL has been shown capable of inhibiting numerous human lipases and covalently modifying 261 lipid esterases in mycobacteria, two of which are Ag85C and Rv3802. To better understand how THL inhibits both enzymes, a library of stereoderivatives was screened against both enzymes. We found that the stereochemistry of the ß-lactone ring is important for cross enzyme reactivity, while the stereochemistry of the peptidyl side arm influences enzyme specificity and stability of the covalent THL-enzyme complex. Observed in vitro inhibition data was rationalized using the Ag85C-THL structure and molecular modeling of Rv3802 and THL. Findings within this dissertation advance the structural and enzymatic understanding of two essential Mtb lipid esterases, while providing a basis for future development of novel inhibitors specific to Ag85C and Rv3802.