One main project goal was to determine the atomic structure of a dual-pumping membrane-bound pyrophosphatase, which pumps both H+ and Na+, to determine how these enzyme differ from their single ion-pumping homologs as well as facilitate structure-based drug design against this class of enzyme. This part of the project began with testing and optimizing the expression of five different dual-pumping membrane-bound pyrophosphatases (mPPases). These five genes were expressed using three different bacterial plasmid constructs in five Escherichia coli expression strains under various growth and expression conditions. Two mPPase proteins, originating from Bacteroides vulgatus (BvPPase) and Clostridium leptum (CpPPase), were successfully expressed in an active form and localized to the membrane fraction. Further optimization of the purification of these two enzymes resulted in a final detergent-solubilized protein with pyrophosphatase activity. Cyrstallization trials were conducted for both proteins, after optimization of promising hit conditions, BvPPase crystals diffracted to 13 Å, whereas CpPPase crystals from initial sparse matrix screens diffracted to 6.8 Å. The 6.8 Å data set of CpPPase was solved using molecular replacement with a homologous protein, in which a monomer search template was placed as a dimer in the correct orientation. This showed that the overall structure of the dual-pumping mPPases is indeed very similar to that of the Na-pumping mPPases, like Thermotoga maritima mPPase (TmPPase). The crystal contacts that allow for packing of the CpPPase protein are evident in the 6.8 Å solution, and this information can be used to optimize the CpPPase construct, to facilitate tighter crystal packing, which often leads to better diffracting crystals.
A second major project goal was to design small molecule inhibitors of mPPase from human pathogens. Though we have not yet solved the structure of a dual pumping mPPase, the structure of a Na+-pumping TmPPase and the H+-pumping mPPase from Vigna radiata, have been solved via x-ray crystallography in various conformations throughout the reaction cycle. We, along with our collaborators, have exploited this to design anti-mPPase drugs that target mPPases, including those found in pathogenic bacteria and protozoan parasites, such as the etiological agents of malaria and leishmaniasis. The TmPPase structures have been used to identify viable binding pockets for structure-based drug design. Two approaches have been implemented: in silico screening of small molecules against specific binding sites to select top hits for in vitro inhibition testing and synthesizing novel small molecules that are designed to bind a specific site on the protein. We are currently screening molecules that are most inhibitory against TmPPase in vitro against other mPPases, such as BvPPase. From the initial top 11 TmPPase hits, we found two compounds, one that is active against BvPPase, but not against Plasmodium falciparum mPPase, and vice versa, at the low M level. This suggests that we will be able to design highly-specific mPPase inhibitors that will reduce the risk of horizontal gene transfer and so the spread of resistance.
Since crystallographic structures only give a snapshot of the enzyme at one point during the enzymatic cycle, we employed molecular dynamics simulations to gain insight into intermediate conformations throughout this cycle and, thus, into the mechanism of mPPases. For our purposes, we needed to set up an atomistic simulation of the 140 kDa TmPPase within a lipid bilayer (Figure), which is not a common task within the molecular dynamics field. In our paper (Shah N. et al. 2016. Struct Dyn), we outline our successful TmPPase atomistic molecular dynamics simulation, and our insights into the loop-closing conformational changes that occur upon substrate binding. This molecular simulation of TmPPase will also be used to assess binding dynamics of different drugs in silico during the small molecule structure optimization steps of drug design. Furthermore, the various protein conformations that are revealed by the simulations can be used to identify pockets that can be targeted via structure-based drug design.
