Chorismate mutase is involved in the biosynthesis of phenylalanine in plants and microorganisms via the shikimate pathway. It has been suggested that inhibitors of bacterial chorismate mutase may act as antibiotics (e.g. against tuberculosis). The rationale is that in the presence of such inhibitors, bacteria cannot carry out de novo synthesis of phenylalanine. The enzyme is absent in humans; mammals do not make phenylalanine via the shikimate pathway but obtain it with food. Despite appearing as a promising antibacterial target based on its metabolic role, chorismate mutase is not well-validated experimentally. Thus, it is likely that inhibitors with strong in vitro activity turn out to be weak antibacterial agents in vivo.
You will be testing the binding affinity of different ligands to E. coli chorismate mutase. The crystal structure of this dimeric protein (1ECM) is shown on the right. Load the PDB entry 1ECM into PyMOL and examine the structure. You can render the protein as a transparent surface by running the prepared script available from http://www.chem.ucsb.edu/~molvisual/chem162.html. Notice that the binding pocket in this protein is completely surrounded by amino acids, suggesting that the protein undergoes a significant conformational change upon ligand binding and product release. In situations like this, successful docking based on the rigid receptor model is possible only if the enzyme-inhibitor complex is available. Also, rigid receptor docking to chorismate mutase would likely miss molecules that are much larger than the ligand present in the binding pocket.
Your goal is to study the binding of chorismate mutase ligands to the enzyme from E. coli. Analyze if docking is capable to reproduce the experimental binding geometry, and ranking of inhibitors. You will also explore if docking supports the notion that chorismate mutase has a high affinity for the transition state of the chorismate-to-prephenate rearrangement reaction. Follow the protocol that was outlined in the tutorial incorporating a few modifications as outlined below. In addition, you may want to analyze how docking parameters such as the grid spacing, and number of configurations generated affect the results.
Under File, Reinitialize PyMOL and reload structure 1ECM. Select Display -> Sequence. Close examination of this structure reveals that some of the residues are not visible in the crystal structure. In particular, chain A starts with asparagine at position 5, and chain B starts with proline at position 6. Show this proline in stick model. The Dock Prep tool in UCSF Chimera gets confused when a chain starts with proline, so you need to add an amino acid to the B chain before docking. Because it is difficult to judge what the position of the asparagine side chain in this chain should be, you will add a glycine residue. To prepend a residue, switch to 3 Button Editing Mode under Mouse, and click on the ring nitrogen of proline. A small sphere appears around this atom. Type Alt-G to build glycine residue (or select glycine from the Build menu). Delete the three hydrogen atoms by switching to 3 Button Viewing Mode under Mouse, and choosing Selecting Atoms mode on the right-side toolbar. Click on the three hydrogen atoms to select them, then Remove Atoms in the selection labeled (sele) in the right-side toolbar. Save this fixed structure as a PDB file 1ECM_fixed.pdb.
Prepare the receptor with UCSF Chimera starting with the 1ECM_fixed.pdb file. Follow the instructions in the tutorial, and write out the MOL2 file with hydrogens, and the PDB file without hydrogens. Also, prepare the ligand from the crystal structure that is needed for the definition of the binding pocket; follow the instructions in the tutorial except that this time two inhibitor molecules carry the same name (TSA) but different chain identifiers. Thus, after removing all waters and protein atoms, select and remove chain B; you are now left with the inhibitor from chain A. The structure already comes with hydrogens, but you still need to add charges with Chimera. Save the prepared ligand as lig_charged_tsa.mol2. Because the binding pocket of chorismate is buried deep within the protein, it may be helpful to reduce the radius of selected spheres from 10 to 6 ångstroms.
You will be docking a virtual library consisting of chorismate dianion in the near-attack conformer (HF/6-31+G(d,p) optimized structure is in the file chor_da_nac_hfo.mol), dianionic transition state of the chorismate to prephenate reaction in endo conformation (PM3-optimized structure in file chr_da_tsen_pm3.mol), and several transition state analogs. The inhibitors are the dianionic forms of the oxa-bicyclic inhibitor, the methylene-bridged bicyclic analog, the molecule that you proposed as an inhibitor of chorismate mutase in your first assignment, and the methyl analog of the oxa-bicyclic inhibitor that I proposed as part of the first homework. You probably have already generated the dianionic analogs of these molecules as part of the previous assignment; if not remove the hydrogen from the second carboxylate and re-optimize at the PM3 level. The dianionic form of the methyl analog is also available from file inh_ox_me_da_pm3.mol. You will dock all these structures to one of the binding pockets of the dimeric chorismate mutase from E. coli. Note that you need to assign partial charges to atoms in each of these molecules using UCSF Chimera, and export the charged ligands as MOL2 files.
Perform docking of these structures to the E. coli chorismate mutase with the program DOCK 6. Use both rigid docking and anchor-and-grow method in DOCK6. Due to the small pocket, the grid calculation is fast for this enzyme. Make sure that you rename ligands appropriately. At the end of the docking, you should have a set of mol2 files with best poses for every structure that you docked via rigid docking, and another set of mol2 file with best poses that you docked via anchor-and-grow method.
Analyze your results: provide electrostatic, van der Waals, and total docking scores for every molecule using the two docking approaches. Compare the docking results with experimentally observed inhibition constants (values are embedded in the cm_inhib.sdf file; use a text editor or PowerMV to get these). Briefly discuss differences between the rigid and flexible docking for these relatively rigid ligands. Did any of the suggested modifications (either yours or mine) bind better than the best known inhibitor (oxa-bicyclic compound)?
Does docking predict that the transition state binds much stronger to the protein than the substrate in the near attack conformer? Recall that kcat/knon for chorismate mutase is about 1*106, and the application of thermodynamic cycle would suggest that the transition state is bound about million-fold stronger than the substrate. Write a brief essay either defending or criticizing the hypothesis that chorismate mutase accelerates the reaction by specifically binding to the transition state.