Difference between revisions of "Team:Virginia/Model"

Overview

We modeled our system in silico to select a sterically feasible protecting group and to optimize a mutant leucyl-tRNA synthetase for complementarity of its catalytic site to protected leucine, and of its editing site to leucine. To select a protecting group, the team used protein-ligand docking software to compare binding affinities of several protected leucine/synthetase complexes. To perform mutagenesis on leucyl-tRNA synthetase, an integrated software script was written in the Linux shell, with inputs including a protein to mutate, a ligand, a list of residues of interest, and binding pocket location. The script runs mutagenesis, assesses mutant protein stability, then performs ligand docking. The outputs are ranked, simulating a streamlined mutagenesis optimization algorithm. We confirmed, using CSM software suites and iGEMDOCK, that leucine-AMP and leucine-AMS yield energetically comparable binding affinities to our synthetase. Lastly, we performed Michaelis-Menten modeling for the enzyme pepsin to gauge activity in nonspecific cleavage enzymes.

Leucine Synthetase Selection

Synthetase File: 4aq7

The leucine tRNA Synthetase (LeuS) used in the following simulations was found using the RCSB Protein Data Bank. The protein, named 4aq7, is a leucyl tRNA synthetase in its aminoacylation conformation. The following is a PyMol visualization of the protein that we compiled.

AMS vs AMP

As stated in its documentation, during the characterization of 4aq7, a leucyl-adenylate analogue was used to determine the crystal structure of the protein. By analyzing the file contents in PDBest, a PDB editor software, we identified the analogue as leucyl-AMS.

We constructed six different molecules with ChemSketch in order to determine whether the analogue significantly altered the conformation of LeuS. The constructed molecules were AMS and AMP adenylations of Cbz-Leu, N-Methoxycarbonyl-Leu, and leucine. Using a variety of protein-ligand docking software like CSM-lig, Autodock Vina and iGEMDOCK, we were able to confirm using the values below that AMS (the adenylating agent found in the literature) and AMP yield similar binding affinities to the synthetase when bound with leucine. Thus, 4aq7 represents an accurate By testing these molecules in CSM-lig, Autodock Vina, and iGEMDOCK, we were able to confirm that AMS and AMP were similar, and that the pdb 4aq7 could be used to accurately model the LeuS.

Protecting Group

Selection - Modeling Considerations

Several programs were used to asses the viability of the different protecting groups in our system. The main programs used for this purpose were Autodock Vina, iGEMDOCK, and CSM-Lig. The initial tests measured residue specific energistic changes between different protecting groups. A major goal was determining which protecting group would be most feasible to work with. Our criteria for measuring feasibility are detailing in the following figure.

Residue Selection

Literature

Half of our selected residues were intended to increase synthetase stability and reaction rates. These mutations were found by searching through current literature. We found four mutations that would improve LeuS. Of the four mutations, one is in the editing site while the other three are in miscilanous locations around the protein. The editing site mutation, Thr 252, allows for the hydrolysis of Leu-tRNA, a crucial step in our overall project.

iGEMDOCK and Ligplot

We used a combination of docking data as simulated in iGEMDOCK and structural data visualized in Ligplot to determine additional mutational focci. iGEMDOCK is able to provide energy values for each residue by conducting an exhaustive docking search. Ligplot is capable of creating a 2D image of the binding site as detailed in the crystolography file. We selected mutations that satisfied two criteria. First, the residue had to have been shown to have a high energy during docking. Residues that returned large values in the protected leucine simulations were believed to have a high impact on the reduced viability of docking. The second criteria was that of location. Residues that protruded around the N terminus of leucine, the location the protecting group binds to, were thought to be possible sources of steric hindrance. a total of four residues that satisfied both criteria were selected.

Final Selections

The following residues were picked for mutagenesis.

Catalytic Site

• M40
• Y41
• L43
• D80

Editing

• T252

Stabilizing

• K186
• A293
• L570

MUT

Purpose

In order to generate and test a large list of mutants we build a software pipeline in linux called MUT. MUT utilizes PyRosetta, Autodock Vina, and FoldX to mutate, test, and rank protein structures based on docking scores. This program was used to querry for the optimal mutant structure for Cbz-Leu docking.

Structure

Upon initiation, MUT asks for four main inputs, the protein pdb that will be analysed, the residues that the program will mutate, the binding pocket for docking simulations, and a ligand pdb file. MUT first performs stability testing and docking to get baseline values for the future tests before it mutates the initial PDB file. At each iteration the residue is mutated to 17 other amino acids, not including Cys or Pro as they induce kinks and are difficult to model, or itself. After mutagenesis is complete files undergo stability testing and docking to determine if the new mutant is stable and a better match for the ligand. Files that fail the tests go to the kennels.

The kennel system allows for two levels of simulation. The first and fastest way is the Reductionist approach. This method disregards all files that fail any test at any point. This method is quick, but because of the variability of docking simulations, it is likely to miss key files. The Exhaustive approach takes into consideration all possible mutations. After the Reductionist approach is complete, all files are mutated to four mutants and re-scored. This method ensures that all possible combinations of mutations is tested, but is computationally intense. This pipeline can be seen in the flowchart bellow.

Output

After a run is complete, all files are automatically ranked and ordered by binding values (more negative the better). Once an initial ranking has be completed, the user has the option to rerun docking on the selected mutants to confirm the simulation. Should the results look bad, mutants from a lower ranking can be selected or new residues can be picked.

Summary

Virginia iGEM designed a Linux-based protein engineering tool called MUT. This program is used to screen protein-ligand docking across possible protein mutants. It is written from a Centros shell, but is designed to be portable across distributions. MUT generates mutant structures, subjects the resultant proteins to stability testing, and ranks the results based on protein-ligand docking data. Three programs are used to accomplish this task: PyRosetta conducts computational mutagenesis, FoldX tests protein stability, and AutoDock Vina performs protein-ligand docking. All three of these programs are available for free with an educational license, and MUT is hosted on an open source site. MUT is designed to be configurable to any iGEM team’s needs. The user inputs to MUT a protein’s PDB file, residues of interest for mutation, a binding pocket, and a ligand. After the simulation runs, the program outputs a top ten list of mutant structures along with their files.

Source
MUT on GitHub

Pepsin Modeling

Purpose

In order to determine the relative efficiency of our enzyme, we modeled another BioBrick. BBa_M1436 codes for pepsinogen, an inactive form of the cleavage enzyme pepsin. Pepsin is a nonspecific cleavage enzyme that cleaves dipeptide bonds between amino acids. Comparisons between pepsin and our enzyme may yield interesting results, specifically regarding the relationship of substrate concentration as it varies over time.

Results

Modeling was conducted by solving for Michaelis-Menten equations in MatLab. Both Equilibrium and Quasi-Steady-State approximation were used to modeling reaction rates.

Conclusions
Programs Used

Autodock Vina

Autodock Vina is a program the performs computational protein-ligand docking. It outputs binding energy as a measure of gibbs free energy (ΔG). Autodock Vina also has other application and is capable of simulating single residue mutations.

PyRosetta

PyRosetta is a software suite that can perform many different simulations. In this project PyRosetta was used to model mutations on specific residues. PyRosetta is a derived from the Rosetta software suite and rewritten to work with python. It is available for free academic licences.

iGEMDOCK

iGEMDOCK is a protein-ligand docking tool meant to allow for large scale ligand screening and analysis. It takes a PDB as input along with any number of ligand files. To model the binding, iGEMDOCK uses a generic evolutionary model to iterate through possible binding conformations. After the run is complete, clusters of data are output along with residue specific energistics.

FoldX

FoldX measures and rates the energy present in a protein structure. It is available by academic license.

LigPlot

LigPlot generates a 2D representation of a protein-ligand interface (among other interfaces, such as protein-protein). This visualization shows both the location of residues and which residues or molecules they interact with. Moleclues can be viewed with or without water present. LigPlot is also compatible with PyMol and Rasplot to create 3D visualizations of the interactions. While the 3D mode is more difficult to use, it can offer unique insights into protein-ligand interactions.

PyMol

PyMol is a visualization tool that can display and edit proteins and molecules. It has many capabilities beyond visualization and is a powerful tool for computational analysis.