| GRID manual | ||
|---|---|---|
| <<< Previous | Next >>> | |
Chapter 62. GLUE Tutorial 02
62.1. Docking a Ligand into a Target Protein.
Adipocyte lipid-binding protein (ALBP) is a 14.6-kDa polypeptide that resides in adipose cells and is responsible for the intracellular solubilization and handling of fatty acids.
The structure ALBP complexed with Hexadecanesulfonic acid has been solved by X-ray crystallography at 1.6 Å resolution. ALBP is a member of a larger family known as intracellular lipid binding proteins or iLBPs. The three-dimensional crystal structures of many members of this family have been determined and include a 10-stranded antiparallel β-sheet arranged in a β-barrel. Within the β-barrel is a cavity formed by the β-strands and two α-helices. The helices are located on the surface covering part of what would be one open end of the barrel.
The binding site for hydrophobic ligands is within the cavity, which is closed at the other end by the packing of side chains. Analysis of the solvent positions in the cavity reveals that a set of ten conserved water-binding sites is maintained regardless of the presence or absence of ligand and the nature of the headgroup. This has implications for two properties: the native protein conformation and the nature of the ligand-binding site.
For this family proteins, standard docking tools often fail to reproduce the experimental complex structures. Conversely, GRID-GLUE procedure seems indeed able to reproduce the experimental findings.
Target: 1LIC.pdb protein
Method: Flexible docking of the 40 ligands using the GRID-GLUE software.
Analysis:
Using GREATER:
Protein preparation
Active site definition
Using GLUE:
Ligand preparation
Automatic docking
Softwares used:
GREATER: is basically a graphical front-end that can help the user dealing with the different tasks involved in the GRID use.
GRID: the force field producing Molecular Interaction Fields around (macro)molecules.
GLUE: is a flexible docking tool using GRID force field to locate the ligand inside the protein cavity. A new algorithm evaluates the protein-ligand interactions (all the intramolecular energies) and selects the most favourable ligand torsion angles to maximise protein-ligand interactions.
62.2. Protein input structure
In this section you will prepare a protein for docking. This job has to be done only once.
GREATER interface will be used to edit, correct and convert macromolecular pdb files ready for GLUE docking tool. The primary input for program Greater will be the xyz coordinate positions of the atoms in the "Target" (macro)molecule, normally in pdb format.
Go from the working directory in the Tutorial_02 directory
cd Tutorial_02
Your computer should be showing its usual prompt on the left-hand margin of the screen. You begin by typing:
Greater
and hitting the RETURN key. You will see this window:
The first thing to do is adding the protein for the computation; to do that select:
Targets->Add single target or press the Add Single button.
Insert 1LIC_SO2.pdb in the File name box shown below, set the Filtering level to no filtering and then press the OK button:
A new line will appear on the Greater window with the 1LIC_SO2.pdb file and status in progress until it turns to yellow colour. Notice the additional information provided in the Greater window like the charge of the imported macro molecule.
The run was a conversion of the input pdb file into the GRID coordinate file called KOUT. During this conversion, the program was completed showing warnings in the pdb structure.
Greater suggests to check and to deeply inspect the structures with severe warnings.
Just click on the yellow line bar of 1LIC_SO2.pdb to select the structure and then click the right mouse button on the 1LIC_SO2 status line. A subwindow appear:
Select view text files and select the GRINLOUT tab on the new window.
Some blocks of yellow warning messages will be reported on the window. In this protein some atoms of residue ASP 47 appear more times in the file; these correspond to alternate positions in the crystal. In this case the User must edit the structure and try to correct it, when possible. However sometimes, like in this case, the problematic residues may be located in regions far distant from those of interest, with a small impact on the focused User region. In the latter case the structure can be accepted as it is.
To check the 3D-location of the problematic residues in the protein structure, click on Plot conflict. This command shows a 3D interactive representation of the selected protein structure, in which the 3D location of the problematic residues is highlighted with yellow crosses.
62.2.1. PDB filtering
PDB filtering automatically detects errors and introduces appropriate modifications in the PDB file. The method can be really useful when a lot of structures are processed as a "set". However, although the method is quite sophisticate, the automatic use is not always sufficient when working with structures in no standard pdb format. The filtering level can be customised (see the GRID manual). In this tutorial we are using the custom filtering, which detects typing errors, non standard nomenclature, atom substitution, non standard position of ATOM or HETATM records and so on, and repairs the PDB file without User assistance. Missing atom in amino acid residues are still considered errors and a message will be reported.
To close the graphic window click on File->Exit. To exit from GRINLOUT editor press EXIT. In the Greater interface click with the right mouse button on 1LIC_SO2 yellow bar and then remove.
The Greater interface is now empty.
To repeat the importing using PDB filtering click on File->Set options(a window appear).
In the tab PDB filtering select custom for Filtering level then unset the following options:
remove HET lines
remove water (we will use the water molecules inside the cavity)
Now press the Add Single button, insert the file 1LIC_SO2.pdb, but this time select custom for the Filtering level and press OK. The protein file is now "repaired" from format error and Greater shows no warnings or serious errors. To exit from Greater interface click on File->Exit. The protein is now ready for docking calculation.
62.3. Site points preparation
1LIC_SO2_in.kout file is the primary input for your first GLUE run.
The first step is to define the type of GRID probes to be used by GLUE. GLUE will generate the Molecular Interaction Fields (interaction between probes and protein) and all the files will be recorded and compressed in a unique file called .bspc (binding site pre-calculation).
Type Glue on your terminal. The GLUE interface will appear.
Select from the menu bar:
Receptor->Receptor and the following window will appear:
On the Working directory section (the directory where results will be saved) click Browse and press OK.
On the Target KOUT filename section click Browse and select the protein input structure 1LIC_SO2_in.kout, then press OK to confirm.
Be sure that Grid Probes Selection is set on Default (8) so GLUE will use standard probes to compute MIFs on protein cavity. The standard probes are:
H2O (to identify hydrophilic regions and evaluate competitions with water molecules)
DRY (to identify hydrophobic regions)
H (to evaluate the protein shape)
N1 (HB donor to evaluate the protein HB-acceptor capabilities)
O (HB acceptor to evaluate the protein HB-donor capabilities)
O:: (partially charged carboxy to evaluate the protein positively charged regions)
N+ (positive charged nitrogen to evaluate the protein negatively charged regions)
O1 (donor-acceptor probe to evaluate promiscuous regions)
Now insert the Binding Site Coordinates:
Bottom X -57 Top X -41 Bottom Y 62 Top Y 78 Bottom Z -18 Top Z -2 |
and click on Start New Computations
GLUE is now computing all the MIFs between the probes and the protein in the cage you have selected. When the job is terminated, the Docking Parameters window is again reported in the screen for a new computation with a new protein. Since we want to work only with 1LIC_SO2 press Cancel to exit from the Docking parameters window.
GLUE is now ready to dock ligands in the protein active site. However, GLUE performs better (and faster) when particular site points are selected in the active site. Although not strictly required, the site points selection helps the User to refine the cavity, to give more weights to particular regions, to constraint solutions; in few words, to add external information to the docking process.
62.4. Site points visualization and selection
Site points are favourable places for chemical probes on the corresponding GRID map and correspond to interaction energy points showing local minima. Their positions define locations at which the selected probe might be able to make favourable strong non-bonded interaction. They also define pharmacophoric features in protein, to be used to make the initial pose of ligand in the protein cavity.
Site points are automatically selected by GLUE, but Users can inspect and modify the proposed selection according to external available information. For instance, GLUE may suggest site points on the protein surface or in locations not important for selective binding. However, these positions may be inspected, deleted and/or modified by the User.
Once site points are stored, they are used to define 4-points pharmacophore features inside the protein, giving up to millions of potential combination of 4-points pharmacophores locations. All these combinations are used by GLUE as initial pose for the docking procedure.
To select the site points select:
Flap->Analyze Receptor
Click Refine on the right of the Receptor Analysis Identification box and a mini editor graphical window will appear.
Open the MIF produced by GLUE and stored in .bspc files clicking on:
File->Open Field (a window will appear). Select the just produced DRY field named 1LIC_SO2_in_DRY.kont an press OK, then OK again
The hydrophobic field is highlighted inside the selected cage.
Change the colour and the energy level using:
Data->Data Levels / Colors (a window will appear). Click on line 1 (surface yellow). Then click on Delete level.
Move the Energy Level of the cyan surface to about -0.20 in order to highlight all the hydrophobic points.
Change the MIF Rendering style to Symbol, then exit with Quit.
Add the crystallographic ligand (used only for reference) in the protein cavity selecting:
Molecules->Molecules Manager then press Add.
Now browse to your tutorial directory and select HDS_Xray.pdb then press OK and Quit.
The hydrophobic field is reported with the crystallographic ligand for reference. Blue crosses represent hydrophobic interaction energies at -0.20 kcal/mol or stronger.
GLUE automatically selects some site points from this interaction maps. To see the selected points for this Dry MIF, use the command:
Mini->Open File then select 1LIC_SO2_in_DRY.mini and press OK
The site points are now visualised with yellow cross symbols. There are about 6 selected points.
The site point can be better visualised using:
Molecules->Molecules Manager then select 1LIC_SO2_in_DRY.mini
Now press Molecules Style and select Ball & Sticks for the Atoms field.
Press Quit then Quit again to exit the molecules manager.
It is important to point out that the automatic selection is trained to select the minimum number of site points able to represent the local minima in a map.
Site points can be added or deleted by the User. In order to add new site points click on:
Mini->Add new points, press the Esc key and the cursor will change from "hand" to "arrow" symbol. Click on a grid-point symbol (cross) in which you would add the point(s).
The selected combination of points can be saved pressing:
Mini->Save then press YES and finally OK
To delete non interesting site points, i.e. site points located in the surface of the cavity, or in micro-cavities non-interesting for the docking:
Mini->Cut selected points then click ESC in the keybord (be sure the arrow is the symbol). Now you are ready to select the points you want to delete.
The selected remaining points can be saved:
Mini->Save->YES->OK
Finally in order to exit, press File->Quit
This procedure should be made at least to check the quality of the automatic selection performed by GLUE. The User can inspect all the six probes, or, like in this case, only one. When the visual inspection and the possible modifications are not performed, GLUE selects the site points automatically. To save your time we provide here the .bspc file in with the site inspection was made by us.
We are still on Flap Receptor window.
Click on the Browse bar of the Output Filename in the Receptor Analysis Identification box and select 1LIC_prep.bspc. Press OK.
Click on Selection (a window appear):
Press on DRY and O:: probes. Then press OK.
In this example only two probes are required for a good docking. If selection is not performed, GLUE will use all the probes, but results will be the same since our ligand contains only DRY and O:: probe.
Click on the "Select Fixed Probe" box, select O:: and then press OK.
Select : 4 as Number of pharmacophoric points. This will force GLUE to start from quadruplets of features.
Click on reject equiv. P. comb. (es. 4-DRY). This option escludes all the poses where all the site points are hydrophobic.
Select Distance Tolerance = 1.25. The Tolerance is the maximum error allowed when the ligand atom probes are positioned in the protein site points.
Select Cavity Expansion = 0. This option simulates an expansion of the protein cavity shape (in Angstrom unit) to relax repulsive interactions due to strong Van der Waals contacts. It is important when X-ray structure quality is not so good or when the shape of the active site is complex. These are not our cases.
Select NO region constrain in the Minimum Number of occupied region box. (It is possible with GLUE to select up to 4 regions).
Finally, press Start Calculations.
GLUE now computes hundred thousands of quadruplet features in which chirality, type of features, interaction energies and shape are recorded. The computation runs only for one or two seconds!! At the end of the computation the active site of the protein is perfectly characterized, and you should be back on GLUE main window.
62.5. Ligand preparation
Standard docking require to generate ligand structure with appropriate information. Atoms types and bonds of the ligand must be stored, the starting ligand conformation must be in low energy state, the atome type nomenclature must be correct, atom charges have to be computed before docking, manual atom charge definition have to be set to discriminate or force specific interactions.
Conversely GLUE does not require any ligand preparation.
The accepted input format are Sybyl Mol2 file (3D file) or GRID kout formats. In the close future the input from SMILES will be provided.
62.6. Automatic docking
How to dock a single molecule
A Hexadecanesulfonic acid was complexed with ALBP and provided for this example. In order to test the reliability of the docking, this compound will be used for docking. The information about the bound ligand position is of course not used. However, since the bio-active conformation is selected for docking, this test can be used to estimate the binding energy and to control the quality of the protein preparation.
62.6.1. Docking the Hexadecanesulfonic acid starting from Xray binding conformation
62.6.1.1. Parameters for ligand docking
Click from menu bar Flap->Run Calculations (a window appear):
On the List of Molecules filename section click on Browse bar, be careful to filter for *.list and then select HDS_Xray.list. The list of molecules may contain only one molecule (like in this case) or many molecules as well.
Select Steric Tolerance = 0.2. This level tolerance accepts some atoms not well positioned in the starting pose. For instance, when superposing the protein site points with the ligand quadruplets, some ligand atoms not belonging to the quadruplets may clash into the protein occupied space. Increasing the tolerance induces GLUE to accept more initial poses. Decreasing tolerance induces GLUE to remove not optimal pose.
Select Maximum Number of Rotatable Bonds = 0 (We are doing a rigid docking).
Select Number of conformers = 0
The zero settings are due to the fact that the ligand is docked as a rigid body.
Press Start Calculations.
After a second the number and type of poses are been saved. During this time GLUE was docking the ligand atoms inside the protein. The docking was made by fitting all the possible ligand quadruplet features with all the protein quadruplet features. Remember that the protein quadruplets are in the range of thousands. The saved solutions are optimal pose, and they represent the starting position for a "real" docking algoritm.
62.6.1.2. Docking
From the GLUE main window press on Docking->Docking (the window docking parameter will appear):
All the information for the docking should be already filled in. Complete the window by selecting the "Ligand File Format" >> kout >> and "Ligands List filename" >> press on Browse bar and select HDS_Xray.list >> OK >> Now press Continue >> (a window will appear):
Click on Use Sites Produced by FLAP (the pose solutions produced on the previous operation) >> Set "Flap-sites:RMSmin." to 0 >> Visualize Docking Dynamics >> Start Computations
A graphic empty window will appear.
Add the ligand for reference in the window:
Molecules->Molecules Manager->Add HDS_Xray.pdb OK->Quit then from the main menu click on View->Full View
The X-ray ligand molecule will be used only as a reference for graphic comparison.
Change its style by: Molecules->Mol. Style->Sticks->Quit and then Molecules->Mol. Colour->by atom type->Quit
Start the docking by pressing:
Docking->Start Computations .
The docking trials will be reported on the graphic window. From each starting pose, there is a full protein-ligand minimization, which yield a final docked solution that is recorded. Only the best solution is stored on the screen.
When the computation is terminated (approx. after few seconds) you can compare the best docked structure with the crystallographic one.
Exit from the graphic window by typing File->Return to Main Applications .
In the blue window of GLUE the energies of the docked solutions are labeled. The best energy, corresponding to the best pose is of -17.533 Kcal/mole and the corrisponding RMSD value is of 0.94.
62.6.2. Docking the Hexadecanesulfonic acid starting from random 3D coordinates
The same procedure will be repeated with a starting conformation of the ligand significantly different from the experimental one.
On the left there is the structure obtained from CONCORD while on the right there is the crystallographic one.
62.6.2.1. Parameters for ligand docking
Click from the main menu bar Flap->Run Calculations .
On the "List molecules filename" section click on Browse bar (a new window appears). In the "Directories" box double click on the second row to come back to the previous directory, then select HDS-concord.list in the Files box. The list of molecules may contain one or more molecules.
Select Steric Tolerance = 0.2
Select Maximum Number of Rotatable Bonds = 15 (this time the ligand is considered flexible).
Select Number of conformers = 2000 (those are the 2000 conformers producing the best poses).
Press Start Calculations. Within three minutes the number and type of poses have been saved.
62.6.2.2. Docking
Docking->Docking (the window docking input will appear).
Select "Ligand File Format" >> kout >> and "Ligands List filename" >> press on Browse bar and select HDS-concord.list >> OK >> Continue (a window will appear).
Check for "Use Sites Produced by FLAP" set ON, then Visualize Docking Dynamics->Start Computations .
A graphic window will appear. Delete the previous docking solution from the graphical interface, click on Molecules->Molecules Manager, then select best_dock.kout Remove->OK .
Start the docking by pressing: Docking->Start Computations .
As one can see, the great majority of docked solutions are very much close to the X-ray solutions.
When the computation is terminated you can compare the best docked structure with the crystallographic one. The best energy, corresponding to the best pose is of -17.046 Kcal/mole and the corrisponding RMSD value is of 1.76. Please note that the ligand has 15 rotatable bonds with 1.3e+14 number of possible conformations (if the angular step is 30°).
Exit from the graphic window by typing File->Return to Main Application
The same procedure will be used to dock 40 ligands into 1LIC_SO2 protein.
62.6.3. Virtual Screening
How to use GLUE for multi-ligand docking and scoring.
62.6.3.2. Parameters for ligand docking
The calculation requires about 30 minutes. To save computational time we have prepared the file with the poses for all the ligands in the dataset.
Docking
Docking->Docking (the window docking input will appear).
On "Working Directory" section, click on Browse bar, click twice on "VIRTUAL_SCREENING" directory and then select OK.
Select "Ligand File Format" >> kout >> and "Ligands List filename" >> press on Browse bar. Inside the Directories box double click on the VIRTUAL_SCREENING and select LISTAKOUT.list in the Files box >> OK >> Continue (a window will appear).
Check for "Use Sites produced by FLAP" set ON >> UnSet "Visualize Docking Dynamics" (it must be OFF), then Start Computations.
When the computation is finished the energies will be reported in the GLUE Main window:
In order to visualize the energy ranking of the obtained solutions in a plot, type in a unix shell: ./script.csh Ris_0003.bind.
The following plot will appear:
Please note that hds and m01 (m01 is a fatty acid) show similar binding and they are the best ligands. m03 and m04 show, in between them, similar docking energy. m02 shows much lower binding energy. About 40% of the dataset shows no solutions.
| <<< Previous | Home | Next >>> |
| GLUE Tutorial 01 | Up | Appendix and Figures |