Part 3a.  Bound Ligand Complex

Download the 1M48 structure file from the Protein Data Bank:  rcsb.org

Setting up the Display:  

• In your VMD application...  VMD Main > File > New Molecule > Browse... 1m48.pdb (file type: PDB) > Load. Close 'Molecule File Browser'.

• Under 'VMD Main' > Display, remember to change perspective view to orthographic and uncheck the Depth Cueing box (atoms in the distance will be too dark). Set Axes to Off. In 'Display Settings', set Near Clip = 0.01.

• Under VMD Main > Graphics > Representations, change 'Drawing Method' to NewCartoon and 'Coloring Method' to Secondary Structure.

 1M48 contains a homodimer structure for the immune system protein interleukin 2 (IL-2), a cytokine that functions almost like a hormone.

By rotating interleukin you will notice that it is narrower in one dimension than others, and that there are two separate backbone chains. We are looking at two identical protein monomers rotated 180 degrees from each other (C2 cyclic symmetry, 2-fold axis). Note how by default the molecular system rotates about the center point of all the atom coordinates.

Customize Graphical Representations:

Interleukin is a small 16 kDa signaling protein that does not actually function as a dimer. Dimerization was an artifact of the experimental crystallization procedure used to determine its structure (when solvent was evaporated).

Bring up the VMD Main > Graphics > Representations window and set up two entries to visualize the binding pocket.

1M48.pdb - Graphical Representations:

DrawStyle  Color   Selected Atoms

Surf     ResType  chain A and not resid 301

Licorice   Name   resid 301

If you open 1M48.pdb in a text editor you will see it contains ATOM entries for two identical protein polymer chains: A and B (we are only interested in the first chain, A).

____________________________________________________________________

                 # Atom Residue Chain Resid        x               y               z   (Angstroms)

ATOM      1  N   SER A   4      22.646   9.384  29.750

ATOM      2  CA  SER A   4      21.314   9.180  30.387

ATOM      3  C   SER A   4      20.460   8.226  29.562

ATOM      4  O   SER A   4      20.913   7.677  28.557

  Atoms 5 thru 996...

TER     997      LEU A 132

ATOM    998  N   SER B   4      17.159  -7.491  -5.025

ATOM    999  CA  SER B   4      17.916  -6.902  -3.878

  Atoms 1000 thru 2019...

ATOM   2020  N   THR B 133      24.591  -8.600   5.343

ATOM   2021  CA  THR B 133      25.328  -7.350   5.221

ATOM   2022  C   THR B 133      26.476  -7.527   4.236

ATOM   2023  O   THR B 133      26.522  -8.513   3.502

ATOM   2024  CB  THR B 133      25.877  -6.908   6.586

ATOM   2025  OG1 THR B 133      24.789  -6.578   7.459

ATOM   2026  CG2 THR B 133      26.649  -5.595   6.448

TER    2027      THR B 133

____________________________________________________________________

Click and drag over the monomer to see how it still rotates about the previous center point of the dimer, which is not helpful. Pick a new center point:

1.  Tap t to enter translate mode (reveals arrows) and left-click drag to center the small ligand in the frame.

2.   Scroll to zoom up close on the center of the ligand so you can see individual atoms.

3.  Select an individual atom to define the new center of the coordinate system. The vertex between each bond edge corresponds to an atom. Hover your pointer over a vertex, tap the  c  key (reveals crosshairs), then click it with your mouse. Zoom out, hit r, and rotate the molecule to see if you picked a good center point.

Now you should be able to rotate into a good view of the binding pocket. Rotating the protein atoms behind your center point will prevent atoms from being clipped from view. The candidate drug compound is seen to be well accommodated in the binding site. This compound is a putative inhibitor of IL-2's interaction with the IL-2 receptor on the surface of immune cells, serving to inhibit its cytokine action.

Back in the terminal, vmd should print Info) for the atom you selected: atom name, type, number (index starting at 0), residue, residue name, chain ID, and xyz coordinates.

Q6:  Describe the ligand. Can you guess from the Licorice structure what functional groups are present in the ligand? Are they acidic or basic? (Nitrogen atoms are colored blue; oxygens are red.) The structure of the ligand is shown in the journal article, "Binding of small molecules to an adaptive protein-protein interface," you downloaded for the 1M48 structure.

Render an image showing the electrostatic and structural complementarity of the binding pocket..

• Zoom in and frame the binding cleft. Select VMD Main > File > Render... > Tachyon (internal, in-memory rendering), Filename = 1m48.tga > Start Rendering.

Part 3b.  Apo Protein Conformation

Download the 3INK structure file for IL-2 from rcsb.org. We will load this structure into the same visualization window. 3INK.pdb is a structure of the same IL-2 protein in a different, apo conformation (in the absence of ligand).

To open 3INK.pdb in the same VMD window:

• VMD Main > File > New Molecule > Browse... select 3INK.pdb (file type: PDB) > click Load. 

In your Display window you should see the new molecule loaded in the same window with 1M48. Rotate the view to see how far this molecule is shifted in Cartesian space (called translational motion) relative to the original protein. This is because the absolute xyz coordinates in each .pdb coordinate file are arbitrary.

Open VMD Maine > Graphics > Representations. Under the drop-down menu for 'Selected Molecule' you should now see your two loaded pdb files (molecules):

0: 1M48.pdb

1: 3INK.pdb

You will have separate graphic representations stored for each pdb molecule. 

Altering the View for 3INK:

VMD Main > Graphics > Representations.. Change 'Selected Molecule' to  1: 3INK.pdb.

Set the Graphical Representation for 3INK.pdb:

DrawStyle   Color   Selected Atoms

Surf  ColorID 8 white   chain C

Part 3c.  RMS Fit and 3D Structural Alignment

In order to compare two structures, or conformations, of the same receptor or of two homologous proteins, we need to overlay them in the same position and rotation. The MultiSeq analysis tool in VMD will do this by automatically dragging one structure while seeking to minimize the RMSD with respect to the first reference structure. The final value of the RMSD is a measure of how similar the two structures are.

1. Overlay the 3INK structure (without ligand) on top of the 1M48 protein-ligand complex.

• Select VMD Main > Extensions > Analysis > MultiSeq.  It may ask you to select one or more folders to store temporary files:  create a scratch folder inside your home directory. It asks you to update to the new database.. It should work if you select "No". It might take a few seconds to download the database files. If you selected a system folder that you don't have write access to, it may not have installed the database and the sequence alignment won't work.

• The first step aligns the two protein sequences. If you scroll to the right you will see the sequence alignment. There a few gaps in the sequence alignment: note how these occur in loop regions between secondary structure elements. They were too flexible so show X-ray density in one or the other structures.

•  Note in Graphics Representations it creates new Reps for each chain in the two dimers so that you'll be able to compare the backbones (NewCartoon ribbons in cyan). Are old Reps are still there: double click on any entry to show/hide. Let's delete the other chains that we're not worried about. In the new 'untitled.multiseq' window, click on the middle of the gray box entry for 3INK_D (the whole line should turn yellow) and hit your delete key (or backspace). This should remove it from multiseq and the yellow chains from your Display. If backspace did not work, select 3INK_D again from the multiseq window and select Edit > Cut.

• Click on the gray multiseq box for  1M48_B  to delete it as well.

2. In the multiseq window, click on 'Tools' and select 'STAMP Structural Alignment'.

• In 'Stamp Alignment Options', click [OK]. In the Display, you should see our two IL-2 conformations overlay each other.

• Hide the 'Surf' representations for each molecule by double-clicking on the Graphics Representation entry to turn it red. Tap c and click on a ligand atom to re-center the system. Now you should be able to rotate the view and compare the fit between the (NewCartoon) backbone elements. The two backbone ribbons nearly coincide except for a flexible linker, or loop, region between two helices.

3INK.pdb - Graphical Representation:

DrawStyle   Color   Selected-Atoms

Surf   ColorID 8 white  chain C (double-click to hide)

NewCartoon  Name    chain C

1M48.pdb - Graphical Representations:

DrawStyle  Color   Selected-Atoms

Surf     ResType  chain A and not resid 301 (hidden)

Licorice   Name   resid 301

NewCartoon Name   chain A

The alignment will never be perfect if the two structures are different. Mathematically, the best fit is done by minimizing the root-mean-square (RMS) deviations of all atom pairings for identical stretches of the lprotein sequence. The average RMS deviation (RMSD) between the two structures is a measure of how identical they are. In Tutorial 2, rmsd.dat contained a time series of a receptor simulation. Each line in the file contained the RMSD in nanometers of the current simulation snapshot compared to the starting PDB structure. The protein transitioned between two dominant conformational basins, or "states". The open conformation was closer, within 2.5 angstroms RMSD to the starting point, but the closed structure was up to 4 angstroms RMSD (a few bond lengths distance). RMSD is typically only calculated for backbone atoms since the sidechains are inherently flexible. For structures that have the same conformation, the RMSD rarely gets below 1 angstrom because of thermal fluctuations.

Part 3d.  Compare Binding Pockets

In de novo drug design, it can be hard to predict where the binding pocket is on a protein. Moreover, the protein sidechains and even backbone segments can move to accommodate the ligand when it binds: this is known as induced-fit. An induced-fit conformational change further strengthens and gives specificity to the binding interaction. Let's see what our binding pocket looks like in the apo protein conformation by trying to visualize/dock the ligand in the same position as it was in the bound structure for the protein-ligand complex. In the bound structure the protein formed a nice binding cleft around the drug candidate.

3INK.pdb - Graphical Representation:

DrawStyle   Color   Selected-Atoms

Surf   ColorID 8 white  chain C (double-click to show)

NewCartoon  Name    chain C (click to hide)

1M48.pdb - Graphical Representations:

DrawStyle  Color   Selected-Atoms

Surf     ResType  chain A and not resid 301 (hidden)

Licorice   Name   resid 301

NewCartoon Name   chain A

You should see that half the ligand is now buried by the white surface, meaning this portion of the binding pocket was not already present in the apo conformation. Also compare how in a few spots the backbone/helixes have moved between the two structures in white/cyan. These observatios reveal that the ligand binds IL-2 via an induced-fit mechanism. This is in contrast to the lock-and-key mechanism where the ligand is a near perfect geometric fit for a rigid bonding pocket.

Induced fit binding has an impact on computer-based drug design methods. If the original 3INK crystal structure was used in a standard program like AutoDock, which uses a 'frozen' template of the protein to screen the binding affinity of many small molecules, this ligand compound would have not been successfully identified as a drug candidate.

Render an image showing the ligand occluded by the 3INK binding surface in white.

• Zoom in and frame the binding cleft. Select VMD Main > File > Render... > Tachyon (internal, in-memory rendering), Filename = occluded.tga > Start Rendering.

Q7:  Discuss the role of lock-and-key and induced-fit models in computer-aided drug design. 

Conclusion

In this VMD tutorial we discussed how to visualize structural representations of molecules. A computer simulation predicts how a molecule changes its structure in time, i.e. molecular "dynamics". VMD stands for Visual Molecular Dynamics: it is a tool for viewing the simulation trajectories generated by available molecular simulation software packages. You can learn how to perform molecular dynamics simulations from these Gromacs tutorials.

Thank you for your participation in our workshop! Please fill out an evaluation form.