Protein-ligand docking with Glide

Workshop in applied structural bioinformatics (2017)


Lars Skjærven, University of Bergen

Overview
In this tutorial we will introduce protein-ligand docking using the Maestro interface to
Glide and related Schrödinger packages. We will use the beta-2 adrenergic receptor
(B2AR) as example protein and predict its interactions with the beta blocker Pindolol.

Beta-2 Adrenergic Receptor complexed with Pindolol

Before we start we will make a directory to store the content of the docking results. In the
terminal make a new directory and change to this directory using the following commands:

cd [will change to you home directory]

mkdir docking [create the docking dir]

cd docking [change to the docking dir]

Start Maestro by typing

/usr/local/schrodinger2016/maestro &

Maestro is a unified interface for all Schrödinger

software packages. We will only use a small part of
the functionality embedded in this package suite.

We will start by loading a protein structure into the

Maestro environment. Click the Get PDB item in the
File menu item, and fetch the PDB of B2AR with
identifier 2RH1 (File → Get PDB → 2rh1 →
Download).

Once the PDB has been dowloaded the protein

structure appears in the visualisation window. You can use the mouse to control the view:

• Rotate by holding down the middle button while moving the mouse,

• Translate by holding down the right button while moving the mouse,

• Zoom by scrolling.

You can additionally

• select atoms/residues by clicking on the structure, or click and drag will select multiple
atoms/residues.

• focus on an atom by middle click.

 Docking Tutorial - WASB 2017

Similarly to PyMOL you can show the protein and ligands in various representations. Use
the representation buttons in the structure hierarchy in the lower left corner to show
ribbons for the protein and the ligands in ball and sticks:

• Click the + button next to Protein, then → Add Ribbons.

• Click the + button next to Ligands, then the ring structure with
balls and sticks icon.

You should now see the protein in cartoon/ribbon representation

and the ligands in sticks.

Before we can dock Pindolol to the structure of B2AR we need to

prepare the protein as well as the ligand for docking. For the
protein this entails adding hydrogen atoms, optimising the
position of the hydrogens, and defining the assumed site of
binding. For the ligand we need to build a three dimensional
structure, add hydrogens, and generate possible protonation/
tautomer states. We will describe these steps in more detail below.

Step 1: Protein preparation

Click the Protein Preparation Wizard button in the task bar. Make sure the “Import and
Process” tab is selected. Click the Preprocess button to add hydrogens (and various other
tasks). Click OK on the warning on missing side chain atoms. You will now see a new entry
(which is automatically selected) in the Entry list (left hand side of the Maestro window).
Zoom in to make sure that the hydrogens has been added.

Go back to the Protein Preparation Wizard window, and select the Review and Modify
tab. Here you see a list of water molecules and non-protein atoms (in the Het name
column). Scroll down the Het name list and click the A:CAU (408) entry. This is ligand
Carazolol, an inverse agonist of B2AR. Note that the when clicking on CAU, the viewer
(Maestro window) has zoomed in on this particularly ligand. Select all other molecules
(except CAU), and hit the Delete button. Go ahead and delete all water molecules as
well. Select the A:CAU entry in the Het name list. Click the Generate states button to
generate possible protonation states of the CAU ligand. When the calculation is complete,
note the checkboxes appearing next to the CAU entry. Toggle the S2 checkbox and view in
the structure the difference between Orig. and S2. Make sure the S2 box is checked before
moving on.

Side-note: Water molecules can be an important contributor to ligand binding as they

can act as a hydrogen bonding bridge between the protein and ligand. In many cases
you will therefore want to keep specific water molecules if they are found in the binding
site.

Page 2 of 6
 Docking Tutorial - WASB 2017

Next, select the Refine tab. In the H-bond assignment section, make sure the Use
simplified rules pH Natural is selected. Continue by clicking the Optimise button. This
will calculate the pKa values and protonate residues accordingly. Note the message
appearing at the bottom of the window in red text while the calculation is running.

In the Restrained minimisation section of the same tab, check the Hydrogens only check
box before you hit the Minimize button. This will start an energy minimisation of the
structure. When the calculation is done the new 2RH1 entry will contain optimised
hydrogen positions.

Step 2: Grid generation

The next step includes binding site definition and the generation of the grid needed for the
subsequent ligand docking. The grid contains the shape and molecular properties of the
defined binding site. Click the Tasks icon in the upper right corner. From the menu that
appears open the grid generation window (Tasks → Browse → Docking → Receptor Grid
Generation). This opens the Receptor Grid Generation dialog with five tabs.

In the Receptor tab, make sure the Pick to identify the ligand checkbox is checked and
Molecule is selected in the dropdown menu. Go to the Maestro window, and zoom in on
the ligand by hitting L on your keyboard. Click the ligand in the structure. Make sure that
the ligands is marked with green boxes (see figure).

In the Site tab, choose the “Centroid of Workspace

ligand” is selected, and toggle the Display box
checkbox. Notice that a box around the ligand
appears.

Now hit the Run button, and wait until the grid has
been generated. Note the progress icon in the lower
right hand side of the window . This will take a
minute or two.

Step 3: Ligand preparation

The structure of the Pindolol compound can be obtained e.g. from ChemSpider. Go to
chemspider.com and write pindolol in the search field. Once the compound has been
identified click the save button and save the “.mol” file to disk.

Side-note: There are several online tools providing functionality to obtain the structure
of small molecule compounds. You can also draw your own compounds e.g. here:
http://www.ambinter.com/moleditor/web and import these compounds to Maestro.

The downloaded mol file contains the 2D structure of the compound (you can open this in
PyMOL to explore it in more details). Ligand preparation in Maestro helps us to build the
3D structure of this compound and assign protonation states (and more). Hit the Ligand

Page 3 of 6
 Docking Tutorial - WASB 2017

preparation button on the taskbar. Click the upper Browse button and select the file with
name 4662.mol, and hit the Run button.

When the preparation step is complete you will see two new entries in the entry list (two
since the compound has a chiral center leading to two stereoisomers). Click the circle next
the empty entry name to view the 3D structure of the ligand.

Step 4: Docking
Now that we have successfully prepared both the protein and ligand we can finally carry
out the final step: the docking experiment to explore a plausible binding mode of Pinolol to
B2AR.

From the main menu open the Ligand Docking dialog (Tasks → Docking → Glide
docking). Here we need to define which “Receptor grid” to use. Click the Browse button
and select the grid we created above (glide-grid_1/glide-grid_1.zip).

Choose Use ligands from: “Workspace (included entries)”. Hit the Run button. When
the job is complete, a message appears in the Maestro viewer window:

Click the View in project link and the protein in complex with Pinolol will show in the
viewer window. If you do not see the ligand, right click on glide-dock_SP_1 in the entry
list, and click View Poses. Then, in the Pose Viewer window appearing on the right hand
side, click the Set Up button. Click the Presets button on the task bar to visualise the
protein as cartoon with residues in lines, and ligand in sticks.

Finally, hit the Ligand Interaction Diagram button to obtain a 2D schematic overview of
the predicted interactions.

Page 4 of 6
 Docking Tutorial - WASB 2017

Figure 1: Ligand interaction diagram of the B2AR in complex with Pindolol. Note the
hydrogen bonding network of the NH2 and OH groups close to Asn312 and Asp113. This
binding pattern is seen in several beta blockers.

Structure-Based Virtual Screening

The search for potential new drug candidates is typically initiated by high-throughput
screening (HTS) of a library of diverse chemical compounds. Its aim is to probe effects of
ligand binding to a target protein and identify small-molecule modulators of protein
function.

Experimental HTS is commonly limited in the chemical space mainly due to high costs of
purchasing or synthesizing thousands of compounds. Computational procedures can
potentially complement experimental screening by rapid and inexpensive virtual screening
of large databases (thousands to millions of compounds). Structure-based virtual screening
depends only on a 3D structure of the target protein, a database of available compounds,
and software (i.e. a docking program) to (ideally) discriminate between active and inactive
compounds.

In structure-based virtual screening we dock each small molecule of our library/database to

the target protein with the aim of identifying the likely binding interaction of the
compound. The compounds in the library is then ranked based on the calculated docking
score. In this section we will download a small library of 20 compounds with 4 known
binders and 16 decoys. We will then use HTVS with Glide to estimate their binding mode to
our B2AR target protein.

Side-note: This small library of 20 compounds is a subset of the DUD (database of useful
decoys; http://dude.docking.org/subsets/gpcr) designed specifically for B2AR. The
selection of 16 decoys was obtained by clustering all compounds with 40 or less atoms.
The four active compounds were collected from PubChem.
Page 5 of 6
 Docking Tutorial - WASB 2017

To get started, download our compound library and save the mylib.mae file in the
docking directory (and unzip it!). In Maestro open the Ligand Docking dialog (Tasks →
Docking → Glide docking). In the same way as for the docking of pindolol above, we must
define the “Receptor grid” to use (we’ll use the same we generated above). Click the
Browse button and select the grid we created above (glide-grid_1/glide-grid_1.zip).
Choose Use ligands from: “Files”. Hit the Browse button and select mylib.mae.

In the settings tab, choose Precision HTVS (high throughput virtual screening). Hit the
Run button to start the docking of the 20 molecules. When the job is complete you will find
a new entry (called glide-dock_HTVS_1_pv..) in the Workspace Navigator on the right
hand side. The compounds are listed below the 2RH1 entry (our target protein) and ranked
according to it’s docking score. All decoys have been named by its ZINC database ID (e.g.
ZINC03271172) while the known actives are name by its PubChem database ID (e.g.
4828). In this case it seems like the VS protocol worked very nicely by ranking the actives
in front of the decoys.

Summary
In this tutorial we have carried out the basic steps of protein-ligand docking – from the
preparation of the protein and ligand structures to the grid generation and the docking
experiment. In this example we docked a known beta blocker to the structure of B2AR
revealing several important interaction points between the protein and ligand.

Page 6 of 6