3-D searching by Pharmacophore

By Scott Lusher and G. Schaftenaar



The definition of a pharmacophore is the minimum functionality a molecule has to contain in order to exhibit activity. For a series of derivatives, the molecules usually have much in common and to derive the true minimum pharmacophore, structurally diverse active molecules are required.
For the ERalpha receptor we will design the following pharmacophore model:

Unity 3D Search window

The Raloxifene molecule, with the pharmacophore model overlayed:

Setup the working environment

If necessary, check the X-windows start-up page for detailed instructions on how to set up the X-windows environment and to access the CMBI's main Unix machine, cheminf.cmbi.kun.nl. Then, from the Unix shell (command prompt):



We will define a pharmacophore for an active ligand of the ERalpha receptor.

We will start with the known active compound Raloxifene.

File >> READ >> raloxifene.mol2

UNITY

Unity >> Select Donor/Acceptor >> Donor Atom >> click on the two hydroxyl oxygens >> Click End Select >> Select Molecular Area 1 (M1) to store this pharmacophore feature.
Respond with Yes to the question: Create spatial constraints, supply the value 1.0 for the Tolerance for spatial constraint.

Unity >> Select Donor/Acceptor >> Donor Atom >> click on the tertiary nitrogen >> Click End Select >> Select Molecular Area 1 (M1) to store this second pharmacophore feature.
Respond with Yes to the question: Create spatial constraints, supply the value 1.0 for the Tolerance for spatial constraint.

To add the last feature, the phenyl ring connected to one of the hydroxyls has to be defined, this will take a little more work:

Build/Edit >> Delete >> Atom .. >> select all aromatic carbons of the phenyl ring >> invert the selections by clicking Invert >> OK

Now we have to save the pharmacophore:

File >> Save as >> RAL_PHARM >> Save




Now let's start with a clean sheet:

Build/Edit >> Zap (Delete) Molecule

and let's read the pharmacophore back in:

File >> Read >> RAL_PHARM >> OK

Now we will search a database of 500 compounds (485 non-drug, 15 drug) with the pharmacophore we have just defined.
First read in the database:

File >> MolecularSpreadsheet >> Open >> 500.tbl >> OK

Now let's do the search:

Unity >> Unity Search.

The Unity Search window will pop up.

Unity 3D Search window

Change the following settings:

In the sybyl textport you will find the number of compounds in the database that fit this pharmacophore: 5 which is 1% of 500.



Now let's have a look a these compounds:

Unity >> Load Results >> From a Search >> take the last entry.

A new spreadsheet will popup.

Unity 3D Search window

Note that all 5 compounds are drugs (serms = Selective Estrogen Receptor Modulators ). There are NO FALSE POSITIVES. But there are 15 drugs in the database, so this search did not find 10 of them. So there are 10 FALSE NEGATIVES.

Inspect the hits for the presence of the pharmacophore features we defined. For each hit:

Select Row >> Show Row Sel.

Notice that raloxifene, the active compound we used to construct our pharmacophore from, is amongst the found hits.



In fact there are two pharmacophore models required to describe all known active ligands of the ERalpha receptor. The second is less stringent than the one we have used upto now. Let's create the new pharmacophore:

Close all spreadsheets withour saving. Zap all the molecules:

Build/Edit >> Zap

Now let's read the original pharmacophore back in and change it:

File >> Read >> RAL_PHARM >> OK

UNITY >> Delete >> Features >> select the hydroxyl oxygen furthest away from the phenyl ring.
A window will pop up with the name Multiple picks found. Select DA_18 and hit the OK button.
The spatial constraint will automatically be removed as well.
>> End Select.

Now let's redo the pharmacophore search described above with this new pharmacophore model. If you have done this correctly, you will find 17 compounds. So this pharmacophore is less stringent than the first we used (we removed a feature).

Inspect the hits for the presence of the pharmacophore features we just defined.
For each hit:

Select Row >> Show Row Sel.

We now found 10 known drugs (serms) but also 5 false positives. Let's check wether these are really false negatives. Let's overlay the first potential false negative with the known active compound we constructed our pharmacophore models from: raloxifene.

In the spreadsheet select the first row >> Show Row Sel.
Read in raloxifene:

File >> Read >> raloxifene.mol2 >> select molecular area 4 (M4) to display it in.

You will find that the molecules don't overlay very well. This is a first indication that we are dealing with a false positive.

A second and better check is to establishing if there are clashes between the ligand and residues of the active site of the receptor.

Build/Edit >> Zap (Delete) Molecule >> All >> OK File >> Read >> ERSTRUCTURE >>

Shown are, the active site residues, raloxifene (Yellow) and the first hit from the pharmacophore search (Red).

NOTE: There are clashes !!

The lesson we should have learned here is that a pharmacophore search is not enough to identify known and new active ligand. We should always try to confirm our hits by docking them into the active site (That is, if the 3D structure of the receptor is known).



MOLCAD is a sybyl module which is capable of displaying the protein as a 3D surface. This is very instrumental in detecting clashes between receptor and ligand.
But first let's get rid of the orginal ligand, raloxifene.

Build/Edit >> Delete >> Atom >> select M1 >> Substructures >> RAL999 >> OK >> OK

Create the surfaces for the active site and the ligand:

View >> MOLCAD surfaces >> Molecular Surfaces
select molecular area 1 (M1) >> Create >> OK
select molecular area 2 (M2) >> Create >> OK change color to Yellow >> Done

Let's go through the protein by displaying only a thin slab of it.

Hit the Slab icon (Left column of icons, 8th from the top). Check the Clip everything option.
Select D1 Connolly, Set Mid Pt to -13.0 and hit the down arrow beneath Mid Pt.
Keep it pressed to move the slab through the protein.