Growing around a fragment using 3D structure-based design

Problem: Starting from a potent fragment, how to generate in Makya novel, synthetically accessible molecules with good binding affinity to the target protein?

In this use case, we follow the design strategy described in a peer-reviewed study:
J. Med. Chem. 2023, 66, 9, 6354–6371

The authors describe the optimization of non-amidine-based C1s inhibitors, starting from a hit compound identified via high-throughput screening. Through successive manual modifications, including the replacement of an aminoisoquinoline with an aminophthalazine, they arrive at compound 4e, which shows improved potency and serves as our starting point for generative design (See reference structure in compound 4e).

Using Makya, we successfully reproduced and extended their results, including the generation of novel structures comparable to compounds described in series 5 and 6 of the paper.

Compound 4e, ref: J. Med. Chem. 2023, 66, 9, 6354–6371

 

Download the Dataset and Structure Files

To follow this use case step-by-step, download the necessary files below:

• Protein file (1ELV)

• Reference ligand file (compound_4e)

These files are required to set up the structure-based generation process in Makya.

Step 1: Create the structure-based parameters

Step 1: We start by setting up a 3D Structure-based configuration that can then be used to guide generation. For more details on the structure-based set-up, see the dedicated section in the documentation. Simply click on the "3D structure-based settings" tab, then add your new 3D docking parameters by clicking on the + logo:

A window appears that lets you choose a name for your parameters, as well as a docking tool. You will then be prompted to upload in Makya the following files:

• the protein file (.pdb format). In this use case, the PBD ID is 1ELV.
• and the reference ligand file (.sdf format). In this use-case, it is compound 4e.

The page will refresh and display the interactions between your reference ligand and target protein. We will tune the weights of each interaction, and give a higher weight to the most crucial interactions, so as to define the contact score.

Choosing the weights is an important step in the set-up that relies on an expert understanding of the problem at hand.

Existing interactions we want Makya to focus on:

Several interactions within the binding pocket were found to be important for activity. Thus, we want newly generated molecules to target and increase the strength of these interactions.

Set the weights as follows:

• Hydrophobic:
  • PRO residue, index 520: weight = 4
• Pi_stacking:
  • HIS residue, index 475: weight = 7
  • Add a new interaction: PHE residue, index 526: weight = 6
• Pi_cation:
  • LYS residue, index 629: weight = 5

New interactions we want to form:

To increase activity and occupy the binding pocket efficiently, we can decide to target new interactions that the reference ligand does not form with the target protein.

Add the following interaction in your structure-based settings:

• Pi_stacking:
  • PHE residue, index 526: weight = 6

Other interactions:

All other interactions are not crucial. Adding a weight to them would "dilute" Makya's focus on the important interactions and make it more difficult to find compounds that fit our objectives. 

You can either let their weight at 1, or even completely remove these interactions from the contact score by putting their weight to 0 / unchecking the check box before the residue name. 

Save

Once the interaction weights have been tuned, click the Save button.

Step 2: Create the generator

Step 2: We will now set up and launch our generator. On the project homepage, go directly to the "Generators" tab and click on New Generator. Give the generator the name of your choice. Under "Generation Engine", select Fragment growing.

On the following pages, specify the settings of the generator:

• Exit vectors: In the text field below "Enter SMILES", copy-paste the following SMILES:

Nc1nncc2cc(Br)ccc12

This is the aminophtalazine that improves compound potency and that we will use as the starting point for our generation. Note that we use a bromo derivative to allow for reactions on the fragment. Set the exit vector as shown on the image below and click on Save.
 

NOTES ON EXIT VECTORS:

• The fragment should not contain charged atoms. Protonation is performed directly inside Makya where needed.

• When you select the exit vectors, the atoms in red are the ones that will be involved in the reaction. The Fragment Growing generator is chemistry-driven and the choice of exit vectors need to be informed by organic chemistry knowledge .

• Building blocks: in the “First building block’s filters” column, click on "Substructures", then on the Add button near the field “Molecules must match at least one of these substructures (matching set)”. Enter the following SMARTS:

This ensures that the generated molecules will all contain a piperazine, as is the core of the reference compound. Click on Save.

• 3D Structure based: Select the 3D structure-based scorer that was previously set up.

A 3D visualization of the reference ligand and target protein pocket appears. Click on the “Anchor” tab. The anchor is a sub-unit of the reference molecule used as a starting point for 3D alignment.

IMPORTANT:

1. This sub-unit has to be present in all the generated molecules for the 3D alignment to be done. Please make sure that your exit vector or building block constraints contain this part.
2. The anchor should contain a minimum of 5 heavy atoms.
3. The anchor should not be too long, otherwise the docking score might not be computed. We recommend choosing the largest rigid ring of your reference ligand.

Enter the following SMILES for the anchor:

Nc1nncc2ccccc12

Note that this aminophtalazine will be present in all generated compounds since it is part of the starting fragment. Click on Save.

The generator is set and ready to run. Go back to the "Generators" tab, and the newly set up generator will show up. Click on the Run button for this generator. The progress bar will indicate the status of generation.

Step 3: Explore the generated molecules

Step 3: while the generation is still running, you can visualize the first results. By clicking on the eye icon, you will be able to browse through the results of the generation.

Molecules generated with a 3D structure-based scorer and docked with the default AutoDock Vina are scored using the Contact score and Vina Docking score. They can be sorted using these scores (see the following page in the documentation).

By clicking on See details on any result, in the "Docking" tab, the generated molecule can be seen docked in 3D in the target protein binding pocket.

To see the overlay with the reference ligand, check the Show Reference Ligand box above the 3D visualization.

Note that the below generated molecule is compound 5g from the paper, which has been shown to have an IC50 equal to 0.37 μM and has been found by Makya generators!

By selecting several molecules and changing the "View" mode to 3D Structure Based Results, you can superpose all the molecules in the pocket simultaneously. You can also export the results in  CSV or SDF format to view the results in your favorite viewer.