In Silico Library Enumeration in Flare™: Using prebuilt and custom reactions to ‘synthesize’ heterobifunctional molecules

SHARE

What is library enumeration?

Library enumeration is a useful tool to build virtual compound libraries for in silico experiments, such as virtual screening, quantitative or qualitative structure activity relationship (QSAR) studies, and free energy perturbation (FEP) calculations. Flare’s Library Enumeration feature1 is reaction-based, i.e., a computational transformation is applied to defined reactants. The process is akin to performing combinatorial chemistry in silico using Reaction SMARTS, or SMIRKS, based on the RDKit method.2-3 With this tool, the user can upload a library of their in-house reagents to Flare, generate their screening compounds, and directly use the products in their subsequent virtual experiments, such as docking or QSAR. The library enumeration tool will run reactions only on the reagents that match the reaction-encoding SMIRKS and skip the others. For example, if your entire chemical inventory is in one sdf or csv file, and you only want to apply Suzuki reactions with boronic acids to another ligand in Flare, the program will only use the boronic acids in that large file to generate a library of coupled products. For more complex molecules, multiple reactions can be queued to set up a reaction cascade that outputs the final desired product. In this article, we will illustrate the utility of the Library Enumeration tool, using both prebuilt and custom reactions to ‘synthesize’ heterobifunctional molecules for targeted protein degradation. Figure 1 highlights the reaction breakdown for the Library Enumeration feature in Flare.

Figure 1. Breakdown of the available reactions in the Library Enumeration feature within Flare V10. There are currently 167 reactions available.

Synthesis of heterobifunctional molecules: an illustrative example

To illustrate the utility of the Library Enumeration feature, we will use the synthesis of degraders 10a-b and 12a-d by Manda, S. et al.,4 highlighted in Scheme 1.

Scheme 1. Synthesis of degraders 10a-b and 12a-d

 

The synthesis begins with a commercially available haloalcohol containing either one or two polyethylene glycol (PEG) units. Steps 1 and 2 are the displacement of the chloride by sodium azide to mask an amine, followed by an SNAr on p-fluoronitrobenzene to generate ethers 4a-b. The azide is then reduced to the free amine in Step 3, which is then reacted via SNAr in Step 4 to the CRBN-binding ligand 6. Reduction of the nitro group in Step 5 gives anilines 8a-b, which are then reacted in Step 6 with the various inhibitors 9 and 11a-b to give degraders 10a-b and 12a-d.

With Flare’s current catalog of reactions, the entire synthesis in Scheme 1 can be completed using the library enumeration tool. However, in Step 6, the cross coupling between the two intermediates produced in Step 5 and the three inhibitors would generate 16 products, as the Library Enumeration feature does not account for chemical reactivity nor sterics. For this last step, we used a custom SMIRKS to avoid these undesired side reactions.

Figure 2 illustrates the reaction cascade in the Library Enumeration panel, as well as the generated heterobifunctional degraders.

Figure 2. The synthesis of degraders with the Library Enumeration feature. Six reactions were set up in Flare, outputting 6 final degrader structures.

Depending on your level of expertise in SMARTS/SMIRKS, you can also condense this entire synthesis with custom reactions. For the first part, illustrated in Figure 3, we can see that instead of the five-step cascade required to go from reagents 1a-b to 8a-b, we have arrived at intermediates 8a-b in two simple reactions. The first custom reaction starts with the PEG halo alcohols (Reagent 1), simultaneously converting the PEG chlorine to nitrogen and automatically attaching the thalidomide ligand to the other side. The second reaction continues with what was originally Step 2 in Scheme 1, attaching the p-aniline to the linker’s terminal hydroxyl group, defined by ‘[O:1]’ (Figure 3). This shortcut would be nearly impossible in a wet-lab scenario but is conveniently allowed virtually.

Figure 4. The last custom reaction of the sequence, reacting a specific carbon on the ligands with a specific nitrogen of the intermediate. The reaction between intermediate 8 and ligand 11b is used as an illustrative example. The transformed atoms are highlighted to correspond to the SMIRKS reaction.

The results now correspond to both the correct number of compounds and correct connectivities of our final degraders. Conveniently, instead of setting up every single reaction in the cascade, we were able to complete the entire synthesis with three custom reactions. Using this virtual approach, we avoid the masking and deprotection steps that would be necessary in the wet lab. The result is the six desired degraders from three ligands and two haloalcohols. Figure 5 shows the 3D structures of the degraders, along with their field points, in which the 2D products from the library enumeration tool have been converted to 3D using the Pop to 3D tool.

Figure 5. The final degrader structures in 3D with their field points.

The final degraders are now ready for subsequent experiments in Flare, such as ligand alignment, docking, or QSAR. They may also be expanded into further analogs with the Hit Expander tool or manually edited into new structures.

Conclusion

With the uptick in research on heterobifunctional molecules as therapies, there is an increasing need for methods to model them computationally. Library enumeration allows for the ‘synthesis’ of large libraries of degraders in silico, avoiding the tedious task of generating virtual structures. With Flare’s 167 reactions (and more likely to be added in future versions), the assembly from common building blocks is easy and facilitates the generation of large libraries for further experimentation.

References:

  1. https://cresset-group.com/software/flare-library-enumeration/
  2. https://www.rdkit.org/docs/RDKit_Book.html#reaction-smarts
  3. https://www.daylight.com/dayhtml/doc/theory/theory.smirks.html
  4. Manda, S. et al. Molecules 2020, 25(8), 1948. https://doi.org/10.3390/molecules25081948

Related Science Resources

No related resources

Subscribe & Don't Miss Out

Receive our newsletter to be among the first to hear about product releases, case studies, opinion articles, events and more.