Accelerated Discovery of Macrocyclic CDK2 Inhibitor QR-6401 by Generative Models and Structure-Based Drug Design
Many cancers feature dysregulated function of complexed cyclin-dependent kinases (CDKs) and their associated cyclin partners. The pursuit of CDK inhibitors has resulted in approval of CDK4/6 inhibitors, but some resistance has been observed. CDK2 amplification in many cancers has been observed, and due to its overlap with the CDK4/6 pathway that results in the release of common transcription factors, CDK2 inhibition may offer a benefit to patients with certain cancers that are resistant to treatments operating through a CDK4/6 mechanism.
In this issue, Yu et al. (DOI: 10.1021/acsmedchemlett.2c00515) used artificial intelligence (AI) to accelerate the development
of macrocyclic CDK2 inhibitors. The authors employed 10 published
CDK2 inhibitors as starting points, and then a Fragment-Based Variational
Auto-Encoder generative model (FBVAE) was developed to fragment hop
and replace the essential hinge binding elements, resulting in 3220
compounds. Glide docking and other filters were used to prioritize
the top 10 molecules, which were then synthesized with some modifications
and tested for CDK1 inhibition. A potent prototype that emerged was
used to solve a CDK2/cyclin E1 co-crystal structure,
thereby revealing a molecular arrangement that might benefit from
rigidification through macrocyclization. Following a survey of linkers
that improved microsomal stability, permeability, selectivity, and
potency, the team generated macrocyclic QR-6401. The compound is potent
and selective for CDK2 and exhibited remarkable antitumor efficacy
in an OVCAR3 ovarian cancer xenograft model. This work showcases how
the creativity and skill of the medicinal chemist could be aided by
AI models to accelerate drug discovery outcomes.
Controlling Ibrutinib’s Conformations about Its Heterobiaryl Axis to Increase BTK Selectivity
Improvement of target selectivity has been an essential topic in drug discovery due to the potential for off-target side effects and toxicities. In this issue, Toenjes et al. (DOI: 10.1021/acsmedchemlett.2c00523) described an efficient strategy to improve the selectivity of a Bruton’s tyrosine kinase (BTK) inhibitor without compromising potency. Here, the authors focused on conformational control of atropisomers which occurred when σ-bond rotation was hindered about an aryl-linked pyrazolopyrimidine core. It is estimated that approximately 33% of FDA-approved small molecules feature a potential atropisomeric axis; therefore, control of this may result in an improved fit within the target protein and reduce off-target binding.
Using ibrutinib
as an example, the authors calculated conformational
energy profiles and showed that addition of ortho-situated methyl groups on the appended phenyl ring near the atropisomeric
axis can favor a conformation that is preferred for BTK binding, thus
improving the selectivity against other kinases that lack this preference.
Upon synthesis of the designed compounds and testing against ibrutinib’s
top 50 targets, the compounds demonstrated a significant increase
in kinase selectivity. Further, these compounds maintained inhibitory
activity against BTK comparable to that of ibrutinib. This work shows
that conformational control is a feasible way to improve the selectivity
toward target proteins.
Passive Membrane Permeability of Sizable Acyclic β-Hairpin Peptides
The disruption of intracellular protein–protein interactions may be better addressed with large-size peptides, as opposed to small molecules which are less suited for this role. However, large biomolecules can lack appropriate drug-like properties, which limits their use in vivo. In this issue, Moxam et al. (DOI: 10.1021/acsmedchemlett.2c00486) developed a library of β-hairpins with improved membrane permeability that target the protein–protein interaction with programmed cell death-1 protein (PD1) and the ligand-1 (PDL1) in intracellular exosomes, as this association is leveraged by cancer cells to attenuate immune responses.
The β-hairpins were synthesized using
solid-phase peptide
synthesis (SPPS), purified by semi-preparative RP-HPLC, and assessed
for permeability using a parallel membrane permeability assay (PAMPA).
A strong correlation was observed between permeability and lipophilicity,
and specific changes to the peptide side chains permitted tuning of
these features. These results provide a potential roadmap to configuring
the properties of these large hairpin peptides to possess high passive
permeability and be useful in the disruption of protein–protein
interactions.