Modeling kinase-ligand structure-activity relationships remains challenging due to kinase conformational flexibility and limitations of computational methods like docking. Here, we introduce Terra, a novel protein-ligand co-folding model fine-tuned on kinase conformations and ligand SAR data. Terra achieves correlation coefficients matching or exceeding FEP methods for known and novel kinase targets, with a 10X lower computational cost. We apply Terra to identify novel allosteric JAK2 inhibitors targeting the V617F mutation. 

Kinases and Phosphatases present similar challenges due to the conservation and polarity of their active sites, which limits the selectivity and drug-likeness of orthosteric inhibitors. Kuano applies QM/MM to simulate transition states of phosphatases in the PTPN family, guiding the design of novel small molecule inhibitors with enhanced specificity, binding complementarity, and potential for covalency. These ‘quantum pharmacophores’ along with AI-assisted design workflows offer transferable solutions to kinase inhibitor design.

Using our ML-driven platform, we engineered a next-generation HER2 inhibitor with exceptional selectivity, broad mutant coverage, brain penetrance, which is currently being evaluated in a Phase 1/1b clinical trial. This new class of compounds bind HER2 in a type II DFG-out conformation, representing the first reported type II HER2 tyrosine kinase inhibitors. This program was enabled by our AI technologies, including PropANE (a precursor to our Enchant platform) and NeuralPLexer. Combined with high-throughput parallel synthesis and screening, these tools accelerated the identification and optimization of highly selective candidates with broad therapeutic potential in HER2-driven cancers.

Pantothenate kinase-associated neurodegeneration (PKAN) is a rare, autosomal recessive disorder caused by inactivating mutations in the human PANK2 gene which encodes the PANK2 protein. In humans, four isoforms of Pantothenate kinase (PANK1α, 1β, 2, and 3) phosphorylate pantothenate (vitamin B5) is the first and rate-limiting step of Coenzyme A (CoA) biosynthesis. To address CoA deficiency in PKAN patients, we developed first in class small molecule PANK1 and 3 activators using lipophilic efficiency and structure guided design. This approach led to the generation of BBP-671, which advanced into clinical trials.

Raf inhibitors trigger paradoxical activation of BRAF at sub-saturating concentrations. Using biophysical techniques tracking BRAF conformation and oligomerization we show that paradoxical activation occurs because the first inhibitor molecule to bind to RAF dimers is much more strongly coupled to dimerization than the second inhibitor molecule. NMR experiments show this arises from inherent asymmetry of the BRAF dimer itself. Our results provide new insight into the mechanisms underlying paradoxical activation.

We developed COOKIE-Pro, a new method to quantify the binding kinetics of covalent inhibitors across the entire proteome. Using mass spectrometry, it determines kinetic values (kinact?/KI?) for both on- and off-target proteins. We validated COOKIE-Pro with BTK inhibitors, reproducing known parameters and identifying novel off-targets. This powerful tool helps optimize the potency and selectivity of covalent drugs, aiding preclinical development by providing a comprehensive view of inhibitor binding.

I will describe sulfonyl-triazoles as an enabling electrophile for developing covalent binders to tyrosine and lysine residues on proteins through sulfur-triazole exchange (SuTEx) chemistry. SuTEx chemistry is highly tunable with respect to protein reactivity and specificity, which can facilitate optimization of potent and selective binders to orthosteric and allosteric sites on kinases. I will conclude my talk by describing efforts to apply SuTEx ligands for modulating kinase function in cells.