The Selectivity Question That Won't Sit Still
Promega Corporation
Publication Date: July 2026
Introduction
You have the selectivity data. The compound looks clean across a broad kinase panel. The question sitting at the back of your mind, the one you might not ask out loud, is whether that's enough.
It's the same question kinase researchers have been asking since the beginning. What changes, generation by generation, is what counts as an answer.
Imatinib and the illusion of a settled question
In 1996, Brian Druker and colleagues showed that a compound designed to inhibit the Abl tyrosine kinase could reduce bcr-abl colonies in CML patient samples by 92–98%, with no effect on normal colony formation. That was the proof of concept the field needed: kinase selectivity was achievable, biochemical assays could document it and a molecule designed with precision could discriminate between cancer cells and healthy ones. CGP 57148, later imatinib, became the founding proof that the target hypothesis was worth pursuing.
The evidence was sufficient for what regulators required, and imatinib's clinical success validated the approach. What the field wouldn't fully appreciate until later is that Abl was a target where biochemical selectivity data and cellular reality happened to align. That alignment wasn't guaranteed.
When biochemical selectivity stopped being the whole answer
What happened next wasn't a failure of the biochemical approach. It was a consequence of success. As kinase drug discovery expanded into more challenging targets, more complex biology and longer timelines, answering the selectivity question required tools the field didn't have yet. Later kinase inhibitor generations learned something the imatinib generation didn't have to reckon with: biochemical selectivity data and cellular selectivity data don't always tell the same story. The assays weren't wrong, but a kinase in its native cellular environment behaves differently than purified protein in a test tube. Those differences turn out to matter.
In the development of rilzabrutinib, a covalent BTK inhibitor, traditional equilibrium ICâ‚…â‚€ measurements couldn't capture what the team needed to know: not just whether a compound bound its target, but whether it stayed bound long enough to matter after the drug was cleared, a critical question for BTK inhibitors intended for autoimmune indications. Answering that question required cellular occupancy assays, measurements of durable target engagement in living cells. Biochemical data and cellular data told different stories, and optimization followed the cellular data. A different set of measurements would have pointed the program somewhere else.
Biochemical data and cellular data told different stories
What cellular target engagement made visible
A 2026 paper from Matthew Binder and colleagues systematically documents the biochemical and cellular divergence. They compared biochemical and cellular selectivity profiles for a diverse set of inhibitors across a panel of 240–300 kinases. For type I inhibitors, cellular assays picked up fewer interactions than biochemical panels did: the intracellular environment appeared to filter out some binding events. For type II inhibitors, the ones that bind the inactive DFG-out conformation, cellular assays picked up additional interactions the biochemical panel missed entirely: not smaller effects, but absent ones.
Matt Robers, a senior research scientist at Promega and a corresponding author on the paper, frames what cellular assays reveal: "A compound that shows modest activity against an isolated protein but potently stabilizes an intracellular complex in cells isn't a false positive. It's telling you the complex is the real drug target." What looks like a biochemical failure can be a cellular signal about the real biology.
"It's telling you the complex is the real drug target."
TPKI-39, one of the compounds Binder and colleagues profiled, looked like a poor probe candidate in biochemical profiling across that same panel: too many off-target interactions to be useful. Run the same compound in living cells using NanoBRET® Target Engagement Assays, and the picture inverts. In cells, TPKI-39 engages DDR1, DDR2 and FLT1 selectively, with 84% DDR2 occupancy against 1% inhibition in the cell-free data.
Many kinases adopt DFG-out states in cells that aren't sampled when you're working with purified, active protein in a biochemical assay, which means the cell isn't just a different container for the same experiment. The practical conclusion from Binder/Robers isn't that biochemical assays are insufficient: they're answering a specific question, and cellular assays are answering a different one. Biochemical selectivity data still guides compound design and narrows the field. Cellular data adds resolution rather than replacing it.
They're still questions about binding, however, and binding isn’t always the full story.
Kinase degraders and the question in a new shape
Targeted protein degraders don't just inhibit a kinase. They recruit an E3 ligase to tag it for destruction, which changes the selectivity question in ways that biochemical and even cellular binding assays weren't designed to answer. It's no longer enough to know what a compound engages. You need to know what gets degraded.
It's no longer enough to know what a compound engages. You need to know what gets degraded.
A 2025 paper from Angela Fan and colleagues demonstrates this directly. Using a series of related pan-kinase warheads with variable target residence times, they built VHL-recruiting degraders and profiled which kinases were degraded. The warheads bound broadly across the kinome. Ternary complexes formed broadly too, but only 25 kinases across the series were degraded. Which kinases got degraded depended on residence time in ways that varied by target. Some kinases were degraded only by fully reversible compounds with short residence times. Others required sustained, covalent engagement; brief occupancy wasn't enough. Binding wasn't the bottleneck for any of them. What limited degradation was whether the ternary complex could be productively ubiquitinated, and that depended on target-specific kinetics that no biochemical or binding assay could predict.
The question the field thought it had answered at the level of binding re-opened at the level of degradation. What counts as selectivity evidence for a PROTAC isn't the same as what counted for an irreversible covalent inhibitor, which wasn't the same as what counted for imatinib. This is the pattern again. Sufficiency isn't fixed.
Conclusions
More than 100 small-molecule kinase inhibitors have been approved in the two decades since imatinib. Every one of those approvals rested on an answer the next generation would find incomplete. That pattern is what drives new tools.
Each new set of tools opened programs the previous set couldn't reach. Biochemical assays opened kinase drug discovery. Cellular target engagement extended what biochemical could show. Ternary complex measurement asked what happens after binding. But all of them describe what a compound does to a cell line, not to a patient. Patient-derived organoids and 3D culture are pushing further still. At that stage, selectivity becomes whether a compound can tell diseased from healthy tissue in the same person. The gap between surrogate models and clinical outcomes is where the question goes next.
Our tools have kept pace because our work lives inside the same pattern the research does. Every time the field asked a harder version of the selectivity question, from biochemical panels to 3D and organoid models, we've built the assays to measure it. As the pattern continues, your next tool is already being built.
References
- Druker, B.J. et al. (1996) Effects of a selective inhibitor of the Abl tyrosine kinase on the growth of Bcr-Abl positive cells. Nat. Med. 2, 561–6. www.nature.com/articles/nm0596-561
- Owens, T.D. et al. (2022) Discovery of reversible covalent Bruton's tyrosine kinase inhibitors PRN473 and PRN1008 (rilzabrutinib). J. Med. Chem. 65, 5300–16. https://pubs.acs.org/doi/10.1021/acs.jmedchem.1c01170
- Mullard, A. (2025) FDA approves 100th small-molecule kinase inhibitor. Nat. Rev. Drug Discov. 24, 892–5. www.nature.com/articles/d41573-025-00188-7
- Binder, M.J., Bashore, F.M., Dunn Hoffman, K.K., Damgaard, C., Slater, M., Drewry, D.H., Robers, M.B. and Axtman, A.D. (2026) Cellular Context Influences Kinase Inhibitor Selectivity. J. Med. Chem. 69, 4208–20. https://pubs.acs.org/doi/10.1021/acs.jmedchem.5c02916
- Fan, A.T., Gadbois, G.E., Huang, H.-T., Chaudhry, C., Jiang, J., Sigua, L.H., Smith, E.R., Wu, S., Poirier, G.J., Dunne-Dombrink, K., Goyal, P., Tao, A.J., Sellers, W.R., Fischer, E.S., Donovan, K.A. and Ferguson, F.M. (2025) A kinetic scout approach accelerates targeted protein degrader development. Angew. Chem. Int. Ed. 64, e202417272. https://onlinelibrary.wiley.com/doi/10.1002/anie.202417272
- Syphers, J.L. et al. (2026) Discovery of APO-50815, a potent WEE1 kinase inhibitor with exceptional efficacy against patient-derived colorectal cancer organoids. Eur. J. Med. Chem. 312, 118838. https://pubmed.ncbi.nlm.nih.gov/42030708/
Learn more about Target-Based Small Molecule Drug Discovery