KINOMEscan is a competitive binding assay that measures how strongly a small molecule interacts with hundreds of kinases in a single experiment. Developed by what was originally Ambit Biosciences (later acquired by DiscoveRx, now part of Eurofins), the platform has become one of the most widely used tools for profiling kinase inhibitor selectivity during early drug discovery. Its value lies in speed and breadth: a single compound can be screened against more than 400 wild-type human kinases plus disease-relevant mutants, producing a kinome-wide binding map in days rather than months. But binding in a test tube is not the same as engagement inside a cell, and the gap between the two is one of the more instructive stories in modern kinase pharmacology.
How the Assay Works
KINOMEscan relies on a competitive displacement principle. Each kinase in the panel is expressed as a fusion protein, either displayed on the surface of T7 bacteriophage or tagged with a short DNA barcode. These tagged kinases are incubated with an immobilized “bait” compound, typically a broad-spectrum kinase-binding ligand such as staurosporine, that captures them out of solution. When a test compound is added, any kinase for which the test compound has affinity gets pulled away from the bait. Kinases that remain bound to the bait are washed away, and whatever the test compound displaced is quantified by reading the phage or DNA tag using quantitative PCR.1Cell Chemical Biology. Global Analysis of Kinome Selectivity and Function of Small Molecule Kinase Inhibitors
The readout is straightforward: the more a test compound displaces a kinase from the bait, the lower the remaining signal. Results are reported as “percent of control” (PoC), where a lower number means stronger binding. A compound that shows 0% of control for a particular kinase is essentially knocking it completely off the bait, while 100% means no detectable interaction. This format lets researchers quickly scan hundreds of kinases and spot hits worth following up on.
Because the assay measures binding rather than enzymatic activity, it does not require ATP or substrate peptides. That is both its strength and its limitation. It means the assay can detect interactions with pseudokinases and kinases that are difficult to express in active form, broadening the panel considerably. But it also means the assay cannot distinguish between a compound that binds tightly and one that actually shuts down kinase function in a biologically meaningful way.
Panel Size and Selectivity Scores
The most commonly used version of the platform, called scanMAX, tests a compound against 403 wild-type human kinases. Additional panels include mutant kinases and atypical members of the kinome. A broader analysis of profiling data in the DiscoveRx format covered 456 kinases across more than 3,300 inhibitors, giving the field an unusually comprehensive picture of which chemical scaffolds hit which parts of the kinase family.2PubMed. Extending kinome coverage by analysis of kinase inhibitor broad profiling data
Selectivity is usually summarized with an “S score.” The most common variant, S(10), represents the fraction of kinases in the panel that a compound binds at less than 10% of control at a given concentration. A low S(10) means the compound is highly selective, hitting only a few kinases. A high S(10) means it is promiscuous. In one study developing chemical probes for understudied kinases linked to neurodegeneration, 21 novel pyrimidine-based compounds were screened on the scanMAX panel at a single micromolar concentration. The most selective compounds hit an average of roughly three out of 403 wild-type kinases, while less selective analogs with a different substitution pattern hit an average of about 45.3Journal of Medicinal Chemistry. Identification of Pyrimidine-Based Lead Compounds for Understudied Kinases Implicated in Driving Neurodegeneration That kind of range illustrates how a single structural change can dramatically shift a compound’s kinome footprint.
For compounds that look promising in the single-concentration screen, the platform offers a follow-up service called Kd ELECT, which generates dose-response binding curves and calculates dissociation constants for selected kinase targets. This two-tier workflow, broad scan first, quantitative follow-up second, has become the standard approach for triaging kinase inhibitor selectivity.
How KINOMEscan Gets Used in Practice
The platform shows up at several stages of the drug discovery pipeline. During early hit-to-lead chemistry, medicinal chemists use it to understand how structural modifications to a compound affect its selectivity profile. Running a scanMAX panel on a handful of analogs can reveal whether a new chemical series is inherently narrow or broad in its kinase binding, guiding decisions about which scaffolds to pursue.
For chemical probe development, KINOMEscan data often serve as a gatekeeper. Initiatives like the Kinase Chemogenomic Set (KCGS) require candidate probes to meet specific selectivity thresholds on the scanMAX panel before they are accepted as tools for biological research.4Journal of Medicinal Chemistry. Identification of Pyrimidine-Based Lead Compounds for Understudied Kinases Implicated in Driving Neurodegeneration The rationale is simple: if you want to study the biology of a specific kinase, you need an inhibitor that hits your target without clouding the picture by hitting dozens of others.
The assay also plays a role in characterizing clinical-stage drugs. When researchers needed to understand the binding profile of crenolanib, a compound active against imatinib-resistant gastrointestinal stromal tumors, they turned to the KINOMEscan Kd ELECT panel to map its recombinant kinase affinities.5Clinical Cancer Research. Crenolanib Inhibits the Drug-Resistant PDGFRA D842V Mutation Associated with Imatinib-Resistant Gastrointestinal Stromal Tumors Similarly, the selectivity of tucatinib for HER2 was measured using a related scanELECT assay, which showed that the drug bound HER2 overwhelmingly more potently than other kinases tested: essentially zero residual signal for HER2 at 200 nanomolar, while other kinases like EGFR and RAF1 retained the majority of their signal even at ten times that concentration.6Molecular Cancer Therapeutics. Tucatinib has Selective Activity in HER2-Positive Cancers and Significant Combined Activity with Approved and Novel Breast Cancer–Targeted Therapies
Researchers developing chemical probes for LIM-kinases used the scanMAX panel to confirm that their type I, type II, and type III inhibitors all showed excellent selectivity alongside low-nanomolar affinity for their intended targets.7Journal of Medicinal Chemistry. Development and Characterization of Type I, Type II, and Type III LIM-Kinase Chemical Probes In another case, selective GSK3 inhibitors showed that only two kinases out of 468 tested were competitively bound above a stringent threshold at a single micromolar dose.8PubMed Central. Elucidation of the GSK3α Structure Informs the Design of Novel, Paralog-Selective Inhibitors These examples show how the platform serves as both a compass and a quality check across diverse kinase targets.
What Broad Profiling Reveals About Polypharmacology
One of the more sobering findings from large-scale KINOMEscan campaigns is that most kinase inhibitors hit far more targets than their developers intended. A recent kinome-wide profiling effort screened 192 carefully chosen small molecules, selected for clinical relevance and chemical diversity, and found that polypharmacology was widespread and independent of whether a drug was approved or still in trials. The “assigned” targets of approved molecules were not necessarily the most potently inhibited, and off-targets included multiple understudied kinases that nobody was paying attention to.9Europe PMC. Polypharmacology is an enduring and nearly universal property of kinase inhibitors
This matters for two reasons. First, it complicates the interpretation of clinical efficacy. If a drug marketed as a BCR-ABL inhibitor also potently inhibits three other kinases, any therapeutic benefit (or side effect) could be driven by those other targets. Second, it opens the door to drug repurposing. When large repositories of binding data reveal that an existing clinical compound potently hits an unexpected kinase, that compound becomes a candidate for testing in diseases linked to the unexpected target.10PubMed Central. New opportunities for kinase drug repurposing and target discovery The sheer scale of kinome-wide profiling data now available, thousands of compounds tested across hundreds of kinases, makes this kind of systematic mining feasible in a way it was not a decade ago.
Where Biochemical Binding Diverges from Cellular Reality
The single biggest caveat with KINOMEscan data is that they describe binding to recombinant proteins in a controlled, ATP-free environment. Inside a living cell, the situation is considerably messier, and potency rankings can shift dramatically.
A study using an energy-transfer technique to measure kinase engagement inside cells found striking discordances with recombinant binding data for crizotinib. Kinases like LCK and the pseudokinase EPHB6, which showed strong binding in biochemical assays (reported affinities of 30 and 6 nanomolar, respectively), were barely occupied in cells at a dose of one micromolar. Meanwhile, relatively weak-affinity targets like MuSK, CASK, and TYRO3 displayed high cellular occupancy under the same conditions. The researchers attributed these discrepancies largely to competition with intracellular ATP, whose effective concentration varies by kinase and is influenced by activation state and subcellular localization.11Cell Chemical Biology. Quantitative Profiling of Kinase Inhibitor Engagement and Selectivity in Live Cells
A separate study measuring cellular binding constants for dasatinib and imatinib across 25 kinases reached a similar conclusion. Cellular affinity values were broadly in line with chemical proteomics measurements from cell lysates but diverged substantially from recombinant KINOMEscan values. The divergences were attributed to ATP competition, autoinhibitory conformations that exist in full-length native proteins but not in recombinant constructs, and membrane-dependent conformational states that only exist in intact cells.12Journal of the American Chemical Society. Quantification of Binding of Small Molecules to Native Kinases by Flow Cytometry Reveals Divergence from Biochemical Affinities
Even broader comparisons between KINOMEscan-style data and chemical proteomics platforms like Kinobeads show poor overall correlation. One large-scale study comparing two-dose Kinobead affinity data against published single-dose results from recombinant assays, including KINOMEscan, found Pearson correlations as low as 0.30 to 0.39 across compound libraries. The correlation improved when looking only at the intended targets of well-characterized tool compounds, but even there it topped out around 0.48 to 0.67. The authors noted that differences in ATP concentration, kinase activation state, the presence or absence of binding partners, and post-translational modifications all contribute to these discrepancies.13Nature Chemical Biology. Chemical proteomics reveals the target landscape of 1,000 kinase inhibitors
None of this means KINOMEscan data are wrong. They accurately report what they measure: binding to a recombinant kinase in the absence of ATP and cellular context. The problem arises when people treat those numbers as though they predict what will happen inside a cell without adjustment. A compound that looks exquisitely selective on the scanMAX panel may turn out to be less selective in cells, or selective in a completely different way, once ATP competition reshuffles the potency rankings.
Making Sense of the Platform’s Strengths and Blind Spots
Researchers who work with KINOMEscan data regularly tend to treat it as a first-pass filter, not a final verdict. The typical workflow looks something like this: screen compounds broadly on the scanMAX panel to find selectivity liabilities and unexpected hits, follow up the most interesting interactions with Kd measurements, and then validate anything that matters in a cellular context using orthogonal assays like NanoBRET target engagement, Kinobead pulldowns from cell lysates, or cellular phosphorylation readouts.
There are a few practical things worth knowing if you are interpreting KINOMEscan results:
- Single-concentration data are coarse: A percent-of-control value at one micromolar tells you whether binding exists, not how strong it is. A PoC of 5% and a PoC of 0.1% look similar on a hit list but could reflect very different affinities. The follow-up Kd determination is needed to rank targets meaningfully.
- Mutant coverage is selective: The panel includes disease-relevant kinase mutations, particularly in targets like EGFR, BRAF, and FLT3, but it does not cover every known mutation. If your program is focused on a resistance mutation not in the panel, you will need a custom assay.
- Pseudokinase hits may not translate: The platform can detect binding to pseudokinases (kinase-domain proteins that lack catalytic activity), which is useful for mapping chemical matter to this underexplored class. But as the crizotinib example illustrates, biochemical affinity for a pseudokinase may have little relationship to cellular occupancy.14Cell Chemical Biology. Quantitative Profiling of Kinase Inhibitor Engagement and Selectivity in Live Cells
- Binding mode matters: The assay can detect type I, type II, and type III inhibitors, which interact with the kinase in different conformations. The LIMK probe study demonstrated that all three binding modes could be profiled effectively on the scanMAX panel.15Journal of Medicinal Chemistry. Development and Characterization of Type I, Type II, and Type III LIM-Kinase Chemical Probes However, allosteric inhibitors that bind far from the ATP site may not displace the immobilized bait and could be missed entirely.
Understudied Kinases and the “Dark Kinome”
Roughly a third of the human kinome has been the subject of very little published research. These understudied or “dark” kinases lack selective chemical probes, validated antibodies, and sometimes even basic functional annotation. KINOMEscan has played a specific role in shining light on this territory, because the panel tests all kinases equally regardless of how well-known they are. When a clinical-stage drug turns up an unexpected hit on an obscure kinase, that observation gets recorded and becomes available for others to pursue.
The kinome-wide profiling of 192 clinical compounds found that off-targets included multiple understudied kinases.16Europe PMC. Polypharmacology is an enduring and nearly universal property of kinase inhibitors Projects specifically aimed at developing probes for dark kinases, like the neurodegeneration-focused pyrimidine series, rely on the scanMAX panel both to confirm on-target activity and to ensure that a new probe does not inadvertently hit well-studied kinases whose biology would confound experiments.17Journal of Medicinal Chemistry. Identification of Pyrimidine-Based Lead Compounds for Understudied Kinases Implicated in Driving Neurodegeneration In this way, the platform serves a dual function: it measures what you are hitting, and it warns you about what else you are hitting that you had not planned on.
Visualizing Kinome-Wide Data
A practical challenge that comes with screening a compound against 400-plus kinases is figuring out how to present the results. The standard visualization in the field maps binding data onto a phylogenetic tree of the human kinome, where kinases are arranged by sequence similarity into major groups like TK (tyrosine kinases), CMGC, AGC, CAMK, and so on. Circles of varying size and color at each kinase position indicate binding strength, giving an at-a-glance picture of where a compound acts across the kinase family.
Producing these figures manually is tedious and error-prone, which led to the development of tools like Kinome Render, a web-accessible application that automates the annotation process. It lets researchers overlay custom text or shape-based annotations at different sizes and colors onto the kinome tree, turning a raw spreadsheet of percent-of-control values into a publication-ready figure in minutes rather than hours.18PubMed Central. Kinome Render: a stand-alone and web-accessible tool to annotate the human protein kinome tree Other tools, like TREEspot provided by the assay vendor itself, and open-source alternatives like CORAL and KinMap, serve similar functions. The kinome tree diagram has become so ubiquitous that it is practically the logo of the kinase inhibitor field. Nearly every paper that reports KINOMEscan data includes one, and the ability to visually compare two compounds’ kinome trees side by side is one of the more intuitive ways to communicate selectivity differences to collaborators who are not staring at spreadsheets all day.
KINOMEscan Versus Other Profiling Platforms
KINOMEscan is not the only way to profile kinase inhibitor selectivity, and understanding where it sits relative to alternatives helps researchers choose the right tool for the question they are asking.
Enzymatic activity assays, offered by providers like Reaction Biology and Nanosyn, measure whether a compound inhibits the catalytic function of a kinase, not just binding. These assays run at a defined ATP concentration (often near the kinase’s apparent Km for ATP), which means they capture some of the ATP-competition effects that KINOMEscan misses. But they require active recombinant enzyme, which limits the panel to kinases that can be purified in catalytically competent form.
Chemical proteomics approaches like Kinobeads and the related Kinativ platform work with cell lysates rather than purified proteins. A bead coated with broad-spectrum kinase-binding ligands captures endogenous kinases from a cell extract, and competition with a test compound is detected by mass spectrometry. Because the kinases come from actual cells, they retain their native post-translational modifications and many of their binding partners. This makes the data more physiologically relevant in some respects, though the lysate environment is still not the same as an intact cell. As noted earlier, the correlation between Kinobead data and KINOMEscan data is surprisingly modest, reinforcing that different assay contexts produce genuinely different selectivity pictures.
Cellular target engagement assays, such as NanoBRET and the split-luciferase approach, measure drug binding inside living cells. They capture the full complexity of intracellular ATP, protein conformations, and compartmentalization. The trade-off is throughput: current cellular platforms can typically profile a compound against dozens of kinases, not hundreds. They are best suited for validating a shortlist of targets identified by broader screens like KINOMEscan rather than for kinome-wide discovery.
The field has increasingly moved toward using multiple platforms in combination. A compound might get its broad kinome map from KINOMEscan, follow-up affinity measurements from enzymatic assays at physiological ATP concentrations, and final target validation from a cellular engagement assay. No single platform gives the complete picture, but the combination covers most of the gaps.