When Affinity Isn’t Enough: How Binding Kinetics Shapes KIT Drug Performance

KINETICfinder^® turns binding kinetics into decisions

Two compounds can display similar affinity (K_d, IC₅₀) values and yet behave very differently in vivo. The difference often lies in binding kinetics. By resolving how fast a compound engages its target (k_on) and how long it stays bound (residence time, τ = 1/k_off), KINETICfinder® reveals how compounds engage their targets over time, providing mechanistic insights that equilibrium measurements alone cannot capture.

For drug discovery teams, this means the ability to:

• Prioritize clinically relevant chemotypes earlier.
• Distinguish compounds with similar affinity profiles.
• Detect resistance mechanisms before they emerge in the clinic.
• Identify kinetic selectivity.
• Detect mechanisms invisible at equilibrium.

One target, several druggable conformations

KIT is a Type III receptor tyrosine kinase that cycles through distinct states: an autoinhibited form in which the juxtamembrane domain blocks the active site, a non-autoinhibited form competent for both active and inactive conformations, and the constitutively active D816V mutant that drives over 90% of systemic mastocytosis and contributes to imatinib-resistant disease.¹ Each state presents a different binding landscape. As a result, the same inhibitor can display dramatically different kinetic behavior depending on which conformation it encounters.  

To understand how these conformational changes influence drug performance, a binding-kinetics dataset covering 172 kinase inhibitors was generated across the KIT activation landscape using KINETICfinder^®.

Different KIT States Demand Different Optimization Strategies

The path to higher affinity is not universal. It depends on both the KIT conformation being targeted and the stage of drug discovery.

For active KIT conformations, including KIT D816V, affinity gains in the hit-to-lead phase are primarily achieved through faster target engagement. As compounds approach the limits of association rate, further improvements come almost exclusively from slower dissociation and therefore longer residence times (Figure 1). In contrast, inactive DFG-out conformations follow a different kinetic logic. Across the entire affinity range, potency is driven predominantly by faster association rates, with little measurable contribution from slower dissociation.

Structural levers for long residence time

Long residence time is not accidental. It emerges from specific conformational and structural interactions that stabilize the inhibitor-target complex. Inhibitors that stabilize DFG-out conformations consistently display higher affinities and residence times extending into the hours range. By contrast, inhibitors targeting DFG-in, αC-helix in conformations associate rapidly but typically exhibit very short residence time.

Binding-site location further shapes kinetics. Front-cleft binders generally associate and dissociate rapidly, whereas back-cleft binders tend to bind more slowly and remain bound for substantially longer periods. A major contributor to this prolonged target engagement is interaction with the R-spine. Compounds that do not engage the R-spine typically exhibit short residence times, whereas inhibitors forming multiple R-spine contacts remain bound up to 400-fold longer (Figure 2).

An alternative route to prolonged target engagement is provided by inhibitors, such as avapritinib and midostaurin, that extend into hydrophobic pockets around the G-loop and αC-helix. This binding mode combines rapid association with unexpectedly long residence times against KIT D816V.

Faster k_off drives KIT D816V resistance

This finding highlights an important limitation of equilibrium measurements. A compound can retain apparently attractive affinity values while simultaneously losing the target engagement required for sustained biological activity.

Against the D816V mutant, 43% of inhibitors exhibited a significant drop in affinity, and in 96% of those cases the loss was driven by a faster off-rate reaching more than 700-fold. The destabilised activation loop and αC-helix of D816V disrupt the F811/E640 contacts that type II, DFG-out back-cleft binders rely on, so these drugs simply fall off faster. Because resistance shows up in k_off, affinity can be misleading.

Two drugs make the point:

References

1. Corrionero A, Prendiville N, Cazorla T, Baena-Nuevo M, Sandra R, Camafeita E, Knapp S, Alfonso P. Kinetic Fingerprints as Mechanistic and Clinical Roadmaps Across KIT Activation States. ChemMedChem. 2026 Jun 15;21(11):e70331.