A really useful summary published recently by Robert Copeland in Nature Reviews Drug Discovery looks back at the concept of drug residence time. The core principle, that the drug-protein target lifetime or residence time is the key to the observed downstream pharmacological effect has been developed, discussed (and variously challenged) over time, and this paper contains many useful references to the wide data sets published. (The contradictory views are not really commented upon – although in my opinion most of these relate to concerns of the over-simplification or generalisation of the principles outlined, rather than fundamental disbelief in the ideas).
All of us who work on drug discovery projects are very familiar with the principles of thermodynamic equilibrium assay readouts – XC50s, Kis etc., which are run under closed conditions within a simplified biochemical assay paradigm. We are also all too aware that the real-life situation is considerably more challenging and that there exist many factors which govern drug concentration at the desired site of action.
In considering the lifetime of a drug bound to its protein, we’re basically looking at one of two simplistic models for all these discussions – either a straightforward model in which a molecule binds directly to protein as b below or a molecule binds, causes a conformation change in the protein to induce a better fit – c below.
From: Nature Reviews Drug Discovery 15, 87–95 (2016) doi:10.1038/nrd.2015.18
Whichever is appropriate, the Koff is key to the residence time and generally the downstream pharmacology. The great strength of this review is the number of examples and references that are included, which make it really helpful as a resource to get (back) into this area. The overall message is that SAR should be built up using both Kd values and equilibrium affinity and, that technologies, such as SPR, are now able to provide these data in a timely and sufficient way.
Measuring the kinetics of saquinavir binding to HIV1 protease and associated resistant mutants, found that the downstream IC50 for viral replication correlated well with Koff (or residence time), although the Kon was found to only vary by 2x across the molecule set.
A series of COX inhibitors was found to have COX2 over COX1 selectivity. The COX2 binding was found to model well to an induced fit model and the COX1 binding to the one step model. A slow value for k4 in picture c above was then found to fit with a greater effect seen for binding to COX2 than COX1.
A series of FabI enoyl reductase inhibitors demonstrated a striking correlation between the numbers of animals surviving infection with the given bacterium and residence time, with no correlation at all present to Ki for the target protein.
Similarly, a series of A2a receptor agonists demonstrated a clear correlation of efficacy with residence time and not with Kd.
A very nice study published on BTK (kinase) inhibitors exploited an active site cysteine to prepare a range of molecules as reversible inhibitors with varied electrophilicity. These inhibitors achieved residence times from minutes to a week which correlated excellently with cellular residence and downstream efficacy.
An ex vivo assay of drug-target residence for HSP90 occupancy identified that longer residence times at HSP90 protein translated into greater duration of action in PD models.
Blog writted by Simon Ward