22 January 2024

Covalent complexities for kinase inhibitors

Covalent drugs are becoming increasingly popular. But as more researchers search for them, they may encounter pitfalls. A new paper in J. Med. Chem. by  David Heppner and collaborators at the State University of New York Buffalo, AssayQuant Technologies, and Eberhard Karls Universität Tübingen provides a nice roadmap for avoiding them.
The researchers focus on covalent inhibitors of epidermal growth factor receptor (EGFR), a kinase that is frequently mutated in cancer. The first drugs against this target, such as erlotinib, were non-covalent, and these have been largely displaced by more effective covalent molecules such as afatinib. Unfortunately, these earlier drugs are not effective against a common mutant (T790M), spurring the development of third generation molecules such as osimertinib, which was approved by the FDA in 2015. Osimertinib has been extensively studied, with more than 2800 references in PubMed. Yet it is not as well understood as you might expect.
The team uses this system to demonstrate how characterizing irreversible inhibitors is not simple. For reversible enzyme inhibitors, researchers frequently discuss IC50 values or, when they are being more precise, inhibition constants (Ki). The latter are in theory absolute values that do not depend on concentrations of cofactors such as ATP. But for irreversible inhibitors, the IC50 values change depending on how long (and at what concentration) incubation occurs. The proper assessment of an irreversible inhibitor is kinact/KI, which takes into account both the irreversible inactivation step (kinact) as well as the inhibition constant (KI). Note that Ki is not the same as KI ; the former describes only the initial reversible association between protein and inhibitor, while KI incorporates the irreversible step. Told you it was complicated!
And it gets worse. The researchers examined three irreversible covalent inhibitors under various conditions. In one condition, the inhibitors were pre-dissolved in 10% DMSO before being added to the assay mixture to give a final DMSO concentration of 1%. In another condition, the inhibitors were dissolved in pure DMSO before being added to the assay. Despite the final concentration of DMSO being the same (1%), the second condition gave kinact/KI values up to 11-times greater (more potent).
If subtle experimental variations in one lab can change values by more than an order of magnitude, you might expect the literature to vary even more, and you’d be right. In the case of osimertinib, the reported values of kinact/KI vary by nearly 500-fold. Some of the experimental parameters the researchers consider are concentrations of reducing agents such as DTT, which can react with covalent inhibitors, and serum albumin, which also contains a free cysteine residue. Although these did not seem to be problematic for osimertinib itself, they could affect other molecules.
Another consideration for kinases in particular is the concentration of the cofactor ATP. The value of kinact/KI itself will vary depending on [ATP], and the researchers describe how to calculate a “true” kinact/KI which could be used to compare the potency of a given inhibitor against the wild-type vs mutant forms of the enzyme. But while this is more theoretically rigorous, it may be less biologically relevant, since physiological ATP concentrations are less variable than differences in the Michaelis constant (KM) for ATP for different kinases and mutants.
There is lots more to digest in this paper, including analyses of structure-kinetic relationships (SKR, akin to structure-activity relationships, or SAR) for different inhibitors and thorough experimental descriptions. The take-home message is that, due in part to different and often incomplete details, “potency measurements are generally difficult to compare among literature studies,” and “any potency assessments should include appropriate controls under the same conditions as the experimental inhibitors.”

1 comment:

Anonymous said...

I hope the point about dissolving compounds in 10% DMSO giving different results is not a surprise to too many people. I think there has been data for many decades that compound solubility does not scale linearly with DMSO concentration and so if you take a 10 mM DMSO solution of a compound and then dilute it 10-fold with water, there is a good chance the compound will precipitate out of solution whereas if you dilute it 1000-fold or 10,000-fold with water (or buffer), it may not. So when doing a dose-response of a small molecule dissolved in DMSO, I think a good practice is to make different dilutions in DMSO and then dilute these all into water/buffer as a final step and not try to make intermediate solutions in mixtures of DMSO/water.

But lots of other good points are raised.