A couple of articles appearing in the last few months brought my attention to a topic that medicinal chemists don't always think about and need to pay more attention to; the important role of kinetics in drug discovery, especially in its early stages.
Anyone who is involved in drug discovery knows the importance of the dissociation constant that signifies the affinity of a therapeutic ligand for a protein. SAR around changing affinities (usually represented by Kd or IC50 values) drive lead design and optimization. But as some of the recent reviews note, the problem with this number is that it's a ratio of the on and off rates of binding of the ligand to the protein. A fast on rate and a fast off rate will give you the same number as a slow on and a slow off rate. But the two situations are not identical.
The most important point emphasized by these reviews is that slow off rates can sometimes lead to prolonged drug efficacy in ways that are not apparent from just the affinity. And this is quite logical if you consider that a slow off rate means that a ligand has a longer residence time in the protein's active site and is spending more time modulating its action. What this means in practice is that even compounds with relatively low affinities can have quite significant efficacies resulting from slow off rates.
So how do you modulate off rates? One good thing about off rates is that unlike on rates, they don't depend on concentration. The benefit of this is that you could have a compound which has a low concentration at the target site and is rapidly cleared away, but which nonetheless spends a lot of time in the protein and therefore provides good efficacy. The other good thing about off rates is that they are essentially dependent on the interactions between ligand and target. So you should in principle be able to improve them just by optimizing these interactions. Again, this won't result in better affinity if you are also slowing down on rates, but it might give you improved efficacy.
In part these studies remind us that we need to clearly distinguish between terms like affinity, IC50, efficacy and the other vocabulary of lingua pharmaceutica. But they also ask an important question; why aren't pharmaceutical scientists paying more attention to kinetic measurements in the early stages of drug discovery? That this is indeed the case became apparent when I looked at the website BindingDB which lists key biological, thermodynamic and kinetic parameters for ligands bound to popular targets. One prominent pharmaceutical target that I looked at had 277 different ligand structures bound to it along with many cases where affinities, IC50s and even free energies had been measured. But out of those 277 I could find only 5 cases (less than 2%) where on and off rates had been recorded. Clearly this is not a focus in preclinical drug discovery.
But as the recent article note, it should be. There are several cases of drugs - HIV protease inhibitors for instance - where differing efficacies for compounds with similar affinities essentially result from differing off rates and residence times. In fact as illustrated by the blood pressure lowering drug amlodipine, off rates can mean the difference between a best-in-class drug and the second-best contender; amlodipine is a better drug than others partly because of its longer residence time in the pocket of the calcium channel protein which it inhibits.
The neglect of kinetic rate measurements reminds me of an almost equal neglect of thermodynamic measurements by ITC. As described in other articles, careful measurement of enthalpy and entropy (and not just free energy) can be very useful in early stage drug discovery. This shouldn't be surprising at all; after all kinetics and thermodynamics are the twin pillars of protein-ligand binding, and you neglect them at your own peril.
Anyone who is involved in drug discovery knows the importance of the dissociation constant that signifies the affinity of a therapeutic ligand for a protein. SAR around changing affinities (usually represented by Kd or IC50 values) drive lead design and optimization. But as some of the recent reviews note, the problem with this number is that it's a ratio of the on and off rates of binding of the ligand to the protein. A fast on rate and a fast off rate will give you the same number as a slow on and a slow off rate. But the two situations are not identical.
The most important point emphasized by these reviews is that slow off rates can sometimes lead to prolonged drug efficacy in ways that are not apparent from just the affinity. And this is quite logical if you consider that a slow off rate means that a ligand has a longer residence time in the protein's active site and is spending more time modulating its action. What this means in practice is that even compounds with relatively low affinities can have quite significant efficacies resulting from slow off rates.
So how do you modulate off rates? One good thing about off rates is that unlike on rates, they don't depend on concentration. The benefit of this is that you could have a compound which has a low concentration at the target site and is rapidly cleared away, but which nonetheless spends a lot of time in the protein and therefore provides good efficacy. The other good thing about off rates is that they are essentially dependent on the interactions between ligand and target. So you should in principle be able to improve them just by optimizing these interactions. Again, this won't result in better affinity if you are also slowing down on rates, but it might give you improved efficacy.
In part these studies remind us that we need to clearly distinguish between terms like affinity, IC50, efficacy and the other vocabulary of lingua pharmaceutica. But they also ask an important question; why aren't pharmaceutical scientists paying more attention to kinetic measurements in the early stages of drug discovery? That this is indeed the case became apparent when I looked at the website BindingDB which lists key biological, thermodynamic and kinetic parameters for ligands bound to popular targets. One prominent pharmaceutical target that I looked at had 277 different ligand structures bound to it along with many cases where affinities, IC50s and even free energies had been measured. But out of those 277 I could find only 5 cases (less than 2%) where on and off rates had been recorded. Clearly this is not a focus in preclinical drug discovery.
But as the recent article note, it should be. There are several cases of drugs - HIV protease inhibitors for instance - where differing efficacies for compounds with similar affinities essentially result from differing off rates and residence times. In fact as illustrated by the blood pressure lowering drug amlodipine, off rates can mean the difference between a best-in-class drug and the second-best contender; amlodipine is a better drug than others partly because of its longer residence time in the pocket of the calcium channel protein which it inhibits.
The neglect of kinetic rate measurements reminds me of an almost equal neglect of thermodynamic measurements by ITC. As described in other articles, careful measurement of enthalpy and entropy (and not just free energy) can be very useful in early stage drug discovery. This shouldn't be surprising at all; after all kinetics and thermodynamics are the twin pillars of protein-ligand binding, and you neglect them at your own peril.
Thanks for the adds to the reading list - while I have the general biochemistry background, I figure it's helpful that I start picking up some things of more direct relevance to pharma/drug discovery.
ReplyDeleteI suppose one issue is variation between on/off rates measured in vitro and in vivo. I could imagine that - for example - this problem could easily arise if you've got a target protein that is membrane associated but the test tube assay involves a soluble construct of said protein. Perhaps that ligand is hydrophobic enough that it might partition into the membrane to some extent (or something).
Regarding ITC - how high-throughput is that technology? My impression is that while strides have been made, it's still not quite at the level of the label-free biosensors out there (I tried my hand at SPR a while back, and even with an older system, I was still able to knock out a number of samples/replicates without too much difficulty).
Yes, that's an important point, knowing the differences between in vitro closed systems and in vivo open systems; the example of a membrane protein sounds especially relevant. You would expect this difference to affect the k(on) rate more though since k(off) depends mainly on the intermolecular interactions. You are probably right that ITC is still not high throughput enough, one reason why drug companies have been reluctant to use it on a regular basis. SPR does seem amenable to higher throughput, especially instruments with 96 wells.
DeleteOn a more general note though, I really liked the second article since it makes a very cogent comparison of the lack of biochemical sophistication in pharma compared to academia where biochemistry is wielded both as a fine scalpel and as an earth mover to uncover all kinds of fascinating details about proteins.
Binding kinetics need to be seen in the pharmacokinetic context if one is arguing for their relevance to Drug Discovery. If binding is faster than distribution then you probably don't need to worry about it too much. It can be difficult to establish the importance of off-rates especially when the kinetics are not separated from the thermodynamics. Put another way, the standard free energy of binding usually contributes to the activation energy for dissociation. The other thing to remember is that for fixed affinity, a slow off-rate means a slow on-rate and equilibrium will not have been attained before the start of the elimination phase.
ReplyDeleteI remain unconvinced that measuring enthalpy and entropy changes associated with binding are particularly relevant to Drug Discovery. These parameters are certainly of scientific interest (as are the volume changes associated with binding) and may ultimately help us make better predictions of affinity. My routine challenge to those who assert the benefits of 'enthalpically-driven' binding is to ask how isothermal systems (i.e. humans taking the drugs) 'sense' these benefits.
I think the papers by Freire illuminated something that's kind of obvious but also hard to control, namely the fact that entropy is much easier to optimize than enthalpy. I think his analysis is most valuable in the rare exceptions where the converse might be true, in which case it could be very valuable to do ITC early on since you could then optimize entropy with scant effort (perhaps by tacking on a hydrophobic aromatic group). Otherwise you might be misled in thinking that you have to optimize enthalpy and actually move away from optimal interactions with the target.
DeleteThe issue for me is whether or not the enthalpy and entropy changes associated with binding have any relevance to drug action. Drugs need to bind in order to act so the changes in Gibbs free energy associated with binding will be relevant to drug action. However, it is much harder to make an analogous case for the relevance of changes in the entropy and enthalpy associated with binding. If one is going to argue that the thermodynamic signature of binding is relevant one has to show how systems are capable of discriminating between compounds with different thermodynamic signatures. When invoking thermodynamics it is important to describe phenomena using the appropriate thermodynamic quantities. Enthalpy changes will certainly be relevant when manufacturing drugs and, since synthetic reactions used in process work are often irreversible, free energy changes will not usually be known.
DeleteThe focus on 'enthalpic binders' reminds me of the excuses made for train delays by the old British Rail. "Snow on the line", said the BR spokesperson. Further interrogation prompted the iconic response: "Wrong type of snow".
So if I understand you correctly, you are asking the following interesting question; how would you distinguish two ligands with the same binding energy but different enthalpic vs entropic contributions? I think you are right that their on-target effect would probably be the same. The entropically driven ligand might be lipophilic however and this would likely contribute to a bad off-target profile.
ReplyDeleteNice quip about the snow. I tried something similar last winter, but my wife didn't buy it.
Then the question would be: Is the off-target affinity of a compound determined more by the enthalpy & entropy changes associated with binding to the target than by its physicochemical properties? There is also the question of the strength of the correlation between promiscuity and lipophilicity...
ReplyDeleteBack to "wrong type of snow" and I'll give you the link since my version was abbreviated in the interest of space. I'll never cease to be amazed by the excuses that rail travel in the UK leads to such as "leaves on the track". My personal favourite was the train driver who had to return to Birmingham because he got "lost".
http://en.wikipedia.org/wiki/The_wrong_type_of_snow