More on the halogen bonding theme
KinasePro has a bit of discussion going on about halogen bonding in his Cl—-O thread. So I thought I could chip in my tuppence worth, being someone who up till now has had little real-life exposure to the whole concept. On series that I have worked, there have been times when adding a chlorine to an aromatic system has resulted in a large (~10-fold) improvement in activity, but this wasn’t ascribed (at the time) to any special halogen bonding effect. Maybe we put it down to the creation of a nice, snug fit with a lipophilic pocket. Maybe that extra slice of LogP went a long way….who knows.
The question I would be interested in answering is this - given a set of SAR data, and without an X-ray structure, how would you determine/suspect if you did have a halogen bonding interaction?
I was reading the latest paper from Bill Denny (DOI) on inhibitors of PI3-kinase alpha, and was checking back through a few of the references. I found this paper (DOI) from Astellas (or Yamanouchi as it was when this work was done), where they had a nice little SAR table with substituents that might help decipher if a halogen bond was involved.
The methyl substituted compound comes in at 6nM, but the chloro analogue beats it hands down at 0.8nM (approx 8-fold improvment), whilst the bromo version is 0.3nM, a whole 20 times better. How much confidence would you have in declaring this a halogen bonding interaction? I don’t know what to make of the nitrile analogue 8h….maybe that indicates some other reason is responsible for the potency increase, unless of course the nitrile is behaving as a pseudohalogen. Whilst no X-ray structures are available, the Denny paper gives some docking images of the bromo compound, looking as if the bromo is within striking distance of the carboxylate side chain of Asp964.
So, what do we make of it all?
I think driving potency with a little extra logP is easy to do — sometimes
it has little to do with the halogen. If you can correlate the Cl/logP with
IC50 and cell potency then you can begin to get a better picture of the actual
halogen effect. Most of the time the potency gains are irrelevant and the halogen
simply inhibits metabolism or forces some sort of conformation. If there is a halogen
bonding actually happening, then the data and several models should tease the data out.
Comment by watchin the wheels — October 23, 2007 @ 4:17 pm
I think there is more of problem both predicting the interaction and determining its true strength since ligand binding is a sum of several parts.
Also, this was just published:
Structural Competition between Hydrogen Bonds and Halogen Bonds
Christer B. Aakeröy,* Meg Fasulo, Nate Schultheiss, John Desper, and Curtis Moore
J. Am. Chem. Soc., ASAP Article 10.1021/ja073201c
Comment by Wavefunction — October 23, 2007 @ 4:23 pm
Differences between chloro and bromo may be more diagnostic because the effect is thought to become more important as the halogen become heavier. Another factor that may influence the strength of the interaction is the electron pull of the rest of the molecule (halogen at C5 of pyrimidine has a real battle on its hands). The case for halogen bonding does not look strong here, especially because D964 appears to be happily interacting with K833 and doesn’t ‘need’ the extra interaction.
Comment by Great Molecular Crapshoot — October 28, 2007 @ 9:26 am