Take a look at the molecule shown below. It looks innocent and fairly drug-like, wouldn’t you say? It is in fact perfectly soluble in buffers and does not contain any electrophilic “red flags”. But if you are equipped with a DLS (differential light scattering) instrument, you will discover the ugly truth that lurks behind the veneer of innocence here. It turns out that at 4 micromolar concentration, this molecule is a nasty aggregator, one that forms gigantic (well, relatively speaking) colloidal particles of 2312 nm in radius.
Many small organic molecules form aggregates in aqueous solution, which has long had a rather perilous effect on drug discovery. Proteins you are trying to assay tend to stick to the surfaces of these particles rather well, which leads to nonspecific inhibition or, curiously, enzyme activation. This results in major aggravation and supplies a never-ending list of false positives in drug discovery. The behaviour of colloidal particles is so “outside the box” that they are often referred to as the fourth state of matter (besides liquid, gas, and solid).
There are additives such as Triton-X100 one can use in order to disrupt aggregation, but Brian Shoichet and colleagues at UCSF recently developed something better – a computational method that uses lipophilicity and similarity to known aggregators to inform on the likelihood that a candidate compound is an aggregator. In my view, this is a very timely study (there are many previous papers on this subject from the Shoichet lab). If not a complete solution, this work reminds us that there are factors other than inherent structural properties of those nicely docked molecules on our computer screens that might be responsible for the effects we observe.