I will cause no controversy by saying that most of the small-molecule compounds that we develop as potential drugs in this business are rather poorly soluble in water. Every organization I’ve worked in has made the standard jokes about “brick dust” and “powdered Teflon”, and for the well-founded standard reasons. A lot of binding sites in proteins tend to be more hydrophobic than the surrounding solution – that’s one of the ways that you get the thermodynamics of compound binding to work, of course, because if everything were joyously happy out there in solution by comparison, why would they bind at all?
There are, of course, really polar binding sites, and those bring in specifically arranged salt bridges (lots of lysines and arginines, if you’re binding something like a phosphoylated species, or lots of glutamate and aspartate if the protein is going the opposite way), but even those have their hydrophobic spots in them. Those polar interactions tend to be a lot snootier in their distance and angle preferences than grease-recognizes-grease, though, so targeting them is generally a lot trickier. It’s easy to miss the preferred orientation, and if you try to stick a really polar/charged part of your molecule into a binding-site region that isn’t recognizing it, the result is generally far worse than neutral. The polar upside is harder to attain while the downside is easier, so medicinal chemists thus tend to rely on hydrophobicity as the default position, with polar functional groups judiciously sprinkled in. That also matches up well with our chemical abilities, as fate would have it. We tend to work best in organic solvents, because there’s a wider variety of bond-forming reactions available under nonaqueous conditions, and purification of the products is more straightforward.
All of that is by way of apology to our colleagues in the formulations labs, who have to deal with the results. It’s often not pretty. Many well-known drugs have pretty hideous aqueous solubilities, and the recent trend towards even larger molecules than usual (protein-protein interaction inhibitors, bifunctional degraders, and so on) is not helping much, either. The list of tricks to mitigate the problem is a long one, and unfortunately it’s pretty darn empirical as well, since the kinetics and thermodynamics of getting into solution (and staying there) can be very complex. There are, for example, compounds that seem pretty freely soluble at first, but will find ways to crash back out if you leave them around for a while (often by forming less-soluble aggregates), and there are others that have pretty decent solubility but take their sweet time to realize it, and everything in between.
This recent paper will give you some idea of the fun that is in store. It’s looking at one of the ways that you can get poorly soluble molecules in at higher concentration, through the addition of “hydrotopes”. Those are added molecular species that somehow stabilize other ones in water. It’s known that these generally work by piling up around the less-soluble molecules and acting as a sort of bridge between them and the bulk watery phase, but you’ll note the hand-waviness of that explanation. Knowing what’s going on in more detail would be of great use in picking likely candidates and designing new ones.
The paper linked to (from Hebrew Univ. in Jerusalem) is studying good old caffeine, which despite its ubiquity in coffee, tea, and soft drinks is actually not all that water-soluble. It has a tendency to form oligomers and aggregates in water, like many drug substances will, and that can give you a range of behavior: outright precipitation, sure, but also still-in-solution-but-less-efficacious, which is quite annoying. There are a range of hydrotopes known to affect caffeine’s solubility, things like urea and thiocyanate which act in many such situations, but they affect the single molecules and the oligomers in the same way.
This work shows an interesting effect of sugar molecules, though: as others had noted, the addition of such species decreases the general solubility of caffeine, but on closer inspection, the team found that the distribution changed: caffeine monomers were much more stabilized in solution than the oligomers. Those oligomers seem to mimic the common crystal form, columnar pi-stacks of caffeine molecules, and it appears that the sugar molecules are interacting more with the two ends of these. That means that they have a much greater effect on such species as they get shorter, and on the individual caffeine molecules most of all. Interestingly, sugars like glucose, sucrose, and fructose fall along a linear relationship for this effect (sucrose, for example, acts pretty much like glucose and fructose together, as it should), while trehalose is a definite outlier. That one’s already been noted as an unusual stabilizer of macromolecules in solution (for reasons that are still being argued about), but in this case it seems that the trehalose is (compared to the others) much more excluded from the “sides” of the oligomer stacks.
This sort of “selective hydrotopy” is a new thing, so the question is how common these effects are and whether they can be anticipated. But now that we know that it’s possible, the search can commence for compounds that both associate with surfaces that are more exposed in solute monomers (such as the “tips” of the caffeine oligomers here), and are also excluded from the surfaces that are formed by the oligomers. It’ll be quite interesting to see how general this turns out to be: is it going to be mostly trehalose (which people already knew to try) or are there more things waiting out there to be found?