There are all sorts of small-molecule drugs that bind to protein targets. Active sites of enzymes are, of course, a big subset of those, but there are plenty of enzymes whose allosteric sites are known to host synthetic ligands as well. Membrane receptor and ion channel proteins get both of those mechanisms too, and then you have the whole area of targeting protein-protein interaction surfaces, which is in the “hard but sometimes workable” category. There’s reversible protein binding, irreversible protein binding, and slowly reversible protein binding. On the computational and theoretical side, huge efforts have gone into modeling protein binding pockets and the interactions that small molecules have with them.
But there are whole other classes of biomolecules out there, and small-molecule drug applications against these drop off rather quickly. Carbohydrates, for example: they’re important far beyond the amount of space that they take up in the thoughts of many chemists and biologists, but they’re a poor fit with small-molecule drug discovery. (For one thing, evolution doesn’t seem to have ever felt much need to make small-molecule binding pockets out of carbohydrate-backbone materials – at least, I can’t think of a single example). So while carbohydrate recognition is very important in immunology and other areas, the surfaces involved tend to be even less friendly to drug researchers than the protein-protein ones.
Then you have nucleic acids. There, the situation is changing, or at least people are trying to change it. With the realization over the last twenty years or so of all sorts of small RNA species that are active in cells, the idea of targeting these things and their pathways has been getting traction. Which kinds of small RNAs are mechanistically best to target, which are most likely to yield useful chemical matter, what those compounds are going to look like and how they’ll behave as drugs – these are all wide-open questions, and there’s still time for you to do as many have been doing and put your money down. It may be that targets that are very hard in protein space become easier to address in RNA space (being a geek, I always make the analogy to Laplace transforms in math, in the way that you can turn a differential equation into an algebraic one). But we just don’t know yet.
And there are drugs that target DNA, too. The only problem is that they tend to target it in the explosive-munitions sense. Cisplatin is an example, a drug that works on DNA if ever there was such. Once inside the cell, it converts to a more reactive species that reacts with nucleophiles, and is particularly suited for making same-strand guanine-guanine connections in DNA. This is not the kind of damage that a cell’s repair system is equipped to reverse, and several pathways lead from there to cell death (apoptosis). It should be noted, though, that the compound isn’t necessarily selective for DNA – it will pick up SH groups from proteins and cause havoc in other directions, too, particularly in mitochondria (which are operating within fairly tight tolerances to begin with).
There are plenty of other compounds that are known to interact with DNA in less catastrophic ways, of course. Famously, there are intercalators – flat molecules that slip in between the stacked base pairs like sliding another card into a deck. Intercalators come in a variety of types, and some of them have bigger effects than others. None of them, to my knowledge, are as bad as something like cisplatin, although you don’t want to go around exposing yourself to them to see if you can find one, either. Note, though, that the hazards of one famous intercalating compound, ethidium bromide, are very much exaggerated, since (among other things) it doesn’t even seem to get into healthy cells in the first place. That one’s not an unknown: many attempts have been made to find toxicity, and it doesn’t seem to be there, fortunately.
Then you have the minor groove binders. DNA often gets represented by logo designers as two even snaky lines curving around each other, but that’s not how it works. It’s more like a curled ribbon hanging down off the bow of a wrapped present, and that gives you two rather different ways to trace around it. The minor groove is apparently more compatible (in size and functionality) for small molecules. Many DNA-binding proteins interact with the major groove, but the known minor-groove small molecule binders outnumber the major-groove ones substantially. You can have slightly larger molecules that do both minor groove binding and intercalation at the same time, like doxorubicin. These things can have all sorts of effects on gene transcription, and many attempts have been made to harness these for therapeutic use.
It ain’t easy. You can get shotgun blasts, as with the chemotherapy agents, but the variations on those are akin to varying the choke on said shotguns. The spread of the damage may change, but it’s still a pretty nasty weapon. It would be very useful indeed to float something more specific and less horrific in there and disrupt the pathways of individual transcription factors according to their DNA binding sites and cofactors, but whether small molecules are up to that job is another one of those open questions. Many are the groups that have tried, so the one thing we’re pretty sure of is that compounds of this sort are, at the very least, thin on the ground.
Why should that be? It’s not for lack of DNA binding: if you take the time to screen any given compound collection, you’ll find a pretty good pile of DNA binders of various kinds in there. (It would be interesting to know what percentage of marketed drugs show up in such assays, although I’ve never seen a figure). Update: here is a paper just today on such profiling against RNA binding! One big problem is selectivity. Transcription factor proteins tend to recognize stretches of DNA that are far larger than druglike molecules can bind to with any specificity. But there’s another difficulty: let’s say that you have such a compound, once that’s reasonably selective for some classic transcription factor recognition sequence, and let’s furthermore stipulate that you can get it into cells without the aid of power tools. What happens when you dose such a compound?
Well, there’s an awful lot of DNA out there. Or in there. Estimates vary, but each human cell has maybe two meters of DNA in it, were it somehow to be stretched out. Multiplying that by the number of cells in a body (and subtracting the erythrocytes!) is an imprecise process, but the result will likely stretch your genome past the Kuiper belt. That is a fearsome number of binding sites, and the result is that many DNA-binding compounds just sort of diffuse out into the body and are never heard from again. It would appear that most DNA-binding compounds, whatever their cellular effects in vitro, don’t have enough of an effect on transcription to be noticed clinically. And the ones that still show effects tend to be, well, things like cisplatin and doxorubicin that have a strong follow-through when and where they do bind.
Can you get a strong beneficial follow-through by binding to DNA? Who knows? We honestly don’t understand transcription that well – it’s wildly complicated, as it would have to be given the three-dimensional nature of the problem, the number of human genes, and the widely varying requirements for them to be transcribed. I’d say that most groups targeting it are looking somewhere other than DNA-binding agents, but it’s a field where people are going to be willing to run with whatever they can get.