Derek Lowe ponders the likelihood of arriving where he started

Here’s an experiment that we’ll never get a chance to run, but the results would be very interesting indeed. What if we were able to give several different discovery organisations the same starting molecule against the same drug target, and tell them to have a go at it? After two or three years had passed, how many different clinical candidates would the various groups be advancing? 

Well, ideally we’d want to run that test more than once (which is too bad, since if doing this once is impossible, doing it over and over is even more so). Different protein targets and different sorts of lead compounds can mean widely varying amounts of structural diversity. On one end of the scale, some proteases are very picky indeed about what sizes and shape of inhibitor they’ll accept. Moving up the scale a bit, many proteins prefer at least some sort of conserved anchor for binding. For example, carbonic anhydrase inhibitors are almost always sulfonamides of some sort, but you can get away with all kinds of variations on the back side of the molecule. Then there are the ligands for the PPAR nuclear receptor subtypes. They generally need some sort of acidic group, but there are a lot of different ones that can serve. And the number of different structures that have been appended to those is completely bewildering (although nowhere near as bewildering as trying to make sense of their biological activities).

Convergence or divergence? 

So, depending on which sort of project we handed out, we could return to find everyone converging on much the same sort of molecule, or we could have a whole set of distantly related ones. It would be up to the biases and experiences of the medicinal chemists involved, and the order in which things were discovered. That last factor is a very large and underappreciated one in drug discovery. One active structure tends to suggest others, and the path for an entire program can be set depending on what gets worked on first. The corollary to that is that some active drug candidates will surely be overlooked in any discovery program. It’s impossible to make everything on the list, and more effort will always go to the compounds that look to be closest to crossing the finish line. 

All this path dependence can lead to situations that are, depending on your perspective and your innate good humour, either amusing, irritating, or downright embarrassing. Some clinical candidate structures turn out to be rather close relatives of the compounds that started off the programme. I’ve worked on two drug discovery efforts (one right after the other, as fate would have it) whose final compounds differed by essentially one methyl group from the starting points of each project. Hundreds of compounds separated the two, but the end of our exploring, as T S Eliot could have told us, was ’to arrive where we started and know the place for the first time’. There were good reasons for this to happen, but explaining them in a confident tone of voice was a challenge. 

The closest thing to the experiment I’ve proposed can be found in the therapeutic areas where several structurally related drugs have entered the market. Depending on the timing, these similarities could be real cases of convergent evolution, near-simultaneous work driven by the constraints of the protein target, or they could be a case of Company B deliberately starting an effort to break Company A’s patents. Like many medicinal chemists, I’ve worked on programs in both categories. That first situation can come up when a new biological target is identified and high-throughput screening turns up structural motifs that several companies have in their files. By contrast, the second situation is a deliberate choice, and a potentially risky one: you never know when a later patent application might emerge that invalidates your own ’fast follower’ programme in one blow. 

Physical chemistry will track you down 

There could be something new to learn from all this. In those cases where different molecules have come along in sequence over a period of years, an interesting effect has been pointed out in the recent literature. Unraveling the various thermodynamic contributions to their binding energies by microcalorimetry shows a tendency for enthalpic factors to increase as the drugs in a given area evolve. That often means that more of the binding is coming from positive interactions between the ligand and its target, and less from hydrophobic-effect displacement of water molecules. I should point out that while those are the most common reasons for such results, there can be others - changes in entropy and enthalpy often compensate for each other in surprising and still-mysterious ways. But we’re probably going to have to come to grips with these, somehow, which is ironic: the medicinal chemistry ranks are full of people who preferred not to go into physical chemistry. But physical chemistry never lost our scent, apparently, and is finally tracking us down to our lairs. It probably serves us right! 

Derek Lowe is a medicinal chemist working on preclinical drug discovery in the US.