Here’s a thorough review of a topic that combines complex drug development issues with complex chemistry: trying to optimize the structures of natural products so that they can be more effective drugs. There are detailed looks at examples like the tetracyclines, the polyene macrolides, pactamycin, geldanamycin, and the thiostrepton-like peptides, along with shorter coverage of several other classes.
A look at some of those structures will show that there’s enough synthetic challenge to keep anyone occupied (and how). Some of these structures are extremely complex – if you can do useful work on scaffolds like vancomycin or thiostrepton, you’re getting up to the limit of what synthetic organic chemistry (and synthetic organic chemists!) can currently accomplish. This is the high-altitude region where a lot of the fancy tools that are coming on drop away. No retrosynthesis software will do much more than choke on such molecules. I’m sure that they could be useful on smaller intermediate structures, but how you figure out which intermediates to make (and how you’ll stitch them together) is going to be your very nontrivial problem. Similarly, you’re not going to have enough material available in some of these cases to try lots of high-throughput reaction discovery or optimization (although if you have it, so much the better).
So compounds like these can take you right back to the heroic days of organic synthesis, although to be sure we have a lot more reactions, disconnections, and transformations available to us than R. B. Woodward did. And that’s a good thing, because the goal in working on such compounds is not just to be the first to climb up the mountain and plant a flag with your name on it – it’s to make a whole list of analogs, to develop the structure-activity relationship, to address pharmacokinetics and off-target tox. In short, to not only deal real hardcore synthetic chemistry, but then to turn around and do real hardcore med-chem as well.
These two fields don’t often intersect. Medicinal chemistry is famous for using the simplest, most reliable methods (many of them, frankly, quite boring for refined total synthesis tastes) to crank out analogs as expeditiously as possible. That’s because chemistry is merely a tool in drug discovery and development (one of a whole list of tools), and if you can keep it from becoming a problem all its own, that’s a victory. But if you’re working on one of these big natural products, chemistry is going to be a big issue immediately, and synthesis becomes what it rarely becomes in most projects: a rate-limiting step. The biological activity to be found in these molecules is the only reason to let that happen to you.
Rarely do you see these things attacked via pure total synthesis, though. Far more common is the semisynthetic route, where you take some advanced intermediate that’s available in reasonable quantity and start making derivatives off that. Let the bugs/plants/sponges, what-have-you, do as much of the work as possible is the principle here. You can even start with the final compound itself and try late-stage functionalization chemistry, which ideally can get around the need for protecting-group schemes and give you new analogs directly. It’s tricky stuff on such complex structures, though, and generally requires a lot of patient dinking around with conditions to get selective reactions. (On the other hand, if the reactions aren’t so selective, there’s something to be said for that if you can separate out your various products and test them individually – although you may regret finding that some weirdo minor isomer turns out to be of great interest!)
The limitations of semisynthetic chemistry, though, are clear: you’ll only be able to work on certain parts of the molecule, and many interesting transformations are just going to be closed off to you. There are no useful reactions on such compounds to (for example) turn an aryl ring into a heteroaryl one via late-stage functionalization or to just drop a nitrogen or oxygen somewhere into the framework. Something like that is going to require going further back in the synthesis, and will also probably cause the rest of the known synthetic sequence (if there is one) to go its own way as well.
Another option requires digging into the biology and modifying the biosynthetic pathways. Tweaking key enzymes and/or providing new substrates to the organisms can give you some really interesting compounds, but that stuff can be even harder than getting the synthetic chemistry to work. And the same limitations apply: some parts of the structure are going to be much more easily varied than others, because the enzyme systems will only put up with so much.
So there’s still an impasse in many cases. It would be very nice indeed to generate a set of a few thousand vancomycin analogs and start screening them, but no such collection exists (and for very good reasons). We’re closer than ever to opening such ideas up, but not quite close enough. I would submit this whole area as a completely appropriate challenge for high-powered organic synthesis and methods development, and judging from this review and other papers in the recent literature, I think many chemists agree. . .