Here's an interesting paper that some of you may have seen in J. Med. Chem.: "Heteroaromatic Rings of the Future". That's an odd title, but an appropriate one.
For the non-chemists in the crowd who made it to this paragraph, heteroaromatic rings are a very wide class of organic compounds. They're flat cyclic structures with one or more nitrogen, oxygen, or sulfur atoms in the ring - I'll leave out explaining the concept of "aromaticity" for now, but suffice it to say that it makes them flat and gives them some other distinct properties. These structures are especially important in medicinal chemistry. If you stripped out all the drugs that contain something from this class, you'd lose a bit under half of the current pharmacopoeia, and that share has lately been increasing.
The authors have sat down and attempted to work out computationally all the possible heteroaromatic systems. If you include a carbonyl group as a component of the ring, you get 23,895 different scaffolds (and only 2986 if you leave the carbonyl out of it). Their methods to define and predict that adjective "possible" are extensive and worth reading if you're curious; they did put a lot of effort into that question, and their assumptions seem realistic to me. (For example, right off, they only considered mono- and bicyclic systems, 5- and 6-membered only, C, H, N, O and S).
At any rate, only 1701 of those 23,985 have ever been reported in the literature. And it looks as if reports of new ring systems reached a peak in the late 1970s, and have either dropped off or (at the very least) never exceeded those heights since then. The authors estimate that perhaps 3,000 of their list are synthetically feasible, with a few hundred of them being notably more likely than the rest. Their paper, in fact, seems to be a brief to alter that publication trend by explicitly pointing out unexplored synthetic territory. It wouldn't surprise me if they go back in a few years to see if they were able to cause an inflection point.
I hope they do. I'm a great believer in the idea that we medicinal chemists need all the help we can get, and if there are reasonable ring systems out there that we're not exploiting, then we should get to them. Adventurous chemists should have a look.
1. Curious Wavefunction on August 27, 2009 1:09 PM writes...
Yes, I saw that one. To me the interesting question from a med. chem. standpoint is if one could come up with a more or less general synthetic procedure that could produce all variants of a particular ring. For instance if we had a general protocol that could produce all variations of, say, three nitrogens around an aromatic bicyclic 6/5 scaffold, that could be very interesting.
Permalink to Comment2. cookingwithsolvents on August 27, 2009 1:42 PM writes...
Thanks for bringing this article to my attention. It's amazing how simple a lot of these are. Lots of reaction space for us to play in!
Permalink to Comment3. MedChem on August 27, 2009 2:20 PM writes...
It doesn't matter. Most likely these rings are already covered in the patent space, even though none might be exemplified. The industrial people here know what I'm talking about.
Permalink to Comment4. CMCguy on August 27, 2009 2:23 PM writes...
This sort of reminds me of combinatorial chem days where one would consider and discuss many scaffold options. Trust it will not suffer the over hype (new discovery paradigm) or under appreciation (library not "drug-like"/rule of 5 compliant) and will spawn new efforts leading to drugs.
Simultaneously I wonder if in spite of past real success are heterocycles by-in-large inherently "too flat/2D" for the "3D nature of biology systems"? I believe you have posted on this topic in the past.
Permalink to Comment5. Orgmed on August 27, 2009 3:11 PM writes...
Hello MedChem you are right, the Markush search of these rings does not miss any from SciFinder.
Permalink to Comment6. MedChemrat on August 27, 2009 4:55 PM writes...
MedChem and Orgmed
Don't bother yourself with the patent space. Use one of these rings for your target and find a drug -> hard : Get the patent -> much easier.
Permalink to Comment7. MedChem on August 27, 2009 5:19 PM writes...
MedChemrat
"find a drug -> hard : Get the patent -> much easier"
Not according to our attorneys. The reverse almost seems to be true.
Permalink to Comment8. Jose on August 27, 2009 7:15 PM writes...
NB- plenty of known heterocycles are synthetically tractable based on what's in the literature. But if you want specified regiochemistry, or substitution, you should mark off 6 months on your calendar to make a couple. There could be many interesting scaffolds out there, but go peruse the current issue of "Heterocycles" and let us know!
Permalink to Comment9. milkshake on August 27, 2009 8:01 PM writes...
#7 Definitely yes - for them. The default position is to cover their asses. (Lot less work and worry in that way). At one point our lawyers were saying we shouldn't put thiazole ring anywhere in the molecule because BMS owns them all - based on some over-broad patent claims. Next time they pull stuff like this you can putting up a slide with few pairs of closely-matched me-too "blockbuster" compounds from competing companies and ask him if he would have recommended to reject these as well.
Permalink to Comment10. Handles on August 27, 2009 9:33 PM writes...
One problem with adding carbonyl groups might be increased sensitivity to nucleophilic attack. I Am Not A Heterocyclic Chemist but my first thought when I saw the example in the middle was: what happens if I use a nucleophilic amine anywhere in the synthesis? Do I get the ring opened urea, analogous to something like carbonyl diimidazole? That pyrimidine is not very basic and might be a reasonable leaving group.
Permalink to Comment11. RB Woodweird on August 28, 2009 6:58 AM writes...
What kind of a lawyer tells you to try and avoid litigation? Isn't that like job security for them?
Permalink to Comment12. chymyst on August 28, 2009 8:23 AM writes...
I would like to see how many variations are left after you take out all the substituents i.e. carbonyl groups?!
Permalink to Comment13. MedChem on August 28, 2009 9:18 AM writes...
Milkshake
My thoughts exactly on all of your points! Funny how we chemists think alike. I tried that but it didn't work. You know somehow I have a feeling you would not work well with our attorneys :)
Permalink to Comment14. FormerMolecModeler on August 28, 2009 9:49 AM writes...
Woodweird,
In-house counsel does not want litigation, it is a big expense. Outside counsel wants litigation, it is a big source of profit.
I would ignore what lawyers say about what you should put into your compounds.
Permalink to Comment15. Sili on August 28, 2009 12:05 PM writes...
Heterocyclic chemistry was one of my worst exams ever (that I cared for, at least). The only positive spin to put on it, is that everyone had a pretty hard time on that particular one. The lecturer and censor had to go over every paper together - they hadn't agreed on any of the marks.
Of course, I've forgotten everything about the subject since then, so perhaps the mark wasn't that far off ...
Permalink to Comment16. Moody Blue on August 28, 2009 12:23 PM writes...
This paper along with another one (Hert et al in Nature Chemical Biology - also blogged here) should provide us opportunity to devise methods to some of the un(der)represented scaffolds.
Permalink to CommentAs to scaffolds being already claimed in patent space, isn't there something called 'reduction to practice'? I mean, one may claim all possible variations of a heterocyclic core, but there needs to be some details on how to actually make them (let alone actual examples), correct? Not all heterocyclic (for example 6,5-heterocycle) can be made by a general route. Kinase inhibitor area is rife with examples of interesting permutations/combinations of bicyclic heteroaromatics. Some patents tend to be very focussed - leaving holes for others to exploit. Others are more like land grab. I wonder how of these get really issued.
17. Histidine on August 31, 2009 1:52 AM writes...
It's probably just me but... I can't find the actual list of "unknown heterocycles" talked about in the paper. Just the methodology of discovering them. Is there supplementary data or something?
Permalink to Comment18. Tom Womack on September 3, 2009 5:03 AM writes...
I would say that this blog entry contains the gist of the paper, I don't think it's worth the $30 J Med Chem access fee.
The list isn't in the paper; fairly vague instructions for manipulating some Accelrys software to produce the list are there, as well as the note that the run takes about five minutes, but I get the vague feeling (from the list of current affiliations of the authors, essentially) that this was the Clever Idea of a small startup company which has now scattered itself to the four winds.
(another interesting paragraph suggests that they also considered half-aromatic half-saturated ring systems 'as these were thought more likely to produce more soluble compounds', but goes no further)
chymyst: 'Without the inclusion of the carbonyl building block, 2986 ring systems are produced; when included this number increases to 24867', and down to 23895 when tautomers are removed.
Permalink to Comment