I've been meaning to mention this paper from John Hartwig (and co-worker Daniel Robbins), because it's just the sort of let's-find-something-new idea that I like. Hartwig has made a name in the field of organometallic catalysis, and is looking for new reactions. So how do you find new reactions?
Most published methods for the high-throughput discovery of catalysts evaluate one of the two catalyst-reactant dimensions. In other words, these methods have been used to examine either many catalysts for a single class of reaction or a single catalyst for many reactions. A two-dimensional approach in which many catalysts for many possible catalytic reactions are tested simultaneously would create a more efficient discovery platform if the reactants and products from such a system could be identified.
Well, this paper details a brutally straightforward technique for doing that. They take a list of seventeen reactants, all around the same rough molecular weight range, each of them with a single functional group. They put a mixture of all seventeen into every well of a 96-well plate. Then they take twelve ligands, dispensed one per column of the plate, and then they take eight different metal catalyst precursors and dispense those across the eight rows. And then they take the plate and heat it up.
Can't get much more straightforward than that, can you? But analyzing the wells by mass spec tells you some interesting things, and you can cover a lot of ground. Seventeen substrates, fifteen metal starting points, and 23 ligand (or lack of ligand) combinations gives you a look into tens of thousands of possible reactions. They simplified the mass spec analysis by combining samples for each row, then combining another site for each column, so you only have to run 20 samples per plate to give you the X-Y coordinated of a well that did something. A test plate containing some combinations of known catalytic reactions showed the expected products in the right wells - and it showed some other reactions, too.
Among those were several wells that indicated an alkyne/aniline addition reaction catalyzed by copper. This turned out to be a hydroamination reaction that no one had observed before. There was also a new product in several Ni-catalyzed wells - a set of deconvolution experiments narrowed that one down, and it turned out to be reaction of arylboronic acids with diphenylacetylene to give a triarylalkene - a reaction not previously catalyzed by such a cheap metal as nickel. And while most of the known reactions are syn, this one gives anti addition, with E/Z ratios that vary depending on the ligand used for the metal.
Not bad - two new reactions in what was, in the end, a pretty simple experiment. And any good chemist should be able to see the ways this protocol could be extended. For example:
This approach to reaction discovery holds considerable potential for purposes beyond those revealed in the current work. For example, this system could be used to explore reactions with additives, such as oxidants, reductants, acids, and bases, and to explore reactions of two substrates with a third component, such as carbon monoxide or carbon dioxide. It could also be used to examine the reactivity of a single class of ligand with various organic substrates and transition metal–catalyst precursors. Thus, we anticipate that this approach to reaction discovery will provide a general and adaptable platform suitable for use by a wide range of laboratories for the discovery of a variety of catalytic reactions.
There's going to be some criticism, though, that this is (a) obvious and (b) not elegant. I regard those as features, not bugs. Never be afraid of the obvious. And organometallic catalysis is so complicated that trying to elegantly reason your way right to the good parts is not always a productive use of your time. Do you want to look like a genius, or do you want to discover new chemistry?
1. lt on September 21, 2011 9:23 AM writes...
Combine everything with everything and heat: that's pretty much as inelegant as chemistry gets - positively medieval...
Can't argue with results though and as there are MS techniques that can analyse a drop/microwell with a diameter in the tens of microns, it really seems like a good time to go massively parallel.
Permalink to Comment2. Anonymous on September 21, 2011 9:27 AM writes...
A possible response to criticism (b):
One of the arguments for total synthesis is that it "discovers new reactions". However this investigation has discovered two novel reactions in one process, compared to the negligible advancements that most 50 step total synthesis routes bring - which is the most elegant would you say?
Permalink to Comment3. SP on September 21, 2011 9:41 AM writes...
David Liu did this with DNA-templated reactions in 2004: Nature 431 p. 545-549 (2004)
Permalink to Comment4. anchor on September 21, 2011 9:44 AM writes...
Our years of training in certain area expects us to be rational in our thinking and formulate ideas thereof. The rationale behind Prof. Hartwig's paper is simply mind boggling. You do not expect this from an individual whose is smart and efficient. The tactic that he employed can be expected of some one who is staring his career in academia. Seems pretty desperate to me!
Permalink to Comment5. Hap on September 21, 2011 10:04 AM writes...
That's only true if you assume that you know everything. Since he's searching for what he (and lots of other people) don't know, searching systematically doesn't make sense - you don't know what's out there, or what rules you haven't seen yet.
Permalink to Comment6. David Formerly Known as a Chemist on September 21, 2011 10:06 AM writes...
Isn't this what Symyx was originally founded to do, way back in the mid-90s?
Permalink to Comment7. milkshake on September 21, 2011 10:30 AM writes...
the hydroamination of alkynes with anilines to form imines is not quite a new reaction, only its Cu-catalyzed version is (which has possible a wider substrate scope and milder conditions).
I remember reading 5+ years back in a review on hydroaminations that aryl acetylenes add reactive primary amines in decent yields thermally, to form acetophenone imines. The addition was by then well known, the discussion went like this : the hydroamination reaction of styrenes is unfavourable thermodynamically: it can work uncatalyzed at fairly reasonable temperatures (below 150C) but to shift the equilibrium towards the product one needs to take advantage of a ring formation, or use O-alkyl hydroxylamine, or use aryl acetylene
Permalink to Comment8. William B Swift on September 21, 2011 10:36 AM writes...
The "obvious" response to complaints of "obviousness" is that since it hadn't been done before, and it is useful, it couldn't be all that obvious.
Permalink to Comment9. LabMonkey on September 21, 2011 10:45 AM writes...
Who needs elegance when you get results - elegance can really be a restrictive ideal that hinders progress. This move by the group is exactly what's needed to explore the "unknown unknowns" - I love their audacity (and ability - to dig through the data to work out what the heck they've made!)
Reminds me a bit of the work carried out at CMLD (Center for Chemical Methodology and Library Development). This paper had a nice summary if memory serves correct: J. Am. Chem. Soc., 2007, 129 (5), pp 1413–1419 http://pubs.acs.org/doi/abs/10.1021/ja0674744
Permalink to Comment10. barry on September 21, 2011 10:58 AM writes...
If Perkin had had these tools, he might have given us quinine (as well as mauve...and who knows what else!)
Permalink to Comment11. barry on September 21, 2011 11:01 AM writes...
If Perkin had had these tools, he might have given us quinine (as well as mauve...and who knows what else!)
Permalink to Comment12. barry on September 21, 2011 11:02 AM writes...
If Perkin had had these tools, he might have given us quinine (as well as mauve...and who knows what else!)
Permalink to Comment13. CMCguy on September 21, 2011 11:19 AM writes...
Sure this is interesting and potentially useful approach however is just application of good old Combichem techniques to a particular area. For drug discovery Combichem is largely perceived as a failure or at least an unfulfilled diversion probably because statements similar to the second quote above which were taken as a total paradigm shift rather than expressing access to a potentially valuable tool for certain occasions to supplement MedChem. I do not know if anything has been published but am aware of a few places that have applied these techniques for Process Development and Optimization studies.
Permalink to Comment14. Tom Womack on September 21, 2011 11:26 AM writes...
I can't quite see where you got the 15 and 23 figures above (rather than 8 for the metal rows and 12 for the ligand columns); it sounds as if you're distinguishing 'this well has copper and no nickel' from 'this well has copper and no cobalt', both of which are true descriptions for a well containing only copper but neither very helpful.
Permalink to Comment15. smurf on September 21, 2011 12:17 PM writes...
Don’t care about elegant as long as it works. But from a drug discovery perspective – and I do NOT want to be provocative - does it really matter, do we really need more reactions?
I am not really a chemist, so I am genuinely curious.
Permalink to Comment16. Derek Lowe on September 21, 2011 12:25 PM writes...
Smurf, I'd say that we do, especially catalytic ones. The field has been utterly changed by the advent of Pd-catalyzed coupling reactions, and work with other metals strongly suggests that there are plenty of other useful chemistries out there that haven't been found yet. They'd be good for lowering the cost of existing transformations, and especially for leading us into new chemical space by allowing us to do reactions that no one would even think about now. But it's true that we don't need fifty more ways to couple phenylboronic acid to iodobenzene; there are some wheels out there that don't have to be reinvented.
Permalink to Comment17. Old Timer on September 21, 2011 1:49 PM writes...
This is how most reaction discovery is done in many top labs (Hartwig, Buchwald, Jacobsen etc.), just not in 96-well plates usually. What makes these guys a cut above the rest is their ability to take a serendipitous discovery and quickly ascertain the underlying mechanism of the reaction to improve it. While their discoveries sound good in the paper, they are more often than not, serendipitous.
Permalink to Comment18. hn on September 21, 2011 5:07 PM writes...
If you do something obvious that no one has ever done before, you're onto something big!
Many years ago, my colleague submitted a paper to Science. One reviewer blasted the work for being trivial and obvious. Fortunately, the other reviewers and editor appreciated the significance of the work. That paper has now collected close to a thousand citations.
Permalink to Comment19. eugene on September 22, 2011 1:31 AM writes...
Echoing Old Timer, this is how I always thought basic research to find new reactions was done in the Hartwig lab. Or the Matthias Beller lab. They just finally got around to publishing a 'method' paper. True, before the advent of 96 well plates with good seals and hooked up to an automatic mass spec, I assumed they used desperate postdocs and 20 vials with shit stirring on a hot plate in parallel.
After you've got the right and very expensive high thoroughput instrumentation you can really run away with it in organometallics I bet.
Permalink to Comment20. Scrofa on September 22, 2011 2:40 AM writes...
I am again amazed at the level of envy I detect in the blog everytime Derek shows up with a new interesting paper from the literature.
Obviously it gets increased with high impact publications, which some of the posters may have great difficulty to obtain.
In my opinion, it all cames to develop new science that could possibly resolve some synthetic problem and eventually make possible to create a new chemical scaffold capable of interact with some receptor inside a body to eventually cure a disease. It is of course a very naive vision of the med chem world, but at the end of the day, it could be perfectly possible.
So please, don't waste time criticising the work of your colleagues and spend your time doing good science, that could be publishable at some point and could make the world a better place to be.
Regards
Permalink to Comment21. eugene on September 22, 2011 6:59 AM writes...
Scrofa --
I don't post usually on other papers that Derek posts. I read this one before Derek posted it. Where do you assume that we are envious and are therefore criticizing this? Maybe some of us work in the field; maybe some of us know how this stuff is actually done and we want to comment on it? Now you're telling us to shut up and you're calling us "high-impact publication lacking loosers"
Fine, I admit (for the sake of disclosure), Hartwig is a competitor on one of my projects and I was in a 'race' with his group at one point to publish. I won that race, but I think they won the relay. Still, I'm pretty sure I don't sit in a cave somewhere with internet, thinking bitter thoughts about Hartwig and how people in his group work in a mindless automoton manner (unless they are doing mechanistic elucidation), all day long.
But don't tell us to not criticize work of our colleagues (we do it all the time anyways during review) and to just accept everything in Science as gospel in pursuit of some utopian better-world dream. I like to think of 'In the Pipeline' as a blog (partly) for chemists who want to comment on recent literature based on their expertise.
Ughhhh.... And to think I actually liked the paper. Now after seeing that comment I wish I actually hated it more. I'll read it a second time to see if there are any bad flaws in it.
Permalink to Comment22. sepisp on September 22, 2011 11:23 AM writes...
Physicists have been doing this for a hundred years or so. It's called a particle collider.
Permalink to Comment23. Sisyphus on September 22, 2011 7:12 PM writes...
If an assistant professor or other lesser known scientist had tried to publish this work, it would have been flatly rejected. Probably ending up in TL or worse. I remember reading a recent blog on the corrupt nature of research in country.....
Permalink to Comment24. Kaleberg on September 24, 2011 1:34 PM writes...
As a computer science type, it looks like a great use of what we would call coding theory to let you study 96 reactions using only 20 tests. Error correcting memory works like this, so you only need log N bits to recover from single bit errors in N bit sequences rather than needing a complete N bit copy. I'm wondering how it scales. Can you do a 96 x 96 array with only 20 x 20 tests? Can you do an 8 x 12 x 10 in a reasonable time?
I know that real chemists just don't throw junk into test tubes and stand back, but this looks like a great way to bite into the combinatorics of reactions. It also looks like something you could automate pretty straight forwardly. If nothing else, you might be finding combinatoric chemistry machines in lots of labs in another five or ten years.
Permalink to Comment25. lab rat on September 27, 2011 9:30 AM writes...
Some folks at Roche used this approach to find multi-component reactions ~15 years ago, where they took a set of reagents, reacted them in a 'combinatorial fashion' - up to ten reagents at once (!) and looked for changes on a HPLC trace:
Permalink to CommentLack, O.; Weber, L. Chimia 1996, 50, 445
26. john.gramophon on November 30, 2011 6:45 PM writes...
I always saw such approaches as clever and elegant. But I am a molecular biologist...
Permalink to Comment