Determining the structure of a new molecule is one of those things that you’d think would be simple – at least, nonscientists often seem surprised at how much of our time we spend on such problems. (It doesn’t help that dramatic depictions involving chemistry almost invariably skip over this problem in the interest of moving the plot along). Most of the time during a new synthesis we pretty much know where we are, true – there’s only one carboxylic acid on the starting material, and we just did something routine to it, and there it is, by gosh. But if something unexpected happens, or some odd side product forms, or if you isolate a new compound from a natural product source, well, that’s a different matter.
The best news you can get in such a situation is that the compound forms a decent crystal, because then you have X-ray crystallography on your side. That’s not foolproof, but it’s a pretty damned strong technique, and a good X-ray structure will pretty much settle most arguments. First, though, grow your decent crystal. That gets harder and harder as your molecules get more complex and your available sample shrinks, as you’d figure, and growing good crystals in general is known far and wide (and correctly!) as a black art. Things have gotten easier as X-ray sources have gotten brighter and the available techniques to work up the data have improved, but there’s a good reason that people have been excited about the “crystalline sponge” technique, which can (in some cases) get rid of the crystal-growing step completely.
NMR can also solve your structure for you, but it has its own limitations. There’s a whole list of neat experiments that can be run on a modern NMR machine, limited mostly by your sample size, which has a direct correlation with your time and patience. You get a lot of connectivity data out of many of these (this is next to this, which is next to this, etc.), but there are many structures that can make the easy experiments less informative or more equivocal. So this paper, which proposes a new protocol entirely, is quite interesting. It’s a joint effort from Merck, Harvard, UConn, and Amsterdam, and it depends heavily on DFT (density function theory) calculations as applied to less-commonly-applied NMR data, specifically residual dipolar coupling (RDC) and residual chemical shift anisotropy (RCSA). Taking these measurements under what the rest of us organic chemists consider weird conditions (letting the molecules align themselves in a gel matrix) and comparing this with solution data allows for the range of possible structures to be narrowed down to the point that doing all the DFT calculations becomes feasible. (To give you an idea, those tend to run to the “several hours per molecule” range, so if you’re down to a dozen structure, you can crank those out in maybe three days.
RDCs and RCSAs, in fact, are only observed when a molecule is aligned, and not when it’s tumbling around freely in solution, which is the usual NMR experiment familiar to most chemists. Comparing these can give you RDCs that report on the relative orientation (for example) of all the C-H bonds in the molecule. The RCSAs, in turn, give information on the relative orientation of carbon atoms, including those that don’t have a hydrogen on them at all. The calculation values for these can be compared to the experimental ones, and it appears that the range of possible values for them is wide enough for a good match to tell you that you’ve hit the right structure. Shown at right is one of several examples from the paper, the natural product aquatolide, whose structure had recently been revised after a great deal of effort. At upper left is the currently accepted structure, and at upper right is the originally proposed one. The lower two structures are alternate ideas generated computationally. One of those isn’t bad, but the fit to the data is by far the best for the upper right structure, and is pretty poor for the original one next to it.
Here are the authors:
Once a reasonable 3D model associated with each candidate is generated, whether via computational methods or investigator deduc- tion, and chemical shielding tensors are calculated by DFT based on this 3D model, RDC and RCSA data can be employed as a sensitive critical measure to evaluate the validity of the structural assignment. The possibility of a false-positive determination—that is, agreement of RDC and RCSA data with an incorrect structure—is substantially lower than that in an analysis using only conventional NMR data, especially when both RDC and RCSA are jointly used. These data can serve as a convenient NMR litmus test of structure and stereochemical validity. As such, the method described in this work has considerable potential to be widely applied, which could help to quell the flow of incorrect structures appearing in the literature.
That would be a good thing, for sure, because there’s a lot of error out there. One problem will be how common such measurements become. Some of the mistake structure in the literature are completely honest, as with the original aquatolide structure, but some of it is the result of carelessness, and those are the ones that are least likely to go on to do a more rigorous technique like the one described. But it’s very good to have something like this to add to the list of structure-solving (and structure-confirming) techniques, because we really do need all the help that we can get. Natural products chemists and medicinal chemists will especially welcome the chance to try it out!
Update: for another application of this approach, from another group entirely, see this paper from February. There are many other reports in the last few years on the use of RDCs and RCSAs, for those wanting to dig into the subject.