Skip to main content

Analytical Chemistry

Bungled Structure, And How

A lot of natural product structures have been misassigned over the years. In the old days, it was a wonder when you were able to assign a complex one at all. Structure determination, pre-NMR, could be an intellectual challenge at the highest level, something like trying to reconstruct a position on a chess board in the dark, based on acrostic clues in a language you don’t speak. The advent of modern spectroscopy turned on the lights, which is definitely a good thing, but many people who’d made their careers under the old system missed the thrill of the old hunt when it was gone.
But even now, it’s possible to get structures wrong – even with high-field 2-D NMR, even with X-ray spectroscopy. Natural products can be startlingly weird by the standards of human chemistry, and I still have a lot of sympathy for anyone who’s figuring them out. My sympathy goes only so far, though.
Specifically, this case. I have to agree with the BRSM Blog, which says: “I have to say that I think I could have done a better job myself. Drunk.” Think that’s harsh? Check out the structures. The proposed structure had two napthalenes, with two methoxys and four phenols. But the real natural product, as it turns out, has one methoxy and one phenol. And no napthyls. And four flipping bromine atoms. Why the vengeful spirit of R. B. Woodward hasn’t appeared, shooting lightning bolts and breaking Scotch bottles over people’s heads, I just can’t figure.

45 comments on “Bungled Structure, And How”

  1. Petros says:

    While microanalysis is no longer routine, and even J Med Chem finally dropped its insistence on characterization by that route, surely MS to confirm the suggested MWt would have been expected?

  2. scientistbymistake says:

    Maybe it really-really wouldn’t ionise for a Mass-Spec to help with those Bromines…or the molecular weight…or have I missed the point too?
    Red faces all round…Next someone will be saying they’ve isolated a marketed drug from some grass or something…oh, hang on!

  3. scientistbymistake says:

    Looks like we were typing at the same time Petros!

  4. Anonymous says:

    I agree with #1, MS should be used.
    I am suprised that Kozlowski et al. did not show the MS of the natural product that have so many bromine. I guess the reviewers of J. Nat. Product. forget about the isotope pattern of bromine!

  5. ScientistSailor says:

    The natural product I worked on in grad school was bungled by the isolation “scientists”. I made what they proposed, figured out what they did wrong, and then made the correct material. Fortunately, we had a sample of the authentic material, which was essential for determining the correct structure. My advisor believes that ~half of the structures in the literature are wrong, and won’t undertake a synthesis of a natural product unless an authentic sample is available.

  6. Wow! The fact the correct structure was a KNOWN compound makes this even more bizarre.

  7. explosci says:

    This is why the first thing I do, even before NMR, is always mass spec.
    Crystal structures are by far my favorite analysis method however….

  8. Christopher R Lee says:

    Back in the 1980’s when I was working in pharma, we patented a rival’s synthetic product because they’d got the structure wrong. I’m still not sure every company has a committee of elders who make sure the compound in development has the stated structure & that it’s the same as the one proposed by discovery. Easy to get it wrong when there are 5 chiral centres and you’ve been through every salt form in the book.
    Nowadays, MS tends to mean atmospheric pressure ionisation or some similar LC-friendly technique that doesn’t ionise quite a lot of mols. So what you see could be a trace impurity (& you can also miss a drug impurity…). Even if you do get ions, they can be wierd adducts of solvent additives & impurities. Ready-ionised compounds like quaternary ammonium ions & sulfonic acids can fool you because they don’t give M+1 or M-1.
    Bring back EI/CI !!

  9. auger says:

    Does anyone use XPS to make sure you at least have the right elements? Probably would have caught those Bromines.

  10. JAB says:

    Missing the bromines is a pretty big deal, and even bigger is the fact that the correct structure had been previously reported from the same sponge genus. The first hypothesis to test with an isolated natural product is that it is a known one, and the tools exist to do this pretty easily. The supposed molecular mass peak in the MS they reported was spot on for the formula, but then it wasn’t the actual molecular mass peak. A pretty good deconstruction of what went wrong is given in the Penn group’s paper. I would disagree with #7, most natural product structures (>90%?) in the lit are OK, but a head to head comparison is always a good thing.

  11. Grad Student says:

    Surely UPENN has an EI instrument available. 70eV will ionize anything and if you have a clean sample the fragmentation pattern will be pretty informative about the structure in addition to the composition. Even supposing for some reason they don’t have access to an EI instrument, I’m pretty sure the revised structure would ionize under APCI or ESI.
    But, seriously, how could you confuse the proton of the actual product for the one proposed?

  12. startup says:

    That’s why when I look at the structure I intend to make I always consider potential scientific ability of the authors. Call me judgemental, but three Chinese students from OleMiss? I probably would have taken a pass.

  13. John Schilling says:

    My dabblings in chemistry mostly involve simple molecules energetically disassembling themselves, so something like this is a bit out of my league. But if I read the structure correctly, the actual compound is a hair over 60% Bromine by weight. There is no way anyone could have failed to notice that they were dealing with mostly Bromine and some hangers-on.
    Unless they didn’t look. It’s organic, so obviously it’s just C, H, O, N and maybe some P and S. Now we have some NMR data that we have to make fit in a conceptual universe that has only those six elements – and some of the NMR peaks almost overlap, giving you a couple of extra degrees of freedom to make the fit. Does that plausibly lead to (S)-2,2′-dimethoxy-[1,1′-binaphthalene]-5,5′,6,6′-tetraol as the least-bad CHONPS-only fit to the NMR data?
    And if so, what happens when we give these guys some, I don’t know, maybe uranium hexafluoride, tell them we found it in a plant, and have them work out the structure for us? Just for grins and giggles.

  14. Anonymous says:

    Aside from the obvious problems, I am curious about the reported CD spectrum. Is it possible that the real natural product, because it is also axially chiral, could have hindered enough rotation to still be a single enantiomer?

  15. antiaromatic says:

    So here’s my take on this situation. Both of my parents are natural products chemists, and I’m an organic chemist. I’ve seen my share of structural assignment work, but this is definitely careless work. What I’m almost certain has happened is that they were willing to ignore the data in front of them BECAUSE of the molecular formula they had obtained from HRMS. What is not clear is whether the original authors had their MS done in house or was sent away for analysis from which they got the molecular formula and didn’t bother to look at the spectrum (could be that whoever did the MS somehow missed out on the pattern). But I think this is a common error to force data into what HRMS is telling you without having very strong indications that the MS peak you observe is indeed the molecular ion and not some fragment. I still think there should’ve been enough clues to suggest that they were barking up the wrong tree, but I think revisiting the MS is usually the last place someone would think to go.

  16. JAB says:

    Two Chinese Ole Miss students. The third Chinese person is the HIF biologist of the group. #11, the work in question was at Ole Miss, not U Penn. And EI instruments are getting harder and harder to find, and while EI ionizes this sort of aromatic compound pretty well, it surely is not very good at many other types – hence the “soft” ionization methods. Also, working with marine sponges, one ALWAYS must consider the possibility of bromine and/or chlorine.

  17. Mike says:

    Even as a lowly MChem this baffles me as to how the group assigned the structure. From even the most basic of organic structure determination course you would expect a first year to be able to figure out the presence of bromine from the peak distributions.
    I guess it goes to show the importance of microanalysis and full proper characterisation of a compound. I bet someone has been having a bad time of it in the publishing group.

  18. LittleGreenPills says:

    # 10 and 15 are pretty spot-on.
    In natural product research you should always assume you have a known compound. In this case it was even previously isolated form the same organism.
    A lot has been said about the MS data, but if you have ever messed with the ionization conditions you know that the results can be highly variable. With this in mind I note that the paper correcting the structure reports the binaphthyl compound ions in positive mode and the bromide in negative mode. The original paper did not report pos or neg (sloppy), but I would suspect that pos was used. The bromide may not have ionized in pos., which is why it is always good to have multiple ions/adducts that fit (M+H, M+NH4, M+Na, etc.)
    Even so, it is still inexcusable.

  19. LittleGreenPills says:

    I should also point out that for a synthetic chemist this is a nuisance, but they still usually get a publication. However, for a natural product chemist this is a nightmare, and probably results in having to publish a correction.

  20. BRSM says:

    Cheers for the link, Derek!
    I think that the guys who made the stuff and published the correction are remarkably measured in their language. I bet they had some fun writing that paper!

  21. McKinsey Inc. says:

    This was from a future McKinsey consultant.

  22. paperclip says:

    @22: Exactly, always check. My graduate work was in natural products, and there was all of one time when a compound I isolated didn’t have published spectral data. So many structures out there these days. They add confidence to your results, but it sounds like it would be fun to discover something new — besides that one time, I wouldn’t know!
    For NMR newbies (and the not-so-newbies) a good spectral predictor software program can be, well, good. Enough to tell you that the proposed structure was problematic.

  23. David Formerly Known as a Chemist says:

    Everything looks simple in hindsight. It’s pretty easy to anonymously heap crap upon this work and post offensive comments about the national origins of the scientists that did the work (does it really matter that three of the authors were Chinese?), now that the conclusions were shown to be false. However, the analysis originally passed muster of a PI (Dale Nagle), the manuscript reviewers, and whoever read the paper since it was first published in 2007.
    Look back at the original paper. They had 10 mg of a colorless gum to work with. Would you have shipped off a minimum of 5 mg (half your entire sample!) for combustion analysis, when you know how the results for gums usually come back? Probably not. The NMR and MS spectra were consistent with the proposed structure. Having never seen the actual MS spectrum, it’s not possible to comment on whether or not any telltale isotope patterns were obvious to suggest presence of halogens. I suspect not, or else they would have followed up on that.
    Lots of stones being thrown here, as usual. No one following this blog would EVER make such a stupid mistake, would they? Such as drawing conclusions based on the absence of any data…

  24. Hap says:

    Argument by competent peer review died the death when the LaClair hexacyclinol “synthesis” was published, and when the 12e-antiaromatic pyridine dimer….errr, Zincke reaction paper was published, and when the anaerobic oxidation of benzylic alcohols by NaH was (almost) published, and when the Science metabolome paper was published with glaring chemical and biological errors, and….
    MS doesn’t take all that much material – it may be hard to find ionization conditions but it doesn’t take much material and is kind of important (they got rotation, and rotation likely takes a lot more material and is a lot less useful, because it can change so much based on solvent, aggregation, etc.). I don’t know if running various types of MS ionization conditions to tell if the results are consistent is standard practice. NMR calculation software is pretty common and not hard to get, and the lack of ortho couplings in any case should probably have been a red flag. The mistakes aren’t dependent on a lack of knowledge of arcane techniques, but of techniques that should be relatively common in a natural product identification group. Hindsight is always 20/20, but when you fly your plane into a mountain on a clear day with a good plane, there’s a good chance that “pilot error” will be the attributed cause.
    My guess is that not too many people cared before now about this paper, and the only reason the Kozlowski group cared was because the proposed structure was amenable to their methodologies. Lack of response may mean “good paper” or it can mean “bad paper…don’t care”.

  25. “Why the vengeful spirit of R. B. Woodward hasn’t appeared, shooting lightning bolts and breaking Scotch bottles over people’s heads, I just can’t figure.”
    For Woodward a Scotch bottle was too precious a thing to waste.

  26. Phi says:

    Mmmmm, I love the smell of casual racism in the morning.

  27. Thomas McEntee says:

    Per @1, “It’s elementary, my dear Watson” Back in the late 1960s, the Chairman of my dissertation committee insisted that I provide elemental analysis for the target compound of a 30+ step convergent synthesis route. No “high-resolution exact mass spectrometry courtesy of Professor K, Biemann, MIT” would satisfy him. He wanted micrograms of water, carbon dioxide, etc.
    Incorrect Structure: C=74.52%, H=5.63%, O=18.85%
    Correct structure: C=29.36%, H=1.52%, O=9.03%

  28. Hap says:

    I wonder if there’s a good qualitative analysis for catechols. Oxidation and reaction with phenylenediamine might give the dibenzopyrazine (don’t know the name of the ring system) or, possibly, oxidation might give the perylenequinone, which ought to be colored like crazy. Metal coordination would also be possible (though iron reacts with phenols, and doesn’t require polydentate ones). Chemical methods usually require lots of material, which they didn’t have, but they are distinctive.
    What does racism smell like? I assume it smells like stannanes and burnt hair, or perhaps illegal fireworks and alcohol, but don’t know.

  29. The Iron Chemist says:

    Add “institutionalism” in there with racism. Yes, I may have just invented a word.
    As somebody who also works at a southern university, I can attest that most of my colleagues take the quality of their research extremely seriously. We don’t just care about football.
    That said, the authors of the 2007 paper did a poor job interpreting their data.

  30. OMFG says:

    Racism aside, the isolation chemists missed FOUR freakin’ bromines! Yeah, I know that aromatic carbons are often difficult to resolve but how could they confuse 13 inequivalent carbons with 11 in a C2-symmetric dimer?

  31. Secondaire says:

    Yeah, let’s can the racism. It was a goof-up based on sloppy analysis and low-quality data.
    That being said, EI/CI would’ve told the difference really quickly.
    And that being said, I did just about spray my tea all over the computer screen when I saw the paper.

  32. Organometallica says:

    #14 You make a good point about the optical activity. I may have missed it, but I believe that the paper with the revised structure did not report an optical rotation. Out of curiosity, I would also like to know if the revised structure has axial chirality.
    With regard to mass spec, I calculated the exact masses of the two structures. The original structure has an exact mass of 378.11, and they get a hit in the original paper for this mass. However, they report 378.11, and they used electrospray ionization, so I would have expected to see a mass correlating to the protonated or sodiated mass, or the deprotonated mass if they ran it in negative ionization mode. It’s also possible that they ran it on and EI instrument, which would give them a mass of 378, and then misreported what ionization method they used.
    The revised structure has an exact mass of 527.72, and the paper with the revised structure reports an exact mass of 526.71, the deprotonated form.
    I haven’t looked at the NMR data, but based on the large differences in the masses from the mass spectra, it’s possible that the lack of agreement between the two reports is a result of a difference in the samples the two groups used. The original authors may have assigned the correct structure for the sample they were given- I’m inclined to believe this due to the optical rotation they observed.

  33. Thomas McEntee says:

    Idiot chemist @27 (yes, that’s me) bungled the molecular formula for the incorrect structure. Revised elemental analysis values (calculated) for the binaphthyl structure are:
    C=69.84% H=4.80% O=25.37%

  34. Jose says:

    “For Woodward a Scotch bottle was too precious a thing to waste.”
    I think they would clearly be *empty* bottles, right?

  35. @23 David Formerly Known as a Chemist says:

    Your diplomatic skills are commendable. Perhaps the authors of the original isolation paper are being unfairly excoriated by a few “racist” and elitist (that’s the word you were looking for, #29) individuals on this thread. However, the simple fact is that bad “science” was published, which led a grad student’s/post doc’s to waste time making a bogus molecule.
    Professional natural products chemists are able to extract identifiable amounts of target compounds from a variety of biological matrices, ranging from puréed sponges to plant resins. It always amazes me as to how often the natural products chemists correctly assign connectivity and regiochemistry. If the original discoverers had mischaracterized a 2000 amu polyketide or peptide, then I and others would be willing to cut some slack. But this type of shoddy work, even if it came from a top-tier lab, is sadly laughable.

  36. startup says:

    Really? You call skipping over inconvenient extra peaks in the C13 “a hindsight”? A hindsight? Well, I call it incompetence. And you call it racism all day long but everyone not drinking PC kool-aid knows that while strong foreign students typically don’t give a crap about composition of the group the lousy ones flock to the groups dominated by their own kind, be it Chinese, Indian, French or, for that matter, American.

  37. newnickname says:

    Who needs NMR and MS? Ever since my 1st year as a grad student, I’ve seen too many examples of structure proof by wishful thinking. Some of the mistakes could have been corrected by undergrads. In every case, the ones who made the made the mistakes were not penalized. It was usually the naysayer or the one who exposed the error that suffered the consequences.
    One of my faves: an “impossible” reaction worked better and better using more and more, up to 1:1 catalyst:substrate. A junior (BS) colleague on the project asked me for some help. I looked at the NMR, asked what catalyst they were using and, sure enough, they were isolating ligand and mistaking it for product. I remember the silent exchange of blank expressions and the slow appearance of his grin of realization and acceptance. That project died soon after but the PhD who led and promoted the project did just fine.

  38. newnickname says:

    Woodward’s “Structure of Tetrodotoxin”, Pure Appl Chem, 1964, 9, 49-74, possibly assisted by a few bottles of Scotch. Pages and pages of incredible puzzle solving structural analysis and then, “At this point, however, the argument was simplified through the obtention of the almost ocular evidence provided by a complete three-dimensional X-ray crystallographic analysis of [a derivative].” I always liked that “almost ocular” description of the X-ray results.
    (Yeah, even some X-ray structures have been misinterpreted.)

  39. mass_speccer says:

    I’d have thought that it might be sensible to require authors of isolation papers to submit supporting information with some useful data in.
    Unless I’m missing something, all of their supporting info is data from biological screens (GI50 values and stuff). Pretty hard for a reviewer to decide if the structure is correct when they never get to see the original NMR/MS….

  40. Jack says:

    “”For Woodward a Scotch bottle was too precious a thing to waste.”
    I think they would clearly be *empty* bottles, right?”
    Yeah, ain’t chronic alcoholism a hoot?

  41. Am I Lloyd says:

    The disappearance of the word “obtention” from the literature always makes me wistful.

  42. Old School says:

    Too much reliance on instumental methods. 1mg and a tiny sliver of Sodium heated together, followed by some Silver Nitrate would probably have showed the presence of the bromine. Sometimes old techniques used by the likes of RB, Gilbert Stork and others are best. This was a fun test I learned in Qualitative Organic Analysis over 30 years ago. Poof! extract, add silver nitrate solution, what color is the ppt. Though that might be hard to do on this scale. Off hand it might be possible to further use chemical methods for the detection of silver bromide at this scale.

  43. dvizard says:

    As I pointed out in the comment on the BRSM blog, they assigned the formula from the HRMS data, but (disregarding that the listed mass has nothing to do with the actual compound of interest) a) they don’t even list the most basic information about their ESI-MS conditions (positive or negative?) and b) they assigned the formula for their molecule from a mass which likely corresponds to [M+H]+. In summary, this means that ALL assignments in the paper are likely wrong, not just this one…

  44. CTUK says:

    Errrr… the MW of their proposed structure is 531 but the real structure is 378. I am only too well aware that some peoples instrumentation is not 100% but I start from GCMS & NMR and go from there (and I suck).

  45. Mrs. Anonymous BMS Researcher says:

    #31, my sympathies for the sprayed tea…I hope you weren’t drinking New Vithanakande or Twinings Blackcurrant, or your equivalently waste-no-microliter favorite.

Comments are closed.