Skip to Content

Analytical Chemistry

X-Ray Sponges Ride Again

The Fujita group has a new paper out on their “crystalline sponge” method for X-ray structure determination. I had the chance to hear Prof. Fujita speak on this recently – a very interesting talk indeed – and I’ve been looking forward to the paper. (Here’s a summary from an equally impressed Quintus).
This time, they’re looking at oxidation reactions of the cyclic terpene humulene. That compound, as a glance at its structure will indicate, is not a crystalline solid, but it soaks into the zinc-containing metal-organic framework just fine. And so do all the epoxide and aldehyde oxidation products when the starting material is treated with MCPBA or selenium dioxide. These structures would not be a lot of fun to figure out from NMR data alone, but the X-ray sponge technique nails them all down.

During his seminar, Fujita said that it’s been his experience that if a parent structure fits will into the metal-organic frameworks, that derivatives of it tend to have much higher success rates as well, and that seems to have been the case here. (The overall success rate, from all sorts of structures, is lower). But I’m not surprised that it doesn’t work every time – what continues to surprise and delight me is that it works as well as it does.
One source of tension during the advent of this technique has been the crystallographic rigor of the results. Crystallographers are nothing if not rigorous with their data – if you’re not, you can get lost so many ways that it’ll make your head spin. (That especially goes for protein/ligand crystals, which are very rarely obtained with the resolution of small molecules). The “crystalline sponge” technique is sort of in-between: protein crystallographers look at it and say “Hmm, not so bad”, while small-molecule crystallographers are more doubtful, especially when they see use of solvent-exclusion routines and other data-cleanup techniques. But the recent guidelines proposed by the Clardy group have probably helped in that regard.
I’m no crystallographer myself, but looking over the Supporting Information file for this paper, I note that it details when crystallographic restraints had to be used during the various structure refinements, which is welcome. You can also see that the group did a lot of refinement of the crystal soaking conditions for each compound, which is something that Fujita emphasized during his seminar as well. Interestingly, in one of the complexes, an impurity showed up also ordered in the crystalline framework, which is probably a phthalate plasticizer (leached out of labware). There’s a possibility that it’s a disordered nitrobenzene, a solvent used to produce the crystals initially, but an infrared spectrum of the crystals showed CO bands. (Clardy’s procedure for making the crystals dispenses with the nitrobenzene entirely, so it’ll be worth watching to see if phtalates show up from that recipe once in a while, too).

9 comments on “X-Ray Sponges Ride Again”

  1. Anonymous says:

    That this technique works so well so often doesn’t surprise me. After all. it’s a bit like soaking heavy metals into existing protein crystals.
    I wonder however if it can be improved further by linking specific antibodies to the crystal lattice, so that binding is more consistent?

  2. Anonymous says:

    The problem is how relevant the structure to intended applications. After all, our cells are not crystal sponges.

  3. Anonymous says:

    @2: That will always be a question no matter the technique, but *any* structure is better than no structure, as long as we keep in mind that question.

  4. MattC says:

    Anon 1:
    it’s kind of a different problem to the heavy atom methods – that’s using the atom’s position to phase the structure, whereas in this case you can just solve the MOF structure (the entire MOF is your heavy atom if you like). it’s also trickier because a heavy atom is much more straightforward to locate and refine than a dozen carbons with fractional occupancy.
    on the antibody issue, changing the binding is of course a good idea – there are tens of thousands of MOFs known (and many many times more possible). the pores are typically on the nanometre scale, so sticking a whopping great protein in pore or on the ligand is probably out of the question in the short term.
    Anon 2. well, yes, Fujita says as much. in most cases the MOF isn’t going to break bonds, so the connectivity and stereochemistry ought to be fine. worth pointing out that this objection is basically always true except for structural information gained in situ.

  5. anon says:

    @anon 1:
    “That this technique works so well so often doesn’t surprise me. After all. it’s a bit like soaking heavy metals into existing protein crystals.”
    Though “works” for fujitas lab is defined as the MOF giving a crystalline lattice, wherease “works” for heavy metal isn’t about gaining crystals, its about how to interpret the phase of the crystal. crystallinity is a much more difficult black box than counter-ion exchanging.

  6. Anonymous says:

    @1 and 5: I meant that heavy metals absorb into a protein crystal lattice at regular points without disrupting the lattice, so the analogy is still valid in principle.

  7. Someone says:

    Oh yeah, I just heard his talk yesterday at the CSC 2015 conference. A very interesting seminar indeed!

  8. ProfessorElectron says:

    If this works routinely, this would be a huge step forward. Absolute configuration from less than 10 microgrammes! The crystallographers I know don’t get out of bed for less than 0.5mg.
    Determining relative configuration by NMR with 10ug is practically impossible unless all stereocentres have protons on them and are near to each other. Try acquiring an NOE spectrum to determine configuration on less than 10ug – you’ll be out of luck unless you’ve a 1mm cryoprobe to hand and a day of acquisition time. You could try dipolar couplings, but then you need HC correlation which is still a challenge on 10ug.
    Note also this example doesn’t have any heavy atoms. If you’ve got halogens, or phosphorous, or sulphur in your molecule, I’d expect it to work even better.
    I do hope other labs take this up and get it working- then I’ll be looking for an XRD rather than another NMR.

  9. ProfessorElectron says:

    PS How many folks here have determined a structure using only 10ug of material?

Comments are closed.