Skip to Content

Invisible Crystals Yield Structure

A crystallographer colleague passes on this new paper, which I find very interesting and just a bit freaky. The authors (a collaboration between UCLA, HHMI, LBNL, and SLAC) are studying alpha-synuclein, the protein that aggregates in Parkinson’s disease pathology, and more particularly, they’re studying a short (11-amino acid) section of it that seems to be responsible for its bad behavior. Mutations and deletions in this region mostly abolish the aggregation behavior and toxicity of the protein, so it really seems to be the sticky anchor that weighs everything down.

The paper reports a very nice structure of the aggregated protein at 1.4 Angstrom resolution. What’s unnerving, though, is that the crystals themselves are invisible, since they’re smaller than optical wavelengths. The various crystallization conditions seemed to produce nothing or just amorphous material, but on inspection by electron microscopy, it became clear that this “amorphous” material was in fact made up of clusters of nanocrystals, with cross sections of only 50 to 300 nm, which would make them invisible in any visible wavelength. (I guess you might begin to detect the larger ones down in the short ultraviolet, although I don’t know if that would be of much help or not).

These crystals are far too tiny for x-ray beam equipment to deal with, so they were solved by cryo-electron microscopy, in what has to be considered a major milestone for that up-and-coming technique. These are, in fact, the first completely unknown protein structures to be determined by micro-electron-diffraction. The core of the structure is shown at right, and it’s the beta-sheet bad news that one would expect from what’s known about other aggregating proteins – everything is lined up perfectly for trouble. (Unusually, there are a couple of water molecules involved – the rest of these amyloid-type structures don’t have anything of the sort. The rest of the paper shows how the protein is likely  to bend over and around this core, and how some of the known mutations are disrupting things.

And here’s some more on the technique used to solve them:

For amyloid crystals, our speculation is that the tiny size is a consequence of the natural twist of β-sheets that form the protofilaments of the fibrils. The crystal lattice restrains the twist, creating a strain in these crystals, which increases as crystals grow. Eventually this strain prevents further addition of β-strands, limiting the thickness of the needle crystals. In our experience, longer segments (for example, 11 residues compared to 9 residues) limit crystal growth even more; in the case of 11-residue NACore and 10-residue PreNAC, the strain produces nanocrystals, invisible by optical microscopy. These crystals are too small for mounting and conventional synchrotron data collection, but are ideally suited for analysis by MicroED. They are ~1010 times smaller than Perutz’s haemoglobin crystals and ~1012 times smaller than von Laue’s CuSO4 crystal, which yielded the first X-ray diffraction pattern. Our structures of NACore and PreNAC demonstrate that MicroED is capable of determining new and accurate structures of biological material at atomic resolutions.

How many other invisible crystals are out there, waiting to be solved? It’s quite a thought. . .

7 comments on “Invisible Crystals Yield Structure”

  1. Kurt Thorn says:

    It’s not true that particles need to be larger than the wavelength of light to be seen by optical techniques – scattering is routinely used to image gold nanoparticles as small as 10 nm in darkfield ( and differential interference contrast microscopy has long been used to see objects as small as single microtubules ( It is true that a light microscope can’t resolve objects that small (i.e. an object smaller than the diffraction limit of the microscope, typically a bit smaller than the wavelength of light, will be blurred to appear as large as the diffraction limit). But very small particles can easily be detected in light microscopy.

  2. Kelvin says:

    Amazing achievement. People have tried, struggled and failed to get crystals of amyloid for decades, and the resolution here is incredible. I’m not bothered by the fact that the crystals are “invisible” as x-rays are much shorter in wavelength than light rays. But seriously, this is a major accomplishment that validates all our thinking and models of aggregated amyloid.

  3. Kelvin says:

    PS. Now to determine their structure while embedded in cell membranes, I’m pretty sure these will form leaky pores of beta-barrels, thus compromising cell membrane integrity and causing the cells to “burn out” with oxidative stress while trying to maintain homeostasis…

  4. anon electrochemist says:

    Let’s not forget that simple structures can be solved ab initio from x-ray powder diffraction, which routinely handles crystals down to maybe 10nm diameter or less. The range they quote (50-300nm) is ideal. The software is now basically push-button. If an initial guess at the structure can be provided, even larger unit cells like these can be Rietveld refined without too much trouble. The trick comes when your crystals have irritating shapes and preferential orientation, and correction-factor errors start to creep in. This is a problem in the TEM too.

  5. Mo Shoreibah says:

    This work on alpha-synuclein is remarkably similar to the work done by another group working on β amyloid and Alzheimers. Freeze fracture electron microscopy studies of β amyloid assemblies inserted in phospholipid membranes were successfully done by Nelson Arispe and colleagues in 2010 ( The model they developed from their work suggests that oligomers of β amyloid form barrel shaped channels that are conductive for cations including calcium and zinc (with obvious implications for the pathology that they lead to). Single channel recordings done on artificial membranes with these assemblies support their channel model hypothesis.

  6. anaaaa says:

    Is this actually cryo-EM? As a layman I’d think diffraction technique used in this paper is fundamentally different from blobology-reconstruction which forms the basis of cryoEM? Can someone clairify?

Comments are closed.