Biocurious is a weblog about biology, quantified.

OIST workshop recap: Cryo-electron imaging

by PhilipJ on 2 May 2006

The OIST Single Molecule Analysis workshop was more or less split down the middle between two different kinds of talks, those by people studying biological systems with optical techniques, either fluorescence or optical/magnetic tweezers (fluorescence and tweezers recap here), and those studying biological molecules via cryo-electron imaging. This recap will focus on cryo-EM measurements, and in particular a talk by Wolfgang Baumeister at the MPI for Biochemistry.

For a bit of background, the resolution with which you can image a system depends on the wavelength of your imaging source, so in the case of optical microscopy it is usually in the visible region of the spectrum, anywhere from 300—700 nm. The problem, of course, is that biological molecules are significantly smaller than this, so detailed information of the structure of individual molecules is essentially impossible to get from light microscopy measurements. On the other end of the photon spectrum, we have had great success using x-rays to probe periodic crystalline structures, including an ever-increasing number of protein samples. The problem, however, is that many proteins are difficult to crystalise, and that proteins are clearly not in a crystalline state in vivo, so while the structures obtained from x-ray diffraction are quite close, it isn’t obvious that this is the conformation these proteins take in a cell. Similarly, because x-ray diffraction requires crystals, there is no easy way to study dynamics.

Enter electrons as the imaging source. In cryo-EM experiments, you take your sample and flash-freeze it on a small grid in water, but the “freezing” step creates amorphous, non-crytalline water instead of ice. Your samples are therefore not crystalline either, but “trapped” in this amorphous water in what is presumably their more natural conformation. Since you don’t need to grow macroscopic crystals, you can study a wider variety of samples that have so far eluded crystallographers.

People have been using this technique to study a wide variety of samples, including the GroEL/GroES chaperone system (a very nice talk by Helen Saibil at Birkbeck College), viral capsids, where unique symmetries allow for good, high resolution reconstructions, actin filaments with bound Myosin heads, etc. The kinds of systems available for study are much broader than for x-ray diffraction studies, but the resolutions are in some cases much worse, with the high end somewhere around 2-4 nanometres (whereas x-ray data is typically now down to Angstroms).

But the one talk that was quite different from the others was on cryo-electron tomography by Baumeister [1]. He is working on 3D reconstructions of the cellular proteome, largely because we still don’t understand things in between the very small and the medium-sized worlds of molecules and cells respectively. He started his talk off with this quote from Philip Ball, a writer for Nature [2]:

[W]e know about molecules; we know about cells and organelles; but the stuff in between is messy and mysterious … yet that is the level of magnification at which much of the action takes place: the scale of perhaps a few to several hundred nanometers.

which makes the case, more or less, for trying to study everything in between. By growing and flash-freezing cells on a cryo-EM grid, it is possible to image an entire cell and reconstruct the entire proteome of an organism. Rather than studying a couple of proteins at a time via FRET, or invididual molecular motors in a tweezers instrument, he is able to directly visualise, at a reasonable resolution, all of the proteins in a cell. By using an organism whose genome has been sequenced and the majority of proteins have structures from either x-ray, NMR, or other cryo-EM measurements, it is then possible to do a 3D map of all the proteins in an organism as they were just seconds ago before freezing. I thought this was remarkable.

The technicalities of these measurements were interesting as well. Basically, the signal to noise ratio in these electron tomograms is not great—you have to keep the electron dose down in order to not destroy your cell, but that means your image is of low contrast. Given this constraint, what is the required resolution the template library of the known proteins that you want to match such that you don’t generate erroneous matches? The example he gave was again of the GroEL/GroES system and another, larger heat shock protein which are of quite similar dimensions (see the orange and red proteins in the template library below), and automating the procedure to pick one versus the other is a non-trivial task. Though this is a difficult problem, the general method of analysing the tomograms is complete, and follows this flowchart, taken from [1]:

With this kind of data, it is possible to start looking at many-body interactions between a wide variety of proteins that are thought to act in highly coordinated ways [3]—what Baumeister called “molecular sociology”, more or less his interpretation of what systems biology is.

All in all the conference was excellent, the venue was outstanding, and my only complaint was that I didn’t get to stay in Japan a little longer. As OIST is a new institute that is trying raise its profile internationally, they’ll be hosting more conferences such as this one in the future, and I highly recommend applying to go!

[1] Nickell S., Kofler C., Leis A.P., & Baumeister W. A visual approach to proteomics, Nature Reviews Molecular Cell Biology 7 225 (2006).

[2] Ball, P. Portrait of a molecule. Nature 421, 421 (2005).

[3] I find it rather amusing that the talk I most enjoyed at a workshop on single molecule analysis basically made the case for no longer analysing single molecules! Now, this doesn’t invalidate the kinds of experiments that tweezers and AFM people do (e.g., André and I), because these tomograms will never give you information about, for example, molecular motor stall forces or the free energies associated with unfolding a protein, but it was definitely an eye opening talk for me.

  1. Andre    4492 days ago    #

    It’s one step closer to the ultimate goal of experimental molecular and cell biology and of biophysics: an atomic scale movie with micro scale time resolution of a cell going about its business. Of course, 100 nm and and 100 ms scale would also be worth doing :) Oh yeah, and if you could also add controlled perturbations to specific regions at those scales, that would be nice too.

    Biophysics would be reduced to interpreting a huge movie, one second of which would have more information than is available in a 1000 papers with current methods. You would just need to know how to ask the right questions of your data the same way molecular dynamicists need to.

    Suggestions on how to accomplish this are welcome… neutron or x-ray holography? Even the spatial resolution of optical methods would be great. Inverse scattering problems anybody?

  2. x-ray fluorescence    3468 days ago    #

    Hi, Resolution in cryo-electron tomography is ultimately limited by the total electron dose used to collect a tilt series. The maximum electron dose that still preserves macromolecular structure to a given resolution has to be distributed among the images of the tilt series…..

  Textile help