# Can protein structure be determined by X-Ray Diffraction in a single image?

I'm reading about the use of x-ray crystallography to determine protein structure. According to my book, data is collected at 30-360 angles (dependent on the symmetry of the protein). An illustration is given with concentric rings labelled with distances - the further out the points are, the higher the resolution.

Is the image a composite (where the angle of the point point from the centre is equivalent to the angle of the reading) or is a separate image taken at each angle? Are there any other reasons why more images would be required?

Thanks.

• I can't answer on the details, but as far as I understand it they take multiple images while changing the angle. Feb 13 '12 at 21:39
• That's right. I was wondering if the images could be superimposed somehow, since that was not very clear to me. Feb 13 '12 at 23:56
• It is not clear from what you write whether you are familiar with the process of analysis of X-ray diffraction. Is it clear to you that the diffraction pattern has to be mathematically analysed to get the final "3D picture" of the crystal?
– nico
Feb 15 '12 at 7:12
• I have to agree with the answers below, it's best to collect light diffraction from the crystal at multiple exposures within the 0 - 180 degree range. Mar 20 '17 at 21:03

You cannot solve a structure with a single frame, even with perfect diffraction.

The reason you need images over a large swath of angles is because the diffraction pattern is also in three dimensions, in the so-called "reciprocal space". At minimum, a 180° rotation of the crystal is needed to sweep the entire reciprocal space sphere with the plane of detection, though symmetry in the crystal structure can reduce this further (some of my proteins only required 90°, I think hexagonal unit cells with 6-fold symmetry can live with 30°). Because crystals are real things (read: non-ideal, doubly so for protein crystals), the sweep is extended to gain some redundancy.

Likely not. While one can get excellent diffraction data from a high quality crystal, it would be extremely difficult to solve the phase problem. The extra angles will help constraint the solutions.

Not by analysing a single protein. There is work with x-ray lasers.

You have to take a simultaneous image of millions of proteins and use that to get a structure. It's not quite prime time. People are also doing this with electron beams in electron microscopes.

These methods will reconstruct 3D models of the molecules, sometimes in states which cannot be obtained from crystallography. Examples being the structure of the many megadalton nuclear pore complex, and the f-actin fiber. The classic study is 3d model of bacteriorhodopsin, the first membrane protein structure which was at molecular resolution (this was a crystalline sample though).

While in principle, it sounds much simpler - get a pure sample of your protein, or complex and freeze it down and zap it with an Xray or Electron beam, its a lot more work to reconstruct the image and can take as long or longer than getting an x-ray structure. The resolution is also usually poor as the crystal will reinforce coherence, that is all the proteins are aligned in the same way and have close to the same 3d shape in a crystal.

• Thanks for your answer. I didn't mention it in my question, but my book does say that a vast number of crystalised proteins are required, and that simultaneous diffraction amplifies the signal. What I was trying to ask is if a type of crystalised protein conformation can be determined from a single image or if many different images would need to be produced for each angle of rotation. From your answer it sounds as if all the information required is stored on one image - is this correct? Feb 13 '12 at 23:36
• I'm confused by your answer, what do you mean by "it's not quite prime time"? Electron microscopic images also don't provide nearly the resolution you can achieve with X-ray crystallography. Feb 14 '12 at 17:18
• @MadScientist I think the implication is that superimposed images are also used in EM to reconstruct the structure. The case is somewhat different of course since EM in general doesn’t use crystallised structures so that the proteins in the image don’t all have the same orientation, which is crucial. Feb 14 '12 at 22:42
• CryoEM also doesn't use any X-ray beams either. Feb 15 '12 at 6:41
• I think that while they are working very hard, the X-ray laser and cryo EM reconstruction doesn't yield as consistant a set of results as crystallography. when the proteins just lay on a surface, they can be shape distorted, and so recombining the images makes can be difficult to interpret. May 17 '12 at 14:27

Can protein structure be determined by X-Ray Diffraction in a single image?

Yes. Using a technique called Laue diffraction, it is possible to obtain sufficient data from a single image to solve a protein crystal structure. One example is the time-resolved study of carbonmonoxymyoglobin dissociating by photolysis (DOI: 10.1107/S090904959501661X). This is not the standard single-wavelength technique that is usually used, but uses "white" X-rays with a range of wavelengths available only at synchrotron beams. For example, the BioCARS user facility provides infrastructure for time-resolved crystallography. It is also used in "diffraction before destruction" when working with free electron lasers, see for example Nature Methods 8, page 283 (2011).

Is the image a composite (where the angle of the point point from the centre is equivalent to the angle of the reading) or is a separate image taken at each angle

The desired structural information (3D electron density in real space) is a Fourier transform of the diffraction data (3D reciprocal space). The word "image" is jargon for a single diffraction image, i.e. the diffraction spots you observe when you point an X-ray beam at a crystal in a certain orientation. Using a different orientation, you get more data (and also measure some spots multiple times).

Are there any other reasons why more images would be required?

The more diffraction images are collected, the higher the completeness and redundancy of the data. Completeness refers to having measured every diffraction spot at least once. Redundancy refers to how often on average a spot was measured, and increasing redundancy increases the quality of the measurement through averaging.

• I accept this answers the specific question "Is it possible?", but this was rather a proof of concept. Do you know whether this has been adopted to any extent — found use in circumstances where it was not possible to obtain more images? Also, could you cite the original reference. Researchgate is not a journal. Jul 2 '19 at 12:32
• This is used routinely in time-resolved crystallography and in "diffraction before destruction" when working with free electron lasers XFEL, Nature Methods 8, page 283 (2011) Jul 2 '19 at 17:21
• Thanks. You should add this to your answer and I will upvote it. Unfortunately this list is dominated by what I call “furry animal” biologists. Nothing wrong with that, I suppose, but the effort put into factual hard scientific answers is seldom rewarded in Brownie points. Jul 2 '19 at 18:43
• @David Yes, I was visiting from StackExchange Chemistry. There are question in Biochemistry that could go either way, just as there are questions that fit in physics or chemistry, but will get a different type of answer (or none). Not that there is anything wrong with organismal biology. Jul 2 '19 at 21:17
• As long as a sufficient amount of the reciprocal space has been collected, then it is of course possible. How the phase problem is handled when doing Laue diffraction I simply don't know, but I would assume Molecular replacement or Experimental phases can be used? Jul 5 '19 at 2:34