Keywords
cryo-EM, X-ray crystallography, computational method, model refinement, structural biology
cryo-EM, X-ray crystallography, computational method, model refinement, structural biology
Following our reviewers' recommendation, we have introduced a sentence at the beginning of the discussion, in order to make clear the meaning of the words “model” and “refinement” throughout the article.
To read any peer review reports and author responses for this article, follow the "read" links in the Open Peer Review table.
In this short communication the word model refers to the macromolecular structure represented by its atomic coordinates, and the word refinement refers to the process of improving the model so that the calculated data based on the model best fit the observed data.
Single-particle cryo-EM has recently joined the circle of techniques that allow macromolecular structure determination at almost atomic resolution. The technique presents a clear advantage over X-ray crystallography in that it allows the structural analysis of particles that are difficult or impossible to grow into crystals, such as very large complexes, membrane proteins or proteins with flexibles portions, and is therefore gaining a prevalent role in the determination of biological macromolecular structures1. As a relatively young method, computational methods to interpret images and derive atomic models continue to be developed and optimized2,3. In that regard, a recent comparison of X-ray and cryo-EM maps calculated at the same resolution, together with the corresponding atomic models, showed that although the appearance of the maps was quite comparable between the two techniques, X-ray crystallography maps were more detailed and the atomic models fitted into them were more accurate4. To make the comparison on a fair basis the X-ray electron densities were calculated with experimental phases (SAD, heavy-atom) and improved by density modification. In this way the maps produced by both techniques were model-free. The accuracy and level of detail of the maps were assessed by fitting the already determined atomic structures into them. These results begged for microscopists to continue to improve the accuracy and performance of methods for map improvement and model refinement, in order to produce atomic models that meet the same quality standards as in X-ray crystallography.
The cryo-EM and X-ray crystallography experimental techniques are very different but the final stages of the structure determination process are similar. An important difference between crystallography and cryo-EM as techniques to reconstruct scattering densities is that in cryo-EM it is possible, in principle, to obtain a reconstruction starting from the raw experimental data without imposing any model, under the assumption that the data are projections of similar particles. If we look at the cryo-EM data in reciprocal space the similarities and differences with X-ray data become apparent. The actual experimental data in cryo-EM are two-dimensional images related to the projections of the particle electrostatic potential, along different directions. According to the projection-slice theorem, the Fourier transform of the two-dimensional experimental image corresponds to a central section of the three-dimensional particle Fourier transform multiplied by the contrast transfer function (CTF), a well-defined mathematical function that introduces a modulation in reciprocal space and restricts the data to annular domains delimited by the zero-crossings of the function. So, while in X-ray crystallography the diffraction experiment gives amplitudes of Fourier coefficients for points in a known reciprocal lattice, in cryo-EM the images give, after CTF correction, complex Fourier coefficients in central planes whose relative orientations in reciprocal space are unknown. The accurate determination of the CTF and the assignment of orientations and centers to the images is the task of the reconstruction procedure which, in principle, does not depend on any model or template. In this regard, the cryo-EM map is somehow equivalent to an experimentally phased X-ray crystallography map in the sense that the required orientations and phases are determined from the sole experimental data. With near atomic resolution data, these maps allow the construction of initial atomic models. In X-ray crystallography, model phases are used to update the map during model building and refinement, keeping the experimental structure factors—the data—unchanged. Thereby, the crystallographic model is improved as the map becomes clearer in both the particle and solvent regions based on feedback from calculated phases. In cryo-EM, the reconstruction is considered as experimental data and kept unchanged during model building and refinement5. In keeping with elementary ideas of data, it seems natural that the central sections, rather than the reconstructed map or its associated Fourier coefficients, should play the role of data in model refinement.
The perspective that atomic structures should be refined against raw cryo-EM images was suggested in conclusion of an excellent recent review on cryo-EM refinement6. We propose here directions for the explicit use of experimental images in cryo-EM refinement. We call Gobsk the two-dimensional Fourier coefficients of the kth image and Gcalck the corresponding calculated quantities, given by
where F is the three-dimensional Fourier transform of the model electrostatic potential, Rk the rotation that specifies the section orientation, ok the origin shift that centers the section, χk the section’s associated CTF and q is a two-dimensional reciprocal vector. The mismatch between Gobsk and Gcalck may be used to define the experimental component of a refinement target function. The refinement target function should thus allow to couple the improvement of the model to that of the orientations, centers and CTF. Accordingly, the mismatch between Gobsk and Gcalck could be used to calculate an R-factor for structure assessment. Note that such criteria remain meaningful even in situations where maps are not of high quality, for example when the distribution of central sections is pronouncedly uneven.
By using experimental images as data, cryo-EM refinement procedures become thus quite similar to those used in X-ray crystallography, in spirit, in that the reconstructed maps are allowed to improve as model building and refinement proceed. A proof-of-principle assessment of the above proposal could be feasible based on available refinement software originally developed by X-ray crystallographers. Such software is mainly implemented in cartesian coordinates, but it can be anticipated that spherical coordinates will be more appropriate not only to handle rotations and interpolation of Fourier coefficients as required by Equation 1, but also to produce more accurate results7,8. Clearly, other not well understood aspects throughout the determination procedures will also need to be improved, such as dealing with errors on the detectors or multiple conformations of molecules6, some of which may benefit from the proposed refinement target to yield statistically more robust structures. Ultimately, this may allow to better exploit the potential of the cryo-EM method and lead to a significant gain of accuracy of the high-resolution refinement protocol.
No data are associated with this article.
We thank our colleagues at the “Conference on methods and applications in the frontier between MX and CryoEM” held in Barcelona in 2017 for discussions, which motivated this communication.
Views | Downloads | |
---|---|---|
F1000Research | - | - |
PubMed Central
Data from PMC are received and updated monthly.
|
- | - |
Is the topic of the opinion article discussed accurately in the context of the current literature?
Yes
Are all factual statements correct and adequately supported by citations?
Partly
Are arguments sufficiently supported by evidence from the published literature?
Yes
Are the conclusions drawn balanced and justified on the basis of the presented arguments?
Yes
Competing Interests: No competing interests were disclosed.
Reviewer Expertise: Structural biology, cryoEM
Is the topic of the opinion article discussed accurately in the context of the current literature?
Yes
Are all factual statements correct and adequately supported by citations?
Yes
Are arguments sufficiently supported by evidence from the published literature?
Yes
Are the conclusions drawn balanced and justified on the basis of the presented arguments?
Yes
Competing Interests: No competing interests were disclosed.
Reviewer Expertise: Structural biology
Alongside their report, reviewers assign a status to the article:
Invited Reviewers | ||
---|---|---|
1 | 2 | |
Version 2 (revision) 28 Nov 19 |
||
Version 1 15 May 19 |
read | read |
Provide sufficient details of any financial or non-financial competing interests to enable users to assess whether your comments might lead a reasonable person to question your impartiality. Consider the following examples, but note that this is not an exhaustive list:
Sign up for content alerts and receive a weekly or monthly email with all newly published articles
Already registered? Sign in
The email address should be the one you originally registered with F1000.
You registered with F1000 via Google, so we cannot reset your password.
To sign in, please click here.
If you still need help with your Google account password, please click here.
You registered with F1000 via Facebook, so we cannot reset your password.
To sign in, please click here.
If you still need help with your Facebook account password, please click here.
If your email address is registered with us, we will email you instructions to reset your password.
If you think you should have received this email but it has not arrived, please check your spam filters and/or contact for further assistance.
Comments on this article Comments (0)