Outcome of the First Electron Microscopy Validation Task Force Meeting
2012; Elsevier BV; Volume: 20; Issue: 2 Linguagem: Inglês
10.1016/j.str.2011.12.014
ISSN1878-4186
AutoresRichard A. Henderson, Andrej Šali, Matthew L. Baker, Bridget Carragher, Batsal Devkota, Kenneth H. Downing, Edward H. Egelman, Zukang Feng, Joachim Frank, Nikolaus Grigorieff, Wen Jiang, Steven J. Ludtke, Ohad Medalia, Pawel A. Penczek, Peter B. Rosenthal, Michael G. Rossmann, Michael F. Schmid, Gunnar F. Schröder, Alasdair C. Steven, David L. Stokes, John D. Westbrook, Willy Wriggers, Huanwang Yang, Jasmine Young, Helen M. Berman, Wah Chiu, Gerard J. Kleywegt, Catherine L. Lawson,
Tópico(s)Force Microscopy Techniques and Applications
ResumoThis Meeting Review describes the proceedings and conclusions from the inaugural meeting of the Electron Microscopy Validation Task Force organized by the Unified Data Resource for 3DEM (http://www.emdatabank.org) and held at Rutgers University in New Brunswick, NJ on September 28 and 29, 2010. At the workshop, a group of scientists involved in collecting electron microscopy data, using the data to determine three-dimensional electron microscopy (3DEM) density maps, and building molecular models into the maps explored how to assess maps, models, and other data that are deposited into the Electron Microscopy Data Bank and Protein Data Bank public data archives. The specific recommendations resulting from the workshop aim to increase the impact of 3DEM in biology and medicine. This Meeting Review describes the proceedings and conclusions from the inaugural meeting of the Electron Microscopy Validation Task Force organized by the Unified Data Resource for 3DEM (http://www.emdatabank.org) and held at Rutgers University in New Brunswick, NJ on September 28 and 29, 2010. At the workshop, a group of scientists involved in collecting electron microscopy data, using the data to determine three-dimensional electron microscopy (3DEM) density maps, and building molecular models into the maps explored how to assess maps, models, and other data that are deposited into the Electron Microscopy Data Bank and Protein Data Bank public data archives. The specific recommendations resulting from the workshop aim to increase the impact of 3DEM in biology and medicine. Structure analysis of macromolecular complexes using three-dimensional electron microscopy (3DEM) has become an essential tool for structural biology research. 3DEM is uniquely able to determine the structural organization of macromolecular complexes not amenable to other methods (Frank, 2006Frank J. Three-dimensional electron microscopy of macromolecular assemblies: visualization of biological molecules in their native state.Second Edition. Oxford University Press, New York2006Crossref Google Scholar, Glaeser et al., 2007Glaeser R.M. Downing K.H. DeRosier D. Chiu W. Frank J. Electron crystallography of biological macromolecules. Oxford University Press, Oxford, New York2007Google Scholar). More than thirty years ago, low-dose imaging and computational averaging of images of two-dimensional (2D) crystals of bacteriorhodopsin produced a density map that revealed protein α helices spanning the lipid bilayer (Henderson and Unwin, 1975Henderson R. Unwin P.N. Nature. 1975; 257: 28-32Crossref PubMed Scopus (854) Google Scholar). Subsequent advances in 3DEM of unstained specimens embedded in vitreous ice (cryo-EM) are increasingly yielding density maps of a wide variety of specimens at near-atomic resolution. Applications to icosahedral viruses and chaperonins already demonstrate that 3DEM maps can be good enough to trace Cα backbones de novo and to visualize some side-chain densities without the aid of X-ray crystallography (Chen et al., 2011Chen D.H. Baker M.L. Hryc C.F. DiMaio F. Jakana J. Wu W. Dougherty M. Haase-Pettingell C. Schmid M.F. Jiang W. et al.Proc. Natl. Acad. Sci. USA. 2011; 108: 1355-1360Crossref PubMed Scopus (73) Google Scholar, Liu et al., 2010Liu H. Jin L. Koh S.B. Atanasov I. Schein S. Wu L. Zhou Z.H. Science. 2010; 329: 1038-1043Crossref PubMed Scopus (138) Google Scholar, Zhang et al., 2010aZhang J. Baker M.L. Schröder G.F. Douglas N.R. Reissmann S. Jakana J. Dougherty M. Fu C.J. Levitt M. Ludtke S.J. et al.Nature. 2010; 463: 379-383Crossref PubMed Scopus (113) Google Scholar, Zhang et al., 2010bZhang X. Jin L. Fang Q. Hui W.H. Zhou Z.H. Cell. 2010; 141: 472-482Abstract Full Text Full Text PDF PubMed Scopus (146) Google Scholar). 3DEM is unusually versatile and can be used to investigate the structures of a wide variety of specimens under conditions close to those in the cell. Specimens can range from highly purified, homogeneous molecular complexes to heterogeneous conformations and may assume different forms with or without symmetry. Subnanometer resolution cryo-EM structures are found to be increasingly useful in providing illustrative snapshots of macromolecular machines such as the ribosome, chaperonins, and viruses bound to various cellular effectors or ligands (Becker et al., 2009Becker T. Bhushan S. Jarasch A. Armache J.P. Funes S. Jossinet F. Gumbart J. Mielke T. Berninghausen O. Schulten K. et al.Science. 2009; 326: 1369-1373Crossref PubMed Scopus (146) Google Scholar, Frank et al., 1995Frank J. Zhu J. Penczek P. Li Y. Srivastava S. Verschoor A. Radermacher M. Grassucci R. Lata R.K. Agrawal R.K. Nature. 1995; 376: 441-444Crossref PubMed Google Scholar, Miyazawa et al., 2003Miyazawa A. Fujiyoshi Y. Unwin N. Nature. 2003; 423: 949-955Crossref PubMed Scopus (849) Google Scholar, Zhang et al., 2010bZhang X. Jin L. Fang Q. Hui W.H. Zhou Z.H. Cell. 2010; 141: 472-482Abstract Full Text Full Text PDF PubMed Scopus (146) Google Scholar). Finally, electron tomography, in which a series of images is collected from a region of the specimen tilted to different viewing angles, can be used to obtain 3D density maps of individual macromolecular particles, including pleiomorphic ones for which whole-particle averaging is inadmissible (Grünewald et al., 2003Grünewald K. Desai P. Winkler D.C. Heymann J.B. Belnap D.M. Baumeister W. Steven A.C. Science. 2003; 302: 1396-1398Crossref PubMed Scopus (264) Google Scholar), as well as sections, or even whole cells, provided they are not thicker than approximately 0.7 μm (Al-Amoudi et al., 2004Al-Amoudi A. Norlen L.P. Dubochet J. J. Struct. Biol. 2004; 148: 131-135Crossref PubMed Scopus (125) Google Scholar, Al-Amoudi et al., 2007Al-Amoudi A. Díez D.C. Betts M.J. Frangakis A.S. Nature. 2007; 450: 832-837Crossref PubMed Scopus (156) Google Scholar, Frank, 2006Frank J. Three-dimensional electron microscopy of macromolecular assemblies: visualization of biological molecules in their native state.Second Edition. Oxford University Press, New York2006Crossref Google Scholar, McIntosh, 2007McIntosh J.R. Cellular Electron Microscopy. Volume 79. Elsevier Science, San Diego2007Google Scholar, Medalia et al., 2002Medalia O. Weber I. Frangakis A.S. Nicastro D. Gerisch G. Baumeister W. Science. 2002; 298: 1209-1213Crossref PubMed Scopus (497) Google Scholar). For an extensive review of 3DEM procedures, see Baker and Henderson, 2012Baker T.S. Henderson R. Electron Cryomicroscopy of Biological Molecules.in: Arnold E. Himmel D.M. Rossmann M.G. International Tables for Crystallography Volume F, Crystallography of Biological Macromolecules. West Sussex, UK: John Wiley & Sons, Ltd, 2012: 593-614Crossref Google Scholar. Interpretation of a 3DEM density map frequently involves building a molecular model. Models may consist of atoms or "coarse-grained" objects representing multiple atoms, such as whole residues, secondary structure segments, and shape-based features. A model of a given macromolecular complex is often computed by assembling experimentally determined atomic structures or homology models of the individual subunits. The subunit models can either be held rigid (Chapman, 1995Chapman M.S. Acta Crystallogr. A. 1995; 51: 69-80Crossref Google Scholar, Jiang et al., 2001Jiang W. Baker M.L. Ludtke S.J. Chiu W. J. Mol. Biol. 2001; 308: 1033-1044Crossref PubMed Scopus (189) Google Scholar, Lasker et al., 2009Lasker K. Topf M. Sali A. Wolfson H.J. J. Mol. Biol. 2009; 388: 180-194Crossref PubMed Scopus (53) Google Scholar, Roseman, 2000Roseman A.M. Acta Crystallogr. D Biol. Crystallogr. 2000; 56: 1332-1340Crossref PubMed Scopus (106) Google Scholar, Rossmann, 2000Rossmann M.G. Acta Crystallogr. D Biol. Crystallogr. 2000; 56: 1341-1349Crossref PubMed Scopus (127) Google Scholar, Volkmann and Hanein, 1999Volkmann N. Hanein D. J. Struct. Biol. 1999; 125: 176-184Crossref PubMed Scopus (132) Google Scholar, Wriggers et al., 1999Wriggers W. Milligan R.A. McCammon J.A. J. Struct. Biol. 1999; 125: 185-195Crossref PubMed Scopus (349) Google Scholar, Wriggers and Chacón, 2001Wriggers W. Chacón P. Structure. 2001; 9: 779-788Abstract Full Text Full Text PDF PubMed Scopus (87) Google Scholar) or allowed to flex (Fabiola and Chapman, 2005Fabiola F. Chapman M.S. Structure. 2005; 13: 389-400Abstract Full Text Full Text PDF PubMed Scopus (77) Google Scholar, Rusu et al., 2008Rusu M. Birmanns S. Wriggers W. Bioinformatics. 2008; 24: 2460-2466Crossref PubMed Scopus (18) Google Scholar, Schröder et al., 2007Schröder G.F. Brunger A.T. Levitt M. Structure. 2007; 15: 1630-1641Abstract Full Text Full Text PDF PubMed Scopus (113) Google Scholar, Tama et al., 2004Tama F. Miyashita O. Brooks 3rd, C.L. J. Mol. Biol. 2004; 337: 985-999Crossref PubMed Scopus (129) Google Scholar, Topf et al., 2005Topf M. Baker M.L. John B. Chiu W. Sali A. J. Struct. Biol. 2005; 149: 191-203Crossref PubMed Scopus (57) Google Scholar, Topf et al., 2008Topf M. Lasker K. Webb B. Wolfson H. Chiu W. Sali A. Structure. 2008; 16: 295-307Abstract Full Text Full Text PDF PubMed Scopus (125) Google Scholar, Trabuco et al., 2008Trabuco L.G. Villa E. Mitra K. Frank J. Schulten K. Structure. 2008; 16: 673-683Abstract Full Text Full Text PDF PubMed Scopus (235) Google Scholar, Trabuco et al., 2011Trabuco L.G. Schreiner E. Gumbart J. Hsin J. Villa E. Schulten K. J. Struct. Biol. 2011; 173: 420-427Crossref PubMed Scopus (21) Google Scholar, Wriggers et al., 2000Wriggers W. Agrawal R.K. Drew D.L. McCammon A. Frank J. Biophys. J. 2000; 79: 1670-1678Abstract Full Text Full Text PDF PubMed Google Scholar, Zhang et al., 2011Zhang J. Ma B. DiMaio F. Douglas N.R. Joachimiak L.A. Baker D. Frydman J. Levitt M. Chiu W. Structure. 2011; 19: 633-639Abstract Full Text Full Text PDF PubMed Scopus (18) Google Scholar) while being fit into the map; precautions need to be taken to avoid over-fitting by introducing too many refinable parameters relative to the data available. At higher resolutions (better than 6 Å for a mostly α-helical structure or 4 Å for a mostly β-stranded structure), it may be possible to recognize known folds of protein subunits (Jiang et al., 2001Jiang W. Baker M.L. Ludtke S.J. Chiu W. J. Mol. Biol. 2001; 308: 1033-1044Crossref PubMed Scopus (189) Google Scholar, Khayat et al., 2010Khayat R. Lander G.C. Johnson J.E. J. Struct. Biol. 2010; 170: 513-521Crossref PubMed Scopus (7) Google Scholar, Saha et al., 2010Saha M. Levitt M. Chiu W. Bioinformatics. 2010; 26: i301-i309Crossref PubMed Scopus (7) Google Scholar). In addition to density map features and protein stereochemistry, modeling may also utilize other types of information, such as symmetry, protein proximities from proteomics experiments, residue proximities from chemical cross-linking, related homologous structures, and SAXS profiles (Alber et al., 2008Alber F. Förster F. Korkin D. Topf M. Sali A. Annu. Rev. Biochem. 2008; 77: 443-477Crossref PubMed Scopus (106) Google Scholar). Increasingly, 3DEM maps and models described in the literature are deposited in public archives, where they can be retrieved for independent assessment, use, and development of new tools for visualization, fitting, and validation. EMDataBank, the Unified Data Resource for 3DEM (http://emdatabank.org; Lawson et al., 2011Lawson C.L. Baker M.L. Best C. Bi C. Dougherty M. Feng P. van Ginkel G. Devkota B. Lagerstedt I. Ludtke S.J. et al.Nucleic Acids Res. 2011; 39: D456-D464Crossref PubMed Scopus (72) Google Scholar; Figure 1), provides joint deposition and retrieval of maps in the Electron Microscopy Data Bank (EMDB) archive as well as coordinates of the models fitted into map volumes in the Protein Data Bank (PDB) archive. Currently, more than 1,000 EM maps and more than 400 map-derived models are available (Figure 2).Figure 23DEM Entries in EMDB and PDB, Cumulative by YearShow full captionStatistics for December 31, 2011: 1322 map entries, 427 model entries.View Large Image Figure ViewerDownload Hi-res image Download (PPT) Statistics for December 31, 2011: 1322 map entries, 427 model entries. Every 3DEM map and model has some uncertainty. Therefore, an assessment of map and model errors is essential, especially when a wide range of techniques are used by a variety of practitioners. In addition, as with all rapidly developing fields, in our enthusiasm to go further and faster, there is a risk that avoidable mistakes, both large and small, may be made in the production or interpretation of maps. Such mistakes may have the adverse effect of undermining the credibility of 3DEM methods in general. It is therefore important to develop methods for checking our conclusions and validating maps and models, with the goal of establishing a set of best practices for the field. Historically, the 3DEM field has not made any notorious blunders but, as with all scientific disciplines, a handful of papers have reported erroneous results. While the rarity of these incidents is heartening, they do provide good justification for being cautious. In the early days of electron crystallography, the resolution of published projection maps was sometimes overly optimistic (Hayward and Stroud, 1981Hayward S.B. Stroud R.M. J. Mol. Biol. 1981; 151: 491-517Crossref PubMed Scopus (36) Google Scholar) before the importance of correcting for beam tilt was realized (Henderson et al., 1986Henderson R. Baldwin J.M. Downing K.H. Lepault J. Zemlin F. Ultramicroscopy. 1986; 19: 147-178Crossref Scopus (496) Google Scholar). It has also proved remarkably easy to get the absolute hand wrong even in subnanometer resolution structures (Böttcher et al., 1997Böttcher B. Wynne S.A. Crowther R.A. Nature. 1997; 386: 88-91Crossref PubMed Scopus (540) Google Scholar, Kühlbrandt and Wang, 1991Kühlbrandt W. Wang D.N. Nature. 1991; 350: 130-134Crossref PubMed Scopus (257) Google Scholar, Li et al., 1997Li H. Lee S. Jap B.K. Nat. Struct. Biol. 1997; 4: 263-265Crossref PubMed Scopus (96) Google Scholar, Zhou et al., 2000Zhou Z.H. Dougherty M. Jakana J. He J. Rixon F.J. Chiu W. Science. 2000; 288: 877-880Crossref PubMed Scopus (212) Google Scholar). Images of tilted and untilted specimens together provide all the information needed to correctly determine this property (Belnap et al., 1997Belnap D.M. Olson N.H. Baker T.S. J. Struct. Biol. 1997; 120: 44-51Crossref PubMed Scopus (34) Google Scholar, Cheng et al., 2002Cheng N. Trus B.L. Belnap D.M. Newcomb W.W. Brown J.C. Steven A.C. J. Virol. 2002; 76: 7855-7859Crossref PubMed Scopus (30) Google Scholar, Conway et al., 1997Conway J.F. Cheng N. Zlotnick A. Wingfield P.T. Stahl S.J. Steven A.C. Nature. 1997; 386: 91-94Crossref PubMed Scopus (301) Google Scholar). In the single particle electron microscopy field, five papers between 2002 and 2005 independently reported different structures of the same receptor complex, the 1.3 MDa inositol phosphate receptor, a tetramer responsible for calcium release from the endoplasmic reticulum. Two of the structures were determined in negative stain and three in amorphous ice (da Fonseca et al., 2003da Fonseca P.C. Morris S.A. Nerou E.P. Taylor C.W. Morris E.P. Proc. Natl. Acad. Sci. USA. 2003; 100: 3936-3941Crossref PubMed Scopus (65) Google Scholar, Hamada et al., 2003Hamada K. Terauchi A. Mikoshiba K. J. Biol. Chem. 2003; 278: 52881-52889Crossref PubMed Scopus (65) Google Scholar, Jiang et al., 2002Jiang Q.X. Thrower E.C. Chester D.W. Ehrlich B.E. Sigworth F.J. EMBO J. 2002; 21: 3575-3581Crossref PubMed Scopus (81) Google Scholar, Sato et al., 2004Sato C. Hamada K. Ogura T. Miyazawa A. Iwasaki K. Hiroaki Y. Tani K. Terauchi A. Fujiyoshi Y. Mikoshiba K. J. Mol. Biol. 2004; 336: 155-164Crossref PubMed Scopus (62) Google Scholar, Serysheva et al., 2005Serysheva I.I. Hamilton S.L. Chiu W. Ludtke S.J. J. Mol. Biol. 2005; 345: 427-431Crossref PubMed Scopus (54) Google Scholar). Although the differences between the maps may be partly explained by differences in biochemical preparation, they are more likely due to errors in the structure determination. A more recent cryo-EM study at ∼10 Å (Ludtke et al., 2011bLudtke S.J. Tran T.P. Ngo Q.T. Moiseenkova-Bell V.Y. Chiu W. Serysheva I.I. Structure. 2011; 19: 1192-1199Abstract Full Text Full Text PDF PubMed Scopus (36) Google Scholar), while being substantially different from the three earlier cryo-EM maps, agrees qualitatively with one of the negative stain structures (Hamada et al., 2003Hamada K. Terauchi A. Mikoshiba K. J. Biol. Chem. 2003; 278: 52881-52889Crossref PubMed Scopus (65) Google Scholar), so there is evidence of convergence. Although each of these studies used methods that were the best available at the time, the absence of appropriate validation tools has meant that it was not possible either to prove the structures were correct or to show they were incorrect. Additionally, it is common practice to fit crystal structures or homology models into cryo-EM maps. A discrepancy in model-based interpretation of the 2.5 MDa ryanodine receptor remains unresolved and illustrates the challenge in fitting molecular fragments into low-resolution maps (Serysheva et al., 2008Serysheva I.I. Ludtke S.J. Baker M.L. Cong Y. Topf M. Eramian D. Sali A. Hamilton S.L. Chiu W. Proc. Natl. Acad. Sci. USA. 2008; 105: 9610-9615Crossref PubMed Scopus (60) Google Scholar, Tung et al., 2010Tung C.C. Lobo P.A. Kimlicka L. Van Petegem F. Nature. 2010; 468: 585-588Crossref PubMed Scopus (66) Google Scholar). In the earlier study, a homology model was docked into the map region indicated by antibody labeling, while in the other study, global fitting with an X-ray structure of a fragment was performed. These differences in the protocol were sufficient to completely alter the final model and illustrate the need to not only identify the best-fitting location for a fragment and determine a confidence interval, but to also consider possible conformational variability of the fragment being docked. Given the current rapid increase in the size, productivity, and impact of the 3DEM community, it is timely to suggest guidelines for validating, annotating, and depositing 3DEM maps and map-derived models. There is an opportunity to synergize experimental and computational efforts by bringing together the respective communities. There is a need to establish standards as well as to share software and databases. Similar efforts in the X-ray crystallography (Read et al., 2011Read R.J. Adams P.D. Arendall 3rd, W.B. Brunger A.T. Emsley P. Joosten R.P. Kleywegt G.J. Krissinel E.B. Lütteke T. Otwinowski Z. et al.Structure. 2011; 19: 1395-1412Abstract Full Text Full Text PDF PubMed Scopus (74) Google Scholar), nuclear magnetic resonance (NMR) spectroscopy (http://www.wwpdb.org/workshop/2010/nmr_validation.html), and modeling communities (Schwede et al., 2009Schwede T. Sali A. Honig B. Levitt M. Berman H.M. Jones D. Brenner S.E. Burley S.K. Das R. Dokholyan N.V. et al.Structure. 2009; 17: 151-159Abstract Full Text Full Text PDF PubMed Scopus (48) Google Scholar) can serve as constructive examples. Twenty eight participants from 19 academic institutions worldwide attended a meeting at Rutgers on September 28 and 29, 2010 (http://vtf.emdatabank.org). The participants discussed issues in the computation and validation of 3DEM maps and models as well as ways to strengthen the collaboration between the experimental and modeling communities. The participants' consensus was formulated as specific recommendations, aimed to increase the impact of 3DEM in biology and medicine. On the first day, 14 presentations focused on computing maps from raw 3DEM data, and computing molecular models from maps were given. On the second day, independent "map" and "model" discussion groups were asked to address specific questions related to the deposition and validation of 3DEM maps and models, respectively, report on their findings, and make recommendations for the future. These two discussion groups are referred to as the Map Group and the Model Group respectively throughout the rest of this review. We now summarize the consensus of recommendations reached among the participants of the meeting. The recommendations are concerned with derivation, annotation, archiving, visualization, distribution, and publication of EM maps and models based on the maps. These recommendations are intended to be a starting point for further refinement by a broader community of scientists interested in EM. The Map Group's discussion began by enumerating the attributes of a 3DEM map, among the most important of which are the method used to prepare the sample and the symmetry of the object examined. The sample can be prepared in many different ways, but here we describe the four most frequently used techniques. First, negatively stained samples are usually prepared by adsorption onto a carbon film, followed by washing with a few drops of negative stain, such as 1% uranyl acetate, and then dried, resulting in images where the molecules of interest are seen as low-density regions from which the stain has been excluded. Second, ice-embedded samples are normally prepared by blotting a thin film of a solution containing the molecules of interest, then plunge-freezing the film into liquid ethane at a temperature just above its freezing point (Dobro et al., 2010Dobro M.J. Melanson L.A. Jensen G.J. McDowall A.W. Methods Enzymol. 2010; 481: 63-82Crossref PubMed Scopus (19) Google Scholar, Dubochet et al., 1988Dubochet J. Adrian M. Chang J.J. Homo J.C. Lepault J. McDowall A.W. Schultz P. Q. Rev. Biophys. 1988; 21: 129-228Crossref PubMed Google Scholar). The images in this case show the structures as regions of higher density against a background of vitreous ice. Third, samples, often 2D crystals, can also be deposited on a continuous carbon film and embedded in ice or a medium other than ice or negative stain, such as glucose, trehalose, or tannic acid (Unwin and Henderson, 1975Unwin P.N. Henderson R. J. Mol. Biol. 1975; 94: 425-440Crossref PubMed Scopus (501) Google Scholar). This treatment frequently preserves the high-resolution diffraction order, but contrast matching at low resolution can obscure the molecular envelope. Finally, sections of tissue or other specimens can be prepared either by plastic embedding (Glauert and Lewis, 1998Glauert A.M. Lewis P.R. Biological Specimen Preparation for Transmission Electron Microscopy.in: Glauert A.M. Practical Methods in Electron Microscopy. Princeton University Press, London1998Crossref Google Scholar) or high pressure freezing and cryo-sectioning (Ladinsky, 2010Ladinsky M.S. Methods Enzymol. 2010; 481: 165-194Crossref PubMed Scopus (9) Google Scholar), followed by tomographic data collection and 3D structure determination. Plastic embedding requires care in interpretation due to possible fixation artifacts, and cryo-sectioning can produce compression artifacts; nonetheless, both are widely used and valuable methods. The nature of the maps that are computed from 2D EM images depends principally on the symmetry of the objects being examined. Thus, there are maps for 2D crystal structures that can be obtained using either crystallographic methods (Henderson et al., 1986Henderson R. Baldwin J.M. Downing K.H. Lepault J. Zemlin F. Ultramicroscopy. 1986; 19: 147-178Crossref Scopus (496) Google Scholar) or single-particle approaches (Frank et al., 1988Frank J. Chiu W. Degn L. Ultramicroscopy. 1988; 26: 345-360Crossref PubMed Scopus (16) Google Scholar); maps for helical structures that can be obtained using either Fourier-Bessel methods (Diaz et al., 2010Diaz R. Rice W.J. Stokes D.L. Methods Enzymol. 2010; 482: 131-165Crossref PubMed Scopus (13) Google Scholar) or single-particle approaches (Egelman, 2010Egelman E.H. Methods Enzymol. 2010; 482: 167-183Crossref PubMed Scopus (26) Google Scholar); single-particle maps, including structures with icosahedral symmetry, other point group symmetries or no symmetry (Rochat and Chiu, 2012Rochat R.H. Chiu W. Cryo-electron microscopy and tomography of virus particles.in: Egelman E.H. Comprehensive Biophysics. Elsevier Science, 2012Crossref Scopus (2) Google Scholar); and finally tomogram and sub-tomogram average maps (Schmid and Booth, 2008Schmid M.F. Booth C.R. J. Struct. Biol. 2008; 161: 243-248Crossref PubMed Scopus (46) Google Scholar). It is clear that the community needs validation methods for assessing the accuracy of 3DEM maps. A satisfactory validation method does not yet exist, and its development remains an open research problem. However, there are a number of conditions that are necessary for map validity, as well as some methods that may detect whether a map is incorrect under certain circumstances. The majority of these methods require at least a 3D reconstruction without any post-processing, such as masking or filtration. Most of these methods also require access to some portion of the raw data and metadata used to produce the reconstruction. A few methods require collection of additional data explicitly for validation. Examples of some validation methods are given below. The absolute hand of a structure cannot be determined without either a tilt experiment or sufficient resolution to resolve chiral features directly in the map. Tilt experiments also offer the opportunity to validate the accuracy of the structure as a whole and can help place limits on orientation accuracy. Such methods include random-conical tilt (Radermacher, 1988Radermacher M. J. Electron Microsc. Tech. 1988; 9: 359-394Crossref PubMed Google Scholar), orthogonal tilt (Leschziner and Nogales, 2006Leschziner A.E. Nogales E. J. Struct. Biol. 2006; 153: 284-299Crossref PubMed Scopus (51) Google Scholar), single-particle tomography (Baumeister et al., 1999Baumeister W. Grimm R. Walz J. Trends Cell Biol. 1999; 9: 81-85Abstract Full Text Full Text PDF PubMed Scopus (206) Google Scholar), and tilt-pair parameter plots (Henderson et al., 2011Henderson R. Chen S. Chen J.Z. Grigorieff N. Passmore L.A. Ciccarelli L. Rubinstein J.L. Crowther R.A. Stewart P.L. Rosenthal P.B. J. Mol. Biol. 2011; 413: 1028-1046Crossref PubMed Scopus (48) Google Scholar, Rosenthal and Henderson, 2003Rosenthal P.B. Henderson R. J. Mol. Biol. 2003; 333: 721-745Crossref PubMed Scopus (342) Google Scholar) for which a web-based service is available (https://cryoem.nimr.mrc.ac.uk/software/). The absolute hand can often be established by comparison of the structures of component subunits whose hands have been determined previously. The availability of structures in a number of such subunits, or subunit domains within a complex, can validate or correct the hand determination (Baker et al., 2003Baker M.L. Jiang W. Bowman B.R. Zhou Z.H. Quiocho F.A. Rixon F.J. Chiu W. J. Mol. Biol. 2003; 331: 447-456Crossref PubMed Scopus (23) Google Scholar, Kanamaru et al., 2002Kanamaru S. Leiman P.G. Kostyuchenko V.A. Chipman P.R. Mesyanzhinov V.V. Arisaka F. Rossmann M.G. Nature. 2002; 415: 553-557Crossref PubMed Scopus (208) Google Scholar, Leiman et al., 2004Leiman P.G. Chipman P.R. Kostyuchenko V.A. Mesyanzhinov V.V. Rossmann M.G. Cell. 2004; 118: 419-429Abstract Full Text Full Text PDF PubMed Scopus (135) Google Scholar). In addition, the hand of an icosahedral capsid with chiral surface lattice (such as T = 7l) can be easily distinguished at low resolution by the arrangement of hexameric capsomeres in images of freeze-fracture, metal shadowed particles (Prasad et al., 1993Prasad B.V. Prevelige P.E. Marietta E. Chen R.O. Thomas D. King J. Chiu W. J. Mol. Biol. 1993; 231: 65-74Crossref PubMed Google Scholar). Additional validation methods used in single-particle reconstruction include ensuring agreement between projections of the 3D structure and raw images or (if generated) class-averages, ensuring that reference-free class-averages are fully represented among the set of model projections, and ensuring sufficient coverage of particle orientations (Orlova et al., 1996Orlova E.V. Serysheva I.I. van Heel M. Hamilton S.L. Chiu W. Nat. Struct. Biol. 1996; 3: 547-552Crossref PubMed Scopus (127) Google Scholar, Tang et al., 2007Tang G. Peng L. Baldwin P.R. Mann D.S. Jiang W. Rees I. Ludtke S.J. J. Struct. Biol. 2007; 157: 38-46Crossref PubMed Scopus (374) Google Scholar). These criteria represent necessary but not sufficient criteria for a reliable 3D reconstruction. Map variance and local resolution determination, such as bootstrap-based variance maps (Penczek et al., 2006Penczek P.A. Frank J. Spahn C.M. J. Struct. Biol. 2006; 154: 184-194Crossref PubMed Scopus (74) Google Scholar) and local Fourier Shell Correlation (FSC) measurements (Ménétret et al., 2007Ménétret J.F. Schaletzky J. Clemons Jr., W.M. Osborne A.R. Skånland S.S. Denison C. Gygi S.P. Kirkpatrick D.S. Park E. Ludtke S.J. et al.Mol. Cell. 2007; 28: 1083-1092Abstract Full Text Full Text PDF PubMed Scopus (60) Google Scholar), can provide additional measures to help interpret structures. For maps with resolution better than 20 Å by the 0.5 FSC criterion, RMEASURE (Sousa and Grigorieff, 2007Sousa D. Grigorieff N. J. Struct. Biol. 2007; 157: 201-210Crossref PubMed Scopus (63) Google Scholar) can be used to estimate resolution and signal-to-noise directly from the map based on correlation of neighboring Fourier Transform terms. Possible bias from a starting model or overfitting of noise should also be estimated and statistics provided where possible. Experimentalists should be encouraged to assess their own maps according to the criteria listed above and report the methods that they used when depositing the maps. To help the community as a whole, EMDataBank should develop a table of existing map validation techniques with a description of what experimental data are required for each technique, the circumstances und
Referência(s)