Available online at www.sciencedirect.com
ScienceDirect The importance of dynamics in integrative modeling of supramolecular assemblies Giorgio E Tamo`1, Luciano A Abriata1 and Matteo Dal Peraro Revealing the atomistic architecture of supramolecular complexes is a fundamental step toward a deeper understanding of cellular functioning. To date, this formidable task is facilitated by an emerging array of integrative modeling approaches that combine experimental data from different sources. One major challenge these methods have to face is the treatment of the dynamic rearrangements of the individual subunits upon assembly. While this flexibility can be sampled at different levels, integrating native dynamic determinants with available experimental inputs can provide an effective way to reveal the molecular recognition mechanisms at the basis of supramolecular assembly. Address Laboratory for Biomolecular Modeling, Institute of Bioengineering, School of Life Sciences, E´cole Polytechnique Fe´de´rale de Lausanne (EPFL) and Swiss Institute of Bioinformatics (SIB), Lausanne, Switzerland Corresponding author: Dal Peraro, Matteo ([email protected])
Current Opinion in Structural Biology 2015, 31:28–34 This review comes from a themed issue on Theory and simulation Edited by Claire Lesieur and Klaus Schulten
http://dx.doi.org/10.1016/j.sbi.2015.02.018 0959-440X/# 2015 Elsevier Ltd. All rights reserved.
Introduction Supramolecular complexes are the cornerstone of cellular architecture and function. Large assemblies of macromolecules are involved in DNA remodeling, translation and transcription, RNA processing, protein synthesis and degradation, import, export and injection of solutes through cell membranes and to different organelles, ATP synthesis, respiration and photosynthesis, packaging of viral nucleic acids, membrane reshaping, just to name some of the systems best characterized to date. Knowing the structure of these complexes at atomistic resolution is essential to understand how they work and to modulate their function in controlled ways. However, such characterization remains a daunting task for conventional techniques [1]. 1
Whereas the atomic structure of single proteins and complexes of relatively low molecular weight can generally be obtained by NMR spectroscopy and/or X-ray crystallography, supramolecular assemblies are not routinely accessible to these techniques [1]. Although recent advances in cryo-electron microscopy (cryo-EM) are enabling near-atomistic resolution of large assemblies [2], a broad array of experimental methods can provide lower resolution data about overall shape, symmetry, composition, contact sites between constituent molecules, angular and distance restraints between domains, as extensively reviewed in [3]. In this context, integrative modeling attempts to consistently combine these heterogeneous, and sometime incomplete, data with the structures of the individual subunits that constitute a complex in order to generate models at near atomistic resolution [4]. However, as we highlight here, the structure of subunits as determined in isolation might differ from the conformation they adopt upon supramolecular assembly [5], further complicating the prediction of native architectures. Such scenario might arise either from artifacts induced by the conditions under which the structure of individual subunits is solved and/or because of actual structural and functional rearrangements taking place upon assembly. One way or the other, these call for the inclusion in integrative modeling protocols of the global and local dynamics of the individual subunits undergoing macromolecular assembly. Notably, the inclusion of dynamic features cannot only lead to more reliable models, but can also reveal fine details of the assembly mechanism and the existence of multiple functional states. In this opinion we discuss the most recent integrative modeling approaches that attempt to determine supramolecular assembly considering the contribution of these dynamic determinants. We show how canonical integrative modeling approaches start to intersect with molecular simulation techniques leading to new powerful hybrid strategies to unveil the structure and function of large cellular complexes.
The quest for spatial restraints and dynamic determinants Integrative modeling is open to all kinds of experimental restraints. In the most common flavor, high-resolution structures of subunits are complemented by coarser details of the assembly such as size, volume and shape
These authors contributed equally to this work.
Current Opinion in Structural Biology 2015, 31:28–34
www.sciencedirect.com
Dynamics in integrative modeling Tamo`, Abriata and Dal Peraro 29
retrieved from small-angle X-ray or neutron scattering (SAXS/SANS) experiments, cryo-EM or tomography. Despite the tremendous progress of these techniques, these volumetric shapes might often not have the resolution required to unambiguously define position and orientation of constituent molecules. Thus, contact information such as residue–residue contacts unveiled through mutagenesis, chemical cross-linking and ion mobility mass spectrometry [6], or even from coevolution analysis [7–9], can help to define assembly rules. Complementarily, chemical footprinting can point at protein surfaces that must remain exposed [10]. Long-distance information can be retrieved from electron paramagnetic resonance (EPR) experiments on samples with localized radical labels [11], as well as from FRET experiments using native or engineered fluorophores [12,13]. The effect of a localized paramagnetic tag on the nuclear relaxation properties of another molecule can also be used to infer spatial proximity through NMR experiments [14]. All these constraints arising from NMR, EPR and FRET usually reflect ensemble distributions and thus internal dynamics. On the more static, but still very informative end, direct interactions between monomers can be inferred from NOEs and from chemical shift perturbations upon binding [15], whereas residual dipolar couplings provide information about relative angular orientations [16,17]. H/D exchange experiments followed by NMR or by mass spectrometry can in turn provide information about solvent exposure, complementing the information about contacts [18,19]. NMR ensembles of individual subunits can moreover explicitly provide information about internal dynamics and conformational heterogeneity that could be important upon assembly. The same holds for solid-state NMR, with the advantage that it is less limited by molecular size therefore increasingly helping in structure determination for large assemblies, as recently shown for the needle structure of a type III secretion system [20].
Integrating structures, restraints and dynamics into models of supramolecular complexes In the best-case scenario, structures of the individual subunits that constitute the complex are available from experiment or can be modeled with high confidence. However, these structures are often static and representative of single states which conformations might differ from those that fit in a supramolecular complex. This discrepancy can be due to the particular conditions in which the structure of the monomer was determined or may arise from the native dynamics underlying the molecular recognition pathways that lead to assembly through mechanisms as diverse as conformational selection or induced fit. By reviewing key works and strategies in the field, we posit that accounting for the flexible and www.sciencedirect.com
dynamic nature of subunits is a key ingredient of the modeling process, which can not only facilitate supramolecular structure prediction, but can also lead to biologically relevant outcomes. On one hand, some integrative modeling approaches use experimental restraints to drive deformations of the subunits upon assembly, gradually improving the fit to the experimental restraints. This typically involves biasing with the given restraints the algorithms that perform backbone/sidechain refinement, normal modes analysis [21] or molecular dynamics (MD) simulations of variable granularity [5,22] (Figure 1a). The other main avenue consists in using the experimental restraints to select conformations from existing ensembles, namely compiled from X-ray and NMR ensembles, homology models, and/ or MD trajectories (Figure 1b). Whereas the first set of strategies resembles and can potentially describe induced-fit mechanisms underlying protein recognition, the second class of approaches seems well positioned to capture structural assembly driven by conformational selection.
Using experimental restraints to drive deformation of subunits into a supramolecular complex A number of methods use experimental spatial restraints, as obtained from low-resolution experiments, to directly drive the physical deformation of starting structures into consistent conformations (Figure 1a). With a very simple and coarse way to represent protein flexibility, Situs [24] performs very well at fitting assemblies to volumetric maps derived from EM experiments within a broad range of resolution (e.g., up to 30 A˚ [25]). Within this approach flexibility is considered by converting starting structures into a skeleton that is allowed to sample conformations following distributions observed in the Protein Data Bank. In this way the fitting protocol uses a knowledge-based force field on which restraints are implemented to penalize shape differences between the assembled model and the experimental volume [24]. Although the dynamics of individual subunits is not completely sampled, this is a simple way to enlarge the conformational space accessible for assembly, which has been successfully applied to several systems [26–29], most notably myosin fibers [25] and full muscle filaments [30]. Similar ways to adjust individual structures into assemblies make use of normal modes computed from a deformable elastic network, as done by DireX [31] and iMODfit [32,33]. Recently, it has been also shown that the combination of diverse flexible fitting protocols of this kind can improve pseudo-atomistic models based on intermediateresolution EM maps, providing in turn a general way to assess the fits [34]. ATTRACT is another software that makes explicit use of flexibility by exploiting normal mode analysis. Specifically, it performs systematic energy Current Opinion in Structural Biology 2015, 31:28–34
30 Theory and simulation
Figure 1
experimental inputs individual subunits • X-ray crystallography • NMR • cryo-EM • homology models
(a) restraints driving subunits deformation
sidechain/backbone sampling
volumetric maps • cryo-EM • electron tomography • SAXS, SANS
experimental restraints
flexible refinement elastic network models
(b) restraints selecting conformations
spatial connectivity NMR ensembles
.... flexible search
....
ensemble of supramolecular assembly
rmsd (Å)
• mutagenesis • chemical cross-linking/MS • proteomics • FRET • DEER EPR • AFM • H/D exchange • ChlP-seq/exo • 3/4/5C, Hi-C • ...
experimental restraints
time (ns)
MD simulations
flexibility treatment Current Opinion in Structural Biology
Integrative modeling strategies accounting for flexibility and dynamics. Integrative modeling requires as input the structures of the individual subunits composing the supramolecular complex. Structures should usually have high-resolution, but, when not available, coarser molecular representations can also be used, along with any experimental data (e.g., experimental inputs at decreasing levels of resolution are indicated in the orange box). Experimental restraints (e.g., cryo-EM maps in this example) can be used upfront to guide the prediction of supramolecular complexes (a), or, alternatively, conformations can be sampled without restraint biases and then filtered to satisfy them (b). In both approaches the flexibility of individual subunits within the complex can be accounted for with different accuracy, from local refinement of intermolecular interactions, to elastic network based structure deformation, to MD-based exploration of the conformational space (see the blue box). As a final result, integrative modeling methods aim at producing biologically relevant supramolecular assemblies consistent with the available set of experimental data. The protein structures used for this example correspond to the RNA polymerase II complex (PDB 4A3D [23]).
minimization along pre-calculated normal modes [35,36], while loop and side-chain flexibility are included using a multi-copy approach [37]. Recent extensions within this framework allow also to fit proteins into low resolution density maps [38]; moreover, the introduction of coarsegrained elastic network models within replica exchange protocols looks as a promising way to enhance the conformational sampling needed to characterize the dynamics of supramolecular assemblies [39]. A possible limitation of this kind of approaches is that conformations generated by normal mode analysis might not always be physically accurate, but further refinement using force field-based MD simulations can improve over this drawback. Bock et al. [40] used this approach to fit atomic X-ray structures of the Escherichia coli ribosome into a set of cryo-EM density maps captured at different stages of the tRNA translocation process. The subsequent MD analysis of the different states was able to reveal their Current Opinion in Structural Biology 2015, 31:28–34
progression along the tRNA translocation pathway, estimating the timescale of transitions and the key driving forces along the process. In the same vein, DireX has been recently used for the refined flexible fitting of cyclic nucleotide-binding potassium channel crystal structure inside EM density maps corresponding to different conformational states. This helped in elucidating the ligandinduced gating mechanism of this protein [41]. At the most accurate end of the dynamic treatment of assembly structures is the direct use of MD simulations restrained by experimental inputs, which profits from an accurate description of the physico-chemical features of the system. Within this group of methods, the molecular dynamics flexible fitting (MDFF) protocol [42] uses the gradient of the electron density distribution to compute forces that are added to those of a classical force field. This method has been widely applied since its introduction [43,44], one of its most recent and striking www.sciencedirect.com
Dynamics in integrative modeling Tamo`, Abriata and Dal Peraro 31
applications being the assembly of the HIV virus capsid into an 8 A˚ cryo-EM map [45]. Illustrating the importance of incorporating protein dynamics in integrative modeling, assembly of the HIV capsid revealed that deformation of the planar hexamer-of-hexamer and pentamer-of-hexamer capsid units is required to properly describe protein–protein contacts upon capsid assembly. Another related way to use specific restraints to drive deformation of the subunits in MD is to use restrainedensemble simulations [46,47]. In a recent application DEER distance distributions allowed to propose functional conformational states of the leucine transporter that were not evident from the available crystallographic data [48]. Similarly, but using as a guide disulfide scanning cross-linking data instead [49], our laboratory disclosed two functional alternating states of the two-component system PhoQ [50], supporting a combination of scissoring and rotational motions for the transduction of the chemical signal through the inner bacterial membrane. Given the continuous growth of the computational power, the strategy to drive atomistic or coarse-grained MD simulations with experimentally derived spatial data is one of the most promising avenues to improve our understanding of the structure and function of large supramolecular assemblies.
Using experimental restraints to select functional states from a conformational ensemble Dealing with a broad variety of experimental inputs as spatial restraints, including cryo-EM, SAXS, NMR and proteomics data, we find the Integrative Modeling Platform (IMP) [51]. In order to satisfy the spatial restraints in an efficient manner, IMP makes use of a large variety of optimization algorithms [51]. This protocol has been successfully used to model the complex assemblies of the 26S proteasome [52], the nuclear pore complex [53–55] and recently the monooxygenase hydroxylase from Methylococcus capsulatus [56] and the 40S-elF1elF3 translation initiation complex [57]. Despite this framework does not directly consider dynamic features of the complex subunits during model prediction, functional states of large assemblies can be discriminated from the experimental data using Bayesian inference [58,59]. Recently, two distinct functional dynamic states of the two-component system PhoQ were found to optimally fit a set of disulfide cross-linking data, supporting thus a scissoring mode of signal transduction in sensor Hiskinase receptors [60]. Other protocols consider flexibility in a more explicit way, like Rosetta [61], where conformational ensembles can be generated from NMR experiments and Monte Carlo sampling, while further refinement of sidechain and backbone torsional angles is used incrementally to improve the quality of the assembled models [61]. In a www.sciencedirect.com
recent advancement, dubbed ‘fold-and-dock’ [62], it is also possible to simultaneously sample the internal flexibility of individual subunits under symmetry constraints, which allowed folding of a symmetric homodimer from its sequence and available experimental data. Improvements in comparative modeling (i.e., RosettaCM [63]) and the inclusion of density maps as fitting guides for local refinement or more extensive model rebuilding now allow reaching near atomic resolutions when starting from cryoEM maps of 4–10 A˚ resolution [64]. In a recent application, this approach has produced multiple atomistic snapshots based on high-resolution cryo-EM data leading to novel proposals on how GTP hydrolysis leads to microtubule dynamic instability and depolymerization [65]. Following similar premises is the well-established integrative modeling framework HADDOCK [66,67]. Initially designed to employ NMR data, it has also been used with evolutionary data predicting inter-protein contacts [8,68] and adapted to handle SAXS data [69]. Conformational flexibility is here accounted in two main ways; that is, by providing ensembles from NMR structures, collections of X-rays structures or simulation snapshots, rather than static structures to the search algorithm [3], and by performing multidomain docking [70]. This latter scheme, like early programs such as FlexDock [71] and RNAbuilder [44], splits a flexible binding partner into subdomains based on an elastic network model and treats the parts independently, but under the strong constraint that the two halves of a hinge must be spatially close in the complex [70]. Using this recipe, the dimeric structure of ubiquitin bound to the deubiquitinating enzyme Josephin was determined, helping to pinpoint its cleavage site [72]. In another recent example, this protocol has been used to build a complex of the interleukin 1 receptor bound to an antagonist, unveiling a conformational change as large as 20 A˚ upon binding [15]. At the most accurate end of this array of methods (Figure 1b), one can sample subunit intrinsic dynamics using explicit MD simulations to determine conformational states which are relevant for the supramolecular complex, and have an optimization algorithm select the best set of conformations by filtering them so as to satisfy the restraints. The great advantage of such approach is that native physico-chemical determinants are more thoroughly accounted for and conformations are not biased a priori. Such strategy is at the core of the power (parallel optimization workbench for enhancing resolution) framework developed in our laboratory [73,74]. In this scheme, MD trajectories are indexed by principal component analysis and evolutionary algorithms, like particle swarm optimization, are used to efficiently select optimal conformations that fulfill the set of experimental restraints. This strategy has already proved successful for the prediction of large assemblies like the HIV1 hexameric capsomer complex [73], the basal body of Yersinia type Current Opinion in Structural Biology 2015, 31:28–34
32 Theory and simulation
III secretion system [75] and the C4b-binding protein [76]. Noteworthy, power was used to elucidate the assembly mechanism of the pore-forming toxin aerolysin [74], for which unbiased MD simulations were able to extensively characterize activated states of the monomeric toxin along the pore formation pathway, which were then used for the prediction of intermediate prepore and mature transmembrane pore states. This study revealed the role of the subunit’s intrinsic flexibility in driving a concerted swirling mechanism that mediates membrane pore insertion. These specific examples, where flexibility of individual subunits upon assembly is important or multiple functional states may exist, are representative for intrinsic pitfalls of methods that do not account for explicit dynamics or treat subunits at too coarse-grained resolution. When dynamics was not considered, we could only assemble models that poorly fitted the given experimental restraints. On the other side, it is however interesting to ask whether unbiased MD simulations of individual subunits, regardless the omnipresent sampling limitations, are able to capture in isolation conformational states relevant for the bound subunits. In a scenario where conformational selection is the main driving mechanism to assembly this approach is conceptually appropriate. When subunits flexibility is limited, static and/or coarser approaches can be equally successful, however if induced fit mechanisms are playing a significant role upon assembly, strategies including flexible refinement or involving restraint-biased MD aided by enhanced sampling techniques might be a more suitable option. Namely, replicaexchange methods or adaptive tempering can be used when variables relevant for assembly are completely unknown, or methods biased by predefined coordinates when experimental evidence suggests specific motions linked to the assembly process. While some methods might be in principle more appropriate than others given a specific system and certain conditions, eventually the success of any integrative approach much relies on the quality and completeness of the set of available experimental restraints.
approach, like, for instance, incomplete sampling and inaccuracy of force fields, can be softened in an integrative modeling framework by restraints provided by a growing array of experimental methods. Therefore, efficient strategies to include dynamics in integrative modeling protocols will certainly contribute in the future to advance our understanding of the principles underlying molecular recognition of supramolecular assembly, facilitating their computational structural prediction.
Conflict of interest Nothing to declare.
Acknowledgements The Laboratory for Biomolecular Modeling is supported by EPFL School of Life Sciences and the Swiss National Science Foundation (SNSF). LAA thanks EMBO and the Marie Curie actions for a Long-Term postdoctoral fellowship.
References and recommended reading Papers of particular interest, published within the period of review, have been highlighted as: of special interest of outstanding interest 1.
Ward AB, Sali A, Wilson IA: Biochemistry. Integrative structural biology. Science 2013, 339:913-915.
2.
Greber BJ, Boehringer D, Leibundgut M, Bieri P, Leitner A, Schmitz N, Aebersold R, Ban N: The complete structure of the large subunit of the mammalian mitochondrial ribosome. Nature 2014, 515:283-286.
3.
Rodrigues JP, Bonvin AM: Integrative computational modeling of protein interactions. FEBS J 2014, 281:1988-2003.
4.
Thalassinos K, Pandurangan AP, Xu M, Alber F, Topf M: Conformational states of macromolecular assemblies explored by integrative structure calculation. Structure 2013, 21:1500-1508.
5.
Spiga E, Degiacomi MT, Dal Peraro M: New strategies for integrative dynamic modeling of macromolecular assembly. Adv Protein Chem Struct Biol 2014, 96:77-111.
6.
Baldwin AJ, Lioe H, Hilton GR, Baker LA, Rubinstein JL, Kay LE, Benesch JL: The polydispersity of aB-crystallin is rationalized by an interconverting polyhedral architecture. Structure 2011, 19:1855-1863.
7.
Morcos F, Pagnani A, Lunt B, Bertolino A, Marks DS, Sander C, Zecchina R, Onuchic JN, Hwa T, Weigt M: Direct-coupling analysis of residue coevolution captures native contacts across many protein families. Proc Natl Acad Sci U S A 2011, 108:E1293-E1301.
8.
Hopf TA, Scharfe CP, Rodrigues JP, Green AG, Kohlbacher O, Sander C, Bonvin AM, Marks DS: Sequence co-evolution gives 3D contacts and structures of protein complexes. eLife 2014, 3:e03430.
9.
Ovchinnikov S, Kamisetty H, Baker D: Robust and accurate prediction of residue–residue interactions across protein interfaces using evolutionary information. eLife 2014, 3:e02030.
Concluding remarks Integrative modeling has started to be routinely adopted to reach a near-atomic description of large supramolecular assemblies, effectively bypassing the current, intrinsic limitations of individual experimental techniques. The dynamics of individual subunits composing large assemblies has emerged as a fundamental ingredient to capture their biological function. Thus, one of the future challenges of integrative modeling will lie on a proper description of these dynamic determinants. While dynamic changes upon assembly can be sampled in different ways, MD simulation, with its abilities to thoroughly characterize the physics of biomolecules, is perfectly poised to describe the native flexibility of relevant functional states in large complexes. The limitations still affecting this Current Opinion in Structural Biology 2015, 31:28–34
10. Poor TA, Jones LM, Sood A, Leser GP, Plasencia MD, Rempel DL, Jardetzky TS, Woods RJ, Gross ML, Lamb RA: Probing the paramyxovirus fusion (F) protein-refolding event from pre- to postfusion by oxidative footprinting. Proc Natl Acad Sci U S A 2014, 111:E2596-E2605. 11. Georgieva ER, Borbat PP, Ginter C, Freed JH, Boudker O: Conformational ensemble of the sodium-coupled aspartate transporter. Nat Struct Mol Biol 2013, 20:215-221. www.sciencedirect.com
Dynamics in integrative modeling Tamo`, Abriata and Dal Peraro 33
12. Akyuz N, Altman RB, Blanchard SC, Boudker O: Transport dynamics in a glutamate transporter homologue. Nature 2013, 502:114-118.
31. Wang Z, Schroder GF: Real-space refinement with DireX: from global fitting to side-chain improvements. Biopolymers 2012, 97:687-697.
13. Kalinin S, Peulen T, Sindbert S, Rothwell PJ, Berger S, Restle T, Goody RS, Gohlke H, Seidel CA: A toolkit and benchmark study for FRET-restrained high-precision structural modeling. Nat Methods 2012, 9:1218-1225.
32. Lope´z-Blanco JR, Chaco´n P: iMODFIT: efficient and robust flexible fitting based on vibrational analysis in internal coordinates. J Struct Biol 2013, 184:261-270.
14. Koehler J, Meiler J: Expanding the utility of NMR restraints with paramagnetic compounds: background and practical aspects. Prog Nucl Magn Reson Spectrosc 2011, 59:360-389. 15. Karaca E, Bonvin AM: Advances in integrative modeling of biomolecular complexes. Methods 2013, 59:372-381. 16. Lemmin T, Dimitrov M, Fraering PC, Dal Peraro M: Perturbations of the straight transmembrane alpha-helical structure of the amyloid precursor protein affect its processing by gammasecretase. J Biol Chem 2014, 289:6763-6774. 17. Pester O, Barrett PJ, Hornburg D, Hornburg P, Probstle R, Widmaier S, Kutzner C, Durrbaum M, Kapurniotu A, Sanders CR et al.: The backbone dynamics of the amyloid precursor protein transmembrane helix provides a rationale for the sequential cleavage mechanism of gamma-secretase. J Am Chem Soc 2013, 135:1317-1329. 18. Emerson SD, Palermo R, Liu CM, Tilley JW, Chen L, Danho W, Madison VS, Greeley DN, Ju G, Fry DC: NMR characterization of interleukin-2 in complexes with the IL-2Ralpha receptor component, and with low molecular weight compounds that inhibit the IL-2/IL-Ralpha interaction. Protein Sci 2003, 12:811822. 19. Mandell JG, Falick AM, Komives EA: Identification of protein– protein interfaces by decreased amide proton solvent accessibility. Proc Natl Acad Sci U S A 1998, 95:14705-14710. 20. Demers JP, Habenstein B, Loquet A, Kumar Vasa S, Giller K, Becker S, Baker D, Lange A, Sgourakis NG: High-resolution structure of the Shigella type-III secretion needle by solidstate NMR and cryo-electron microscopy. Nat Commun 2014, 5:4976. 21. Bakan A, Meireles LM, Bahar I: ProDy: protein dynamics inferred from theory and experiments. Bioinformatics 2011, 27:15751577. 22. Spiga E, Alemani D, Degiacomi MT, Cascella M, Dal Peraro M: Electrostatic-consistent coarse-grained potentials for molecular simulations of proteins. J Chem Theory Comput 2013, 9:3515-3526. 23. Cheung AC, Sainsbury S, Cramer P: Structural basis of initial RNA polymerase II transcription. EMBO J 2011, 30:4755-4763. 24. Wriggers W, Milligan RA, McCammon JA: Situs: a package for docking crystal structures into low-resolution maps from electron microscopy. J Struct Biol 1999, 125:185-195. 25. Wriggers W: Using Situs for the integration of multi-resolution structures. Biophys Rev 2010, 2:21-27. 26. Abe K, Tani K, Friedrich T, Fujiyoshi Y: Cryo-EM structure of gastric H+, K+-ATPase with a single occupied cation-binding site. Proc Natl Acad Sci U S A 2012, 109:18401-18406. 27. Tolun G, Makhov AM, Ludtke SJ, Griffith JD: Details of ssDNA annealing revealed by an HSV-1 ICP8-ssDNA binary complex. Nucleic Acids Res 2013, 41:5927-5937. 28. Cuervo A, Pulido-Cid M, Chagoyen M, Arranz R, GonzalezGarcia VA, Garcia-Doval C, Caston JR, Valpuesta JM, van Raaij MJ, Martin-Benito J et al.: Structural characterization of the bacteriophage T7 tail machinery. J Biol Chem 2013, 288:26290-26299. 29. Birol M, Enchev RI, Padilla A, Stengel F, Aebersold R, Betzi S, Yang Y, Hoh F, Peter M, Dumas C et al.: Structural and biochemical characterization of the Cop9 signalosome CSN5/ CSN6 heterodimer. PLoS One 2014, 9:e105688. 30. Alamo L, Wriggers W, Pinto A, Bartoli F, Salazar L, Zhao FQ, Craig R, Padron R: Three-dimensional reconstruction of tarantula myosin filaments suggests how phosphorylation may regulate myosin activity. J Mol Biol 2008, 384:780-797. www.sciencedirect.com
33. Lopez-Blanco JR, Garzon JI, Chacon P: iMod: multipurpose normal mode analysis in internal coordinates. Bioinformatics 2011, 27:2843-2850. 34. Pandurangan AP, Shakeel S, Butcher SJ, Topf M: Combined approaches to flexible fitting and assessment in virus capsids undergoing conformational change. J Struct Biol 2014, 185:427-439. 35. May A, Zacharias M: Accounting for global protein deformability during protein–protein and protein–ligand docking. Biochim Biophys Acta 2005, 1754:225-231. 36. May A, Zacharias M: Energy minimization in low-frequency normal modes to efficiently allow for global flexibility during systematic protein–protein docking. Proteins 2008, 70:794-809. 37. Bastard K, Prevost C, Zacharias M: Accounting for loop flexibility during protein–protein docking. Proteins 2006, 62:956-969. 38. de Vries SJ, Zacharias M: ATTRACT-EM: a new method for the computational assembly of large molecular machines using Cryo-EM Maps. PLoS One 2012, 7:e49733. 39. Ostermeir K, Zacharias M: Hamiltonian replica exchange combined with elastic network analysis to enhance global domain motions in atomistic molecular dynamics simulations. Proteins 2014 http://dx.doi.org/10.1002/prot.24695. 40. Bock LV, Blau C, Schroder GF, Davydov II, Fischer N, Stark H, Rodnina MV, Vaiana AC, Grubmuler H: Energy barriers and driving forces in tRNA translocation through the ribosome. Nat Struct Mol Biol 2013, 20:1390-1396. This work revealed at atomistic resolution structures of the ribosome in different states along a step of spontaneous tRNA translocation. Largescale atomistic MD simulations coupled with cryo-EM maps allowed connecting these structures into a kinetic model able to dissect the causality of the translocation mechanism. 41. Kowal J, Chami M, Baumgartner P, Arheit M, Chiu PL, Rangl M, Scheuring S, Schroder GF, Nimigean CM, Stahlberg H: Ligandinduced structural changes in the cyclic nucleotidemodulated potassium channel MloK1. Nat Commun 2014, 5:3106. 42. Trabuco LG, Villa E, Schreiner E, Harrison CB, Schulten K: Molecular dynamics flexible fitting: a practical guide to combine cryo-electron microscopy and X-ray crystallography. Methods 2009, 49:174-180. 43. Li QF, Wanderling S, Paduch M, Medovoy D, Singharoy A, McGreevy R, Villalba-Galea CA, Hulse RE, Roux B, Schulten K et al.: Structural mechanism of voltage-dependent gating in an isolated voltage-sensing domain. Nat Struct Mol Biol 2014, 21:244-252. 44. Flores SC, Sherman MA, Bruns CM, Eastman P, Altman RB: Fast flexible modeling of RNA structure using internal coordinates. IEEE-ACM Trans Comput Biol Bioinform 2011, 8:1247-1257. 45. Zhao G, Perilla JR, Yufenyuy EL, Meng X, Chen B, Ning J, Ahn J, Gronenborn AM, Schulten K, Aiken C et al.: Mature HIV-1 capsid structure by cryo-electron microscopy and all-atom molecular dynamics. Nature 2013, 497:643-646. A complete structure of the HIV capsid was built at atomistic resolution using MDFF based fitting of capsid proteins into a cryo-EM map. MD simulations were crucial to introduce curvature in the hexameric and pentameric subassemblies, while the atomistic treatment was fundamental to reveal native subunits intermolecular contacts, which can be exploited in the future for therapeutic development. 46. Islam SM, Stein RA, McHaourab HS, Roux B: Structural refinement from restrained-ensemble simulations based on EPR/DEER data: application to T4 lysozyme. J Phys Chem B 2013, 117:4740-4754. Current Opinion in Structural Biology 2015, 31:28–34
34 Theory and simulation
47. Roux B, Islam SM: Restrained-ensemble molecular dynamics simulations based on distance histograms from double electron–electron resonance spectroscopy. J Phys Chem B 2013, 117:4733-4739. 48. Kazmier K, Sharma S, Quick M, Islam SM, Roux B, Weinstein H, Javitch JA, McHaourab HS: Conformational dynamics of liganddependent alternating access in LeuT. Nat Struct Mol Biol 2014, 21:472-479. DEER spectroscopy was used to measure distance distributions between spin labels engineered into a leucine/sodium symporter. Analysis of these distributions, using a new restrained-ensemble MD method [46,47], defined a ligand-dependent conformational equilibrium and identified the structural elements that mediate alternating access of the two substrates. The dynamic component of this analysis was essential to propose a new mechanism that differs from what was previously suggested from static X-ray structures. 49. Goldberg SD, Soto CS, Waldburger CD, Degrado WF: Determination of the physiological dimer interface of the PhoQ sensor domain. J Mol Biol 2008, 379:656-665. 50. Lemmin T, Soto CS, Clinthorne G, DeGrado WF, Dal Peraro M: Assembly of the transmembrane domain of E. coli PhoQ histidine kinase: implications for signal transduction from molecular simulations. PLoS Comput Biol 2013, 9:e1002878. 51. Russel D, Lasker K, Webb B, Velazquez-Muriel J, Tjioe E, Schneidman-Duhovny D, Peterson B, Sali A: Putting the pieces together: integrative modeling platform software for structure determination of macromolecular assemblies. PLoS Biol 2012, 10:e1001244. 52. Lasker K, Forster F, Bohn S, Walzthoeni T, Villa E, Unverdorben P, Beck F, Aebersold R, Sali A, Baumeister W: Molecular architecture of the 26S proteasome holocomplex determined by an integrative approach. Proc Natl Acad Sci U S A 2012, 109:1380-1387. 53. Sampathkumar P, Kim SJ, Upla P, Rice WJ, Phillips J, Timney BL, Pieper U, Bonanno JB, Fernandez-Martinez J, Hakhverdyan Z et al.: Structure, dynamics, evolution, and function of a major scaffold component in the nuclear pore complex. Structure 2013, 21:560-571. 54. Alber F, Dokudovskaya S, Veenhoff LM, Zhang W, Kipper J, Devos D, Suprapto A, Karni-Schmidt O, Williams R, Chait BT et al.: The molecular architecture of the nuclear pore complex. Nature 2007, 450:695-701. 55. Shi Y, Fernandez-Martinez J, Tjioe E, Pellarin R, Kim SJ, Williams R, Schneidman D, Sali A, Rout MP, Chait BT: Structural characterization by cross-linking reveals the detailed architecture of a coatomer-related heptameric module from the nuclear pore complex. Mol Cell Proteomics 2014 http:// dx.doi.org/10.1074/mcp.M114.041673. 56. Politis A, Stengel F, Hall Z, Hernandez H, Leitner A, Walzthoeni T, Robinson CV, Aebersold R: A mass spectrometry-based hybrid method for structural modeling of protein complexes. Nat Methods 2014, 11:403-406. 57. Erzberger JP, Stengel F, Pellarin R, Zhang SY, Schaefer T, Aylett CHS, Cimermancic P, Boehringer D, Sali A, Aebersold R et al.: Molecular architecture of the 40S center dot eIF1 center dot eIF3 translation initiation complex. Cell 2014, 158:1123-1130. 58. Habeck M, Rieping W, Nilges M: Weighting of experimental evidence in macromolecular structure determination. Proc Natl Acad Sci U S A 2006, 103:1756-1761. 59. Rieping W, Habeck M, Nilges M: Inferential structure determination. Science 2005, 309:303-306. 60. Molnar KS, Bonomi M, Pellarin R, Clinthorne GD, Gonzalez G, Goldberg SD, Goulian M, Sali A, DeGrado WF: Cys-scanning disulfide crosslinking and bayesian modeling probe the transmembrane signaling mechanism of the histidine kinase, PhoQ. Structure 2014, 22:1239-1251. Disulfide crosslinking data analyzed using Bayesian inference allowed identifying two distinct functional states of two-component system PhoQ.
Current Opinion in Structural Biology 2015, 31:28–34
This work thus showed how optimal fitting of the experimental dataset can discover multiple functional states that are relevant for the conformational landscape of a given supramolecular assembly. 61. Das R, Baker D: Macromolecular modeling with rosetta. Annu Rev Biochem 2008, 77:363-382. 62. DiMaio F, Leaver-Fay A, Bradley P, Baker D, Andre I: Modeling symmetric macromolecular structures in Rosetta3. PLoS One 2011, 6:e20450. 63. Song Y, DiMaio F, Wang RY, Kim D, Miles C, Brunette T, Thompson J, Baker D: High-resolution comparative modeling with RosettaCM. Structure 2013, 21:1735-1742. 64. DiMaio F, Tyka MD, Baker ML, Chiu W, Baker D: Refinement of protein structures into low-resolution density maps using rosetta. J Mol Biol 2009, 392:181-190. 65. Alushin GM, Lander GC, Kellogg EH, Zhang R, Baker D, Nogales E: High-resolution microtubule structures reveal the structural transitions in ab-tubulin upon GTP hydrolysis. Cell 2014, 157:1117-1129. 66. Dominguez C, Boelens R, Bonvin AMJJ: HADDOCK: a protein– protein docking approach based on biochemical or biophysical information. J Am Chem Soc 2003, 125:1731-1737. 67. Kastritis PL, Rodrigues JP, Bonvin AM: HADDOCK(2P2I): a biophysical model for predicting the binding affinity of protein–protein interaction inhibitors. J Chem Inf Model 2014, 54:826-836. 68. van Dijk AD, Ciofi-Baffoni S, Banci L, Bertini I, Boelens R, Bonvin AM: Modeling protein–protein complexes involved in the cytochrome c oxidase copper-delivery pathway. J Proteome Res 2007, 6:1530-1539. 69. Karaca E, Bonvin AMJJ: On the usefulness of ion-mobility mass spectrometry and SAXS data in scoring docking decoys. Acta Crystallogr Sec D Biol Crystallogr 2013, 69:683-694. 70. Karaca E, Bonvin AMJJ: A multidomain flexible docking approach to deal with large conformational changes in the modeling of biomolecular complexes. Structure 2011, 19:555-565. 71. Mashiach E, Schneidman-Duhovny D, Peri A, Shavit Y, Nussinov R, Wolfson HJ: An integrated suite of fast docking algorithms. Proteins 2010, 78:3197-3204. 72. Nicastro G, Todi SV, Karaca E, Bonvin AM, Paulson HL, Pastore A: Understanding the role of the Josephin domain in the PolyUb binding and cleavage properties of ataxin-3. PLoS One 2010, 5:e12430. 73. Degiacomi MT, Dal Peraro M: Macromolecular symmetric assembly prediction using swarm intelligence dynamic modeling. Structure 2013, 21:1097-1106. The native dynamics of individual subunits of a complex is sampled within this approach using explicit MD simulations. This enlarged conformational ensemble is then used to predict symmetrical assemblies satisfying a set of experimental inputs using swarm intelligence optimization. The approach, which has been applied in [74,75], has thus the ability to explore and select states of the subunits that are functionally relevant for the assembly conformation. 74. Degiacomi MT, Iacovache I, Pernot L, Chami M, Kudryashev M, Stahlberg H, van der Goot FG, Dal Peraro M: Molecular assembly of the aerolysin pore reveals a swirling membrane-insertion mechanism. Nat Chem Biol 2013, 9:623-629. 75. Kudryashev M, Stenta M, Schmelz S, Amstutz M, Wiesand U, Castano-Diez D, Degiacomi MT, Munnich S, Bleck CK, Kowal J et al.: In situ structural analysis of the Yersinia enterocolitica injectisome. eLife 2013, 2:e00792. 76. Hofmeyer T, Schmelz S, Degiacomi MT, Dal Peraro M, Daneschdar M, Scrima A, van den Heuvel J, Heinz DW, Kolmar H: Arranged sevenfold: structural insights into the C-terminal oligomerization domain of human C4b-binding protein. J Mol Biol 2013, 425:1302-1317.
www.sciencedirect.com
ScienceDirect The importance of dynamics in integrative modeling of supramolecular assemblies Giorgio E Tamo`1, Luciano A Abriata1 and Matteo Dal Peraro Revealing the atomistic architecture of supramolecular complexes is a fundamental step toward a deeper understanding of cellular functioning. To date, this formidable task is facilitated by an emerging array of integrative modeling approaches that combine experimental data from different sources. One major challenge these methods have to face is the treatment of the dynamic rearrangements of the individual subunits upon assembly. While this flexibility can be sampled at different levels, integrating native dynamic determinants with available experimental inputs can provide an effective way to reveal the molecular recognition mechanisms at the basis of supramolecular assembly. Address Laboratory for Biomolecular Modeling, Institute of Bioengineering, School of Life Sciences, E´cole Polytechnique Fe´de´rale de Lausanne (EPFL) and Swiss Institute of Bioinformatics (SIB), Lausanne, Switzerland Corresponding author: Dal Peraro, Matteo ([email protected])
Current Opinion in Structural Biology 2015, 31:28–34 This review comes from a themed issue on Theory and simulation Edited by Claire Lesieur and Klaus Schulten
http://dx.doi.org/10.1016/j.sbi.2015.02.018 0959-440X/# 2015 Elsevier Ltd. All rights reserved.
Introduction Supramolecular complexes are the cornerstone of cellular architecture and function. Large assemblies of macromolecules are involved in DNA remodeling, translation and transcription, RNA processing, protein synthesis and degradation, import, export and injection of solutes through cell membranes and to different organelles, ATP synthesis, respiration and photosynthesis, packaging of viral nucleic acids, membrane reshaping, just to name some of the systems best characterized to date. Knowing the structure of these complexes at atomistic resolution is essential to understand how they work and to modulate their function in controlled ways. However, such characterization remains a daunting task for conventional techniques [1]. 1
Whereas the atomic structure of single proteins and complexes of relatively low molecular weight can generally be obtained by NMR spectroscopy and/or X-ray crystallography, supramolecular assemblies are not routinely accessible to these techniques [1]. Although recent advances in cryo-electron microscopy (cryo-EM) are enabling near-atomistic resolution of large assemblies [2], a broad array of experimental methods can provide lower resolution data about overall shape, symmetry, composition, contact sites between constituent molecules, angular and distance restraints between domains, as extensively reviewed in [3]. In this context, integrative modeling attempts to consistently combine these heterogeneous, and sometime incomplete, data with the structures of the individual subunits that constitute a complex in order to generate models at near atomistic resolution [4]. However, as we highlight here, the structure of subunits as determined in isolation might differ from the conformation they adopt upon supramolecular assembly [5], further complicating the prediction of native architectures. Such scenario might arise either from artifacts induced by the conditions under which the structure of individual subunits is solved and/or because of actual structural and functional rearrangements taking place upon assembly. One way or the other, these call for the inclusion in integrative modeling protocols of the global and local dynamics of the individual subunits undergoing macromolecular assembly. Notably, the inclusion of dynamic features cannot only lead to more reliable models, but can also reveal fine details of the assembly mechanism and the existence of multiple functional states. In this opinion we discuss the most recent integrative modeling approaches that attempt to determine supramolecular assembly considering the contribution of these dynamic determinants. We show how canonical integrative modeling approaches start to intersect with molecular simulation techniques leading to new powerful hybrid strategies to unveil the structure and function of large cellular complexes.
The quest for spatial restraints and dynamic determinants Integrative modeling is open to all kinds of experimental restraints. In the most common flavor, high-resolution structures of subunits are complemented by coarser details of the assembly such as size, volume and shape
These authors contributed equally to this work.
Current Opinion in Structural Biology 2015, 31:28–34
www.sciencedirect.com
Dynamics in integrative modeling Tamo`, Abriata and Dal Peraro 29
retrieved from small-angle X-ray or neutron scattering (SAXS/SANS) experiments, cryo-EM or tomography. Despite the tremendous progress of these techniques, these volumetric shapes might often not have the resolution required to unambiguously define position and orientation of constituent molecules. Thus, contact information such as residue–residue contacts unveiled through mutagenesis, chemical cross-linking and ion mobility mass spectrometry [6], or even from coevolution analysis [7–9], can help to define assembly rules. Complementarily, chemical footprinting can point at protein surfaces that must remain exposed [10]. Long-distance information can be retrieved from electron paramagnetic resonance (EPR) experiments on samples with localized radical labels [11], as well as from FRET experiments using native or engineered fluorophores [12,13]. The effect of a localized paramagnetic tag on the nuclear relaxation properties of another molecule can also be used to infer spatial proximity through NMR experiments [14]. All these constraints arising from NMR, EPR and FRET usually reflect ensemble distributions and thus internal dynamics. On the more static, but still very informative end, direct interactions between monomers can be inferred from NOEs and from chemical shift perturbations upon binding [15], whereas residual dipolar couplings provide information about relative angular orientations [16,17]. H/D exchange experiments followed by NMR or by mass spectrometry can in turn provide information about solvent exposure, complementing the information about contacts [18,19]. NMR ensembles of individual subunits can moreover explicitly provide information about internal dynamics and conformational heterogeneity that could be important upon assembly. The same holds for solid-state NMR, with the advantage that it is less limited by molecular size therefore increasingly helping in structure determination for large assemblies, as recently shown for the needle structure of a type III secretion system [20].
Integrating structures, restraints and dynamics into models of supramolecular complexes In the best-case scenario, structures of the individual subunits that constitute the complex are available from experiment or can be modeled with high confidence. However, these structures are often static and representative of single states which conformations might differ from those that fit in a supramolecular complex. This discrepancy can be due to the particular conditions in which the structure of the monomer was determined or may arise from the native dynamics underlying the molecular recognition pathways that lead to assembly through mechanisms as diverse as conformational selection or induced fit. By reviewing key works and strategies in the field, we posit that accounting for the flexible and www.sciencedirect.com
dynamic nature of subunits is a key ingredient of the modeling process, which can not only facilitate supramolecular structure prediction, but can also lead to biologically relevant outcomes. On one hand, some integrative modeling approaches use experimental restraints to drive deformations of the subunits upon assembly, gradually improving the fit to the experimental restraints. This typically involves biasing with the given restraints the algorithms that perform backbone/sidechain refinement, normal modes analysis [21] or molecular dynamics (MD) simulations of variable granularity [5,22] (Figure 1a). The other main avenue consists in using the experimental restraints to select conformations from existing ensembles, namely compiled from X-ray and NMR ensembles, homology models, and/ or MD trajectories (Figure 1b). Whereas the first set of strategies resembles and can potentially describe induced-fit mechanisms underlying protein recognition, the second class of approaches seems well positioned to capture structural assembly driven by conformational selection.
Using experimental restraints to drive deformation of subunits into a supramolecular complex A number of methods use experimental spatial restraints, as obtained from low-resolution experiments, to directly drive the physical deformation of starting structures into consistent conformations (Figure 1a). With a very simple and coarse way to represent protein flexibility, Situs [24] performs very well at fitting assemblies to volumetric maps derived from EM experiments within a broad range of resolution (e.g., up to 30 A˚ [25]). Within this approach flexibility is considered by converting starting structures into a skeleton that is allowed to sample conformations following distributions observed in the Protein Data Bank. In this way the fitting protocol uses a knowledge-based force field on which restraints are implemented to penalize shape differences between the assembled model and the experimental volume [24]. Although the dynamics of individual subunits is not completely sampled, this is a simple way to enlarge the conformational space accessible for assembly, which has been successfully applied to several systems [26–29], most notably myosin fibers [25] and full muscle filaments [30]. Similar ways to adjust individual structures into assemblies make use of normal modes computed from a deformable elastic network, as done by DireX [31] and iMODfit [32,33]. Recently, it has been also shown that the combination of diverse flexible fitting protocols of this kind can improve pseudo-atomistic models based on intermediateresolution EM maps, providing in turn a general way to assess the fits [34]. ATTRACT is another software that makes explicit use of flexibility by exploiting normal mode analysis. Specifically, it performs systematic energy Current Opinion in Structural Biology 2015, 31:28–34
30 Theory and simulation
Figure 1
experimental inputs individual subunits • X-ray crystallography • NMR • cryo-EM • homology models
(a) restraints driving subunits deformation
sidechain/backbone sampling
volumetric maps • cryo-EM • electron tomography • SAXS, SANS
experimental restraints
flexible refinement elastic network models
(b) restraints selecting conformations
spatial connectivity NMR ensembles
.... flexible search
....
ensemble of supramolecular assembly
rmsd (Å)
• mutagenesis • chemical cross-linking/MS • proteomics • FRET • DEER EPR • AFM • H/D exchange • ChlP-seq/exo • 3/4/5C, Hi-C • ...
experimental restraints
time (ns)
MD simulations
flexibility treatment Current Opinion in Structural Biology
Integrative modeling strategies accounting for flexibility and dynamics. Integrative modeling requires as input the structures of the individual subunits composing the supramolecular complex. Structures should usually have high-resolution, but, when not available, coarser molecular representations can also be used, along with any experimental data (e.g., experimental inputs at decreasing levels of resolution are indicated in the orange box). Experimental restraints (e.g., cryo-EM maps in this example) can be used upfront to guide the prediction of supramolecular complexes (a), or, alternatively, conformations can be sampled without restraint biases and then filtered to satisfy them (b). In both approaches the flexibility of individual subunits within the complex can be accounted for with different accuracy, from local refinement of intermolecular interactions, to elastic network based structure deformation, to MD-based exploration of the conformational space (see the blue box). As a final result, integrative modeling methods aim at producing biologically relevant supramolecular assemblies consistent with the available set of experimental data. The protein structures used for this example correspond to the RNA polymerase II complex (PDB 4A3D [23]).
minimization along pre-calculated normal modes [35,36], while loop and side-chain flexibility are included using a multi-copy approach [37]. Recent extensions within this framework allow also to fit proteins into low resolution density maps [38]; moreover, the introduction of coarsegrained elastic network models within replica exchange protocols looks as a promising way to enhance the conformational sampling needed to characterize the dynamics of supramolecular assemblies [39]. A possible limitation of this kind of approaches is that conformations generated by normal mode analysis might not always be physically accurate, but further refinement using force field-based MD simulations can improve over this drawback. Bock et al. [40] used this approach to fit atomic X-ray structures of the Escherichia coli ribosome into a set of cryo-EM density maps captured at different stages of the tRNA translocation process. The subsequent MD analysis of the different states was able to reveal their Current Opinion in Structural Biology 2015, 31:28–34
progression along the tRNA translocation pathway, estimating the timescale of transitions and the key driving forces along the process. In the same vein, DireX has been recently used for the refined flexible fitting of cyclic nucleotide-binding potassium channel crystal structure inside EM density maps corresponding to different conformational states. This helped in elucidating the ligandinduced gating mechanism of this protein [41]. At the most accurate end of the dynamic treatment of assembly structures is the direct use of MD simulations restrained by experimental inputs, which profits from an accurate description of the physico-chemical features of the system. Within this group of methods, the molecular dynamics flexible fitting (MDFF) protocol [42] uses the gradient of the electron density distribution to compute forces that are added to those of a classical force field. This method has been widely applied since its introduction [43,44], one of its most recent and striking www.sciencedirect.com
Dynamics in integrative modeling Tamo`, Abriata and Dal Peraro 31
applications being the assembly of the HIV virus capsid into an 8 A˚ cryo-EM map [45]. Illustrating the importance of incorporating protein dynamics in integrative modeling, assembly of the HIV capsid revealed that deformation of the planar hexamer-of-hexamer and pentamer-of-hexamer capsid units is required to properly describe protein–protein contacts upon capsid assembly. Another related way to use specific restraints to drive deformation of the subunits in MD is to use restrainedensemble simulations [46,47]. In a recent application DEER distance distributions allowed to propose functional conformational states of the leucine transporter that were not evident from the available crystallographic data [48]. Similarly, but using as a guide disulfide scanning cross-linking data instead [49], our laboratory disclosed two functional alternating states of the two-component system PhoQ [50], supporting a combination of scissoring and rotational motions for the transduction of the chemical signal through the inner bacterial membrane. Given the continuous growth of the computational power, the strategy to drive atomistic or coarse-grained MD simulations with experimentally derived spatial data is one of the most promising avenues to improve our understanding of the structure and function of large supramolecular assemblies.
Using experimental restraints to select functional states from a conformational ensemble Dealing with a broad variety of experimental inputs as spatial restraints, including cryo-EM, SAXS, NMR and proteomics data, we find the Integrative Modeling Platform (IMP) [51]. In order to satisfy the spatial restraints in an efficient manner, IMP makes use of a large variety of optimization algorithms [51]. This protocol has been successfully used to model the complex assemblies of the 26S proteasome [52], the nuclear pore complex [53–55] and recently the monooxygenase hydroxylase from Methylococcus capsulatus [56] and the 40S-elF1elF3 translation initiation complex [57]. Despite this framework does not directly consider dynamic features of the complex subunits during model prediction, functional states of large assemblies can be discriminated from the experimental data using Bayesian inference [58,59]. Recently, two distinct functional dynamic states of the two-component system PhoQ were found to optimally fit a set of disulfide cross-linking data, supporting thus a scissoring mode of signal transduction in sensor Hiskinase receptors [60]. Other protocols consider flexibility in a more explicit way, like Rosetta [61], where conformational ensembles can be generated from NMR experiments and Monte Carlo sampling, while further refinement of sidechain and backbone torsional angles is used incrementally to improve the quality of the assembled models [61]. In a www.sciencedirect.com
recent advancement, dubbed ‘fold-and-dock’ [62], it is also possible to simultaneously sample the internal flexibility of individual subunits under symmetry constraints, which allowed folding of a symmetric homodimer from its sequence and available experimental data. Improvements in comparative modeling (i.e., RosettaCM [63]) and the inclusion of density maps as fitting guides for local refinement or more extensive model rebuilding now allow reaching near atomic resolutions when starting from cryoEM maps of 4–10 A˚ resolution [64]. In a recent application, this approach has produced multiple atomistic snapshots based on high-resolution cryo-EM data leading to novel proposals on how GTP hydrolysis leads to microtubule dynamic instability and depolymerization [65]. Following similar premises is the well-established integrative modeling framework HADDOCK [66,67]. Initially designed to employ NMR data, it has also been used with evolutionary data predicting inter-protein contacts [8,68] and adapted to handle SAXS data [69]. Conformational flexibility is here accounted in two main ways; that is, by providing ensembles from NMR structures, collections of X-rays structures or simulation snapshots, rather than static structures to the search algorithm [3], and by performing multidomain docking [70]. This latter scheme, like early programs such as FlexDock [71] and RNAbuilder [44], splits a flexible binding partner into subdomains based on an elastic network model and treats the parts independently, but under the strong constraint that the two halves of a hinge must be spatially close in the complex [70]. Using this recipe, the dimeric structure of ubiquitin bound to the deubiquitinating enzyme Josephin was determined, helping to pinpoint its cleavage site [72]. In another recent example, this protocol has been used to build a complex of the interleukin 1 receptor bound to an antagonist, unveiling a conformational change as large as 20 A˚ upon binding [15]. At the most accurate end of this array of methods (Figure 1b), one can sample subunit intrinsic dynamics using explicit MD simulations to determine conformational states which are relevant for the supramolecular complex, and have an optimization algorithm select the best set of conformations by filtering them so as to satisfy the restraints. The great advantage of such approach is that native physico-chemical determinants are more thoroughly accounted for and conformations are not biased a priori. Such strategy is at the core of the power (parallel optimization workbench for enhancing resolution) framework developed in our laboratory [73,74]. In this scheme, MD trajectories are indexed by principal component analysis and evolutionary algorithms, like particle swarm optimization, are used to efficiently select optimal conformations that fulfill the set of experimental restraints. This strategy has already proved successful for the prediction of large assemblies like the HIV1 hexameric capsomer complex [73], the basal body of Yersinia type Current Opinion in Structural Biology 2015, 31:28–34
32 Theory and simulation
III secretion system [75] and the C4b-binding protein [76]. Noteworthy, power was used to elucidate the assembly mechanism of the pore-forming toxin aerolysin [74], for which unbiased MD simulations were able to extensively characterize activated states of the monomeric toxin along the pore formation pathway, which were then used for the prediction of intermediate prepore and mature transmembrane pore states. This study revealed the role of the subunit’s intrinsic flexibility in driving a concerted swirling mechanism that mediates membrane pore insertion. These specific examples, where flexibility of individual subunits upon assembly is important or multiple functional states may exist, are representative for intrinsic pitfalls of methods that do not account for explicit dynamics or treat subunits at too coarse-grained resolution. When dynamics was not considered, we could only assemble models that poorly fitted the given experimental restraints. On the other side, it is however interesting to ask whether unbiased MD simulations of individual subunits, regardless the omnipresent sampling limitations, are able to capture in isolation conformational states relevant for the bound subunits. In a scenario where conformational selection is the main driving mechanism to assembly this approach is conceptually appropriate. When subunits flexibility is limited, static and/or coarser approaches can be equally successful, however if induced fit mechanisms are playing a significant role upon assembly, strategies including flexible refinement or involving restraint-biased MD aided by enhanced sampling techniques might be a more suitable option. Namely, replicaexchange methods or adaptive tempering can be used when variables relevant for assembly are completely unknown, or methods biased by predefined coordinates when experimental evidence suggests specific motions linked to the assembly process. While some methods might be in principle more appropriate than others given a specific system and certain conditions, eventually the success of any integrative approach much relies on the quality and completeness of the set of available experimental restraints.
approach, like, for instance, incomplete sampling and inaccuracy of force fields, can be softened in an integrative modeling framework by restraints provided by a growing array of experimental methods. Therefore, efficient strategies to include dynamics in integrative modeling protocols will certainly contribute in the future to advance our understanding of the principles underlying molecular recognition of supramolecular assembly, facilitating their computational structural prediction.
Conflict of interest Nothing to declare.
Acknowledgements The Laboratory for Biomolecular Modeling is supported by EPFL School of Life Sciences and the Swiss National Science Foundation (SNSF). LAA thanks EMBO and the Marie Curie actions for a Long-Term postdoctoral fellowship.
References and recommended reading Papers of particular interest, published within the period of review, have been highlighted as: of special interest of outstanding interest 1.
Ward AB, Sali A, Wilson IA: Biochemistry. Integrative structural biology. Science 2013, 339:913-915.
2.
Greber BJ, Boehringer D, Leibundgut M, Bieri P, Leitner A, Schmitz N, Aebersold R, Ban N: The complete structure of the large subunit of the mammalian mitochondrial ribosome. Nature 2014, 515:283-286.
3.
Rodrigues JP, Bonvin AM: Integrative computational modeling of protein interactions. FEBS J 2014, 281:1988-2003.
4.
Thalassinos K, Pandurangan AP, Xu M, Alber F, Topf M: Conformational states of macromolecular assemblies explored by integrative structure calculation. Structure 2013, 21:1500-1508.
5.
Spiga E, Degiacomi MT, Dal Peraro M: New strategies for integrative dynamic modeling of macromolecular assembly. Adv Protein Chem Struct Biol 2014, 96:77-111.
6.
Baldwin AJ, Lioe H, Hilton GR, Baker LA, Rubinstein JL, Kay LE, Benesch JL: The polydispersity of aB-crystallin is rationalized by an interconverting polyhedral architecture. Structure 2011, 19:1855-1863.
7.
Morcos F, Pagnani A, Lunt B, Bertolino A, Marks DS, Sander C, Zecchina R, Onuchic JN, Hwa T, Weigt M: Direct-coupling analysis of residue coevolution captures native contacts across many protein families. Proc Natl Acad Sci U S A 2011, 108:E1293-E1301.
8.
Hopf TA, Scharfe CP, Rodrigues JP, Green AG, Kohlbacher O, Sander C, Bonvin AM, Marks DS: Sequence co-evolution gives 3D contacts and structures of protein complexes. eLife 2014, 3:e03430.
9.
Ovchinnikov S, Kamisetty H, Baker D: Robust and accurate prediction of residue–residue interactions across protein interfaces using evolutionary information. eLife 2014, 3:e02030.
Concluding remarks Integrative modeling has started to be routinely adopted to reach a near-atomic description of large supramolecular assemblies, effectively bypassing the current, intrinsic limitations of individual experimental techniques. The dynamics of individual subunits composing large assemblies has emerged as a fundamental ingredient to capture their biological function. Thus, one of the future challenges of integrative modeling will lie on a proper description of these dynamic determinants. While dynamic changes upon assembly can be sampled in different ways, MD simulation, with its abilities to thoroughly characterize the physics of biomolecules, is perfectly poised to describe the native flexibility of relevant functional states in large complexes. The limitations still affecting this Current Opinion in Structural Biology 2015, 31:28–34
10. Poor TA, Jones LM, Sood A, Leser GP, Plasencia MD, Rempel DL, Jardetzky TS, Woods RJ, Gross ML, Lamb RA: Probing the paramyxovirus fusion (F) protein-refolding event from pre- to postfusion by oxidative footprinting. Proc Natl Acad Sci U S A 2014, 111:E2596-E2605. 11. Georgieva ER, Borbat PP, Ginter C, Freed JH, Boudker O: Conformational ensemble of the sodium-coupled aspartate transporter. Nat Struct Mol Biol 2013, 20:215-221. www.sciencedirect.com
Dynamics in integrative modeling Tamo`, Abriata and Dal Peraro 33
12. Akyuz N, Altman RB, Blanchard SC, Boudker O: Transport dynamics in a glutamate transporter homologue. Nature 2013, 502:114-118.
31. Wang Z, Schroder GF: Real-space refinement with DireX: from global fitting to side-chain improvements. Biopolymers 2012, 97:687-697.
13. Kalinin S, Peulen T, Sindbert S, Rothwell PJ, Berger S, Restle T, Goody RS, Gohlke H, Seidel CA: A toolkit and benchmark study for FRET-restrained high-precision structural modeling. Nat Methods 2012, 9:1218-1225.
32. Lope´z-Blanco JR, Chaco´n P: iMODFIT: efficient and robust flexible fitting based on vibrational analysis in internal coordinates. J Struct Biol 2013, 184:261-270.
14. Koehler J, Meiler J: Expanding the utility of NMR restraints with paramagnetic compounds: background and practical aspects. Prog Nucl Magn Reson Spectrosc 2011, 59:360-389. 15. Karaca E, Bonvin AM: Advances in integrative modeling of biomolecular complexes. Methods 2013, 59:372-381. 16. Lemmin T, Dimitrov M, Fraering PC, Dal Peraro M: Perturbations of the straight transmembrane alpha-helical structure of the amyloid precursor protein affect its processing by gammasecretase. J Biol Chem 2014, 289:6763-6774. 17. Pester O, Barrett PJ, Hornburg D, Hornburg P, Probstle R, Widmaier S, Kutzner C, Durrbaum M, Kapurniotu A, Sanders CR et al.: The backbone dynamics of the amyloid precursor protein transmembrane helix provides a rationale for the sequential cleavage mechanism of gamma-secretase. J Am Chem Soc 2013, 135:1317-1329. 18. Emerson SD, Palermo R, Liu CM, Tilley JW, Chen L, Danho W, Madison VS, Greeley DN, Ju G, Fry DC: NMR characterization of interleukin-2 in complexes with the IL-2Ralpha receptor component, and with low molecular weight compounds that inhibit the IL-2/IL-Ralpha interaction. Protein Sci 2003, 12:811822. 19. Mandell JG, Falick AM, Komives EA: Identification of protein– protein interfaces by decreased amide proton solvent accessibility. Proc Natl Acad Sci U S A 1998, 95:14705-14710. 20. Demers JP, Habenstein B, Loquet A, Kumar Vasa S, Giller K, Becker S, Baker D, Lange A, Sgourakis NG: High-resolution structure of the Shigella type-III secretion needle by solidstate NMR and cryo-electron microscopy. Nat Commun 2014, 5:4976. 21. Bakan A, Meireles LM, Bahar I: ProDy: protein dynamics inferred from theory and experiments. Bioinformatics 2011, 27:15751577. 22. Spiga E, Alemani D, Degiacomi MT, Cascella M, Dal Peraro M: Electrostatic-consistent coarse-grained potentials for molecular simulations of proteins. J Chem Theory Comput 2013, 9:3515-3526. 23. Cheung AC, Sainsbury S, Cramer P: Structural basis of initial RNA polymerase II transcription. EMBO J 2011, 30:4755-4763. 24. Wriggers W, Milligan RA, McCammon JA: Situs: a package for docking crystal structures into low-resolution maps from electron microscopy. J Struct Biol 1999, 125:185-195. 25. Wriggers W: Using Situs for the integration of multi-resolution structures. Biophys Rev 2010, 2:21-27. 26. Abe K, Tani K, Friedrich T, Fujiyoshi Y: Cryo-EM structure of gastric H+, K+-ATPase with a single occupied cation-binding site. Proc Natl Acad Sci U S A 2012, 109:18401-18406. 27. Tolun G, Makhov AM, Ludtke SJ, Griffith JD: Details of ssDNA annealing revealed by an HSV-1 ICP8-ssDNA binary complex. Nucleic Acids Res 2013, 41:5927-5937. 28. Cuervo A, Pulido-Cid M, Chagoyen M, Arranz R, GonzalezGarcia VA, Garcia-Doval C, Caston JR, Valpuesta JM, van Raaij MJ, Martin-Benito J et al.: Structural characterization of the bacteriophage T7 tail machinery. J Biol Chem 2013, 288:26290-26299. 29. Birol M, Enchev RI, Padilla A, Stengel F, Aebersold R, Betzi S, Yang Y, Hoh F, Peter M, Dumas C et al.: Structural and biochemical characterization of the Cop9 signalosome CSN5/ CSN6 heterodimer. PLoS One 2014, 9:e105688. 30. Alamo L, Wriggers W, Pinto A, Bartoli F, Salazar L, Zhao FQ, Craig R, Padron R: Three-dimensional reconstruction of tarantula myosin filaments suggests how phosphorylation may regulate myosin activity. J Mol Biol 2008, 384:780-797. www.sciencedirect.com
33. Lopez-Blanco JR, Garzon JI, Chacon P: iMod: multipurpose normal mode analysis in internal coordinates. Bioinformatics 2011, 27:2843-2850. 34. Pandurangan AP, Shakeel S, Butcher SJ, Topf M: Combined approaches to flexible fitting and assessment in virus capsids undergoing conformational change. J Struct Biol 2014, 185:427-439. 35. May A, Zacharias M: Accounting for global protein deformability during protein–protein and protein–ligand docking. Biochim Biophys Acta 2005, 1754:225-231. 36. May A, Zacharias M: Energy minimization in low-frequency normal modes to efficiently allow for global flexibility during systematic protein–protein docking. Proteins 2008, 70:794-809. 37. Bastard K, Prevost C, Zacharias M: Accounting for loop flexibility during protein–protein docking. Proteins 2006, 62:956-969. 38. de Vries SJ, Zacharias M: ATTRACT-EM: a new method for the computational assembly of large molecular machines using Cryo-EM Maps. PLoS One 2012, 7:e49733. 39. Ostermeir K, Zacharias M: Hamiltonian replica exchange combined with elastic network analysis to enhance global domain motions in atomistic molecular dynamics simulations. Proteins 2014 http://dx.doi.org/10.1002/prot.24695. 40. Bock LV, Blau C, Schroder GF, Davydov II, Fischer N, Stark H, Rodnina MV, Vaiana AC, Grubmuler H: Energy barriers and driving forces in tRNA translocation through the ribosome. Nat Struct Mol Biol 2013, 20:1390-1396. This work revealed at atomistic resolution structures of the ribosome in different states along a step of spontaneous tRNA translocation. Largescale atomistic MD simulations coupled with cryo-EM maps allowed connecting these structures into a kinetic model able to dissect the causality of the translocation mechanism. 41. Kowal J, Chami M, Baumgartner P, Arheit M, Chiu PL, Rangl M, Scheuring S, Schroder GF, Nimigean CM, Stahlberg H: Ligandinduced structural changes in the cyclic nucleotidemodulated potassium channel MloK1. Nat Commun 2014, 5:3106. 42. Trabuco LG, Villa E, Schreiner E, Harrison CB, Schulten K: Molecular dynamics flexible fitting: a practical guide to combine cryo-electron microscopy and X-ray crystallography. Methods 2009, 49:174-180. 43. Li QF, Wanderling S, Paduch M, Medovoy D, Singharoy A, McGreevy R, Villalba-Galea CA, Hulse RE, Roux B, Schulten K et al.: Structural mechanism of voltage-dependent gating in an isolated voltage-sensing domain. Nat Struct Mol Biol 2014, 21:244-252. 44. Flores SC, Sherman MA, Bruns CM, Eastman P, Altman RB: Fast flexible modeling of RNA structure using internal coordinates. IEEE-ACM Trans Comput Biol Bioinform 2011, 8:1247-1257. 45. Zhao G, Perilla JR, Yufenyuy EL, Meng X, Chen B, Ning J, Ahn J, Gronenborn AM, Schulten K, Aiken C et al.: Mature HIV-1 capsid structure by cryo-electron microscopy and all-atom molecular dynamics. Nature 2013, 497:643-646. A complete structure of the HIV capsid was built at atomistic resolution using MDFF based fitting of capsid proteins into a cryo-EM map. MD simulations were crucial to introduce curvature in the hexameric and pentameric subassemblies, while the atomistic treatment was fundamental to reveal native subunits intermolecular contacts, which can be exploited in the future for therapeutic development. 46. Islam SM, Stein RA, McHaourab HS, Roux B: Structural refinement from restrained-ensemble simulations based on EPR/DEER data: application to T4 lysozyme. J Phys Chem B 2013, 117:4740-4754. Current Opinion in Structural Biology 2015, 31:28–34
34 Theory and simulation
47. Roux B, Islam SM: Restrained-ensemble molecular dynamics simulations based on distance histograms from double electron–electron resonance spectroscopy. J Phys Chem B 2013, 117:4733-4739. 48. Kazmier K, Sharma S, Quick M, Islam SM, Roux B, Weinstein H, Javitch JA, McHaourab HS: Conformational dynamics of liganddependent alternating access in LeuT. Nat Struct Mol Biol 2014, 21:472-479. DEER spectroscopy was used to measure distance distributions between spin labels engineered into a leucine/sodium symporter. Analysis of these distributions, using a new restrained-ensemble MD method [46,47], defined a ligand-dependent conformational equilibrium and identified the structural elements that mediate alternating access of the two substrates. The dynamic component of this analysis was essential to propose a new mechanism that differs from what was previously suggested from static X-ray structures. 49. Goldberg SD, Soto CS, Waldburger CD, Degrado WF: Determination of the physiological dimer interface of the PhoQ sensor domain. J Mol Biol 2008, 379:656-665. 50. Lemmin T, Soto CS, Clinthorne G, DeGrado WF, Dal Peraro M: Assembly of the transmembrane domain of E. coli PhoQ histidine kinase: implications for signal transduction from molecular simulations. PLoS Comput Biol 2013, 9:e1002878. 51. Russel D, Lasker K, Webb B, Velazquez-Muriel J, Tjioe E, Schneidman-Duhovny D, Peterson B, Sali A: Putting the pieces together: integrative modeling platform software for structure determination of macromolecular assemblies. PLoS Biol 2012, 10:e1001244. 52. Lasker K, Forster F, Bohn S, Walzthoeni T, Villa E, Unverdorben P, Beck F, Aebersold R, Sali A, Baumeister W: Molecular architecture of the 26S proteasome holocomplex determined by an integrative approach. Proc Natl Acad Sci U S A 2012, 109:1380-1387. 53. Sampathkumar P, Kim SJ, Upla P, Rice WJ, Phillips J, Timney BL, Pieper U, Bonanno JB, Fernandez-Martinez J, Hakhverdyan Z et al.: Structure, dynamics, evolution, and function of a major scaffold component in the nuclear pore complex. Structure 2013, 21:560-571. 54. Alber F, Dokudovskaya S, Veenhoff LM, Zhang W, Kipper J, Devos D, Suprapto A, Karni-Schmidt O, Williams R, Chait BT et al.: The molecular architecture of the nuclear pore complex. Nature 2007, 450:695-701. 55. Shi Y, Fernandez-Martinez J, Tjioe E, Pellarin R, Kim SJ, Williams R, Schneidman D, Sali A, Rout MP, Chait BT: Structural characterization by cross-linking reveals the detailed architecture of a coatomer-related heptameric module from the nuclear pore complex. Mol Cell Proteomics 2014 http:// dx.doi.org/10.1074/mcp.M114.041673. 56. Politis A, Stengel F, Hall Z, Hernandez H, Leitner A, Walzthoeni T, Robinson CV, Aebersold R: A mass spectrometry-based hybrid method for structural modeling of protein complexes. Nat Methods 2014, 11:403-406. 57. Erzberger JP, Stengel F, Pellarin R, Zhang SY, Schaefer T, Aylett CHS, Cimermancic P, Boehringer D, Sali A, Aebersold R et al.: Molecular architecture of the 40S center dot eIF1 center dot eIF3 translation initiation complex. Cell 2014, 158:1123-1130. 58. Habeck M, Rieping W, Nilges M: Weighting of experimental evidence in macromolecular structure determination. Proc Natl Acad Sci U S A 2006, 103:1756-1761. 59. Rieping W, Habeck M, Nilges M: Inferential structure determination. Science 2005, 309:303-306. 60. Molnar KS, Bonomi M, Pellarin R, Clinthorne GD, Gonzalez G, Goldberg SD, Goulian M, Sali A, DeGrado WF: Cys-scanning disulfide crosslinking and bayesian modeling probe the transmembrane signaling mechanism of the histidine kinase, PhoQ. Structure 2014, 22:1239-1251. Disulfide crosslinking data analyzed using Bayesian inference allowed identifying two distinct functional states of two-component system PhoQ.
Current Opinion in Structural Biology 2015, 31:28–34
This work thus showed how optimal fitting of the experimental dataset can discover multiple functional states that are relevant for the conformational landscape of a given supramolecular assembly. 61. Das R, Baker D: Macromolecular modeling with rosetta. Annu Rev Biochem 2008, 77:363-382. 62. DiMaio F, Leaver-Fay A, Bradley P, Baker D, Andre I: Modeling symmetric macromolecular structures in Rosetta3. PLoS One 2011, 6:e20450. 63. Song Y, DiMaio F, Wang RY, Kim D, Miles C, Brunette T, Thompson J, Baker D: High-resolution comparative modeling with RosettaCM. Structure 2013, 21:1735-1742. 64. DiMaio F, Tyka MD, Baker ML, Chiu W, Baker D: Refinement of protein structures into low-resolution density maps using rosetta. J Mol Biol 2009, 392:181-190. 65. Alushin GM, Lander GC, Kellogg EH, Zhang R, Baker D, Nogales E: High-resolution microtubule structures reveal the structural transitions in ab-tubulin upon GTP hydrolysis. Cell 2014, 157:1117-1129. 66. Dominguez C, Boelens R, Bonvin AMJJ: HADDOCK: a protein– protein docking approach based on biochemical or biophysical information. J Am Chem Soc 2003, 125:1731-1737. 67. Kastritis PL, Rodrigues JP, Bonvin AM: HADDOCK(2P2I): a biophysical model for predicting the binding affinity of protein–protein interaction inhibitors. J Chem Inf Model 2014, 54:826-836. 68. van Dijk AD, Ciofi-Baffoni S, Banci L, Bertini I, Boelens R, Bonvin AM: Modeling protein–protein complexes involved in the cytochrome c oxidase copper-delivery pathway. J Proteome Res 2007, 6:1530-1539. 69. Karaca E, Bonvin AMJJ: On the usefulness of ion-mobility mass spectrometry and SAXS data in scoring docking decoys. Acta Crystallogr Sec D Biol Crystallogr 2013, 69:683-694. 70. Karaca E, Bonvin AMJJ: A multidomain flexible docking approach to deal with large conformational changes in the modeling of biomolecular complexes. Structure 2011, 19:555-565. 71. Mashiach E, Schneidman-Duhovny D, Peri A, Shavit Y, Nussinov R, Wolfson HJ: An integrated suite of fast docking algorithms. Proteins 2010, 78:3197-3204. 72. Nicastro G, Todi SV, Karaca E, Bonvin AM, Paulson HL, Pastore A: Understanding the role of the Josephin domain in the PolyUb binding and cleavage properties of ataxin-3. PLoS One 2010, 5:e12430. 73. Degiacomi MT, Dal Peraro M: Macromolecular symmetric assembly prediction using swarm intelligence dynamic modeling. Structure 2013, 21:1097-1106. The native dynamics of individual subunits of a complex is sampled within this approach using explicit MD simulations. This enlarged conformational ensemble is then used to predict symmetrical assemblies satisfying a set of experimental inputs using swarm intelligence optimization. The approach, which has been applied in [74,75], has thus the ability to explore and select states of the subunits that are functionally relevant for the assembly conformation. 74. Degiacomi MT, Iacovache I, Pernot L, Chami M, Kudryashev M, Stahlberg H, van der Goot FG, Dal Peraro M: Molecular assembly of the aerolysin pore reveals a swirling membrane-insertion mechanism. Nat Chem Biol 2013, 9:623-629. 75. Kudryashev M, Stenta M, Schmelz S, Amstutz M, Wiesand U, Castano-Diez D, Degiacomi MT, Munnich S, Bleck CK, Kowal J et al.: In situ structural analysis of the Yersinia enterocolitica injectisome. eLife 2013, 2:e00792. 76. Hofmeyer T, Schmelz S, Degiacomi MT, Dal Peraro M, Daneschdar M, Scrima A, van den Heuvel J, Heinz DW, Kolmar H: Arranged sevenfold: structural insights into the C-terminal oligomerization domain of human C4b-binding protein. J Mol Biol 2013, 425:1302-1317.
www.sciencedirect.com
