Multiscale Simulations of Ubiquitin Chains : Linkage and Chain Behavior

Cite This

Files in this item

Checksum: MD5:37d29927f27eb76e452e5510b25c6165

BERG, Andrej, 2021. Multiscale Simulations of Ubiquitin Chains : Linkage and Chain Behavior [Dissertation]. Konstanz: University of Konstanz

@phdthesis{Berg2021Multi-53072, title={Multiscale Simulations of Ubiquitin Chains : Linkage and Chain Behavior}, year={2021}, author={Berg, Andrej}, address={Konstanz}, school={Universität Konstanz} }

<rdf:RDF xmlns:dcterms="" xmlns:dc="" xmlns:rdf="" xmlns:bibo="" xmlns:dspace="" xmlns:foaf="" xmlns:void="" xmlns:xsd="" > <rdf:Description rdf:about=""> <dcterms:rights rdf:resource=""/> <dc:language>eng</dc:language> <dc:creator>Berg, Andrej</dc:creator> <dcterms:title>Multiscale Simulations of Ubiquitin Chains : Linkage and Chain Behavior</dcterms:title> <dc:rights>Attribution-NonCommercial-NoDerivatives 4.0 International</dc:rights> <dcterms:hasPart rdf:resource=""/> <dcterms:available rdf:datatype="">2021-03-04T11:56:29Z</dcterms:available> <dspace:hasBitstream rdf:resource=""/> <dc:contributor>Berg, Andrej</dc:contributor> <dspace:isPartOfCollection rdf:resource=""/> <dcterms:isPartOf rdf:resource=""/> <dcterms:abstract xml:lang="eng">A full thermodynamically weighted description of conformational states for all relevant Ub chains under physiological conditions is required for the understanding of the Ub signalling system. Although, such detailed data are not available yet, results presented in this thesis yield valuable knowledge about the dynamic Ub chain behaviour and will facilitate further research on this topic. This was achieved by systematically performing a set of simulations of differently linked Ub oligomers (linkage type) of different chain length (diUb and triUb) on two levels of resolution (atomistic and CG). Furthermore, an analysis framework was developed, implemented and validated, which is suitable to characterise the conformational space of Ub oligomers from massive trajectory data sets as they were required and obtained for diUb and triUb. By this, the impact of linkage type and chain length on the conformation of Ub chains was studied qualitatively and quantitatively. In (Berg et al. 2018) large-scale CG simulations of all differently linked diUb were carried out. An initial version of an analysis framework was presented for the conformational characterisation of diUb by the use of a suitable set of CVs which are sensitive to the relative orientation of two Ub subunits to each other. These CVs are internal and can be used to compare protein-protein interaction patterns from diUb conformations with different linkage position and from models with different levels of resolution (atomistic vs. CG). When used in combination with an MDS based dimensionality reduction technique (Sketch-map), two-dimensional landscapes were obtained for all eight Ub dimers which allowed to compare the conformational space of these diUb topologies on a qualitative, but also quantitative level. From these maps low energy conformations for each linkage type were systematically extracted and back-mapped to the atomistic level as a starting point for additional MD simulations. Since no unphysical behaviour was observed for these back-mapped conformations when an atomistic force field was applied to them, CG results were confirmed to give a reasonable prediction on the conformation of diUb. Based on results from this dual-scale simulation ansatz and in relation to other studies on the conformation of diUb (Castañeda et al. 2016; Kniss et al. 2018), one can conclude, that all Ub dimers, independent of linkage type, have a dynamic conformational behaviour in solution with transitions between multiple meta-stable states. Furthermore, the linkage type was confirmed to have an impact on the conformational space of diUb in solution thus affecting the interface character and stability between the Ub subunits and the nature of the surface exposed to possible binding partners. However, some linkage types appear to be very similar from a conformational point of view, e.g. K6 and K11. Also the distal subunit, i.e. the Ub subunit which is linked by its C-tail to the second Ub, has a linkage type independent residue-wise coverage pattern. This might indicate that the Ub code has a redundant character to some extent and that the very last moiety inside a Ub chain contributes less to the specific recognition of these chains than Ub moieties which are ubiquitylated themself. Based on the finding that some diUb linkage types appear to have a similar conformational space, while other are significantly different, in two most diverse diUb linkage types were selected. For K11 and K27-linked diUb representative structures were extracted and used to verify and interpret experimental NMR data on a residue-by-residue basis. By this, the structural and dynamical features of these two diUb types were investigated and a structural model for each linkage type was obtained. These models show that these two linkage types have indeed very distinct conformational features which might play a crucial role for the specific interaction with potential binding partners, and thus be the basis for their linkage specificity, and finally their involvement in different cellular functionalities. Methods and results, which were presented in , were used as a basis and extended in by CG simulations of two unlinked Ub proteins. A low-dimensional representation (Sketch-map) was obtained for this system, where the two proteins were allowed to obtain configurations without any covalent constrain. Due to the missing covalent bond the conformational space of two unlinked Ub proteins is larger than of each diUb and can therefore be used as basis for the Sketch-map setup, i.e. landmark selection and parameter optimisation. This representation was used to project diUb conformations on top of this larger map, which was obtained from the simulation of two unlinked Ub proteins. By this, it was possible to extend the capabilities of the dimensionality reduction and to study the immediate impact of covalent bond formation on the interaction patterns between two Ub proteins. Interestingly, some linkage types tend to populate and amplify states which are already present in unlinked Ub (K6, K11, K48, K63) but in other cases – K27-linked diUb is the most remarkable candidate – new states do appear which were not significant in unlinked Ub. Such differences might play a role in the mechanism of Ub ligation, i.e. the formation of Ub chains, and in the evolution of differently linked Ub chains as signalling proteins in cellular communication cascades. In the CG data set was further extended by large scale simulations of triUb and three unlinked Ub proteins. To allow the analysis of such massive data sets, as they were obtained for these systems, dimensionality reduction was carried out with EncoderMap, a machine learning autoencoder ANN based technique. Additionally, HDBSCAN clustering was used to identify conformational states and to determine their thermodynamic weight. The hypothesis, that the conformational ensembles of Ub chains have a linkage specific, highly dynamic character, was confirmed for all triUb linkage types, as well (Kniss et al. 2018). By applying EncoderMap and HDBSCAN on data of two unlinked Ub proteins a very detailed picture of Ub aggregation in solution was obtained which served as a basis for a closer examination of protein-protein interaction patterns in diUb and triUb. It turned out, that states found in the conformational maps of two unlinked Ub are relevant in Ub dimers but also in the interaction patterns found between adjacent Ub moieties in triUb. The impact of chain elongation on these interaction patterns is linkage dependent, namely they are altered in triUb compared to diUb if the linkage position is close to the contact interface primarily formed in diUb. This effect appears to be most significant for K6, K11 and K48-linked chains, but not in the case of K63-linked chains. Finally, for the overall conformation of triUb, it was found that all eight linkage types have a tendency to form compact triangular like conformations. However, Ub trimers have also a linkage type dependent fraction of elongated rod like states. Development and implementation of the analysis framework presented in allows a systematic characterisation of large-scale MD trajectory data. Based on this characterisation, simulation results obtained on the CG level were validated on the atomistic level. This indicates, that the Ub oligomer CG model1 yields valuable predictions on the interaction patterns between proteins. Therefore, this CG model, which was tailored for Ub oligomers, should be extended to simulate ubiquitilated proteins and UBDs in the future, although ongoing work of the MARTINI developers and other groups should be considered and integrated (Larsen et al. 2020; Jussupow et al. 2020). Since the linkage type and Ub chain length determines the conformational states of Ub oligomers, selective binding of Ub chains by UBDs is, at least to some extent, presumably driven by the conformational selection mechanism (Lange et al. 2008; Kniss et al. 2018). The linkage type dependent impact of chain elongation, which turned out to be most significant for K6, K11 and K48 linked chains, should be considered in the interpretation of (experimental) results. Very recently, this effect was experimentally studied by Lutz et al. and Jussupow et al. for K27, K29, K33 and M1-linked chains, and it was shown that the impact of chain elongation is linkage type dependent (Lutz et al. 2020; Jussupow et al. 2020). These results and the analytical framework presented in might be a starting point to obtain a more detailed picture about the properties of longer Ub chains in solution. It should be noted, that the conformational space of Ub chains under real physiological conditions (in cells) might be significantly different from the results obtained from simulations of a single Ub chain in CG water. This should be considered and investigated in the future by studying the effect of crowded environments on the protein-protein conformation (Bülow et al. 2019). Furthermore, it might be suitable in the future to apply an experimental data based refinement on the MD ensembles presented in this work (Köfinger et al. 2019). Beyond that, a powerful computational platform was presented to obtain a detailed understanding of multi-domain protein conformations from massive simulated ensembles. The ability to detect and extract conformational states from these ensembles allows to extend the analysis by a MSM, which may yield an even more sophisticated protein-protein interaction model for Ub chains (Husic and Pande 2018). Although optimised and applied to Ub oligomers, all analysis steps should, in principle, work for other systems, as well, which makes it a valuable tool for conformational characterisation of complex biological systems. This will accelerate the investigation of protein-protein conformation in the future and contribute to the understanding of cell (mal)function, which plays a seminal role in the emergence and treatment of diseases in general.</dcterms:abstract> <foaf:homepage rdf:resource="http://localhost:8080/jspui"/> <bibo:uri rdf:resource=""/> <dcterms:issued>2021</dcterms:issued> <dcterms:isPartOf rdf:resource=""/> <dc:date rdf:datatype="">2021-03-04T11:56:29Z</dc:date> <void:sparqlEndpoint rdf:resource="http://localhost/fuseki/dspace/sparql"/> <dspace:isPartOfCollection rdf:resource=""/> </rdf:Description> </rdf:RDF>

Downloads since Mar 4, 2021 (Information about access statistics)

Berg_2-15ig57nsed45e4.pdf 77

This item appears in the following Collection(s)

Attribution-NonCommercial-NoDerivatives 4.0 International Except where otherwise noted, this item's license is described as Attribution-NonCommercial-NoDerivatives 4.0 International

Search KOPS


My Account