Thematic Meeting, Berlin 2017: Session III

Welcome back to my post-conference note. Let’s jump right into the third session on “Interpreting Experiments Through Molecular Simulations“ including talks by Ceclilia Clementi, Michael Feig and Arianna Fornili on Saturday afternoon and with Shang-Te Danny Hsu, Jana Selent and Massimiliano Bonomi on Sunday morning.

The challenges of computational biophysics aiming to bridge molecular and cellular studies is among others due to the characterization of macromolecular systems by sets of different timescales, separated by large gaps. Celilia Clementi tackles this challenge by combining coherent state analysis and Markov State Modeling. Furthermore, she introduced a theoretical framework for the optimal combination of simulation and experiments in the definition of simplified coarse-grained Hamiltonian protein models. As those simplified models loose information, the combination with experimental data makes them more realistic and provides larger time-scales. She further illustrated the method and mentioned as application the coarse-graine model of FIP35.

Next, Michael Feig investigated protein dynamics and stability with crowders in simulations and experiments. He states protein destabilization in Villin due to crowding, in order to explain the differing results from NMR vs. MD. With Villin under dilute conditions, he observed oligomer formation in MD, whereas transient oligomers form on timescales longer than rotational translation diffusion. Rotational diffusion is slowed down by additional factors. In addition to his study on Villin, he could identify transitions between conformations connecting all states of Bacterial genomic DNA, based on targeted MD.

The last talk of this Saturday was given by Arianna Fornili about identification of rescue sites for protein function. As rescue mutations can be mimicked by drugs, their locations is of interest for drug design wherefore she developed a method. Double Force Scanning (DFS) mimics mutations by external forces, using an elastic network model to represent protein dynamics. In oder to model structural perturbations, linear response theory is used. In detail, she used the Fibonacci lattice for uniform distribution for force vectors. The performance of DFS was tested on p53, predicting 80% non-rescue sides correctly, and on an evolutionary dataset, where 79% of evolutionary rescue sites where predicted correctly.

Subsequently after Ariannas talk and a short coffee break, the first poster session was started. Although it was very crowed (which might have been a live experiments from the organizers, fitting to the last session), posters of high interests have been presented, leaving the presenters no break for a small sip of water. Afterwards, the program was open for all those aiming to explore the city of Berlin. On Sunday morning, the third session was continued with talks from Shang-Te Danny Hsu, Jana Selent and Massimiliano Bonomi.

The first session of this Sunday started with a rather biological topic on the structural basis of substrate recognition and chaperone activity of ribosome-associated trigger factor (TF) regulated by monomer-dimer-equilibrium from Shang-Te Danny Hsu. As the structure contains highly dynamic regions, those parts could not be easily resolved. This was also confirmed by solid state NMR studies, showing ribosome binding-induced conformational change in the ribosome binding domain of TF (TF-RBD). Shang-Te revealed TF substrate specificity by peptide array analysis, originating from recognition of averaged property rather than an exact sequence. Produced SAXS data showed a misfit between those and the crystal structure of TF. Further studies using pulse dipolar ESR spectroscopy revealed multiple dimer configurations of TF, which could be identified by chemical cross-linking. Furthermore, he modeled a dynamical system using several different techniques such as crystallography, NMR, SAXS. Comment: good luck for your PhD student 😉

Next, Jana Selent presented her work on the functional dynamics of the distal C-tail of arrestin. As introduced by her, phosphorylation of the GPCR C-tail triggers the arrestin pre-complex, including a partial C-tail displacement of arrestin. To investigate the role of the distal C-tail, she performed all-atom MD simulations and site-directed mutagenesis studies. She described its conformational space with preference to bind to positively charged residues. Tryptophan-induced dynamic quenching was increased for some residues. Furthermore, she investigated the mechanism of IP6-induced displacement, where IP6 displaces residue 393 to 400 but not further down upon binding to GPCR. As second topic, she investigated the C-edge loop of arrestin. By MD simulations, she showed that C-edge loop of pre-activated arrestin was able to penetrate the membrane, in contrast to activated arrestin. This was confirmed by quenching mutagenesis experiments. Finally, she postulates a step-wise binding process from unbound GPCR over an arrestin-GPCR pre-complex including the displacement of the distal C-tail and the penetration of the C-edge into the membrane, followed by the high affinity complex.

The last talk for this session was hold by Massimiliano Bonomi on integrative structural and dynamical biology with PLUMED-ISDB. As computational and experimental technique have their challenges or errors, a hybrid or integrative method could provide a more realistic view. By providing the module PLUMED-ISDB, he presents a way to investigate heterogeneous systems by including experimental data with a priori information. It uses a Bayesian inference method which accounts for data noise and averaged ensembles. He applied this metainference approach to cryo-EM data, able to explain the data better with a lower resolution as a mixture of dynamics and noise.

At last, he addresses challenges to the community, to those I could not agree more and therefore will end my post with his wishes: More distribution of ensembles of the community like structural models, model populations and protocols; the establishment of robust methods for ensemble comparison and validation; and a way to facilitate comparison of different ensemble modeling approaches by sharing methodologies.




Thematic Meeting, Berlin 2017: Session I & Session II

Just a blink and this fantastic conference was over. Equally my time for writing some posts passed by, which preparation I fairly underestimated. So, now I would like to share my notes with you. Hope you enjoy reading my post-conference posts, to remember the great talks or get an impression, what you missed! Let’s start with session I & II!

After planning this meeting thoroughly together with Helen Berman, Andrea Cavalli and Gerhard Hummer for several years, Kresten Lindorff-Larsen addressed the opening remarks emphasizing what they aimed for this program and how much they looked forward to tackle it. And they were not alone! Finally, the first session on “Disordered Protein Ensembles” started, including talks by Adriaan Bax, Tanja Mittag, Teresa Head-Gordon and Paul Robustelli on Friday and Martin Blackledge and Birthe Kragelund on Saturday morning.
For me it was the first time I heard such comprehensive collection of talks about (intrinsically) disordered proteins (IDP) and this session had a great mixture of experimental methods to computational strategies and the benefit of their conjunction.

Starting with Adriaan Bax, he observed protein folding and misfolding by pressure jump NMR with special focus on HIV-1 Protease. The challenge in there is indeed the pressure but by managing this, he observed pressure-induced fully reversible unfolding including slow exchange time scale. Additionally, monomer exchange faster than stable dimer. He further looked into single residue cis-trans isomerisation. By investigating hydrogen bonding network stability, he sees a significant shift towards low pressure. For Abeta(1-40) fibril formation, real time NMR signals disappear after dropping pressure from 2.4 kbar to 1 bar, which is then heterogeneous, but rapidly reappears after jumping to high pressure. For his third system (Ubiquitin), the showed the influence of pressure by revealing an intermediate state, leaving open that this might also be due to mutations.

Next Tanja Mittag studied the effect of multi-site phosphorylation on conformations of intrinsically disordered proteins, exemplary S. cerevisiae transcription factor Ash1. Ash1 is expected to be collapsed and to expand upon multi-site phosphorylation but SAXS shows only little differences. By following an ensemble optimization method, she identified preferences for expanded conformations to be insensitive to screening of long-range electrostatic interactions, but reacting to the presence of weak local structural preferences. She identifies sequence features, mainly a relationship between proline and charged amino acids, deriving intrinsic sequence code expansion. She argues against chain compaction upon multi-site phosphorylation due to proline isomerization or presence of pSer/Arg or pThr/Arg salt bridges. Additionally, all-atom simulations could reproduce the experimental observations. Finally, she states that a large enough concentration of Prolines distributed along a sequence can buffer changes in net charge per residue while changes in local conformation occur.

After a refreshing coffee break, Teresa Head-Gordon presented new methods for generating and evaluating conformational ensembles. She looked at secondary structure propensities for Abeta42 and Abeta43 and observed differences in N-terminus and central hydrophobic core with REMD simulation, indicating either bad force-fields, poor sampling or both. There she introduced Temperature Cool-Waking (TCW) using annealed importance sampling and could indicate poor sampling. She states that TCW and polarizable forcefields work for folded proteins and are robust for IDPs, whereas Boltzmann priors may be questionable for IDPs. Monte-Carlo side chain ensemble (MC-SCE) has implemented a sophisticated energy function to distinguish different side chain packings based on Boltzmann factor. She concluded that MC-SCE and entropy expansions are informative about improved catalysis.

For the last talk of the first day, Raul Robustelli from D.E. Shaw Research gave insights in ongoing force field development, focusing on ordered and disordered protein states. The problem of current force fields for IDPs is that they tend to being structurally too compact relative to experiments. In this context, he stated the importance of an improved water model, as water dispersion energies have been systematically underestimated. Therefore, he introduced the TIP4P-D water model. Enhanced hydrogen bond potential and ‘fractional’ charges showed improved fits of non-bonded parameters to quantum data. As those analysis fit nicely, still further improvements have to be conducted, e.g. to challenge the over-representation of beta sheets.

With Paul Robustelli, an interesting first day at the “Conformational Ensembles from Experimental Data and Computer Simulations” conference ended and everybody went over to the welcome reception to enjoy a beautiful evening on the terrace outside. Perfect time for some networking. You can find some impression on twitter (@BiophysicalSoc).

Day two started with Martin Blackledges talk about “Large-scale Protein Conformational Dynamics from NMR and Molecular Simulation. From Fundamental Biophysics to Biological Function”. He aims to understand large-scale domain motion in influenza, where temperature is a key factor. Problematically is that its crystal conformation cannot bind to importin alpha. For NMR, its 2 domains do not interact with each other, rather having an open and a closed form. He finds that highly conserved salt bridges are involved in stabilizing the closed formation. He uses chemical exchange saturation transfer (CEST) to measure interconversion rate and population of substates. Using integrated structurally dynamics (NMR, smFRET, SAXS) CEST analyses are supported and dynamic interconversion between open and closed form of H5Na influenza 627 are revealed .
To explain how IDPs interact with their physiological partners, Martin introduced Asteroids, a selection tool of ensemble descriptions of intrinsically disordered systems, which was tested using target ensembles. In order to understand the mechanism of highly dynamic IDPs interactions, first their intrinsic dynamics have to be understood and therefore he developed a physical framework by comparing conformational sampling between 274 -298 K.
Using the ABSURD (average block selection using relaxation data) procedure, he could reproduce the experimental data, overcoming current limitations of MD simulations of IDPs. It identifies ensembles of trajectories of IDPs and can thereby map extent of inter-residue dynamic correlation.

The last talk for this session was held by Birthe Kragelund on dynamics and disorder in class 1 Cytokine receptors. She states that disorder is an important and integral part of membrane proteins as intrinsically disordered regions (IDRs) co-structure them as regulatory platforms with the need of structural information on bound state. Still, they are often not recognized. For membrane proteins, other players come into account. Disordered lipid binding motifs still need to be identified and understood, especially short linear motifs. In general, she combined experimental and computational techniques, including NMR, X-ray scattering, mass spectrometry, simulation and molecular modeling, solving the prolactin receptor monomeric structure. As the transmembrane domain is a weak dimer with highly specific interactions, its study is more challenging. She could identify two conformations, providing more insights into its dynamical function, e.g. that specificity can be modified by one single methyl group. Finally, she describes the multiple transmembrane domain states which are influenced by lipid composition, kinase binding and dimerization.

After the coffee break, the second session started on “Integrative and Hybrid Methods” with Andrej Sali, Alexandre Bonvin and Ji-Joon Song started.

Andrej Sali introduced us his research on integrative structure determination, where he uses experiments, physical theory and statistical inference to maximize accuracy, resolutions, completeness and efficiency. He presented 4 general steps of his integrative modeling platform (IMP), first naming gathering of information by experimental data, statistical inference and physical principles, second designing system representation and scoring, third followed by sampling and finally analysis and validation. Following this workflow for the Spindle Pole Body, he obtained a validated structure including insights into functional implications of the model.

Next, Alexandre Bonvin presented his integrative modeling platform HADDock (high ambiguity driven docking). It incorporates ambiguous and low-resolution data to aid the docking of up to 6 molecules and has a powerful algorithm to handle flexibility at the interface including refinement in explicit solvent. Using RNA-polymerase II, he introduced DisVis for explorative modeling and consistency quantification of information content of distance restraints solely based on geometric considerations. It can provide information about possible interfaces and calculable information to guide modeling but does not account for conformational changes and energetics.

Ji-Joon Song focused with his talk on the human importin4_histone H3/H4 Asfla complex, which is a perfect example for a successful integrative structural approach, as he was able to solve the whole complex by combining several techniques, such as x-ray crystallography, SAXS, EM, Mass spectrometry, biological application and modeling. He also stated that C-importin4 directly interacts with histon H3 peptide via a highly acid patch. The validation of the complex via single particle electron microscopy, small angle x-ray scattering and cross-linking mass spectrometry confirmed conformational flexibility.

Closing his talk, many discussions were extended during a manifold but time-wise rather tight lunch outside at the venue. Enjoying the sun we were waiting for the next session, curious what surprises there might be… You want to know what happened? Check out my next post on the third session about “Interpreting Experiments through Molecular Simulations”!



Berlin calling – let’s get thrilled!

Hey folks! I am Johanna Tiemann and as you can read I got the honor to post some comments and personal perspectives of this Biophysical Society thematic meeting, ‘Conformational Ensembles from Experimental Data and Computer Simulations’ that will be held in Berlin, Germany from the 25th to the 29th of August 2017.

Originally from Munich, Germany, I studied Bioinformatics in Berlin and joined Peter Hildebrand’s lab at Charité Medical University already during my Bachelor and Master thesis. The cluster of knowledge within the lab and institute (e.g. Klaus Peter Hofmann and others) stimulated my fascination about G-protein coupled receptors (GPCR), modeling, simulating and coding. So here I am in my second year PhD that I started in Berlin and now continue in Leipzig. First time I saw the announcement for this meeting I got excited as the Harnack house is located at the green and cozy campus of my Alma Mater, the Free University of Berlin. Although it is rather off the beaten track, it is great to get out of the hectic city center of Berlin and focus on science. As a venue for a conference it is pleasant as you will not be that much distracted by the busy city life but rather can get in touch intensively with each other. If you still have energy to spare after a day of very interesting talks and posters, the night life of Berlin is definitely worth to visit, especially during the weekend (plus there is public transport during the whole night). If you need some advice what to see in Berlin – I am happy to share my experience ;-).

During my time as student assistant and PhD student, I already attended several interesting conferences and workshops but this will be my first BPS meeting and the first time as a blogger at such an event. Of course, I got excited when I noticed the organizing committee, as I already had the pleasure of meeting Andrea Cavalli and Kresten Lindorff-Larsen at a CECAM meeting in Lausanne 2016, leaving no doubt that this conference will be full of highlights with a great mixture of computational and experimental topics.

So what else is there to say about me? I am an enthusiastic, always smiling structural biologist. Within the lab of Peter Hildebrand and with great collaborations, I develop tools for protein modeling (especially loop prediction) and the placement of internal water molecules. Oh, and I like moving pictures, so I put GPCRs and their binding partners in a box and simulate them. To enhance understanding of complex dynamic processes and promote scientific transparency, I want to make analysis and especially visualization of those dynamics with our tools more accessible to others. When I am not driving water molecules in my simulations crazy, I try to see the world while hopping from one conference to another or I can be found abroad with friends, cycling, climbing or trying some new sports. I also like taking pictures, so be warned – you might end up in one with me 😉

Coming to the end, I am looking forward to attending this very interesting BPS meeting and see you all later!

Cheers, Johanna Tiemann