We develop and apply methods related to molecular dynamics simulations to understand the function and behavior of (bio)molecular systems. Below you find a brief summary of previous and ongoing projects.

Refining protein structures against SAXS/WAXS data using MD simulations

SAXS/WAXS refinement

The interpretation of SAXS/WAXS data in terms of conformational transitions of proteins is challenging due to the low information content of the signals. We are developing methods to refine atomistic protein structures against experimental SAXS, WAXS, and SANS data fully based on explicit-solvent MD. The MD simulation adds a lot of chemical and physical knowledge (or information) to the low-information SWAXS data, thus drastically reducing the risk of overfitting. The method has been awarded with the 2015 Paper of the Year Award by the Biophysical Society and Biophysical Journal. More recently, we embedded our method into a Bayesian framework, and, in collaboration with Jan Lipfert (LMU Munich), we derived the first accurate atomic models of detergent micelles against SAXS data. For more details, and for a modified Gromacs code that implements the SAXS calculations, see here.

Temperature-dependent atomic models of detergent micelles refined against small-angle X-ray scattering data
Miloš T. Ivanović, Linda Bruetzel, Jan Lipfert, and Jochen S. Hub
Angew. Chem. Int. Ed., doi:10.1002/anie.201713303
Bayesian refinement of protein structures and ensembles against SAXS data using molecular dynamics
Roman Shevchuk and Jochen S. Hub, PLoS Comp. Biol., 13, e1005800 (2017) [www] [pdf]
Interpreting solution X-ray scattering data using molecular simulations
Jochen S. Hub, Curr. Opin. Struct. Biol., 49, 18-26 (2018) [www] [pdf]
Interpretation of solution X-ray scattering data by explicit-solvent molecular dynamics, Po-chia Chen and Jochen S. Hub, Biophys. J., 108, 2573–2584 (2015) [pdf] [www] [supporting info] [main text and supporting info]

Viral fusion protein

Lipid-protein interactions of viral fusion proteins

The anchoring of viral fusion proteins at the host membranes depends on the lipid composition. In collaboration with Félix Rey (Institute Pasteur), we investigate how fusion proteins recognize their target membrane by specific lipid-protein interaction sites.

A glycerophospholipid-specific pocket in the RVFV class II fusion protein drives target-membrane insertion
Pablo Guardado-Calvo, Kalina Atkovska, Scott A. Jeffers, Nina Grau, Marija Backovic, Jimena Pérez-Vargas, S.M. de Boer, M.A. Tortorici, Gérard Pehau-Arnaudet, Jean Lepault, Patrick England, Peter J. Rottier, Berend J. Bosch, Jochen S. Hub, Félix A. Rey
Science, 358, 663-667 (2017) [www] [pdf] [supporting info] [supporting movies]

Efficient and reliable free-energy calculations of pore formation in lipid membranes

Membrane pore

The formation pores over lipid membranes plays an important role in various biophysical and biotechnological processes, including membrane fusion, electroporation, or antimicrobial peptide activity. A quantitative understanding of pore formation, however, has remained limited to a lack of reliable free-energy calculations of pore formation. We have developed a new reaction coordinate for pore formation, which overcomes problems with previous methods (such as hysteresis problems). Notably, PMFs along the new reaction coordinate reveal for the first time a barrier for pore nucleation, thus demonstrating metastability of pores in certain membranes, and confirming a nearly 40 year-old hypothesis.

Metastable prepores in tension-free lipid bilayers Christina L. Ting, Neha Awasthi, Marcus Müller, and Jochen S. Hub, Phys. Rev. Lett., 120, 128103 (2018) [www] [pdf] [supporting info] [supporting movie]
Simulations of pore formation in lipid membranes: reaction coordinates, convergence, hysteresis, and finite-size effects, Neha Awasthi and Jochen S. Hub, J. Chem. Theory Comput., 12, 3261-3269 (2016) [www] [pdf]
Probing a continuous polar defect: A reaction coordinate for pore formation in lipid membranes, Jochen S. Hub and Neha Awasthi, J. Chem. Theory Comput., 13, 2352-2366 (2017) [www] [pdf]

Structural details of protein/detergent complexes

SAXS/WAXS refinement

Membrane proteins are frequently solubilized in a corona of detergent molecules for various types of experiments. The structural parameters of such protein/detergent complexes (PDCs) could in principle be extracted from SAXS data. We showed that explicit-solvent SAXS calculations are capable of extracting the number of detergent molecules from experimental data. In contrast, implicit-solvent methods tend to overfit the hydration layer density, thereby loosing information on the correct detergent number. Remarkably, we also found that an MD ensemble yields better agreement with experimental data than any single structure.

Structural Properties of Protein-Detergent Complexes from SAXS and MD Simulations, Po-chia Chen and Jochen S. Hub, J. Phys. Chem. Lett., 6, 5116–5121 (2015) [www] [pdf] [supporting info]

How (in)homogeneous are cholesterol-containing membranes?

Cholesterol domain

Cholesterol modulates many membrane properties such as permeability, propensity of pore formation, bending moduli, etc. To understand the properties of macroscopic cholesterol-containing membranes, knowledge about the lateral distribution (or lateral inhomogeneity) of cholesterol is mandatory. However was known about how inhomogeneous cholesterol-containing membranes are, or how unsaturations of the phospholipid might modulate the inhomogeneity of the membrane. Using a larges set of coarse-grained simulations, we have systematically quantified the lateral inhomogeneity of binary cholesterol-containing membranes, and deciphered the role of lateral entropy vs. cholester-cholesterol interaction in the lateral ordering. Surprisingly, we found that the unsaturation of the phospholipid has only a small effect on the lateral inhomogeneity.

Quantifying lateral inhomogeneity of cholesterol-containing membranes, Celsa Díaz-Tejada, Igor Ariz-Extreme, Neha Awasthi, and Jochen S. Hub, J. Phys. Chem. Lett., 6, 4799-4803 (2015) [www] [pdf] [supporting info]

Anisotropic WAXS patterns from explicit-solvent MD

Anisotropic WAXS

Time-resolved solution scattering experiments have reached sub-picosecond time resolution, thanks to X-ray free-electron lasers. The wide-angle X-ray scattering patterns detected by such experiments are anisotropic because excitation probability of photoactive proteins (used in such experiments) depend on the orientation of the protein with respect to the excitation laser. The anisotropy of the patterns contains valuable additional information, which could not be used so far because there was no method to compute anisotropic WAXS patterns from structural models or proteins. We have closed this gap and developed a method that predicts anisotropic WAXS patterns from explicit-solvent MD.

Anisotropic time-resolved solution X-ray scattering patterns from explicit-solvent molecular dynamics, Levin U.L. Brinkmann and Jochen S. Hub, J. Chem. Phys., 143, 104108 (2015) [www] [pdf]

Accurate predictions of SAXS/WAXS patterns from explicit-solvent MD

WAXS curves

Small- and wide-angle X-ray scattering (SAXS/WAXS) is a well-established technique providing structural information on proteins in solution. However, the interpretation of the SAXS/WAXS patterns is difficult because the data has only a low information content. In addition, the hydration layer and excluded solvent contribute to the scattering signals. Such solvent contributions are typically fitted against the experimental data, thereby (i) further reducing the amount of information available to draw structural conclusions; and (ii) increasing the risk of overfitting. We have therefore developed methods for the prediction of SWAXS curves from explicit-solvent MD simulations, which do not require any solent-related fitting parameters and also account for effects from thermal fluctuations.

We have set up a web server for the calculation and fitting of SAXS/WAXS curves based on explicit-solvent molecular dynamics simulations. The new service, termed WAXSiS (WAXS in Solvent), is available at:

Christopher J. Knight and Jochen S. Hub, Nucleic Acids Res., 43, W225-W230 (2015) [www] [pdf]
Po-chia Chen and Jochen S. Hub, Validating solution ensembles from molecular dynamics simulation by wide-angle X-ray scattering data, Biophys. J. 107, 435-447 (2014)

Solute permeation across cholesterol-containing membranes

Small molecule inside a membrane

Most molecules including drugs cross biological membrane by passive diffusion across the lipid bilayer. We have systematically studied the energetics and the molecular interactions involved in permeation across cholesterol-containing membranes, as cholesterol in abundant in all animals membranes. The simulations suggest that cholesterol reduced the partitioning into the bilayer more strongly than previously expected. This effect is achieved by the formation of large Van-der-Waals contacts between cholesterol and phospholipid tails, which much break upon an permeation event. Equilibrated structures of cholesterol-containing membranes are available for download here.

Large Influence of Cholesterol on Solute Partitioning into Lipid Membranes, Christian L. Wennberg, David van der Spoel, and Jochen S. Hub
J. Am. Chem. Soc. 134, 5351–5361 (2012) [pdf] [www] [supporting info]
Local partition coefficients govern solute permeability of cholesterol-containing membranes , Florian Zocher, David van der Spoel, Peter Pohl, and Jochen S. Hub, Biophys. J., 105, 2760-2770 (2013) [pdf] [www]

Should I add counter ions to neutralize my simulation?

PME artefacts with background charge

The particle-mesh Ewald (PME) method is the standard method used to compute the electrostatic interactions during MD simulations. A feature of PME is that it adds a uniform background charge if the simulation system is not neutralized by counter ions. Is this a problem? Or is a uniform background charge even desirable because it might mimic a converged distribution of counter ions? And does it have any effect at all - a uniform charge density has no gradients and should hence not induce any trouble? We clarified that a uniform background charge may impose severe artefacts in inhomogeneous systems, such as a lipid membrane surrounded by water. Indeed, the uniform background charge is highly unphysical because it adds counter charge density also to regions of low dielectric, where no counter charge should be. We present a set of simple equations that quantify these artefacts, allowing the user to decide if such artefacts are an acceptable approximation for their system.

Quantifying artifacts in Ewald simulations of inhomogeneous systems with a net charge, Jochen S. Hub, Bert de Groot, Helmut Grubmüller, and Gerrit Groenhof, J. Chem. Theory Comput., 10, 381-390 (2014) [pdf] [www]

Ions and small molecules at the water surface

Rhesus protein simulation box

Ions like water. Ions typically tend to become fully solvated because the large water dipoles can orient with respect to an ionic charge, allowing strong electrostatic interactions. However, recent simulations and later experiments indicated remarkable exceptions: large halide ions, that is Cl, Br, and I instead prefer to rest on the water/air interface (see image). Whether H3O+ or OH are surface bound or not is controversial. Using MD simulations and polarizable ion and water models we derived the full energetics of the solvation of a set of ions into a water droplet. In contrast to previous assumptions we found that large halide ions and H3O+ are driven to the surface enthalpically by water-water (and not by ion-water) interactions, whereas F and OH is driven into bulk water by entropy. More recently, we extended the analysis to organic molecules that prefer to be solvated at the water surface, showing that surface solvation can enhance the solubility of such molecules by orders of magnitude.

· Carl Caleman, Jochen S. Hub, Paul J. van Maaren, and David van der Spoel, Atomistic simulation of ion solvation in water explains surface preference of halides, Proc. Natl. Acad. Sci. USA 108, 6838-6842 (2011) [pdf] [www] [supporting info]
· Jochen S. Hub, Maarten Wolf, Carl Caleman, Paul van Maaren, Gerrit Gronhof, and David van der Spoel, Thermodynamics of hydronium and hydroxide surface solvation Chemical Science, in press [www]
· Jochen S. Hub, Carl Caleman, and David van der Spoel, Organic molecules on the surface of water droplets - an energetic perspective PCCP, DOI: 10.1039/C2CP40483D [www]

Selectivity of Rhesus protein channels

Rhesus protein simulation box

The Rhesus (Rh) family of membrane channel proteins have been known for a long time as the blood group antigens on the red blood cell surface (“Rhesus factor”), yet their physiological role remains poorly understood. Rhesus proteins are homologue to bacterial ammonia transporters (Amt) and have therefore been implicated in ammonia transport, but recent experiments also proposed their role as carbon dioxide channels. We have studied the energetics and rates for ammonia, carbon dioxide, and water flux across a Rhesus channel using MD simulations. The simulations provide a detailed quantitative picture of solute permeation across the Rhesus channel and lipid membranes, and they allow one to detect the significance of channel-mediated flux versus passive flux across the lipid bilayer.

J.S. Hub, F.K. Winkler, M. Merrick, and B.L. de Groot, Potentials of mean force and permeabilities for carbon dioxide, ammonia, and water flux across a Rhesus protein channel and lipid membranes, J. Am. Chem. Soc. 132, 13251-13263 (2010) [pdf] [www] [ supporting info]

Allosteric transitions of Hemoglobin


Understanding the allosteric transition in hemoglobin is one of the grand challenges in biophysics, and has drawn tremendous scientific research for many decades. Yet the molecular mechanisms underlying the allosteric interactions in hemoglobin are still not understood. Armed with nowadays computer power, we have observed for the first time full quaternary T-R transitions of hemoglobin in full-atomistic simulations. The simulations provide a detailed picture on the interplay between the quaternary and the subunit transitions.

J.S. Hub, M.B. Kubitzki, and B.L. de Groot, Quaternary and tertiary T-R transitions of human hemoglobin in molecular dynamics simulations, PLoS Comp Biol 6(5), e1000774 (2010) [pdf] [supporting info] [www]

Identifying collective protein motions related to protein function

Collective motions

Proteins are flexible nanomachines that frequently accomplish their biological function by collective atomic motions. Such motions may be characterized by hinge, shear, or rotational motions of entire protein domains, loop movements, or subtle rearrangements of amino acid side chains. In many cases it is far from obvious how collective motions are related to a particular biological task. Therefore, have developed a novel technique termed Functional Mode Analysis (FMA) that aims to identify the collective atomic motion that is directly related to a specific protein function.

More details and the source code of an FMA implementation are provided here, on a separate FMA website

J.S. Hub and B.L. de Groot, Detection of functional modes in protein dynamics, PLoS Comp Biol, 5(8), e1000480 (2009) [download]

The selectivity mechanism of aquaporins


Aquaporins are universal water channels in biological membranes. Molecular dynamics simulations have contributed significantly to our current understanding of the mechanism of water permeation and selectivity against protons. In this project we aim at understanding the molecular determinants underlying the mechanism of selectivity: why are some water channels solely permeable to water, whereas others may also permeate other small solutes, gases, or even ions? More recently, we found that water flux across aquaporins can be regulated by an electrostatic membrane potential.

· J.S. Hub, C. Aponte-Santamaría, H. Grubmüller, and B.L. de Groot Voltage-regulated water flux through aquaporin channels in silico, Biophys. J. 99(12), L97-L99 (2010) [www]
· J.S. Hub and B.L. de Groot, Mechanism of selectivity in aquaporins and aquaglyceroporins, Proc. Natl. Acad. Sci. USA 105:1198-1203 (2008) [download]
· E.M. Müller, J.S. Hub, H. Grubmüller, and B.L. de Groot, Is TEA an Inhibitor for human Aquaporin-1?, Pflügers Arch 456:663-669 (2008) [download]
· J.S. Hub and B.L. de Groot, Does CO2 permeate through Aquaporin-1? , Biophys J 91, 842-848 (2006) [download]
· J.S. Hub, H. Grubmüller, and B.L. de Groot, Dynamics and Energetics of Permeation through Aquaporins. What do we learn from Molecular Dynamics Simulations?, in 'Aquaporins', E. Beitz (ed.), Handbook of Experimental Pharmacology 190, Springer 2009 [pdf]
· C. Aponte-Santamaría, J. S. Hub, and Bert L. de Groot, Dynamics and energetics of solute permeation through the Plasmodium falciparum aquaglyceroporin, Phys Chem Chem Phys 12, 10246-10254 (2010) [www]

Static and dynamic properties of lipid bilayers

Lipid membrane

In collaboration with the Institute for x-ray physics at the University of Göttingen we investigate short range structure and collective dynamics of lipid bilayer membranes. As x-ray experiments usually probe only the intensity of the Fourier transform of the electron density of the lipid bilayer, MD simulations can complement such experiments as they provide (a) complete information in both Fourier and real space, and (b) enable to decompose the data into contributions from, e.g. the lipid tails, the head groups, and water. Finally, critical comparison of experimental and simulation data can support and stimulate further development of lipid force fields.

J.S. Hub, T. Salditt, M.C. Rheinstädter, and B.L. de Groot, Short range order and collective dynamics of DMPC bilayers. A comparison between molecular dynamics simulations, x-ray, and neutron scattering experiments , Biophys J 93, 3156-3168 (2007) [download]