Abstract Motivation Hydrogen bonds (H-bonds) play an essential role for many molecular interactions but are also often transient, making visualising them in a flexible system challenging. Results We provide pyHVis3D which allows for an easy to interpret 3D visualisation of H-bonds resulting from molecular simulations. We demonstrate the power of pyHVis3D by using it to explain the changes in experimentally measured binding affinities for three T-cell receptor/peptide/MHC complexes and mutants of each of these complexes. Availability and implementation pyHVis3D can be downloaded for free from http://opig.stats.ox.ac.uk/resources. Contact firstname.lastname@example.org Supplementary information Supplementary data are available at Bioinformatics online. 1 Introduction Hydrogen bonds (H-bonds) are non-covalent interactions (5 to 30 kJ/mol) that play an essential role in the stabilisation of protein, DNA and RNA structures as well as interfaces between them. An H-bond is an electrostatic attraction occurring between a hydrogen atom (H) bound to an electronegative atom (donor) and another electronegative atom (acceptor). Distance and angle constraints are commonly used to determine the presence of an H-bond [e.g. DSSP (Kabsch and Sander, 1983) or GROMACS (Pronk et al., 2013)]. For example, a frequently used distance cut-off between acceptor and donor is < 3.5 Å and an angle between hydrogen––donor–acceptor < 30° (Pronk et al., 2013). As proteins are dynamic structures the existence of an H-bond can be transient and is not a binary decision. Computational techniques such as Molecular Dynamics (MD) simulations can now provide us with dynamic views of biological systems but visualising H-bond information from these techniques is currently not straightforward. Here, we present pyHVis3D, a convenient way to illustrate H-bonds deduced from molecular simulations. Our approach allows the visualisation of individual trajectories as well as the comparison of mutants. We illustrate the usability of our approach by explaining experimentally measured binding affinities in the T-cell receptor (TCR), peptide and Major Histocompatibility Complexes (MHC) interface. 2 Materials and methods Any simulation package can be used to create all atom molecular simulations of the system of interest. The output trajectory format needs to be *.xtc, or *.pdb. We have implemented a Python 3 based software package to calculate pair-wise H-bonds between all atoms of all frames of the simulation trajectory. A grid-based algorithm calculates an n*n matrix, where n is the number of donor/acceptor atoms of the simulation and each matrix element contains the average presence of an H-bond between two atoms over time. The matrix can be compressed on a per residue basis to save memory and inflated for all atoms for representation purposes. Alternatively only user specified chains and/or residues can be analysed. If the aim is to compare two simulations then the above procedure is carried out for both trajectories and a difference matrix can be calculated. For each matrix element above a user specified threshold, a 3-dimensional cylinder is drawn by Python’s matplotlib. The diameter of the cylinder represents the presence of an H-bond over all frames i.e. a value of 0 Å means that an H-bond is never present and is not visualised and 1 Å means that an H-bond is present in 100% of all frames. The value can exceed 1 Å if visualisation is done on a per residue instead of per atom basis and two amino acids have on average more than one H-bond between them. In the case of a difference matrix, a blue cylinder indicates more H-bonds for simulation one (e.g. wild-type) while a red cylinder indicates more H-bonds for the other simulation (e.g. mutant). The relative difference between wild-type and mutant is illustrated i.e. the cylinder can only be red or blue but not both. The radius of the cylinder is proportional to the amount of difference. The radius is 0 Å if an H-bond is equally present in the wild-type and mutation simulation and the radius is 1 Å if one H-bond is present 100% of the time in one of the sets but 0% in the other set. In addition a heatmap of the H-bond matrix is shown and a text file containing numerical values (including a significance test; Supplementary Material) about each H-bond is given. For high quality 3D representations a VMD (Humphrey et al., 1996) readable file containing 3D plotting commands is created. PyHVis3D can be run via an easy-to-use graphical user interface or directly via python commands to allow batch processing. 3 H-bonds in the TCR/peptide/MHC interface The interaction between T-cells and antigen presenting cells is a crucial process in the human immune system. Antigen presenting cells use MHCs to present fragments of potentially harmful proteins on their cell surface. These peptide/MHC (pMHC) complexes are scanned and bound by TCRs. Depending on the combination of TCR, peptide and MHC (TCRpMHC) an immune response against a peptide can take place. The binding between pMHC and TCR is of low affinity (KD 1–100 µM) and somewhat degenerate in that one TCR can recognise multiple pMHCs and one pMHC can be recognised by multiple TCRs. Here, we used pyHVis3D to understand the effects of single point mutations in the MHC class I HLA-A2 on its binding affinity to three different TCRs. We extracted A6/LLFGYPVYV/HLA-A*02: 01 (1AO7; KD = 0.88 μM), JM22/GILGFVFTL/HLA-A*02: 01 (1OGA; KD = 5.29 μM) and 1G4/SLLMWITQC/HLA-A*02: 01 (2BNR; KD = 3.19 μM) from the Protein Data Bank. We computationally introduced the mutation R65AMHC into the MHC α-chain of all three wild-type complexes using DeepView (Guex and Peitsch, 1997). The R65AMHC mutation leads to a decreased binding affinity for the 1G4 TCR (+1.81 ΔΔG kcal/mol) and A6 TCR (>+3.16 ΔΔG kcal/mol) but to an increased binding affinity for the JM22 TCR (–0.36 ΔΔG kcal/mol) (Zhang et al., 2016). These different effects on different TCRs make 65MHC an interesting case for investigation. We ran MD simulations for all six complexes including their constant TCR domains (3 TCRpMHC wild-type complexes and the R65AMHC mutant of each) for 100 ns each. Ten replicas (identical parameters but different seeds) of each complex were performed. This led to 60 simulations with a total runtime of 6000 ns. The detailed simulation protocol is described in the Supplementary Material. In JM22 the R65AMHC mutation increases the binding affinity (ΔΔG –0.36 kcal/mol) (Zhang et al., 2016). It is not clear why changing a positively charged side-chain that is not in contact with the peptide and not closely packed (Fig. 1A) to an alanine increases the binding affinity between pMHC and TCR. Fig. 1. View largeDownload slide Visualisation of the JM22 H-bond network in reaction to the R65AMHC mutation. (A) Spatial arrangement between MHC (white), peptide (green), TCR α-chain (transparent orange), TCR β-chain (transparent black) and MHC residue 65 (yellow sphere). (B) H-bonds that are more frequently present in the wild-type simulation. The inlay shows a magnified and rotated view of the hedgehog-like H-bond pattern between R65MHC and CDR2β (C) H-bonds that are more frequently present in the R65A mutant. (D) Overlay of (B) and (C) (Color version of this figure is available at Bioinformatics online.) Fig. 1. View largeDownload slide Visualisation of the JM22 H-bond network in reaction to the R65AMHC mutation. (A) Spatial arrangement between MHC (white), peptide (green), TCR α-chain (transparent orange), TCR β-chain (transparent black) and MHC residue 65 (yellow sphere). (B) H-bonds that are more frequently present in the wild-type simulation. The inlay shows a magnified and rotated view of the hedgehog-like H-bond pattern between R65MHC and CDR2β (C) H-bonds that are more frequently present in the R65A mutant. (D) Overlay of (B) and (C) (Color version of this figure is available at Bioinformatics online.) Using pyHVis3D to analyse the H-bond network of the wild-type and the mutant allows us to shed light on this question. R65MHC binds unstably to several different residues of CDR2β (N55-K59; blue hedgehog-like structure in the inlay of Fig. 1B) and thereby destabilises the CDR2β loop increasing the entropy and lowering the binding affinity between TCR and MHC. In the R65A mutant this destabilisation is not present allowing Q52CDR2β to act as stabilisation center (Fig. 1C). The disruptive interaction between R65MHC and CDR2β is not present for the 1G4 TCR (Supplementary Fig. S1). Here, a stable H-bond is formed between R65MHC and D55CDR2β. This H-bond is not possible in the A65MHC mutant explaining the lower binding affinity of the R65AMHC mutant (ΔΔG +1.81 kcal/mol). In the case of the A6 TCR the wild-type R65MHC also forms a stable H-bond. In contrast to the 1G4 TCR, however, the H-bond is mainly formed between R65MHC and D93CDR3α (Supplementary Fig. S2). In the R65AMHC mutant this H-bond to D93 is not possible explaining the lower binding affinity of the R65A mutant (ΔΔG +3.16 kcal/mol). Our computationally deduced H-bond networks visualised by pyHVis3D allows interesting insights as to why the mutation of MHC residue 65 from Arginine to Alanine increases the binding affinity to the JM22 TCR but decreases the binding affinity to 1G4 and A6. A full research article investigating the wild-type TCRs experimentally and computationally can be found in (Zhang et al., 2016). Alternative allosteric network interaction analysers are discussed in (Stolzenberg et al., 2016). For example (Stolzenberg et al., 2015) was applied for MHC class II states (Wieczorek et al., 2016). Funding Engineering and Physical Sciences Research Council (EPSRC) (EP/I017909/1), EPSRC & Medical Research Council (MRC) CDT (EP/L016044/1), MRC Program Grant to AM (G9722488). The authors acknowledge the use of the Oxford Advanced Research Computing (ARC) facility. Conflict of Interest: none declared. References Guex N. , Peitsch M.C. ( 1997 ) SWISS-MODEL and the Swiss-PdbViewer: an environment for comparative protein modeling . Electrophoresis , 18 , 2714 – 2723 . Google Scholar CrossRef Search ADS PubMed Humphrey W. et al. . ( 1996 ) VMD: visual molecular dynamics . J. Mol. Graph ., 14 , 33 – 38 . Google Scholar CrossRef Search ADS PubMed Kabsch W. , Sander C. ( 1983 ) Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features . Biopolymers , 22 , 2577 – 2637 . Google Scholar CrossRef Search ADS PubMed Pronk S. et al. . ( 2013 ) GROMACS 4.5: a high-throughput and highly parallel open source molecular simulation toolkit . Bioinformatics , 29 , 845 – 854 . Google Scholar CrossRef Search ADS PubMed Stolzenberg S. et al. . ( 2016 ) Computational approaches to detect allosteric pathways in transmembrane molecular machines . Biochim. Biophys. Acta , 1858 , 1652 – 1662 . Google Scholar CrossRef Search ADS PubMed Stolzenberg S. et al. . ( 2015 ) Mechanism of the association between Na+ binding and conformations at the intracellular gate in neurotransmitter: sodium symporters . J. Biol. Chem ., 290 , 13992 – 14003 . Google Scholar CrossRef Search ADS PubMed Wieczorek M. et al. . ( 2016 ) MHC class II complexes sample intermediate states along the peptide exchange pathway . Nat. Commun ., 7 , 13224 . Google Scholar CrossRef Search ADS PubMed Zhang H. et al. . ( 2016 ) The contribution of major histocompatibility complex contacts to the affinity and kinetics of T cell receptor binding . Sci. Rep ., 6 , 35326 . Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: email@example.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
Bioinformatics – Oxford University Press
Published: Jan 10, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera