3rd Annual CCP-BioSim Conference:

Frontiers of Biomolecular Simulation

21-23
rd
May 2014
University of Edinburgh



Programme and
Abstract Booklet
2

Welcome to the 3
rd
Annual conference of CCP-Biosim, we hope you have a very enjoyable
time here. Here are a few housekeeping details:
1. Conference venue:
All talks will take place in the Prestonfield room on the first floor of the John
McIntyre Conference Centre (JMCC), Pollock Halls of residence.

Pollock Halls, 18 Holyrood Park Rd, Edinburgh EH16 5AY

See the maps below. The following links may also be useful to plan your trip to
Edinburgh.

http://www.edinburghfirst.co.uk/getting-here

http://www.edinburghfirst.co.uk/locations


2. Catering:

On Wednesday and Thursday sandwich lunches, refreshments and coffee breaks will
be served in the Centro room, JMCC, 1
st
floor.

There will be a drinks reception on Wednesday afternoon after the end of the poster
pitches session.

Cooked lunch on Friday will be served in the JMCC restaurant, ground floor.

PLEASE NOTE THAT NO DINNER IS PROVIDED FOR WEDNESDAY NIGHT

The nearest restaurant is the Salisbury Arms, on Dalkeith road opposite the
Royal Commonwealth Pool. There are several other restaurants in Minto street,
in the Newington neighbourhood, ca. 10 min walk away from the conference
venue (see map).
Alternatively, the city centre is only ca. 30 min walk away from Pollock Halls.
See the transport section for info about nearby bus stops.

3. Accommodation:
For those who have booked it, bed and breakfast accommodation is provided in
Masson House, which is about a 5-minute walk across the campus from the John
McIntyre Conference Centre. Check-in on Wednesday 21
st
is from 2 pm. On Friday
23
rd
you must have checked out by 10.30 am. Luggage can be stored in the reception
centre (see map).

4. Posters:
All posters should be put up after registration on the 21
st
, and must be taken down by
11.15 am on the 23
rd
. Poster boards are available in the rooms Salisbury, Holyrood
and Duddingston in JMCC 1
st
floor. Please check the poster number you have been
assigned (page 9/10). There are two two minute pitch sessions for poster presenters
please make sure you know which one you have been assigned to!

3

5. Conference dinner:
The conference dinner, included in your registration fee, is in the Pentland room and
will start at 7.00 pm. Wine and soft drinks will be provided during the dinner. In
addition there will be a bar open on the JMCC ground floor until 11.00pm. IF FOR
ANY REASON YOU ARE UNABLE TO ATTEND THE DINNER, PLEASE
INFORM A MEMBER OF THE ORGANISING COMMITTEE AS SOON AS
POSSIBLE.

6. Internet access:

Free Wifi access is available in JMCC. Please use the keysurf network and follow
the instructions on your web-browser to register on the network.

7. Transport:
For those staying at Masson House, car parking is limited and available on a
first come first served basis, no spaces can be reserved. The nearest NCP car
park is about 10 min walk away, down St Leonards Street.
Edinburgh has an excellent bus service. The best place to catch buses into the
city centre is from Dalkeith road (see map). Royal mile is a 25 minute walk
away.
Taxi A taxi rank is available at the reception centre.

If you have any queries, please dont hesitate to contact a member of the organising
committee.
Sponsors
CCP-BioSim is grateful for the support of the following organisations:
- The Center for Numerical Algorithms and Intelligent Software (NAIS)
http://www.nais.org.uk/
- The Molecular Graphics and Modelling Society
http://www.mgms.org/
- The Computational Chemistry list
http://www.ccl.net/chemistry/

Organising committee
Philip Biggin, University of Oxford
Richard Henchman, University of Manchester
Charles Laughton, University of Nottingham
Julien Michel, University of Edinburgh
Martyn Winn, Daresbury Laboratory

4

Maps
John McIntyre Conference Centre, 1
st
floor.

5


John
McIntyre
Conference
Centre
Masson
House
6


Programme

Wednesday 21st May

11:00 Registration Open

13:00 Lunch (JMCC Centro)

Session 1 - Chair Julien Michel

14:00 Michael Gilson (UCSD)
Plumbing the Depths of Entropy and Enthalpy in Molecular Recognition

14:45 Christopher Baker (University of Cambridge)
Simulating Coupled Folding and Binding in Intrinsically Disordered Proteins

15:15 Coffee / Tea (JMCC Centro)

Session 2 - Chair Philip Biggin

15:45 Bert de Groot (MPI Gttingen)
Molecular dynamics of inhibition, permeation and recognition

16:30 Hideaki Fujitani (University of Tokyo)
Molecular dynamics simulations for pharmaceutical target proteins with refined AMBER
force field

17:00 Poster pitches - Group A - Chair Martyn Winn

18:00 Posters and wine reception (JMCC Centro)

7

Thursday 22nd May

Session 3 - Chair Richard Henchman

09:00 Frauke Graeter (Heidelberg Institute for Theoretical Studies)
Protein allosteric regulation from force distribution analysis
09:45 Ferrucio Palazzesi (ETH Zurich)
The allosteric communication pathways in KIX domain

10:15 Poster pitches - Group B - Chair Martyn Winn

10:45 Coffee / Tea (JMCC Centro)

Session 4 - Chair Martyn Winn

11:15 Alexandre Bonvin (Utrecht University)
Modelling structure, affinity and specificity of biomolecular complexes
12:00 Michela Candotti (IRB Barcelona)
Urea-unfolded ubiquitin: from NMR to MD simulation
12:30 Antonija Kuzmanic (IRB Barcelona)
X-ray refinement significantly underestimates the level of microscopic heterogeneity in
biomolecular crystals

13:00 Lunch and posters (JMCC Centro)

13:30 CCP-BioSim Lunchtime Bytes
James T. Gebbie (Daresbury) High-End Computing for Biomolecular Simulation
Hannes H. Loeffler (Daresbury) Automating Free Energy Simulations with FESetup

Session 5 - Chair Charles Laughton

14:30 Jochen Hub (Georg-August-University Gttingen)
Coupling atomistic simulations to wide-angle X-ray scattering data
15:00 Markus Lill (Purdue University)
Including Ligand Induced Protein Flexibility into Protein Tunnel Prediction

15:30 Coffee / Tea (ESLC Atrium)

Session 6 - Chair Adrian Mulholland

16:00 Michael Mazanetz (Evotec)
Maximising the impact of structure-based in silico design at Evotec - highlights and lessons
learned
16:45 Odin Kvam (AstraZeneca)
Free energy perturbation for relative binding energy prediction: 2,4-bisanilinopyrimidine
inhibitors of the tyrosine kinase EphB4

17:15 CCPBioSim annual meeting
19:00 Conference Dinner (Pentland)
8

Friday 23rd May

Session 7 - Chair Francesco Gervasio

09:00 Irina Tikhonova (Queens University Belfast)
Tackling drug selective polypharmacology using molecular simulations

09:45 Jemma Trick (University of Oxford)
Designing Hydrophobic Gates into Biomimetic Nanopores

10:15 Marieke Schor (University of Edinburgh)
Exploring the role of multiple docked states in amyloid fibril formation of TTR

10:45 Coffee / Tea (JMCC Centro)

Session 8 Chair Christopher Woods

11:15 Edina Rosta (King's College London)
Catalytic Mechanism of Phosphate Cleavage Reactions

11:45 Mark Waller (WWU Mnster)
A Density Based Adaptive QM/MM Approach for Complex (Bio-)Chemical Systems

12:15 Gerhard Hummer (MPI Frankfurt)
Molecular simulation of protein dynamics and function

13:00 Lunch (JMCC restaurant, ground floor)
14:00 Departure

9

Posters - Group A
1. Philip Biggin - Regulatory ions bound at the iGluR ligand binding domain dimer
interface a shared property of GluK2 and AvGluR1?
2. Michael Bodnarchuk - The Application of Constant-pH Molecular Dynamics to
Polyamino Acids
3. Juan Bueren Calabuig - Impact of ligand binding on the N-terminal MDM2 lid dynamics
explored by accelerated Molecular Dynamics and Umbrella Sampling simulations
4. Robin Corey - Molecular Dynamics simulations of the Sec protein translocon with its
cytosolic partner SecA
5. Benjamin Cossins - Exploring IgE with molecular dynamics
6. Remi Cuchillo - Protein Druggability: the JEDI Approach
7. Erin Cutts - Evaluating crystallographic interaction interfaces through MD
8. Jacek Czub - Accumulation and transmission of energy during the rotary catalytic
cycle of F1-ATPase
9. Ioanna Danai Styliari - Simulating the coating procedure of indomethacin
nanoparticles with mPEG-PCL diblock copolymers
10. Callum Dickson - Molecular dynamics simulation of lipid membranes with AMBER
and application to the study of radioimaging compounds
11. Charis Georgiou - Rational design of isoform specific ligands
12. George Gerogiokas - Biomolecular hydration thermodynamics via grid cell theory
aids prediction of ligand-protein binding affinities
13. Alexander Goetz - Investigating dynamic motifs in amyloid precursor protein mutants
as impact factor for Alzheimers disease
14. Jonathan Higham - Multicell theory to calculate hydration entropy
15. David Huggins - Improved Entropy Estimation Using The k-Nearest Neighbors
Algorithm
16. Mika Ito - A Rational Method for Quantum Chemical Prediction of Key Residues in
Enzymatic Reactions: The Case of Proton Abstraction in Ketosteroid Isomerase
17. Pablo Jambrina - Computational Study of the phosphorylation of RAF dimers
18. Dominika Jankowska - Modelling of mechanisms of reactions catalyzed by quorum
quenching enzyme isolated from Ochrobactrum species
19. Nathjanan Jongkon - In-depth understanding into the reaction mechanism of Di-
Methyl-Malate Lyase (DMML)
20. Outi Kamarainen - Dynamics of protein ligand interactions impact on drug discovery
21. Mateusz Kogut - Molecular basis of the stability of G-quadruplexes - molecular
dynamics simulation study
22. Tomas Kubar - Multi-scale simulation of biological electron transfer
23. Joanna Lee - Molecular Dynamic Simulations of the fucose transporter, FucP
24. Mickael Lelimousin - Highly enhanced conformational sampling of the
transmembrane domain of EGF receptor sheds light on the activation mechanism
25. Filip Leonarski - RedMDStream: A Tool to Design Coarse-Grained Molecular
Dynamics Models and Force Fields for Proteins and RNA
26. Pete Leung - The NorM MATE transporter from N. Gonorrheae: insights into drug &
ion binding from atomistic molecular dynamics simulations
27. Greg Lever - Benchmarking large-scale DFT calculations on the chorismate mutase
enzyme
28. Valeria Losasso - Molecular dynamics study of Epidermal Growth Factor Receptor
ectodomain
29. Silvia Lovera - Dynamic fingerprint of imatinib sensitive kinases
30. Robbert Mackenzie - Multiscale Modelling of Drug-Polymer Nanoparticle Assembly
10

Poster - Group B
31. Dmitry Osolodkin - Elucidation of intrinsic determinants of fusion initiation in
flavivirus envelope proteins through molecular dynamics simulations
32. Shashank Pant - Effect of attractive interactions on the water-like anomalies of a
core-softed model potential
33. Jamie Parkin - Probing the outer membrane of Pseudomonas aeruginosa using
molecular dynamics simulations
34. Kevin Pinto Gill - Solvent Extension of MDmix methodology and its impact on the
quality of the predictions
35. Giorgo Saladino - Accurate Prediction of the Thermodynamics and Kinetics of Drug
Binding
36. Maysaa Saleh - Development of a New Series of Bis-Triazoles as Anti-Tumour
Agents
37. Twana Salih - Investigating the reliability of the MM-GBSA method for predicting
protein-ligand binding free energies
38. Firdaus Samsudin - Improving drug delivery: computational studies of proton
dependent oligopeptide transporters
39. Lars Schaefer - Scrutinizing Hybrid AA/CG Force Fields through Free Energy
Calculations
40. Christina Scharnagl - How dynamic transmembrane helices can influence
peptide/lipid interactions: a molecular dynamics study
41. Pedro Sfriso - Breaking down protein conformational transitions with coarse-grained
models
42. Adrita Shkurti - Towards faster-to-implement and extensible python-based tools for
molecular simulation data analysis
43. ABSTRACT WITHDRAWN
44. Thana Sutthibutpong - Molecular dynamics study on the structural transition of dna
under superhelical stress
45. Siri van Keulen - Dynamics of retinal chromophore in rhodopsin: from cis/trans
isomerisation to activation
46. Milosz Wieczor - A Molecular Dynamics study of mechanisms of sequence
recognition and DNA binding in two telomeric proteins, TRF1 and TRF2
47. Nathalie Willems - Lipase enzyme interactions with lipid bilayers
48. Rilei Yu - Exploring the Interaction of Agonists with the human alpha1 Glycine
Receptor


11

TALKS













12


Plumbing the Depths of Entropy and Enthalpy in Molecular Recognition

Mike Gilson
UCSD Center for Drug Discovery Innovation, University of California San Diego, La Jolla,
California, 92093-0736, USA

Molecular recognition is of fundamental importance in biology, and targeted molecules are
widely used as drugs and biochemical probes. However, the design of drugs and other
targeted molecules still involves a great deal of experimental trial and error. Our lab aims to
speed this process by developing a deeper understanding of molecular recognition and
building this understanding into new computational tools. I will discuss our progress in
computing and interpreting changes in entropy and enthalpy in noncovalent binding,
including the considerations of protein flexibility, motional correlations, and structured water.
13

Simulating Coupled Folding and Binding in Intrinsically Disordered Proteins

Christopher M. Baker

University of Cambridge, Department of Chemistry, Lensfield Road, Cambridge, CB2 1EW
Current Address: Syngenta, Jealotts Hill International Research Centre, Bracknell, Berkshire,
RG42 6EY

chris.baker@syngenta.com


Over the last twenty years, the traditional structure-function paradigm has undergone a fundamental
re-evaluation. While we have long been taught that proteins rely on their three dimensional
structure for their biological function, we now know that this is not necessarily the case.
Intrinsically disordered proteins (IDPs) are a class of protein that, in the native state, possess no
well-defined secondary or tertiary structure, existing instead as dynamic ensembles of
conformations. IDPs are also biologically important, with more than 20% of eukaryotic proteins
significantly disordered (1). The biological functions of IDPs are mediated by the process of
coupled folding and binding: while unstructured in solution, IDPs typically fold into well-defined
three-dimensional structures upon interaction with binding partners.
Despite its biological importance, the mechanism of coupled folding and binding is not well
understood, principally because it is very hard to study experimentally. At present, computer
simulation offers perhaps the best opportunity to understand this process, but is itself very
challenging, principally due to the size and complexity of the molecules involved (2). One way to
reduce this size and complexity is to use a multiscale approach, in which simplified, or coarse-
grained, computational models are combined with more detailed atomistic models. Here, I use this
approach to determine the mechanism of the sequence-specific binding of homodimers of the IDP
jun (a component of the AP-1 transcription factor) to DNA.
This work in turn raises another question: can we use results from molecular dynamics simulations
to design bioactive small molecules that control the coupled folding and binding of IDPs?

References
1. Pansca, R.; Tompa, P. PLoS One, 7, e34687, 2012
2. Baker, C. M.; Best, R. B. WIREs Comput. Mol. Sci., in press, 2014
14

Molecular dynamics of inhibition, permeation and
recognition.
Bert de Groot
Max Planck Institute for Biophysical Chemistry
Gttingen, Germany

bgroot@gwdg.de
Can we design specific membrane channel inhibitors? What is the antimicrobial mechanism of
the human antibiotic dermcidin? What are the molecular determinants of channel permeation and
gating? What is the molecular basis of protein-protein recognition, and can we alter protein
binding affinity by computational design? These are some of the questions that are addressed at
the atomic level by molecular dynamics simulations.


[1] Sren J. Wacker, Camilo Aponte-Santamaria, Per Kjellbom, Soren Nielsen, Bert L. de Groot,
Michael Rtzler. The identification of novel, high affinity AQP9 inhibitors in an intracellular
binding site. Molecular Membrane Biology 30:246-260 (2013).
[2] Ulrich Zachariae, Robert Schneider, Rodolfo Briones, Zrinka Gattin, Jean-Philippe Demers,
Karin Giller, Elke Maier, Markus Zweckstetter, Christian Griesinger, Stefan Becker, Roland
Benz, Bert L. de Groot, and Adam Lange. Beta-barrel mobility underlies closure of the voltage-
dependent anion channel. Structure. 20:1540-1549 (2012).
[3] Chen Song, Conrad Weichbrodt, Evgeniy S. Salnikov, Marek Dynowski, Bjrn O. Forsberg,
Burkhard Bechinger, Claudia Steinem, Bert L. de Groot, Ulrich Zachariae, and Kornelius
Zeth.Crystal structure and functional mechanism of a human antimicrobial membrane channel.
Proc. Nat. Acad. Sci. 110: 4586-4591 (2013).
[4] Carsten Kutzner, Helmut Grubmller, Bert L. de Groot, Ulrich Zachariae. Computational
Electrophysiology: The Molecular Dynamics of Ion Channel Permeation and Selectivity in
Atomistic Detail. Biophys. J. 101: 809-817 (2011).
[5] Jan-Henning Peters and Bert L. de Groot. Ubiquitin dynamics in complexes reveal molecular
recognition mechanisms beyond induced fit and conformational selection. PLoS Comp. Biol. 8:
e1002704 (2012).

15

Molecular dynamics simulations for pharmaceutical target proteins with refined
AMBER force field

Hideaki Fujitani

Research Center for Advanced Science and Technology, The University of Tokyo

fjtani@nifty.com

Owing to the latest advance in the computational technology like microprocessors, high-speed
inter-processor connection, and parallelization algorithm, all-atom molecular dynamics (MD)
simulations of microsecond time scale are getting popular to study the solvated pharmaceutical
target proteins such as protease, kinase, and antigen and antibody systems. An antibody binds to its
antigen with structure changes at the binding interface including complementarity-determining
regions (CDRs). A ligand moves around the target protein and finds an entrance to gets into the
binding site. These phenomena can be observed in microsecond simulations, but an important issue
is whether the adopted force field is enough accurate to correctly describe these phenomena. We
developed FUJI force field using general AMBER (GAFF) atom types and AMBER94 van der
Waals parameters in order to describe arbitrary organic molecules in a unified manner including
proteins and nucleic acids. The point charges of AMBER94 are used for amino acids and nucleic
acids. The dihedral torsion parameters of protein backbone were determined to agree with the
torsion energy profiles calculated by high-level quantum mechanical (QM) theory DF-LCCSD(T0)
for the model systems of protein backbone. Conformational preferences of alanine dipeptides in
water were recently measured by vibrational spectroscopy. MD simulations with FUJI force field
gave distribution of Ramachandran angles / of alanine dipeptide, which agrees well with the
experimental observation, while various force fields like AMBER, CHARMM, OPLS-AA and
semi-empirical QM methods cannot reproduce the experimental distribution of / angles of
alanine dipeptide in water. As the torsion parameters play the most important role in the backbone
rigidity, all-atom MD simulations of microsecond time scale with FUJI force field might reveal new
flexible and complex behaviours of proteins, which could be not observed by other force fields.
The absolute binding free energies are so sensitive to the protein rigidity that other force fields
might give wrong values. We performed MD simulations to examine interfacial structures and
interaction characters of antigen and antibody systems such as an immunotherapeutic target protein
for hepatocellular carcinoma and a ligand to the epidermal growth factor receptor (EGFR), which
stimulates the proliferative signalling especially in colon cancer cells. We also performed
massively parallel computation for absolute binding free energy (MP-CAFEE) for pharmaceutical
target proteins and small molecules. These results were compared with experiments such as X-ray
crystal structures, binding constants, thermal enthalpy and entropy measured by isothermal titration
calorimetry (ITC) and surface plasmon resonance (SPR).

References

1. Fujitani, H. ; Matsuura, A. ; Sakai, S. ; Sato, S. ; Tanida, Y. J. Chem. Theory Comput. 5, 1155-
1165, 2009
2. Fujitani, H. ; Tanida, Y. ; Matsuura, A. Phys. Rev. E 79, 021914, 2009
3. Vymtal, J. ; Vondrek. J. Chem. Phys. Lett. 503, 301-304, 2011
4. Fujitani, H. ; Shinoda, K. ; Yamashita, T. ; Kodama, T. J. Phys.: Conference Series 454, 012018,
2013
16

Protein allosteric regulation from force distribution analysis

Frauke Graeter, Jing Zhou, Christian Seifert

Heidelberg Institute for Theoretical Studies, Schlosswolfsbrunnenweg 35, 69118 Heidelberg,
Germany

frauke.graeter@h-its.org


Tight regulation of proteins is critical for cellular life. Typical mechanisms to allosterically regulate
and thereby reversibly activate proteins are ligand binding or covalent modifications. An additional
mode of protein regulation is recently emerging, namely mechanical force. We have gained
intriguing insight into the dynamics and allosteric control of proteins using Force Distribution
Analysis (FDA). FDA is based on Molecular Dynamics simulations and reveals the propagation of
an external perturbation, being it a bound ligand or mechanical stretching, through the protein
scaffold, resulting in a connected allosteric network. Here, I will present two applications:
Hsp90 is a dimeric chaperone, which upon activation undergoes large conformational changes
triggered by nucleotide binding and release. Using FDA, we have determined the allosteric network
between the nucleotide binding site and the hinge region driving the conformation change [3].
Focal adhesion kinase is a pivotal signaling protein at focal adhesion sites, right on the spot of stress
transmission between the cell and its exterior. Our simulation results suggest a novel activation
mechanism triggered by both ligand binding and mechanical force, which leads to the detachment
of an inhibitory domain of the kinase, allowing subsequent phosphorylation events [2].
Being based on molecular forces instead of molecular coordinates, our results propose FDA to be a
promising method to elucidate the mechanism underlying protein allosteric regulation.


References

1. C. Seifert and F. Grter (2013). Biochim. Biophys. Acta - General Subjects, 1830(10):4762-8.
(review)
2. Zhou J.; Lietha D; Grter F, in preparation
2. Seifert C., Grter, Biophy. J., 103(10):2195-2202., (2012)
3. Palmai Z, Seifert C, Grter F, Balog E (2014) PLoS Comput Biol 10(1): e1003444 (2014)


17

The allosteric communication pathways in KIX domain
Palazzesi F.
a
, Barducci A.
b
, Tollinger M.
c
and Parrinello M.
a

a. Department of Chemistry and Applied Biosciences, ETH Zurich, and Facolt di Informatica,
Istituto di Scienze Computazionali, Universit della Svizzera Italiana Via G. Buffi 13, 6900
Lugano, Switzerland
b. Laboratory of Statistical Biophysics, Institute of Theoretical Physics, cole Polytechnique
Fdrale de Lausanne, Lausanne, Switzerland
c. Institute of Organic Chemistry, Center for Molecular Biosciences Innsbruck (CMBI), University
of Innsbruck, Innrain 80/82, 6020 Innsbruck, Austria
ferruccio.palazzesi@phys.chem.ethz.ch
Decades after the discovery of the allosteric phenomenas, the classical models built on static images
of proteins are being reconsidered with the knowledge that dynamics plays an important role in
their function
1
. Here we present an investigation on the origins of the allosteric dynamical changes
that are caused by the interaction between MLL protein and KIX domains.
In previous NMR studies
2
, the dynamic ensemble of the accessible states of this binary complex
(MLL:KIX) were studied using relaxation dispersion techniques. The dispersion profiles indicated
the presence of a conformational transition (on the milliseconds timescale) between two
configurations: a so called ground state which is highly populated, and an excited state which is
scarcely populated. However, it was not possible to resolve the structure of this latter and,
moreover, this study did not provided any description of the allosteric mechanism.
Molecular dynamics (MD) is a natural choice for understanding the molecular origins of dynamics
allostery in this binary complex. This simulation technique allows also an accurately
characterization of the structural behaviors of the two conformational states. However, this time
demanding conformational transition is inaccessible for the typical MD simulation timescale. In
order to avoid this limitation, we have employed enhanced MD methodologies such as
Metadynamics. Indeed, to achieve the sampling of the conformations ensemble of the MLL:KIX
complex, Well-Tempered Ensemble
3
combined with Parallel Tempering (WTE-PT) simulations
were performed. Using this advanced technique we have properly characterized the atomistic
configuration of the excited state for the KIX domain in the binary complex
4
. Other important
outcomes were obtained from the understanding of the structural nature of the communication
pathway and the energetics associated with them. Indeed we shed a light on the signal transmission
along the allosteric pathway and how the residues pass the information from one site to the other.
From these results we had also provide a new simulation protocol useful to give a better description
of many important biological process, such as allosteric phenomena.
References
1. Gunasekaran, K.; Ma, B.; Nussinov, R. Proteins 57, 433443, 2004
2. Bruschweiler, S.; Schanda, P.; Kloiber, K.; Brutscher, B.; Kontaxis, G.; Konrat, R.; Tollinger, M.
J. Am. Chem. Soc., 131, 3063-3068, 2009
3. Bonomi, M.; Parrinello, M. Phys. Rev. Lett. 104, 190601, 2010
4. Palazzesi, F.; Barducci, A.; Tollinger, M.; Parrinello, M. PNAS 110, 14237-14242, 2013
18

MODELLING STRUCTURE, AFFINITY AND SPECIFICITY OF
BIOMOLECULAR COMPLEXES.

Alexandre M.J.J. Bonvin

Utrecht University, Faculty of Science - Chemistry, Padualaan 8, 3584 CH Utrecht, the Netherlands

a.m.j.j.bonvin@uu.nl


Biomolecular interactions underlie most cellular processes, including signal transduction and
apoptosis. Understanding how the cell works requires describing these at molecular level, which is
bound to have a dramatic impact on current and future structure-based drug design. Computational
methods may assist in this task, particularly when some experimental data can be obtained.

I will describe our information-driven docking approach HADDOCK (http://haddock.science.uu.nl),
illustrating it with various examples including results from the CAPRI blind docking experiment. I
will then discuss the problem of binding affinity prediction, showing that current scoring functions in
macromolecular docking fail at predicting the affinity of protein-protein complexes. For binding
affinity calculation, the surface buried upon complexation is not the absolute determinant and
inclusion of additional structural parameters, previously neglected is deemed mandatory for near-
accurate predictions. Related to affinity, understanding the structural determinant of specificity is
another challenging problem which I will illustrate showing how a conserved Asp to Glu mutation can
switch the specificity profile of ubiquitination enzymes. In conclusion, current biophysical models are
far more adequate in predicting accurate conformations of protein-protein complexes rather than
assessing the affinity and specificity of their interactions.

References

1. Karaca, E.; Bonvin, A.M.J.J. Methods 59, 372-381, 2013.
2. Kastritis, P.L.; Bonvin, A.M.J.J. Curr. Opin. Struct. Biol. 23, 868-877, 2013.
3. de Vries, S.J.L.; Melquiond, A.S.J.; de Vries, S.J.; Timmers, H.Th.M; Bonvin, A.M.J.J. PLoS
Comp. Biol., 8(11), e1002754, 2012.



19


Urea-unfolded ubiquitin: from NMR to MD simulation.

Candotti M, Esteban-Martin S, Salvatella X, Orozco M

Joint Research Program in Computational Biology, Institute for Research in Biomedicine and
Barcelona Supercomputing Center, C/ Baldiri Reixac 10, 08028, Barcelona, Spain,
michela.candotti@irbbarcelona.org


The delicate equilibrium between the folded and functional structure of a protein and its unfolded state
is highly dependent on environmental variables such as the solvent. For example the co-solvent urea is
a well-known protein denaturant that displaces the equilibrium towards unstructured and non-
functional conformations of proteins. However the molecular mechanism behind its ability remains an
enigma and the interpretation of the experimental data is still ambiguous. In this work we present the
characterization of the structural, dynamics, and energetics of properties of the urea-denatured state of
ubiquitin, a small prototypical soluble protein. By combining molecular dynamics simulations with
nuclear magnetic resonance and small-angle X-ray scattering data, we were able to: (i) define the
unfolded state ensemble, (ii) understand the energetics stabilizing unfolded structures in urea, (iii)
describe the differential nature of the interactions of the fully unfolded proteins with urea and water,
and (iv) characterize the early stages of protein refolding when chemically denatured proteins are
transferred to native conditions. The results provide a new picture of the chemically unfolded state of
proteins and contribute to deciphering the mechanisms that stabilize the native state of proteins, as
well as those that maintain them unfolded in the presence of urea.


References
Candotti M, Esteban-Martin S, Salvatella X, Orozco M. Towards an atomistic description of the urea-denatured
state of proteins. Proc Natl Acad Sci U S A ; 110 (15) 59338, 2013


20

X-ray refinement significantly underestimates the level of microscopic
heterogeneity in biomolecular crystals

Antonija Kuzmanic,
1
Navraj S. Pannu,
2
Bojan Zagrovic
3

1
Structural and Computational Biology Department, IRB Barcelona, C/ Baldiri Reixac 10, 08028
Barcelona, Spain
2
Biophysical Structural Chemistry, Leiden University, PO Box 9502, 2300 RA Leiden, The
Netherlands
3
Department of Structural and Computational Biology, Max F. Perutz Laboratories, University of
Vienna, Campus Vienna Biocenter 5, 1030 Vienna, Austria

bojan.zagrovic@univie.ac.at


Biomolecular X-ray structures typically provide a static, time- and ensemble-averaged view of
molecular ensembles in crystals. In the absence of rigid-body motions and lattice defects, B-factors are
thought to accurately reflect the structural heterogeneity of such ensembles. In order to study the
effects of averaging on B-factors, we employ molecular dynamics simulations to controllably
manipulate microscopic heterogeneity of a crystal containing 216 copies of villin headpiece. Using
average structure factors derived from simulation, we analyze how well this heterogeneity is captured
by high-resolution molecular-replacement-based model refinement. Remarkably, both isotropic and
anisotropic refined B-factors often significantly deviate from their actual values known from
simulation: even at high 1.0- resolution and R
free
of 5.9%, B-factors of some well-resolved atoms
underestimate their actual values even six-fold. Our results suggest that conformational averaging and
inadequate treatment of correlated motion considerably influence estimation of microscopic
heterogeneity via B-factors, and invite caution in their interpretation.



21

Coupling atomistic simulations to wide-angle X-ray scattering data

Jochen S. Hub, Po-chia Chen

Georg-August-University Goettingen, Institute for Microbiology and Genetics,
Justus-von-Liebig-Weg 11, 37077 Gttingen, Germany

jhub@gwdg.de


Wide-angle X-ray scattering (WAXS) of biomolecules is a promising experimental technique that in
principle provides structural information on biomolecules in solution down to a resolution of a few
Angstrm, even in a time-resolved fashion. However, the structural interpretation of the measured
signals has remained problematic for two reasons, which are currently limiting a wider application of
WAXS. First, accurate calculations of WAXS spectra from structural models have remained
challenging. Second, because the information content of WAXS spectra is very low, a narrow prior
distribution is required to avoid drastic overfitting when deriving a structural interpretation. Here, we
present solutions to overcome those two problems.
We present the methodology to compute WAXS spectra from atomistic molecular dynamics
(MD) simulations. We validated our calculations against experimental SAXS/WAXS spectra from five
different proteins, and we found excellent agreement. Our calculations require only a single fitting
parameter that accounts for experimental uncertainties due to the buffer subtraction and/or due to
detector dark currents. We further applied this new method to systematically analyse the role of the
solvation shell and of protein dynamics on WAXS spectra, and demonstrated the importance of
accurately modelling both the solvent and the atomic fluctuations in solution X-ray scattering.
In order to interpret alterations of WAXS signals due to conformational transition of
biomolecules, we have developed the methodology to couple MD simulations to experimental
SAXS/WAXS signals. In our simulations, an additional WAXS-derived potential drives the
simulations into conformations that agree with the experimental data. The MD force field ensures
physically correct conformations, thus generating the required narrow prior distribution and avoiding
overfitting. We demonstrate the new method using three different proteins: a periplasmic binding
protein, ATCase, and seminal RNase. A novel solution state of ATCase is predicted.
References

1. Hub, J.S.; Chen, P.C, Validating solution ensembles from molecular dynamics simulation by
wide-angle X-ray scattering data, submitted
2. Chen, P.C; Hub, J.S, Coupling molecular simulations to X-ray solution scattering data, in
preparation

22


Figure 1. IterTunnel method
Including Ligand Induced Protein Flexibility into Protein Tunnel Prediction
Markus A. Lill, Laura J. Kingsley

Department of Medicinal Chemistry and Molecular Pharmacology, Purdue University, 575 Stadium
Mall Drive, West Lafayette, IN 47906, USA

mlill@purdue.edu

Understanding how ligands migrate through tunnels that connect the exterior of the protein to the active
site can shed light on substrate specificity and enzyme function. We have developed a novel tunnel
prediction and evaluation method named IterTunnel
1
, which includes the influence of ligand-induced
protein flexibility, guarantees ligand egress, and provides detailed free energy information as the ligand
proceeds along the egress route. IterTunnel combines geometric tunnel prediction with steered MD in an
iterative process to identify tunnels that open as a result of ligand migration and calculates the potential of
mean force (PMF) of ligand egress through a given tunnel.
Our method uses the geometric tunnel prediction program,
MolAxis
2
, to initially identify tunnels leading out of the binding
site (Fig. 1A). Using these tunnels as a guide the ligand is then
pulled from the binding site along each initial tunnel using steered
MD (Fig. 1B). After a pre-defined time, the steered MD simulation
is stopped and tunnels are recalculated from the new position of
the ligand within the tunnel (Fig. 1C). This allows for the
identification of any new tunnels that may open or close as a result
of ligand migration. Steered MD is then resumed along the three
highest-ranked tunnels as well as the original tunnel (Fig. 1D).
This process is repeated until the ligand exits the protein at which
point the simulation is terminated. Using the steered MD
trajectories, umbrella sampling is then used to calculate the PMF
(Fig. 1E) of ligand transit.Applying this new method to
cytochrome P450 2B6 (CYP2B6), we demonstrate the influence of
protein flexibility on the shape and accessibility of tunnels. More
importantly, we demonstrate that the ligand itself, while traversing
through a tunnel, induces conformational changes of the protein
and thus can reshape tunnels due to its interaction with the protein.
This process results in the exposure of new enegetically favorable
tunnels and the closure of pre-existing tunnels as the ligand
migrates from the active site.
References

1. Kingsley, L. J.; Lill, M. A. J. Comput. Chem., submitted 2014.
2. Yaffe, E.; Fishelovitch, D.; Wolfson, H. J.; Halperin, D.;
Nussinov, R. Proteins 2008, 73, 72-86.
23

Maximising the impact of structure-based in silico design at Evotec - highlights
and lessons learned.

Michael P Mazanetz
Evotec (UK) limited, 114 Innovation Drive, Milton Park, Abingdon, Oxfordshire, OX14 4RZ, UK
michael.mazanetz@evotec.com

The suite of computational tools available to assist molecular modelling is vast both in terms of
computational complexity and the scope of the method. A strategy to employ the most appropriate
method to have the greatest positive impact on a drug discovery campaign within tight design-make-
analyse cycles is therefore a required skill. This talk will focus on recent structure-based approaches
developed at Evotec to guide ligand design and protocols developed to enable greater access to tools.
Specific examples will demonstrate how timely access to methods can drive ligand design and
enhance the medicinal chemists understanding of the structure-activity relationships, and the
importance of method development and the transfer of knowledge.




24

Free energy perturbation for relative binding energy prediction: 2,4-bis-
anilinopyrimidine inhibitors of the tyrosine kinase EphB4

Odin Kvam
1
, Derek Ogg
1
, Martin J. Packer
1
, Daniel Robinson
2
1
AstraZeneca, Mereside, Alderley Park, Macclesfield, Cheshire SK10 4TG, UK
2
Schrdinger, Quatro House, Frimley Road, Camberley GU16 7ER, UK
odin.kvam@astrazeneca.com



Direct prediction of relative binding affinity is emerging as a powerful tool for computer-aided drug
design, but currently suffers from a lack of validation. Using FEP (free energy perturbation) theory,
relative binding free energies for a set of 36 homologous 2,4-bis-anilinopyrimidine EphB4-binding
ligands
[1,2]
were predicted, and canonical ensembles of bound ligand conformations generated,
allowing quantitative and qualitative assessment of FEP performance. A consensus protein system was
constructed from four X-ray crystal structures of EphB4 bound to ligands in the 2,4-bis-
anilinopyrimidine series, and used for all simulations. The full ligand series was aligned assuming
analogous binding modes, and a sparse simple graph of mutations generated based on ligand
similarity. FEP simulations were carried out using Desmond with the REST enhanced conformational
sampling algorithm
[3]
. Relative binding energies correlated well with experimental values (RMSE =
0.8, pIC50 4.4 to 7.4), and simulated conformational ensembles indicated a characteristic dual
occupancy binding mode for a subset of ligands, supported by experimental X-ray electron density
maps. By delivering high-accuracy affinity and binding mode predictions for drug design FEP enables
efficient exploration of chemical space, as well as allowing rapid mapping of often synthetically
challenging ligand series such as heterocycle permutations.
References
1. Bardelle, C. ; Cross, D., Davenport, S. ; Kettle, J. G. ; Ko, E. J. ; Leach, A.G. ; Mortlock, A. ; Read,
J. ; Roberts, N. J. ; Robins, P. ; Williams, E. J. Bioorg. Med. Chem. Lett., 18, 2776-2780, 2008
2. Bardelle, C. ; Coleman, T. ; Cross, D. ; Davenport, S. ; Kettle, J. G. ; Ko, E. J. ; Leach, A.G. ;
Mortlock, A. ; Read, J. ; Roberts, N. J. ; Robins, P. ; Williams, E. J. Bioorg. Med. Chem. Lett., 18,
5717-5721, 2008
3. Wang, L. ; Deng, Y. ; Knight, J. ; Wu, Y. ; Kim, B. ; Sherman, W. ; Shelley, J. C. ; Lin, T. ; Abel,
R. J. Chem. Theory Comput., 9, 1282-1293, 2013

25

Tackling Drug Selective Polypharmacology Using Molecular Simulations
Irina Tikhonova, Balaji Selvam, Simon L. Porter
School of Pharmacy, Queen's University Belfast, Northen Ireland
i.tikhonova@qub.ac.uk

Selective polypharmacology, where a drug acts on multiple rather than single molecular targets
involved in a disease, emerges to develop a structure-based system biology approach to design drugs
selectively targeting a disease-active protein network. We focus on the bioaminergic receptors that
belong to the group of integral membrane signalling proteins coupled to the G protein and represent
targets for therapeutic agents against schizophrenia and depression. Among them, it has been shown
that the serotonin (5-HT
2A
and 5-HT
6
), dopamine (D
2
and D
3
) receptors induce a cognition-enhancing
effect (group 1), while the histamine (H
1
) and serotonin (5-HT
2C
) receptors lead to metabolic side
effects and the 5-HT
2B
serotonin receptor causes pulmonary hypertension (group 2). Thus, the problem
arises to develop an approach that allows identifying drugs targeting only the disease-active receptors,
i.e. group 1. The recent release of several crystal structures of the bioaminergic receptors, involving
the D
3
and H
1
receptors provides the possibility to model the structures of all receptors and initiate a
study of the structural and dynamic context of selective polypharmacology. In this work, we use
molecular dynamics simulations to generate a conformational space of the receptors and subsequently
characterize its binding properties applying molecular probe mapping. All-against-all comparison of
the generated probe maps of the selected diverse conformations of all receptors with the Tanimoto
similarity coefficient (Tc) enable to separate the receptors of group 1 from group 2. The
pharmacophore built based on the Tc-selected receptor conformations, using the multiple probe maps
discovers structural features that can be used to design molecules selective towards the receptors of
group 1. The importance of several predicted residues to ligand select ivity is supported by the
available mutagenesis and ligand structure-activity relationships studies. In addition, the Tc-selected
conformations of the receptors for group 1 show good performance in isolation of known ligands from
a random decoy. Our computational structure-based protocol to tackle selective polypharmacology of
antipsychotic drugs could be applied for other diseases involving multiple drug targets, such as
oncologic and infectious disorders.

26

Designing Hydrophobic Gates into Biomimetic Nanopores

Jemma L. Trick
1
, E. Jayne Wallace
2
, Hagan Bayley
3
and Mark S. P. Sansom
1


1
SBCB Unit, Department of Biochemistry, University of Oxford, Oxford. OX1 3QU
2
Oxford Nanopore Technologies, Edmund Cartwright House, 4 Robert Robinson Avenue,
Oxford Science Park. OX4 4GA
3
Chemistry Research Laboratory, University of Oxford, 12 Mansfield Road, Oxford. OX1
3YA

jemma.trick@bioch.ox.ac.uk

The use of nanopores is fast being a major scientific tool in molecular analysis and detection
due to their ability to detect polynucleotides, proteins and small molecules. Biomimetic
modelling of pores allows for a specific function to be incorporated into the molecular
structure of the nanopore, based on amino acid motifs found in existing protein structures.

An initial beta barrel model was built computationally, based on the transmembrane domain
of 14-stranded beta-barrel pore, alpha-hemolysin. Hydrophobic and hydrophilic residues
were built in a specific arrangement within the structure to replicate an hourglass shape cavity
with a central constriction. From this, pore conductions were observed via Molecular
Dynamics (MD) and selected models were transformed into hybrid pores in which the
location of hydrophobic residues differed to give constricting regions surrounded by
hydrophilic residues. From All Atom MD simulations, a hydrophobic gating mechanism has
been established within these toy models with intermittent water currents through the pore
giving an insight into possible biomimetic motifs which could be biochemically integrated
into the wild type protein.



27

Exploring the role of multiple docked states in amyloid fibril formation of
TTR
105-115

Marieke Schor
1
, Antonia S.J.M. Mey
2
and Cait E. MacPhee
1

1 School of Physics and Astronomy, University of Edinburgh, West Mains Road, Edinburgh
EH9 3JJ, UK
2 Department of Mathematics, Freie Universitt Berlin, Arnimallee 6 D-14195 Berlin,
Germany

mschor@ed.ac.uk

Amyloid fibrils are long, highly ordered, insoluble protein assemblies. They are most
commonly associated with diseases like Alzheimer's and diabetes type II. However, more
recently functional amyloid-like fibrils have been discovered and material scientists have
started exploring their potential use as biocompatible nanomaterials [1].

Fibril formation is generally thought of as a nucleation and growth process but the steps
involved are not well understood at the molecular level. Once a stable nucleus (or seed) has
formed, fibrils are thought to grow through the incorporation of peptide monomers one at a
time at the growing end(s). Monomer addition is essentially a two step process that is referred
to as the dock-lock mechanism [2,3]. In the first docking step, the peptide forms an initial
contact with the fibril. In the subsequent, much slower, locking step the peptide changes
conformation to fit the fibril template. In principle, multiple docked states could lead to a
correctly fibril-incorporated peptide. On the other hand, one could imagine incorrect docking
leading to off-pathway metastable states, which would slow down the dock-lock transition
significantly.

Here, we present a Markov State Model (MSM) constructed from extensive all-atom MD
simulations aimed at assessing the role of multiple docked states in monomer addition to
TTR
105-115
fibrils. Our MSM indicates that there are indeed multiple docked conformations
which can transition to the locked state via distinct pathways. We also find various
offpathway metastable states. Some of the slowest transitions in the system are between these
metastable states and the fibril-compatible docked states.

References

[1] I. Cherny and E. Gazit, Angew. Chem. Int. Ed., 2008, 47, 4062.

[2] W.P. Esler, E.R.Stimson, J.M. Jennings, H.V. Vinters, J.R. Ghilardi, J.P.Lee, P.W.
Mantyh and J.E. Maggio, Biochemistry, 2000, 39, 6288.

[3] P.H. Nguyen, M.S. Li, G. Stock, J.E. Straub and D. Thirumalai, Proc. Natl. Acad. Sci.
USA, 2007, 104, 111.


28

Catalytic Mechanism of Phosphate Cleavage Reactions

Edina Rosta

Department of Chemistry, Kings College London, London, SE1 1DB

edina.rosta@kcl.ac.uk


The formation and cleavage of phosphate bonds is essential in most biological processes
including nucleic acid processing. Many enzymes that catalyze phosphate hydrolysis require
bound divalent metal ions. Most commonly, Mg
2+
ions are required for catalysis, while
similar Ca
2+
ion abolishes the catalytic activity. To elucidate the poorly understood
mechanism of these ubiquitous metal ion catalyzed reactions, we carry out hybrid quantum-
classical QM/MM free energy simulations. In our calculations, we focus on several systems,
including Ribonuclease H (RNase H) [1], dUTPase [2], and RAF kinases. RNase H is a
prototypical member of a large family of enzymes that use two-metal ion catalysis to process
nucleic acids. The active site of RNase H is almost identical across species with respect to
sequence and structure, including the human enzyme and the HIV Reverse Transcriptase
(HIV-RT) RNase H domain. HIV-RT is essential to viral replication, which makes it an
important target in HIV drug research. In our simulations, we combine [1] Hamiltonian
replica exchange with a finite-temperature string method to calculate the QM/MM free
energy surface underlying the catalytic reaction. We use a histogram-free reweighting method
to obtain this surface from combined multidimensional string simulations. Our method allows
us to search for the optimal pathway in multiple dimensions and, therefore, to identify the
detailed sequence of steps in the phosphate cleavage reactions. From our calculations,
coupled proton transfer reactions emerge as central factors in the catalytic phosphate cleavage
reactions.

References

1. E. Rosta, M. Nowotny, W. Yang and G. Hummer, J. Am. Chem. Soc., 133:8934, 2011
2. Barabs O, et al., Nucleic Acids Research DOI: 10.1093/nar/gkt756, 2013


29

A Density Based Adaptive QM/MM Approach for
Complex (Bio-) Chemical Systems

Mark P. Waller, Sadhana Kumbhar, Jack Yang
Organic Chemistry Institute, WWU Mnster, Corrensstr. 40 Mnster, Germany

m.waller@uni-muenster.de

QM/MM modeling has become one of the methods of choice for modeling
biomacromolecular systems. This is evidenced by the awarding on the 2013 Nobel Prize in
Chemistry "for the development of multiscale models for complex chemical systems"
1

However, one of the limitations of QM/MM modeling is the traditional rigid partitioning of a
given system into a QM and MM region. For instance, this becomes problematic during
QM/MM-MD simulations, as the initial partitioning on the system eventually becomes invalid.
There has been much work spanning more than almost two decades on an adaptive variant,
i.e. where the partitioning occurs during the simulation. The whole-body of work can be
roughly classified into two main families based on the partitioning criteria employed. Firstly,
there are distance-based
2
approaches, see Figure 1a, which are empirical in nature because
the cut-offs must be fitted. Secondly, a number based approach
3
, see Figure 1b, which can
circumvent this crude distance based metric, by enabling a pre-determined integer number
of molecules to surround the QM core region that can be, set to experimentally known
hydration numbers if, and only if, such values are available. Our approach presented here is
to develop a method whereby no fitted parameters are needed to partition the system;
instead the system is analyzed and partitioned based on physical arguments. This neatly
circumvents the aforementioned empiricism, and makes the adaptive-QM/MM method more
generally applicable. Our new adaptive-QM/MM method is based on employing an auxiliary
atom-centered spherical density, which is analyzed to detect non-covalent interactions
between the QM-core and the rest of the system. If non-covalent interactions are detected
between a fragment and any QM-core atom, then the fragment is placed into the QM region.
Based on this definition, all non-interacting fragments are placed into the MM region, see
Figure 1c

a. Distance based partitioning b. Number based partitioning c. Density based partitioning.*
Figure 1: A schematic partitioning of a small water cluster: QM(vdW radii)/Transition(ball and
Stick)/MM(thin wire). *The green iso-surfaces indicate the presence of non-covalent interactions.

References
1. http://www.nobelprize.org/nobel_prizes/chemistry/laureates/2013/press.html
2. Kerdcharoen, T.; Liedl, K. R.; Rode, B. M. Chem. Phys., 211, 313323, 1996.
3. Takenaka, N.; Kitamura, Y.; Koyano, Y.; Nagaoka, M. Chem. Phys. Lett., 524, 5661, 2012.
30

Molecular simulation of protein dynamics and function
Gerhard Hummer
1
, Edina Rosta
2
, Ville R. I. Kaila
3
, Kei-ichi Okazaki
1

1
Department of Theoretical Biophysics, Max Planck Institute of Biophysics, Frankfurt,
Germany (Email: Gerhard.Hummer@biophys.mpg.de)
2
Department of Chemistry, Kings College London, London SE1 1DB, United Kingdom
3
Department Chemie, echnische Universitt nchen, arching, ermany

Modern simulation methods make it possible to study the molecular function of proteins in
unprecedented detail. We have been using string simulations to elucidate the molecular
mechanisms and chemical reaction principles underlying the enzymatic catalysis of complex
multistep reactions, with the processing of nucleic acids using two-metal ion catalysis [1,2] as
a paradigmatic example. A combination of quantum chemical calculations and molecular
simulations also helped clarify the early time evolution of the molecular structure of the light-
activated signaling protein PYP, in combination with picosecond time-resolved X-ray
crystallography [3,4]. On larger scales of space and time, simulations help us explore the
motions and function of the molecular machines involved in biological energy transduction,
including the F
1
rotary motor ATP synthase [5] and the proton pump Complex I [6].
Remarkably, common physical principles emerge in the function of these proteins, despite
large variations in their structure and function, including water and hydration effects,
extended protein motions, and long-range electrostatic couplings.

1. E. Rosta, M. Nowotny, W. Yang, G. Hummer, J. Am. Chem. Soc. 133, 8934 (2011).
2. E. Rosta, W. Yang, G. Hummer, J. Am. Chem. Soc. 136, 3137 (2014).
3. F. Schotte, H.-S. Cho, V. R. I. Kaila, H. Kamikubo, N. Dashdorj, E. Henry, T. Graber, R.
Henning, M. Wulff, G. Hummer, M. Kataoka, P. A. Anfinrud, Proc. Natl. Acad. Sci. USA
109, 19256 (2012).
4. V. R. I. Kaila, F. Schotte, H.-S. Cho, G. Hummer, P. A. Anfinrud, Nature Chemistry 6, 258
(2014).
5. K.-I. Okazaki, G. Hummer, Proc. Natl. Acad. Sci. USA 110, 16468 (2013).
6. V. R. I. Kaila, M. Wikstrm, G. Hummer, submitted.


31

LUNCHTIME
BYTES

32

High-End Computing for Biomolecular Simulation

James T. Gebbie, Hannes Loeffler, Martyn Winn.

Scientific Computing Deparment, STFC Daresbury Laboratory, Keckwick Ln, Warrington,
WA4 4FS.

james.gebbie@stfc.ac.uk


Simulations have proved important in aiding the development of scientific understanding in a
wide array of areas. Biomolecular simulation is a vibrant and internationally important field
that is making highly significant contributions to biology. The need for biolmolecular
simulations in order to understand biological problems is growing substantially, with this
need the computational resource demand is also growing. This is in part due to the fact that
scientists are becoming increasingly ambitious in which systems are being studied via
simulation.

The High-End Computing Consortium for Biomolecular Simulation (HECBioSim) was
formed by CCPBioSim as a result of the ongoing and increasing demand for access to
national scale facilities. The HECBioSim consortium aims to provide simplified access to
time on the Archer national facility. Working closely with CCPBioSim the consortium aims
to significantly increase participation in HEC level simulation from the wider community
including amongst those members of the community that are not of the traditional user base.
In order to accomplish this, the consortium will deliver a number of software packages aimed
at simplifying access and the use of these resources specific to biosimulation. A new website
has been constructed www.hecbiosim.ac.uk that will provide community contributed material
and knowledge sharing through the use of forums, wiki's and software repositories. The main
goal is to increase participation and knowledge sharing throughout the biosimulation
community to make HEC more accessible. A selection of these community oriented tools will
be presented during a lunchtime byte during the poster session on the second day of the
conference.



33

Automating Free Energy Simulations with FESetup

Hannes H Loeffler
1
, Julien Michel
2
, Christopher Woods
3


1
Scientific Computing Department, STFC Daresbury, Keckwick Lane, Warrington WA4 4AD

2
School of Chemistry, University of Edinburgh, West Mains Road, Edinburgh EH9 3JJ

3
School of Chemistry, University of Bristol, Cantocks Close, Bristol BS8 1S

Hannes.Loeffler@stfc.ac.uk


The setup for Molecular Dynamics (MD) or Monte Carlo (MC) simulations can be very
tedious to do because large numbers of small organic ligand molecules may need to be
parameterised. In the case of mutational free energy approaches appropriate mappings
between many morph pairs may have to be drawn up. FESetup assists in automating these
steps as much as possible through simple control via a INI style input file. The aim of the
software is to minimise the human bottleneck by helping the researcher to focus more on
scientific problems and much less on software control.

FESetup[1] is a an automatic setup tool currently targeted mainly at proteinligand free
energy simulations like thermodynamic integration (TI) and MMPBSA. The program
currently supports simulations with Sire[2] and AMBER. Supported force fields are all
modern AMBER type force fields including GAFF for the ligand. The ligand charge model
is currently AM1/BCC.

Future goals are the support of other popular biomolecular simulation packages (Gromacs
support currently worked on), support for other free energy methods, additional force fields
and parameterisation schemes. User friendliness of the interface, robustness of the code and
generally facilitating the access to simplified simulation setup are central to our software
development efforts.

A complete, readytorun package of the FESetup code is available via [1]. FESetup is
developed as part of the software support project effort within the Collaborative
Computational Project for Biomolecular Simulation (CCPBioSim) [1].

FESetup will be presented in a Lunchtime Byte during the poster session on the second day
of the conference.

References

1. http://www.ccpbiosim.ac.uk/flagship
2. http://www.siremol.org



34

POSTERS

35

Regulatory ions bound at the iGluR ligand binding domain dimer interface a shared property
of GluK2 and AvGluR1?

Maria Musgaard, Jack Barber, M. Khadeesh bin Imtiaz and Philip C. Biggin

Structural Bioinformatics and Computational Biochemistry, Department of Biochemistry,
University of Oxford, South Parks Road, Oxford, OX1 3QU, United Kingdom

Email: Philip.biggin@bioch.ox.ac.uk


Ligand-gated ion channels activated by glutamate binding, ionotropic glutamate receptor (iGluRs),
play crucial roles in our central nervous system (CNS), e.g. in learning and memory. Furthermore,
iGluRs are implicated in many CNS disorders. Binding of glutamate to the iGluR ligand-binding
domain (LBD) triggers opening of the transmembrane cation channel. In the overall tetrameric
structure, the LBDs are organized in a dimer of dimers. As opposed to other vertebrate iGluRs,
kainate-selective iGluRs (KARs) require binding of extracellular sodium and chloride ions to the
LBD dimer interface in addition to agonist binding for activation. We have recently shown that the
regulatory cations for the GluK2 KAR determine the onset of desensitization, with desensitization
not occurring until the cation site is vacated [1]. IGluRs are also found in other kingdoms of life,
structurally exemplified by AvGluR1 from a primitive eukaryote. Surprisingly, this LBD dimer
structure shows four chloride ions bound at the dimer interface. With atomistic molecular dynamics
and steered molecular dynamics simulations, we have studied the binding of these chloride ions to
elucidate whether they appear to have a regulatory effect on AvGluR1, and whether such a
regulatory mechanism could be a shared property between KARs and AvGluR1. Furthermore, we
have studied the structural changes believed to be linked to desensitization in the presence and
absence of interface-bound ions for both structures. The LBD dimer interface is thought to open up
around the ion binding sites in conjunction with desensitization. Interestingly, this interface is
packed closer in the AvGluR1 structure than observed for LBD dimer structures of vertebrate
iGluRs. Our results illustrate how this feature influences the dynamical changes of desensitization.

1. Defining the structural relationship between kainate-receptor deactivation and desensitization.
Dawe GB, Musgaard M, Andrews ED, Daniels BA, Aurousseau MR, Biggin PC, Bowie D.
Nat Struct Mol Biol. 2013 9:1054-61.

2. Anions mediate ligand binding in Adineta vaga glutamate receptor ion channels.
Lomash S1, Chittori S, Brown P, Mayer ML. Structure. 2013 21:414-25.

36

The Application of Constant-pH Molecular Dynamics to Polyamino Acids

Michael Bodnarchuk, David Heyes and Daniele Dini

Department of Mechanical Engineering, Imperial College London, Exhibition Road, London, SW7,
UK.


m.bodnarchuk@imperial.ac.uk


The assignment of fixed protonation states to protein residues in Molecular Dynamics (MD)
simulations can lead to erroneous results when the pK
a
of a critical residue is close to biological pH,
as the charged state of residues may be a function of conformational state and therefore time.
Constant-pH Molecular Dynamics (CpHMD) reduces this problem by allowing the protonation
state of protein residues to change with time during the simulation.
1,2
CpHMD also facilitates
improved sampling in regions where the protonation state is difficult to assign, and leads to the time
average pK
a
from the probability the residue is protonated as a function of pH.
3
Literature results
obtained using CpHMD have primarily focussed on calculating the pK
a
of single amino acid side
chains in proteins with various degrees of success.
4,5
The extension of this method to consider
multiple identical titratable residues in the same molecule (here a polyamino acid) is described and
preliminary results presented. CpHMD calculates residue-specific pK
a
values along the chain, the
average of which is in very good agreement with experimental pK
a
values. Analysis of the
individual pK
a
values for each residue along the chain gives insight into the mechanism by which
polyamino acids can change their conformation, highlighting the potential for such a method to be
incorporated into protein folding models.


References
1. J. Mongan, D. A. Case and J. A. McCammon, J. Comput. Chem. 2004, 25, 2038-2048
2. R. Burgi, P. A. Kollman and W. F. van Gunsteren, Proteins: Struct. Funct. Gen. 2002, 47,
469480
3. S. Donnini, F. Tegeler, G. Groenhof and H. Grubmller, J. Chem. Theory Comput. 2011, 7(6),
19621978
4. S. L. Williams, C. A. F. de Oliveira and J. M. McCammon, J. Chem. Theory Comput. 2010,
6, 560-568
5. S. L. Williams, P. G. Blachly and J. A. McCammon, Proteins: Struct. Funct. Bioinf. 2011,
79(12), 33813388

37

Impact of ligand binding on the N-terminal MDM2 lid dynamics explored by
accelerated Molecular Dynamics and Umbrella Sampling simulations

Juan A. Bueren-Calabuig and Julien Michel

EaStCHEM School of Chemistry, Joseph Black Building, he Kings Buildings
University of Edinburgh, Edinburgh, EH9 3JJ, UK

jbueren@exseed.ed.ac.uk


The oncoprotein MDM2 is a negative regulator of the tumor suppressor protein p53. The
disruption of the p53-MDM2 complex induced by ligand binding constitutes a very promising
strategy in cancer research.
1
The N-terminal domain of MDM2 is partially folded (residues 25-119),
whereas the first 24 amino acids form a disordered lid region in the apo state that competes for
the p53 binding site via a pseudo-substrate mechanism.
2
Several structural, biochemical and
theoretical experiments have shown that the lid can adopt two possible conformations: one open
conformation in which p53 is able to bind MDM2, and one closed that occludes the p53 binding
site. In addition, the lid can also undergo a disorder-to-order transition upon binding of small
molecules to MDM2.
3
However, the exchange between the different lid conformations take place
on a very slow time scale (>10-ms)
4
which is often outside the reach of canonical MD simulations.
To achieve sufficient sampling of lid conformations with reasonable computing resources, we have
combined two different enhanced sampling techniques: accelerated molecular dynamics (aMD) and
umbrella sampling (US). With the combined aMD/US protocol we have completed extensive
simulations of MDM2 with a complete lid (residues 1-119) in the absence and presence of several
ligands of pharmaceutical relevance. Our studies provide new insights into the interplay between
MDM2 lid dynamics and ligand interactions, and may contribute to the design of new and more
effective p53-MDM2 inhibitors.












MDM2 structures showing the N-terminal lid domain (green) conformations in the open state (left, apo-MDM2) and
in the closed state (right, in the presence of Nutlin 3a)

References
1. Zhang, Z., Li, M., Wang, H., Agrawal, S. & Zhang, R. Proc. Natl. Acad. Sci. U.S.A. 100, 1163611641, 2011
2. McCoy, M. A., Gesell, J. J., Senior, M. M. & Wyss, D. F. Proc. Natl. Acad. Sci. U.S.A. 100, 16451648,2003
3. Michelsen, K. Jordan J.B., Lewis J. et al. J. Am. Chem. Soc. 134, 1705917067, 2012
4. Showalter, S. A., Bruschweiler-Li, L., Johnson, E., et al. R. J. Am. Chem. Soc 130, 64726478, 2008

38

Molecular Dynamics simulations of the Sec protein translocon with its
cytosolic partner SecA

Robin Corey, William Allen, Ian Collinson, Richard Sessions

School of Biochemistry, University of Bristol, Bristol, UK

r.corey@bristol.ac.uk


The ubiquitous and highly conserved Sec translocon acts as a site of protein translocation
across or into a variety of cellular membranes, from the bacterial inner membrane to the
eukaryotic endoplasmic reticular membrane. Atomistic structures of the Sec translocon
with and without the cytoplasmic ATPase SecA suggest a possible mechanism of
activation, however the full mechanism of activation and translocation remains unknown.

Using the SecYEG-SecA crystal structure embedded in a POPC bilayer as a starting
structure(1), a series of 100 ns all-atom molecular dynamics simulations were run to
investigate the effects of ATP hydrolysis on the system, as well as the effect of applying
both a well-characterized super-active mutant of SecYEG and a mutant previously created
and tested in our group, designed to mimic a substrate activated conformation of the
SecYEG channel (2). The output provided both biologically plausible and interesting
results. The data has subsequently been supported using biochemical techniques.

Now that these relatively simple simulations have been run and confirmed experimentally,
we are extending their complexity and breadth to investigate different features of protein
translocation. Using both the crystal structure and structures based on recent EM data
(2,3) as input models, simulations will be run with short peptide fragments in SecYEG or
with a full peptide substrate manually threaded through the complex(4). Different
conditions will be applied including a bidirectional force on the substrate and the
membrane proton motive force, allowing us to test recent experimental results and
confirm/amend our working model of SecA and SecYEG mediated translocation


References

1. Zimmer J, Nam Y, Rapoport TA. Structure of a complex of the ATPase SecA and the protein-
translocation channel. Nature. 2008 Oct 16;455(7215):93643.
2. Hizlan D, Robson A, Whitehouse S, Gold VA, Vonck J, Mills D, et al. Structure of the SecY
complex unlocked by a preprotein mimic. Cell Rep. 2012 Jan 26;1(1):218.
3. Park E, Mntret J-F, Gumbart JC, Ludtke SJ, Li W, Whynot A, et al. Structure of the SecY
channel during initiation of protein translocation. Nature. 2013 Oct 23.
4. Zimmer J, Rapoport TA. Conformational Flexibility and Peptide Interaction of the
Translocation ATPase SecA. J Mol Biol. 2009 Dec;394(4):60612.


39

Exploring IgE with molecular dynamics

Benjamin P. Cossins

CADD, UCB Celltech
216 Bath Road
Slough
SL1 3WE


Ben.Cossins@ucb.com



Crystallographic and solution studies have shown that IgE molecules are acutely bent in their Fc
region. Antibody capture and crystallography now shows us that IgE-Fc is able to access less-bent
and extended conformations when interacting with Fab molecules
1
. We have used large
metadynamics simulations of IgE-Fc without binding Fab molecules to explore the conformational
transition from bent to extended, the possibly large conformational space of extented IgE-Fc and the
possibility of a flip between symmetrical bent conformations
1
. These simulations suggest that the
bent state of IgE is ~20 kJ/mol more stable than any unbent/extended states, something which is
borne out in the difficulty to detect extended states experimentally. The barrier to initial IgE-Fc
unbending is very large yet the various lower-free-energy unbent conformations are separated by
barriers which are relatively small. Some of the various extended lower-free-energy conformations
are very similar those captured by anitibodies and crystallography.
We have also carried out large metadynamics and MSM analyses focused on the bent state of IgE-
Fc. There is great flexibility within the bent state which seems to have three different conformations
separated by relatively small barriers. Interestingly, one of these conformations seems to be primed
for receptor FcRI binding.
The extraordinary flexibility and conformational diversity of IgE-Fc offers a new perspective on
IgE function in allergen recognition, as part of the B cell receptor and as a therapeutic target in
allergic disease.


References

1. Drinkwater, N. ; Cossins, B. P. ; Keeble A. H. ; Wright M. ; Cain K. ; Hailu H. ; Oxbrow
A. ; Delgado J. ; Shuttleworth L. K. ; Kao M. ; McDonnell J. M. ; Beavil A. J. ; Henry A. J.
; Sutton B. J. Nature SMB., in press, 2014



40

Protein Druggability: the JEDI Approach

Rmi Cuchillo, Julien Michel

University of Edinburgh, EaStCHEM school of Chemistry, Joseph Black Building, West Mains
Road, Edinburgh, Scotland EH9 3JJ

remi.cuchillo@ed.ac.uk


Several scoring functions have been developed over the last decade to evaluate the druggability -or
ligandability- of a protein structure. The majority of existing methods, such as Fpocket, were
designed to assess the druggability of crystallographic structures and were not developed to be
tightly coupled to molecular dynamics (MD) simulations, in spite of the fact that post-processing of
MD trajectories is possible.
1
We present JEDI, a novel computational approach for druggability
assessment using a combination of empirical descriptors that can be collected "on-the-fly" during
MD simulations.
The Druggable Cavity Directory (http://fpocket.sourceforge.net/dcd) was used to build a data set of
64 diverse proteins in order to parameterize the JEDI scoring function. JEDI is a grid-based
approach able to perform the druggability assessment of a binding site in only a few seconds
making it one of the fastest methodologies in the field. Agreement between computed and
experimental druggability estimates is comparable to literature alternatives. In addition, our
estimator is less sensitive than existing methodologies to small structural rearrangements and gives
consistent druggability predictions for similar structures of the same protein.
To facilitate evaluation of the druggability of a target at each step of a MD simulation, the JEDI
scoring function has been integrated within the PLUMED free energy plugin that supports a broad
range of MD packages.
2
Preliminary results show that druggability estimates can be computed for
each step of a MD simulation with modest overheads. A unique feature of the approach is that
because our druggability function is sufficiently rapid and continuously differentiable, it is possible
to: 1) supplement typical potential energy functions with a druggability potential, and 2) exploit
information provided by the druggability force to bias molecular dynamics simulations with a
variety of free energy methods. Progress towards the identification of cryptic druggable
conformations in a range of systems will be reported.


References

1. Schmidtke, P. P.; Barril, X. X. J. Med. Chem., 53, 5858-5867, 2010
2. Bonomi, M.; Branduardi, D.; Bussi, G.; Camilloni, C. Comp. Phys. Commun., 180, 1961-1972,
2009



41

EVALUATING CRYSTALLOGRAPHIC INTERACTION INTERFACES THROUGH MD

Erin Cutts, Leanne Slater, Justyna Wojdyla, Phillip Stansfeld, Ioannis Vakonakis

Department of Biochemistry, University of Oxford, South Parks road, Oxford, OX13QU

erin.cutts@wolfson.ox.ac.uk


Approximately 90% of protein structures in the protein data bank are derived using X-ray
crystallography, but how trustworthy are the interactions observed between molecules in these
structures? Presented here is our analysis of three recent crystallographic protein structures that
show interactions leading to dimers or higher order oligomers:
PFI1780w, a Plasmodium falciparum protein, whose structure was resolved as a helix
swap dimer despite biophysical data suggesting it is monomeric.
TolR, an E. coli inner membrane protein that displays an intertwined beta strand swap
dimer.
The structure of G-box from Danio rerio CPAP, a novel domain type, which was
resolved as filaments of anti-parallel beta sheets and exhibits oligomerisation in
solution.
In each case MD simulations were performed on the monomeric and higher order structures to
investigate their stability and to provide insight into experimental observations. Although clear
conclusions can not always be drawn, MD simulations generally supported experimental
obversations and helped to resolve confusion caused by molecular interactions in crystallographic
structures.


42

Accumulation and transmission of energy during the rotary catalytic cycle of
F
1
-ATPase

Jacek Czub
1
, Milosz Wieczor
1
, Pawel Wityk
1
, Helmut Grubmueller
2

1
Department of Physical Chemistry, Gdansk University of Technology, Narutowicza 11/12, Gdansk,
Poland
2
Department of Theoretical and Computational Biophysics, Max Planck Institute for Biophysical
Chemistry, Am Fassberg 11, Goettingen, Germany
jacczub@pg.gda.pl


F
1
-ATPase (F
1
) is the catalytic portion of ATP synthase, a rotary motor protein that couples the
proton gradient to ATP synthesis. When driven by a proton flux, the F
1
asymmetric subunit (-
shaft) undergoes a discrete rotation inside the
3

3
catalytic headpiece and causes the binding sites
located at the subunits to cycle between states of different affinity for nucleotides. These
concerted transitions drive the synthesis of ATP from ADP and phosphate.

It was suggested that elastic power transmission with transient storage of energy in some compliant
part of the -shaft is required for the high turnover rate to occur. To investigate the proposed
mechanism in atomistic detail we used MD simulations. Analysis of the rotational fluctuations
revealed that the elasticity of the F
1
-shaft, as sensed by F
o
, arises from two distinct contributions:
the intrinsic elasticity and an effective potential imposed by the catalytic subunit. Separation of
these two contributions provided a quantitative description of the dynamic coupling between the
rotor and the stator and enabled us to propose a minimal model of the F
1
energetics near the crystal
structure resting state.

To directly study the energy transmission between the rotor and stator subunits of F
1
during the
catalytic cycle we employed a force-probe MD in which the central -shaft is driven to rotate by
externally applied torque within the
3

3
hexamer. With this approach we show that the nucleotide-
free subunit, initially in the open, low-affinity state, undergoes a fast closing transition in response
to the -shaft rotation in the synthesis direction. By computing a free energy profile, we further find
that the initial transition to the half-open state is driven by the intrinsic elasticity of , dominated by
the electrostatic interactions between elements of the active site. Therefore, our data suggest that
ADP binding to F
1
occurs via conformational selection and is preceded by the transition of to the
half-open state a previously unknown functional state that complements the conformational
landscape of subunits. Elastic properties of imply that the -shaft is unlikely to be pulled into the
position seen in the x-ray structure by spontaneous opening of the empty . Instead, the observed
position is stabilized by interactions with the two other subunits and keeps the empty in the fully
open conformation. This finding supports the notion that the -shaft acts by coupling the extreme
conformational states of subunits within the
3

3
headpiece and therefore is responsible for high
efficiency of the coordinated catalysis.

To obtain a complete picture of the F
1
rotary cycle, we determined the free energy profile for the -
shaft rotation within the
3

3
headpiece using umbrella sampling approach. These simulations
revealed the presence of a second minimum of the free energy at +80 with respect to the crystal
structure resting state, in agreement with single-molecule experiments. By decomposing the
observed free energies into individual interactions, we analyzed the major determinants of the
stability of this state likely corresponding to the so-called ATP-dependent pause of F
1
-ATPase.
43

SIMULATING THE COATING PROCEDURE OF INDOMETHACIN
NANOPARTICLES WITH MPEG-PCL DIBLOCK COPOLYMERS

Ioanna Danai Styliari
1
, Andrew Theophilus
2
, Cameron Alexander
1
, Martin Garnett
1
and
Charles Laughton
1


1
Centre for Doctoral Training in Targeted Therapeutics and Formulation Sciences, School of
Pharmacy, University of Nottingham, University Park, NG7 2RD, Nottingham
2
Product Development, R&D GlaxoSmithKline, Stevenage
paxids@nottingham.ac.uk


Nanoparticles hold a great promise as drug delivery carriers with a wide range of therapeutic
applications. Coating nanoparticles with functional entities like polymers can enhance their target
properties and their protection from the immune system
1
. Diblock copolymers in particular,
consisting of a hydrophobic and a hydrophilic part, have interesting potential as tunable coating
agents. In this study we are examining the interaction properties of diblock copolymers formed of
the hydrophilic monomethoxy poly(ethylene)glycol (mPEG) and hydrophobic polycaprolactone
(PCL)
2
in different molecular weight ratios (MwPCL/MwMPEG). We are studying, both by
experiment and simulation, a) how these polymers interact with drug nanoparticles and b) how the
coated nanoparticles interact with each other. In order to investigate this using molecular dynamics
simulations, GROMACS and AMBERtools have been used to construct different lengths of
polymer chains, in accordance with the experimental values. Indomethacin, a non-steroidal anti-
inflammatory drug has been selected as the first compound. A two-phase simulation system has
been set up, consisting of a model of an acetone drop containing an indomethacin nano-crystal
and the polymers, surrounded by the water phase. The simulation, in which these two solvent
regions mix, thus models the coating method that is used experimentally. The results of these
nanoparticle coating simulations will be compared with the experimental observations and will be
the first step towards the future investigation of nanoparticle nanoparticle interactions.



References

1. Elsabahy, M. Wooley, K. L. Chem. Soc. Rev, 41, 2545-2561, 2012
2. Hou, J.; Qian, C.; Zhang, Y.; Guo, S. J. Biom. Nanotech., 9, 231-237, 2013



44

Molecular dynamics simulation of lipid membranes with
AMBER and application to the study of radioimaging
compounds

Callum J. Dickson
1
, Antony D. Gee
2
and Ian R. Gould
1

1


Department of Chemistry and Institute of Chemical Biology, Imperial College London, South Kensington,
SW7 2AZ, United Kingdom

2
Division of Imaging Sciences, King's College London, St Thomas' Hospital, London, SE1 7EH, United
Kingdom

callum.dickson09@imperial.ac.uk
Positron emission tomography (PET) scanning is a molecular imaging technique allowing the
detection and analysis of biological processes, including metabolism and disease. PET scanning is
regularly implemented in the imaging and study of diseases such as cancer, Alzheimers and
Parkinsons disease and is also becoming increasingly popular to aid the drug discovery process.
To perform a PET scan, the patient is administered a small molecule radiotracer, which emits a
trace amount of radiation and binds to the site of interest, allowing an image to be constructed. In
order to image new targets, novel radiotracers must be designed. However a limitation in the design
of new radiotracers is non-specific binding, whereby the tracer binds to off-target species, such as
cell membranes, creating an uninformative image with poor contrast. An in silico indicator, able to
predict the level of non-specific binding a new radiotracer may undergo in vivo prior to synthesis
and testing, would be extremely beneficial to the PET community.
In this work we investigate non-specific binding using molecular dynamics (MD) to study the
interaction of PET radiotracers with lipid bilayers, a simple model for the cell membrane, using the
AMBER MD package. To accurately model lipid bilayers, suitable parameters were first
constructed.[1] These parameters are currently being combined with the AMBER Lipid11 modular
lipid force field [2] to create Lipid12, an updated unified AMBER lipid force field. The potential of
mean force (PMF) method has been inserted into AMBER, in order to calculate the free energy of
transfer of radiotracers through a lipid membrane. The PMF method is currently being applied to a
dataset of well characterised PET radiotracers to investigate the relationship between membrane
permeability and non-specific binding.
References
[1] C. J. Dickson, L. Rosso, R. M. Betz, R. C. Walker and I. R. Gould, Soft Matter, 2012, 8, 9617-9627.
[2] . A. Skjevik, B. D. Madej, R. C. Walker and K. Teigen, The Journal of Physical Chemistry B, 2012,
116, 11124-11136.

45

Rational design of isoform specific ligands

Charis Georgiou

, Malcolm Walkinshaw

, Julien Michel

Institute of Structural and Molecular Biology,

EastCHEM Shool of chemistry, The
University of Edinburgh, Edinburgh, EH9 3JR

C.Georgiou@sms.ed.ac.uk


Cyclophilins are proteins able to catalyze the interconversion of trans/cis isomers of proline
and belong to the peptidyl-prolyl isomerases family (PPIase).
1
In addition to their PPIase
activity, cyclophilins have diverse biological roles and have been implicated in a number of
different diseases such as HIV-1 and HCV. Although several cyclophilin inhibitors have been
reported in the literature, none are able to inhibit with high specificity selected cyclophilin
isoforms.

To facilitate the development of isoform-specific cyclophilin ligands, we are pursuing
detailed studies of cyclophilin dynamics and binding thermodynamics using molecular
simulations, biophysical assays and protein x-ray crystallography. Initial efforts are focussed
on molecular dynamics (MD) simulations of the cyclosporine A (CsA) cyclophilin A
(CypA) complex, for which extensive biophysical measurements and structural data have
been reported in the literature, with the goal to be the development of novel small molecule
Cyp inhibitors. Models of the apo and holo CypA were set up using the Amber and
AmberTools software and MD simulations were performed using Sire/OpenMM.
2, 3, 4

Analysis of the MD simulations are performed using custom scripts and the software package
Gromacs.
5


To facilitate the development of novel Cyp inhibitors, x-ray crystallographic studies were
carried out using small fragments, for which binding affinities were reported from Surface
Plasmon Resonance (SPR) experiments. Intermolecular interactions of CypA complexes
observed from x-ray crystallography and MD simulations have been characterised. Those
interactions include: hydrogen bonding interactions between the ligand and protein,
interactions with water molecules. Efforts towards the computation of binding free energies
and changes in conformational entropies will be reported.

References

1. Davis, T. L. et al, Plos Biology 2010, 8 (7), e1000439.
2. Case, D.A. et al, AMBER 12. University of California: San Francisco, 2012.
3. www.siremol.org.
4. Eastman, P.; Pande, V. S., Computing in Science & Engineering 2010, 12 (4), 34 -39.
5. Hess, B. et al, Journal of Chemical Theory and Computation 2008, 4 (3), 435 - 447.


46

Biomolecular hydration thermodynamics via grid cell theory aids prediction of
ligand-protein binding affinities

Georgios Gerogiokas, Richard H. Henchman, Michelle W. Y. Southey, Richard J. Law, Michael
Mazanetz, Julien Michel

EastCHEM School of Chemistry, Joseph Black Building, The King's Buildings, Edinburgh, EH9 3JJ, UK
anchester Institute of Biotechnology, he University of anchester, 131 Princess Street, anchester 1
7DN, United Kingdom and School of Chemistry, The University of Manchester, Oxford Road, Manchester
M13 9PL, United Kingdom
Evotec (UK) limited, 114 Innovation Drive, Milton Park, Abingdon, Oxfordshire, OX14 4RZ, UK
g.gerogiokas@sms.ed.ac.uk
Prediction of binding free energy can aid acceleration of the drug discovery and optimisation
process. However, many current predictors do not account for the hydration thermodynamics. A novel,
efficient methodology has been developed to quantify water energetics through an analysis of explicit
solvent molecular simulations of organic and biomolecular systems. The approach, grid cell theory (GCT),
relies on a discretization of the cell theory methodology on a three-dimensional grid to spatially resolve the
density, enthalpy and entropy of water molecules in the vicinity of system(s) of interest for qualitative and
semi-quantitative analyses [1].
Using GCT we investigate ligand-protein systems by computing numerical estimates of the
energetics of desolvation from suitable thermodynamic cycles. This involves simulating protein, ligand and
complex structures of your system of interest, in our case of a congeneric series of Hsp90 ligands [2]. The
hydration energetics of each state gives a physically intuitive understanding of the water reorganisation cost
of binding incurred by the protein and ligand desolvation energies as well as complex solvation energy.
These thermodynamic states can be visualised as isosurfaces allowing identification of stable and unstable
water sites. These GCT desolvation terms coupled with interaction energies between the protein and ligand
can be used to compute a predictive protein-ligand score which should prove useful during ligand
optimisation.

1 2
Figure. The spatial resolved grid of the hydration G in the protein-ligand structures are shown as
isosurfaces. ASP 93, THR 184, and SER 52 are shown in smaller VDW spheres while the selected ligands, 1
and 2, are shown in larger VDW spheres. Hydrogen bonds are displayed as red dotted lines while the red
isocontours of each hydration grid is generated using an isovalue of -16 kcal mol
-1

-3
.

References
1. Gerogiokas, G.; Calabro, G.; Henchman, R. H.; Southey, M. W. Y.; Law, R. J.; Michel, J. J.
Chem. Theo. Comput. 10, 35-48, 2014
2. Woodhead, A. J.; et al, J. Med. Chem. 53, 5956-5969, 2010
47

INVESTIGATING DYNAMIC MOTIFS IN AMYLOID PRECURSOR
PROTEIN MUTANTS AS IMPACT FACTOR FOR ALZHEIMERS DISEASE

Alexander Gtz
#
, Hannes Uhrmann
#
, Christina Scharnagl
*
, Dieter Langosch
#


#
Lehrstuhl fr Chemie der Biopolymere, Technische Universitt Mnchen, Freising, Germany
*
Fakultt fr Physik, Technische Universitt Mnchen, Freising, Germany

alexander.goetz@mytum.de


A major step in the development of Alzheimer's disease is the enzymatic cleavage of the Alzheimer
Precursor Protein (APP) transmembrane domain (TMD) by -secretase. The proteolysis is assumed
to take place in the water-filled cavity of the enzyme's active site. The enzyme itself is located in the
membrane bilayer. Multiple successive cleavage pathways along the TMD
1
lead to the formation of
so-called A-peptides. Depending on their length, these cleavage fragments can form toxic cerebral
plaques. Genetic and epidemiological studies have identified a variety of disease-associated mutants
in the APP TMD. Their chemical nature leads to the assumption, that they modify the flexibility of
the substrate TMDs. Therefore, it is assumed, that conformational flexibility of the different
substrates has an impact on recognition and cleavage and might determine efficiency and
processivity of the substrates.
We use all-atom molecular dynamics simulations to investigate their global and local flexible
motifs in membrane patches as well as in membrane mimicking solvents, containing 20 % water, to
emulate the enzyme cleavage cavity. Our simulations use the CHARMM27 force field and are
running on the SuperMUC PetaFlop/s system operated by Leibniz Computing Center (Garching,
Germany). We use NAMD2.9, running on 512 Intel Sandy Bridge-EP computing cores, to produce
trajectories for ~30 000 atoms over 200 ns with an output of 45 ns/day. To increase the sampled
conformational space and enhance statistical convergence we run multi-copy simulations. Start
structures will be generated by a Monte-Carlo based approach. All simulations are validated against
in-vitro experiments like deuterium/hydrogen exchange, NMR, CD-spectroscopy and functional
assays.
Our results reveal a gradient of local backbone dynamics along the helix: A highly dynamical N-
terminal domain is connected by a di-glycine hinge to a rather rigid C-terminal domain which
harbours the cleavage sites
2
. Unexpectedly, the cleavage region of the APP-TMD shows a reduced
flexibility compared to non-substrate TMD helices. Our results reveal significant differences in the
overall dynamic of several familial mutants. We find that the helix dynamics is constrained by side-
chain to main-chain backbonding of threonine residues. Removing backbonding by mutation to
valine do indeed increase helix dynamics but not at the site of initial cleavage. However, these
mutations profoundly alter global bending and rotational motions at the hinge
3
. We thus propose
that such global motions within the substrate TMD may affect the way by which it is recognized by
the enzyme and/or presented to its catalytic site.

References

1. Chvez-Gutirrez, L. et al.; EMBO J. 31, 226174, 2012
2. Pester, O. et al. J. Am. Chem. Soc. 135, 131729, 2013
3. Scharnagl, C. et al. Biophys. J.,106, in Press, 2014

48


MULTICELL THEORY TO CALCULATE HYDRATION ENTROPY

Jonathan Higham and Richard H Henchman

Manchester Institute of Biotechnology, The University of Manchester, 131 Princess Street,
Manchester, M1 7DN and School of Chemistry, The University of Manchester, Oxford Road,
Manchester, M13 9PL

jonathan.higham@manchester.ac.uk


A generalised method is presented to calculate the entropy of hydration for a range of small solutes
in water. The entropy of each molecule is evaluated by determining an effective potential, a so-
called cell, for the molecule in the mean-field of its neighbours. As in earlier theory,
1,2
the
vibrational and librational entropy is evaluated from the forces and torques measured in a molecular
dynamics simulation of the solution. Previous work on solutions had only considered two kinds of
water molecules, either in the solute's first hydration shell or in bulk, and had determined the
number of molecular arrangements from the cell's average coordination number. The development
in this work, as has already been done for pure water,
3
is to have multiple cells for each molecule.
Such cells differ by the molecule's coordination number and atom-type, leading to so-called
multicell theory. The cells here are identified using a general, parameter-free hydrogen-bond
definition, and new theory is presented to determine the associated number of molecular
arrangements. As well as hydration entropy, the method gives detailed insight into its components.

References

1. Henchman, R. H. J. Chem. Phys., 126, 064504, 2007.
2. Irudayam, S. J., Plumb, R. D. and Henchman, R. H. Faraday Discuss., 145, 467-485, 2010.

3. Henchman, R. H. and Irudayam, S. J. J. Phys. Chem. B, 114, 16792-16810, 2010.

49

Improved Entropy Estimation Using The k-Nearest Neighbors Algorithm

David J. Huggins
1,2,3

1
University of Cambridge, Theory of Condensed Matter Group, Cavendish Laboratory, 19 J J Thomson
Avenue, Cambridge CB3 0HE, United Kingdom
2
University of Cambridge, Cambridge Molecular Therapeutics Programme, Hutchison/MRC Research
Centre, Hills Road, Cambridge, CB2 0XZ, United Kingdom
3
University of Cambridge, Department of Chemistry, Lensfield Road, Cambridge, UK CB2 1EW, United
Kingdom

djh210@cam.ac.uk
Histogram methods for entropy estimation suffer from two fundamental and related
problems. The first problem is that the widths of the histogram bins must be sufficient to
capture the underlying probability density function (PDF). Bins that are too large are unable
to describe sharply peaked PDFs and will underestimate the entropy. Conversely, bins that
are too small require vast amounts of sampling to reach convergence and will otherwise
overestimate the entropy. One alternative to this histogram method is to estimate the
entropy using the k-nearest neighbors (KNN) algorithm. KNN provides an asymptotically
unbiased estimate of the entropy and can deal effectively with sharply peaked PDFs.[1] We
are interested in thermodynamic entropy estimation using inhomogeneous fluid solvation
theory (IFST), a statistical mechanical method for calculating solvation free energies by
quantifying the effect of a solute acting as a perturbation to bulk solvent.[2] The solvation
entropy is calculated in terms of translational and orientational ordering of solvent
molecules in the solute reference frame (solute-water terms) and translational and
orientational ordering of solvent molecules relative to one another (water-water terms). In
IFST, the solute-water orientational entropy has generally been estimated by integrating
correlation functions using a histogram method and recent work has highlighted the
problems with this approach.[3] Here we incorporate a KNN approach to compute solvation
free energies for 15 small molecules and nine singly charge ions. To enable this, we study a
number of distance metrics for molecular orientations and identify quaternion metrics as
superior to those based on the Euler angles. Importantly, it is demonstrated that samples
from molecular dynamics simulation must be independent for effective use of the KNN
algorithm. This finding impacts any application of the KNN algorithm to time series data.
References
[1] Hnizdo, V., J. Tan, B.J. Killian, and M.K. Gilson, J. Comput. Chem., 2008. 29(10): p. 1605-1614.
[2] Lazaridis, T., J. Phys. Chem. B, 1998. 102(18): p. 3531-3541.
[3] Huggins, D.J. and M.C. Payne, J. Phys. Chem. B, 2013. 117(27): p. 82328244.
50

A Rational Method for Quantum Chemical Prediction of Key Residues in
Enzymatic Reactions: The Case of Proton Abstraction in Ketosteroid Isomerase

Mika Ito, Tore Brinck
Department of Chemistry, KTH Royal Institute of Technology, 100 44 Stockholm, Sweden

mikaito@kth.se

Enzymes are potent catalysts which are substrate specific and achieve enormous enhancement of
reaction rates. To design an ideal enzyme for a synthesis of a compound of interest, we should
know the residues of a native enzyme that play key roles in its reaction. One can easily predict that
residues which interact directly with a substrate should be of vital importance. However, residues
without direct substrate interactions can also be important. In fact, such a case was indicated by an
experimental study
1
on an enzymatic reaction in which ketosteroid isomerase (KSI) catalyzes the
isomerization of a 5-3-ketosteroid (5-androstene-3,17- dione) to a 4-3-ketosteroid (4-androstene-
3,17-dione). According to the experimental results, Tyr14 and Asp99, which form hydrogen bonds
with the substrate, have primary catalytic roles, and Tyr55, which does not form hydrogen bonds
with the substrate, has a secondary catalytic role. This suggestion was made by comparing reaction
rate constants of wild type KSI and its mutants (Y14F, D99L, and Y55F). Similar results were
obtained from molecular dynamics simulations
2
on wild type KSI and a few mutants. However, it
is difficult to estimate rate constants of many mutants due to the preparation time in experimental
studies and computational cost in theoretical studies. Thus, you might overlook some key residues
which were not targeted for mutation. To overcome this problem, we propose a new approach by
which we can estimate the energy contributions of many residues simultaneously by calculations of
only a native enzyme, e.g. wild type KSI. In our approach, we define the energy of a deletion
form and calculate it by utilizing the procedure of the fragment molecular orbital (FMO) method,
which divides a large molecule into small fragments. This energy corresponds to the predicted
energy of a special mutant in which a mutated residue does not interact with other residues and a
substrate. In order to examine the reliability of our approach, we evaluated activation energies for
the KSI proton abstraction (Figure 1), which is a prerequisite step for the KSI Diels-Alder
reaction
3
, by using the energies of deletion forms, and compared them to activation energies
calculated from experimental rate constants by transition state theory. Through this study, it was
demonstrated that our approach is powerful for quantitatively evaluating the degree of contribution
of each residue to the enzymatic reaction and for rationally predicting the key residues.



Figure 1. The KSI proton abstraction: Asp38 of KSI abstracts a proton from the keto substrate and
generates the dienolate substrate.

1. Kim, D.-H. et al. Biochemistry, 39, 4581-4589, 2000. 2. Chakravorty, D. K.; Hammes- Schiffer, S.
J. Am. Chem. Soc., 132, 7549-7555, 2010. 3. Linder, M.; Johansson, A. J.; Manta, B.; Olsson, P.;
Brinck, T. Chem. Commun., 48, 5665-5667, 2012.
51

Computational Study of the phosphorylation of RAF dimers.

P. G. Jambrina
1
, N. Rauch,
3
K. Rybakova,
3
N.-V. Buchete,
2
W. Kolch,
3,4,5
E. Rosta
1,*

1
Department of Chemistry. Kings College London.
2
School of Physics, University College Dublin, Dublin, Ireland
3
Systems Biology Ireland, University College Dublin, Dublin, Ireland
4
Conway Institute, University College Dublin, Dublin, Ireland
5
School of Medicine & Medical Sciences, University College Dublin, Dublin, Ireland


edina.rosta@kcl.ac.uk


Although protein phosphorylation is one of the most common post-translational
modifications in cell regulatory mechanism, little is known about the structural and mechanistic
details of how phosphorylation affects the activity of kinases. In order to fill this gap we have
carried out MD simulations of phosphorylated and non-phosphorylated BRAF homodimers. In
particular, we are interested in finding a structural explanation for the paradoxical activation
mechanism observed in RAF kinases, whereby an activator kinase forms a dimer with a receiver
kinase, and thus significantly enhances its catalytic activity
1
. Our study also focus on the role
played by the N-terminal acidic motif (NtA), specific of RAF, that has been suggested to be
involved in the dimerization of RAF. To the best of our knowledge this motif has not been
experimentally resolved, and the phosphorylation of at least one of its residues is required for RAF
activation.


Coulombic interactions in which the phosphorylated S446 residue is involved for BRAF dimers.


References

1. Hu, J., et al., Allosteric Activation of Functionally Asymmetric RAF Kinase Dimers. Cell, 2013.
154(5): p. 1036-1046.

52

Modelling of mechanisms of reactions catalyzed by quorum quenching
enzyme isolated from Ochrobactrumspecies

Dominika Jankowska
1
, Marc W. van der Kamp
3
, Christopher J. Woods
3
, Dorota Krzyanowska
2
, Sylwia Jafra
2
,
Rajmund Kamierkiewicz
1
and Adrian J. Mulholland
3

1
Laboratory of Biomolecular Systems Simulations, Intercollegiate Faculty of Biotechnology, University of Gdansk and Medical
University of Gdansk, ul. Kadki 24, 80-822 Gdask , Poland

2
Laboratory of Plant Protection and Biotechnology, Department of Biotechnology, Intercollegiate Faculty of Biotechnology,
University of Gdansk and Medical University of Gdansk, ul. Kadki 24, 80-822 Gdask , Poland

3
Centre for Computational Chemistry, School of Chemistry, University of Bristol, Bristol BS8 1TS, U.K.
dominika.jankowska@biotech.ug.edu.pl
Many Gram positive and Gram negative bacteria have developed a genetic regulatory mechanism which
allows them to communicate and thus to adapt to current conditions of environment. That mechanism
called quorum sensing controls such processes as vilurency, biofilm formation, antibiotics synthesis or
sporulation and is based on production of small signal molecules. It is also known that quorum sensing
signalling may be disrupted for instance by degradation of signal molecules by enzymes produced by other
bacteria. That process, called quorum quenching, is very interesting and promising as it allows to reduce or
block the bacterial infection or inhibit the biofilm formation [1]. One of lately discovered quorum
quenching enzymes produced by Ochrobactrum species is reported to hydrolise N-acylohomoserine
lactones (AHLs), a wide family of signal molecules used by many types of bacteria. Two independent
research groups have published results of their studies in which they show that this enzyme acts as acylase
[2] and lactonase [3] separately. With reported presence of catalitic triad within the structure, capable of
undertaking both of those reactions it is highly possible that the enzyme has dual hydrolitic activity. These
enzymes have been investigated by a combination of computational methods that include molecular
dynamics and QM/MM simulations [4] and WaterSwap [5]. Every of those techniques allowed to obtain
results that gave insight into different aspects of the study. Simulations were preceded with AHLs
molecules' docking into protein active site and molecular dynamics. Docking process is based on fitting
structure of the molecule in different conformations into desired area and then ranking them. That allowed
to detect different binding modes of signal molecules that seem to be one of the milestones on path
leading to acylase- or lactonase-like reaction to happen. It also provided a different starting structures for
molecular dynamics simulations which showed how AHLs move within protein's binding site, how they
interact with residues and pointed the most stable of AHLs' conformations that became a subject of
QM/MM calculations- a method that is based on quantum mechanics and allows to simulate exact stages
of reactions and estimate how likely those stages are to happen. Water-swap is a novel method developed
at University of Bristol that bases on a swap of bound ligand with equivalent volume of water. It allows to
estimate ligand-protein binding free energy, gives details on contribution of particular residues to binding
the ligand and finally influence of structure of the ligand on its binding.
References
[1] Y.-H. Dong and L.-H. Zhang, J. of Microbiology (Seoul, Korea), 2005, 43 Feb., 1019.
[2] R. Czajkowski, D. Krzyanowska, J. Karczewska, S. Atkinson, J. Przysowa, E. Lojkowska, P. Williams, and S.
Jafra, Env. Microb. Rep., 2011, 3, 1, 5968.
[3] A. Gao, G. Mei, S. Liu, P. Wang, Q. Tang, Y. Liu, H. Wen, X. An, L. Zhang, X. Yan, and D. Liang, A. Cryst.
Sec. D Biol. Cryst., 2012, 69, 1, 8291.
[4] M. W. van der Kamp and A. J. Mulholland, Biochemistry, 2013, 52, 16, 270823.
[5] C.J. Woods, M. Malaisree, S. Hannongbua, A.J. Mulholland, J.Chem. Phys., 2011, 134,054114.
53

In-depth understanding into the reaction mechanism of Di-Methyl-Malate Lyase
(DMML)

Nathjanan Jongkon
1
, Warot Chotpatiwetchkul
2
, Matthew Paul Gleeson
2


1
Department of Social and Applied Science, College of Industrial Technology. King Mongkut's
University of Technology North Bangkok, Bangkok, 10800, Thailand
2
Department of Chemistry, Faculty of Science,Kasetsart University, Chatuchak, Bangkok 10903,
Thailand


email: paul.gleeson@ku.ac.th


Di-Methyl-Malate Lyase (DMML) is a protein from the isocitrate synthase family that catalyses the
conversion of (2R,3S)-2,3-dimethylmalate into propanoate and pyruvate. A number of
closemembers of this protein family are found in the bacteria Aspergillus niger, a common source
of bacterial infection in fruit, vegetables and processed food. In this study the catalytic mechanism
of the bacterial enzyme DMML enzyme have been investigated using quantum
mechanics/molecular mechanics (QM/MM) combined with Molecular dynamics simulations (MD).
To gain insight into the catalytic reaction of DMML, we investigate the natural substrate and a
range of inhibitors using X-ray structural data previously reported. We assess which of the different
acids and bases present in the active site contribute to the reactivity of the protein. We also assess
the effect two different cofactors (Mn
2+
and Mg
2+
) have on the energetic barriers and how these
relate to the known rate constants. Our results suggest that the catalytic mechanism involves
Arg114 acting as the general base and Cys124 acting as the general acid. The proposed mechanism
involving Tyr44 was predicted to be energetically unfavourable.

References

1. Narayanan, B.; Niu, W.; Joosten, H. J. ; Li, Z.; Kuipers, R. K.; Schaap, P. J.; Dunaway-Mariano,
D.; and Herzberg, O. J. Mol. Biol, 386, 486-503, 2009
2. Narayanan, B. C., Niu, W., Han, Y., Zou, J., Mariano, P. S., Dunaway-Mariano, D., and
Herzberg, O. Biochemistry, 47, 167-182. 2008
3. Huang, K. ; Li, Z., Jia, Y.; Dunaway-Mariano, D.; and Herzberg, O. Structure, 7, 539-548, 1999






54

DYNAMICS OF PROTEIN LIGAND INTERACTIONS IMPACT ON DRUG DISCOVERY

Outi Kamarainen, Sarah Harris, Anastasia Zhuravleva, Geoff Holdgate, Andrew Scott,
Steve Homans

Softmatter Physics, School of Physics and Astronomy, University of Leeds, E C Stoner Building,
Leeds LS2 9JT

s.a.harris@leeds.ac.uk (corresponding author)
o.kamarainen13@leeds.ac.uk (presenting author)


Introducing a new drug to market is a lengthy and expensive process (10-15 years and $1.2billion).
Better understanding of how and why a drug molecule binds to a target and what changes in the
structure and chemistry could improve the binding affinity may help to shorten the drug design
process. Whilst calorimetric methods can quantify the total free energy and entropy change, it is
difficult to estimate contributions from the different components of entropy, one of the largest
unknowns being the magnitude of the configurational entropy change. Molecular dynamics
simulations of the drug and target protein can provide more details of the different atomistic
movements contributing to the total entropy change, thus potentially providing valuable clues for
lead optimisation. The use of molecular dynamics simulations for estimating vibrational entropy
components for the binding of different antibiotics to HSP90 are described. Different methods for
calculating entropies which will help to quantify dynamics during simulations are evaluated and
methods for investigating the length of simulations and number of replicates required as well as
how well the data fits with wet lab values are discussed.



55

Molecular basis of the stability of G-quadruplexes - molecular dynamics
simulation study

Mateusz Kogut, Jacek Czub

Department of Physical Chemistry, Gdansk University of Technology, Gabriela Narutowicza 11/12
80-233 Gdansk Poland
Matkogut@student.pg.gda.pl
Native DNA usually folds to form the canonical double helix, however under certain conditions it
can also fold into various other inter- and intramolecular secondary structures. Undoubtedly, an
interesting segment of DNA with this properties is telomeric DNA, which contains structures called
G-quadruplex (G4). G4 are secondary DNA structures in which four guanine residues are held in a
plane by Hoogsteen bonds. These noncanonical nucleic acid structures are further stabilized by the
presence of a cation, mainly K+, which fits in the central channel between pairs of tetrads.
Formation of G-quadruplexes has been reported for different regions of the genome, such as
telomeres, oncogenic promoters, immunoglobulin switches, introns, and 5-untranslated regions of
RNA. However, G4 mainly occur at the ends of linear chromosomes (telomeres), terminating which
terminate in a 3'-single-stranded overhang of TTAGGG tandem repeats. The formation of G4 at
telomeres has been shown to decrease the activity of telomerase, an enzyme responsible for
maintaining length of telomeres and involved in the majority of all cancers. One possible way of
interfering with its telomere maintenance activity is to expose cells to G4-stabilizing agents. There
is a particular interest in G4 as targets for therapeutic intervention. The thermodynamic stability of
various forms of G4, expressed as the unfolding free energy, was determined experimentally,
however the molecular basis of G4 structure stabilization has not been fully elucidated.
The aim of our research was to determine the microscopic origin of the G4 thermodynamic stability
and structural diversity. Using equilibrium molecular dynamics (MD), steered MD and free energy
methods, we focused on comparing the properties of parallel, antiparallel and mixed G4 structures.
There are also some suggestions on G4 folding pathways. Some models assume that G-triplex is an
in-pathway intermediate in the folding of the G-quadruplex [1,2]. We decided to investigate the
final phase of the process. For the propeller structure, free energy profiles of single guanine residue
out-of-plane dissociation were determined by detailed thermodynamical analysis of folding substeps
and subsequent decomposition of the overall unfolding free energies into enthalpic and entropic
contributions. Similar studies were performed to determine the free energy profile for the formation
of a single G-tetrad from a fully dissociated state. This allowed us to identify the possible driving
forces governing the process of G4 folding.
References
1. Wei Li, Xi-Miao Hou, Peng-Ye Wang, Xu-Guang Xi, and Ming Li J. Am. Chem. Soc, 135, 6423-
6426, 2013
2. Tomoko Mashimo, Hirotaka Yagi, Yuta Sannohe, Arivazhagan Rajendran, Hiroshi Sugiyama J.
Am. Chem. Soc, 132, 1491014918, 2010
56

MULTI-SCALE SIMULATION OF BIOLOGICAL ELECTRON TRANSFER

Tom Kuba, Marcus Elstner

Institute of Physical Chemistry, Karlsruhe Institute of Technology, 76131 Karlsruhe, Germany

tomas.kubar@kit.edu


Theoretical description of biomolecular electron transfer (ET) reactions is challenging because large
systems require a quantum-mechanical treatment and a broad range of time scales are involved.
We have developed a multi-scale computational scheme to perform direct simulation of ET in
biomolecular systems. Combination of semi-classical propagation schemes with a linear-scaling
quantum chemical method and molecular mechanics yields an unbiased dynamics of the excess
electron, while the computation is notably efficient. Thus, the ET reactions occurring in systems
with up to hundreds of quantum atoms on nanosecond time scales can be simulated, connecting the
domains of applicability of non-adiabatic ab initio simulations and of model approaches such as the
classical Marcus theory. Importantly, the involved approximations are less restrictive than the
assumptions inherent to the Marcus theory. Thus, the current method may be applied for instance in
the difficult cases of sequential ET reactions where the polarization response of the surrounding
medium occurs on a time scale that overlaps with the time scale of the transfer itself.
The initial application is the hole transfer in DNA in an aqueous medium, and further work deals
with the hole transfer in proteins, peptides and organic semi-conductors.










References

1. Kuba, T.; Elstner, M. J. Royal Soc. Interface, 10, 20130415, 2013.
2. Kuba, T.; Gutirrez, R.; Kleinekathfer, U.; Cuniberti, G.; Elstner, M. Phys. Status Solidi B,
250, 22772287, 2013.
3. Kuba, T.; Elstner, M. Phys. Chem. Chem. Phys., 15, 57945813, 2013.


57

Molecular Dynamic Simulations of the fucose transporter, FucP.

Joanna M. Lee and Philip C. Biggin

Structural Bioinformatics and Computational Biochemistry, University of Oxford, South Parks
Road, Oxford, OX1 3QU, United Kingdom.

joanna.lee@bioch.ox.ac.uk


The Major Facilitator Superfamily (MFS) is thought to transport via an alternating access
mechanism. For FucP the 2 6-helix transmembrane (TM) domains are sequentially open to the
periplasm, occluded and finally open to the cytosol permitting fucose release into the cell [1].
FucP has been identified as a proton/fucose symporter and so the protonation of 2 charged residues;
D46 and E135 in the TM cavity are thought to be key for initiating this mechanism via closure of
the outward-open state. A crystal structure of the mutated, outward-open FucP-N162A, with -
nonylglucoside (BNG) has been solved [1] and many residues key to fucose transport have now
been identified [2]. Atomistic simulations were conducted on the different protonation states of
FucP, with ligand bound to explain the dynamics required to instigate the initial step of the
alternating access mechanism. That is, the closure of the outward-open conformation seen in the
crystal structure.

References

1. Dang, S., et al., Nature, 467, 734-738, 2010.
2. Madej, MG., et al., PNAS, 110, 15, 5870-5874, 2013.



58

HIGHLY ENHANCED CONFORMATIONAL SAMPLING OF THE
TRANSMEMBRANE DOMAIN OF EGF RECEPTOR SHEDS LIGHT ON
THE ACTIVATION MECHANISM

Mickal Lelimousin
1
, Vittorio Limongelli
2
, Mark S.P. Sansom
1


1
Department of Biochemistry, University of Oxford, South Parks Road, OX1 3QU Oxford (U.K.)
2
Department of Medicinal Chemistry and Toxicology, University of Naples Frederico II (Italy)

mark.sansom@bioch.ox.ac.uk

The epidermal growth factor receptor (EGFR) is an in-depth studied receptor tyrosine kinase (RTK)
due to its importance in cell signalling and cancers. Yet the mechanism of activation that initiates
signal transduction cascades in the intracellular region is incompletely understood, although the
transmembrane (TM) domain is known to play an active role in both dimerization and allosteric
regulation of the receptor. Here we combine coarse-grained molecular dynamics [1] and well-
tempered metadynamics [2] simulation techniques to characterize the free energy landscape of the
TM domain of EGFR. The method (CG-MetaD) presents highly enhanced sampling features as
estimated by comparison to experimental kinetics measurements. The second timescale simulation
enables identification of dimerization pathways between TM -helices, which individual molecular
dynamics trajectories stochastically follow. The multidimensional free energy landscape suggests
two possible mechanisms of activation through either pivot or rotation motions of the TM helices.
Complementary modelling of the juxtamembrane (JM) domain supports the two proposed
mechanisms, in which the reversible dissociation of antiparallel interaction between JM segments
might be essential for reorientation of the kinase domains in the allosteric regulation. Overall the
mechanistic findings provide new important insights on EGFR activation to help the design of
novel therapeutics.



References

1. Bond, P. J. ; Wee, C. L. ; Sansom M. S. P. Biochemistry, 47(43) , 11321-11331, 2008
2. Barducci, A. ; Bussi, G. ; Parrinello, M. Phys Rev Lett, 100(2) , 020603, 2008
59

RedMDStream: A Tool to Design Coarse-Grained Molecular Dynamics Models
and Force Fields for Proteins and RNA

Filip Leonarski, Joanna Trylska

Centre of New Technologies, University of Warsaw,
ul. Zwirki i Wigury 93, 02-089 Warszawa, Poland
F.Leonarski@cent.uw.edu.pl


Coarse-grained molecular dynamics (CG MD) models help in overcoming spatial and temporal
limitations of full-atomistic MD methods in studying time evolution of biomacromolecules.
Reducing systems representation and implicit treatment of solvent results in orders of magnitude
decrease of computational cost. Performance gain is however accompanied by the loss of
universality. Although general-purpose CG MD models exist, they involve at least 5-8 pseudo-
atoms to describe a single residue (an amino acid or nucleotide). Simpler models, with 1-3 pseudo-
atoms per residue, are usually designed with a particular aim in mind, e.g., for tertiary structure
prediction of proteins and RNAs or for simulations of the dynamics of large complexes like
ribosome and nucleosome. However, each of these problems requires a highly specific model, built
solely for that purpose.

Although many models are available in the literature, they are often difficult to apply: (1) usage of
custom representation and functional forms usually binds the model to a particular CG MD engine,
(2) there are no methods to help in adapting one CG MD to a different task or molecule class by re-
optimizing the numerical parameters describing the potential energy function.

RedMDStream is a set of tools to overcome the mentioned problems:
a) An XML format defining a coarse-grained representation. User can easily (i) define the
transformation from full-atomistic representation to a set of pseudo-atoms, (ii) assign
topology and functional forms (including custom equations) of bonds, angles and dihedrals
connecting pseudo-atoms and (iii) numerical values of force field parameters. This
assignment can be also based on external data from programs like RNAView or DSSP. The
XML Format is oriented to resemble the way scientists describe force fields, rather than
focusing on computational details of how the transformation is performed.
b) A distance distribution calculation tool to apply the Boltzmann Inversion method to CG MD
models distance distributions can be calculated for CG trajectories but also as a reference
for structures gathered from the PDB database or full-atomistic simulation.
c) RedMD[1] CG MD simulation engine, providing Newtonian, Brownian and Langevin
dynamics with SHAKE/RATTLE algorithms.
d) CG MD model parameters optimization tools, including the evolutionary algorithm[2]
method, to automatically tune the numerical parameter values for a model, and to identify
correlations between these parameters. This can help in re-adaptation of existing CG MD
models to new tasks or molecule classes.
Application will be available on the laboratory website: http://bionano.cent.uw.edu.pl/Software.

References

1. Grecki, A. ; Szypowski M. ; Dugosz, M. ; Trylska, J. J. Comput. Chem., 30, 2364-2373, 2009
2. Leonarski, F ; Trovato, F. ; Tozzini, V. ; Le, A. ; Trylska, J. J. Chem. Theory Comput., 9, 4874
4889, 2013
60

THE NORM MATE TRANSPORTER FROM N. GONORRHEAE: INSIGHTS INTO DRUG & ION BINDING
FROM ATOMISTIC MOLECULAR DYNAMICS SIMULATIONS

Yuk Ming Leung
1
, Daniel A Holdbrook
1,2
Thomas J Piggot
1,3
and Syma Khalid
1*


1
School of Chemistry, University of Southampton, Highfield, Southampton, SO17 1BJ, UK.
2
Current address: Bioinformatics Institute. (A*STAR), 30 Biopolis Street, #07-37 Matrix, Singapore.
3
Current address: Detection Department, Defence Science and Technology Laboratory, Porton Down,
Salisbury, Wiltshire, SP4 0JQ
*
To whom correspondence should be addressed at:
E-mail: S.Khalid@soton.ac.uk Telephone: +44-2380-594176 Fax: +44-2380-593781


The multidrug and toxic compound extrusion (MATE) transporters extrude a wide variety of
substrates out of both mammalian and bacterial cells via the electrochemical gradient of protons and
cations across the membrane. The substrates transported by these proteins include toxic metabolites
and antimicrobial drugs. These proteins contribute to multidrug resistance in both mammalian and
bacterial cells and are therefore extremely important from a biomedical perspective. While specific
residues of the protein are known to be responsible for the extrusion of solutes, mechanistic details
and indeed structures of all the conformational states remain elusive. Here we report the first
simulation study of the recently resolved X-ray structure of the MATE transporter, NorM from
Neisseria gonorrheae (NorM_NG) [1]. Multiple, atomistic simulations of the unbound and bound
forms of NorM in a phospholipid lipid bilayer allow us to identify the nature of the drug-
protein/ion-protein interactions, and secondly determine how these interactions contribute to the
conformational rearrangements of the protein. In particular we identify the molecular
rearrangements that occur to enable the Na
+
ion to enter the cation-binding cavity even in the
presence of a bound drug molecule.


References
1. Lu, M., et al., Structures of a Na+-coupled, substrate-bound MATE multidrug transporter.
Proceedings of the National Academy of Sciences of the United States of America, 2013.
110(6): p. 2099-104.



61

Benchmarking large-scale DFT calculations on the chorismate mutase enzyme

Greg Lever*
a
, Daniel J. Cole
a,b
, Richard Lonsdale
c
, Kara E. Ranaghan
c
, David J. Wales
d
, Adrian J.
Mulholland
c
, Chris-Kriton Skylaris
e
and Mike C. Payne
a

a
Theory of Condensed Matter group, Cavendish Laboratory, 19 JJ Thomson Ave, Cambridge CB3
0HE, UK,
b
Department of Chemistry, Yale University, New Haven, Connecticut 06520-8107,
United States,
c
Centre for Computational Chemistry, School of Chemistry, University of Bristol,
Bristol BS8 1TS, UK,
d
University Chemical Laboratory, Lensfield Rd, Cambridge CB2 1EW, UK
and
e
School of Chemistry, University of Southampton, Highfield, Southampton SO17 1BJ, UK.

*gl319@cam.ac.uk


A benchmark study on a large portion of the enzyme chorismate mutase (CM) using linear-scaling
density functional theory (DFT) will be discussed. Combined quantum mechanics/molecular
mechanics (QM/MM) methods, where only the substrate and a few residues are treated quantum
mechanically have become important within computational enzymology. A recent QM/MM study
(Org. Biomol. Chem., 9, 157, (2011)) identified reaction pathways for the chorismate to prephenate
rearrangement in solution and catalysed by CM. However, recent advances in linear-scaling DFT
now allow the accurate prediction of transition state geometries and energetics whilst treating
systems, comprising thousands of atoms, fully quantum mechanically. Such an approach allows a
comparison with QM/MM approaches, using methods free from inaccuracies due to force field
parameterisation and coupling between QM and MM regions. The DFT code ONETEP uniquely
combines near-complete basis set accuracy with a computational cost that scales linearly with atom
number, allowing an accurate QM description of the enzyme. Large-scale DFT calculations on
structures from the CM pathways described above have been performed to address convergence of
energies of activation and reaction with the number of protein atoms surrounding the active site.
The calculations demonstrate the need for a DFT treatment in order to accurately determine
interaction energies between substrate and enzyme, including the strain induced in the enzyme.
Further understanding of enzymatic principles from an atomistic perspective will allow improved de
novo computational enzyme design, enabling biomimetic design principles to be drawn from
biological catalysts, utilising their properties to advance industrial catalytic processes.

62

MOLECULAR DYNAMICS STUDY OF EPIDERMAL GROWTH FACTOR RECEPTOR
ECTODOMAIN

Valeria Losasso
1
, Marisa L. Martin-Fernandez
2
and Martyn D. Winn
1,2


1
Scientific Computing Department, Science and Technology Facilities Council, Daresbury Laboratory,
Warrington WA4 4AD, UK
2
Central Laser Facility, Science and Technology Facilities Council, Research Complex at Harwell, Rutherford
Appleton Laboratory, Didcot OX11 0QX, UK.


valeria.losasso@stfc.ac.uk


Epidermal growth factor receptor (EGFR) belongs to the ErbB family of receptor tyrosine kinases,
activated by binding with growth factors of the EGF-family of proteins. This binding leads to
activation of signal transduction pathways that are involved in regulating cellular proliferation,
differentiation and survival. Deregulation of EGFR activity is frequently linked to the development
and progression of several human tumors.
The EGFR protein consists of an extracellular (EC) domain, a transmembrane (TM) alpha-helical
domain and an intracellular (IC) domain, composed of a juxtamembrane region, a kinase domain
and a C-terminal regulatory domain. Ligands bind to the EC domain, inducing a conformational
change which leads the IC kinase to form active asymmetric dimers.
Models of individual EGFR components, as well as of near-complete EGFR models (monomer,
inactive dimer, active dimer) have recently been reported and investigated by microseconds-long
molecular dynamics (MD) simulations [1]. Here we start from these models to undertake further all-
atom MD studies on EGFR EC and TM domains. We focus in particular on the role of the linker
connecting EC and TM (residues 605-618) on the conformational flexibility of the EC domain,
which has been never been addressed before due to its unstructured nature of this region.

References

1. Arkhipov, A. ; Das, R. ; Endres, N. F. ; Eastwood, M. P. ; Wemmer, D. E. ; Kuriyan, J. ; Shaw D.
E. Cell, 152, 557 - 569, 2013


63


DYNAMIC FINGERPRINT OF IMATINIB SENSITIVE KINASES

Silvia Lovera, Giorgio Saladino, Nicole Dlker, Francesco L. Gervasio
Department of Chemistry, University College London, London WC1E 6BT, United Kingdom.
silvia.lovera.12@ucl.ac.uk

Tyrosine kinases (TKs) are of crucial importance for signal transduction and their deregulation is
related to numerous human diseases, including cancer.
Among kinases, c-Abl and c-Src are of particular interest for cancer research. The BCR-Abl fusion
protein, is the initial cause of chronic myeloid leukemia disease (CML) and is successfully treated
with the powerful anti-cancer drug imatinib, which acts as ATP-competitor. On the contrary, c-Src
is not inhibited by the drug, despite they share a 45% of sequence homology and have the same
bound crystal structures.
In a previous work we identified a different profile of flexibility for the two tyrosine kinases Abl
and Src. The greater flexibility of Abl was found to facilitate the adoption of the druggable DFG-
out conformation
1
.
Imatinib, however, is not so selective and targets numerous kinases, beside c-Abl and c-Src, like c-
KIT, PDGFR, Lck, Met and many others, showing towards some of them an apparently
inexplicable lack of activity.
Moreover, an increasing problem concerning imatinib is the lost of activity towards Abl, due to the
onset of drug resistant mutations in CML patients under treatment.
Understanding the mechanisms which lead to these behaviours is of paramount interest to finally
develop new effective anticancer drugs, overcome drug resistance and gain a deeper insight into
kinase regulation and allostery.
In this multidisciplinary work, we first study the different affinity of imatinib towards TKs and Abl
resistant mutants by combining advanced sampling methods (e.g. PTMD, WTE) and multi-s
classical molecular dynamics simulations.
Then we experimentally validate the computational results with a rational mutagenesis of the kinase
c-Src. The chosen mutations were planned in order to change the dynamics nature of crucial
elements of the structure. The mutants have been expressed and purified in vivo and their K
D
for
imatinib finally recovered by means of ITC essays.
From our results, we could recognize flexibility as a possible hallmark of all the studied kinases
sensitive to Imatinib. These findings are confirmed not only by the big changes discovered in the
dynamics of the resistant mutants of c-Abl but also by the enhanced sensitivity towards the drug
gained by the designed mutants of c-Src.

References

1. Lovera S; Sutto L.; Boubeva R.; et al. Journal of the American Chemical Society, 134, 2496-
2499, 2012.


64


Multiscale Modelling of Drug-Polymer Nanoparticle Assembly

R. Mackenzie, C. Laughton, M. Garnett, C. Alexander, J. Booth

School of Pharmacy, University of Nottingham, University Park, Nottingham, NG7 2RD, UK

paxrm@nottingham.ac.uk


Drug molecules with poor stability and solubility can be encapsulated by biodegradable polymers to
form nanoparticles. These drug delivery systems provide slow release properties and a reduction in
off-target effects whilst potentially providing targeting to tumour sites via the EPR effect.

The potential of polymeric drug delivery systems is thwarted by a difficulty in encapsulating large
amounts of drug within the nanoparticles as they assemble. To further study how polymers interact
with drugs in these systems and attempt to improve their encapsulation efficiency we have utilised
molecular dynamics simulations.

Using GROMACS we have developed a multiscale model using the GROMOS and MARTINI
forcefields that allows for the simulation of the nanoparticle assembly process called
nanoprecipitation. Polymer dissolved in acetone is added drop-wise to water containing drug.
Acetone disperses in the water and this causes the polymer to aggregate and encapsulate
surrounding drug molecules.

In this poster we present our findings on two different polymers with dexamethasone phosphate.
We found that differences in encapsulation efficiency and drug loading between the two polymers
were in agreement with experimental data. In the future we hope to use our model to alleviate the
laborious experimental trial and error process in which new drug-polymer combinations are
discovered.


65

ELUCIDATION OF INTRINSIC DETERMINANTS OF FUSION
INITIATION IN FLAVIVIRUS ENVELOPE PROTEINS THROUGH
MOLECULAR DYNAMICS SIMULATIONS

Dmitry I. Osolodkin
1,2
, Liubov I. Kozlovskaya
2
, Evgenia V. Dueva
1,2
,
Galina G. Karganova
2
, Vladimir A. Palyulin
1
, Nikolay S. Zefirov
1


Department of Chemistry, Moscow State University, Moscow 119991 Russia
Chumakov Institute of Poliomyelitis and Viral Encephalitides, Moscow 142782 Russia
dmitry_o@qsar.chem.msu.ru

Flaviviruses form a widespread genus of arthropod-borne pathogens, including such socially
important ones as Dengue virus (DENV), West Nile virus (WNV), tick-borne encephalitis virus
(TBEV), Yellow fever virus (YFV), etc. Effective methods of treatment are not developed for these
diseases, and available therapeutic options include only symptomatic treatment.
Application of structure-based drug discovery methods requires deep understanding of the
molecular machinery underlying important steps of viral life cycle. One of such steps is fusion of
viral and cellular membranes, occuring in the acidic environment of the endosome. The fusion is
mediated by envelope glycoproteins E, forming the outer shell of the virion along with membrane
proteins M. Widely recognised 'histidine switch' hypothesis suggests that histidine residues are the
ones that initiate conformational rearrangement of the proteins upon protonation.
Recent advances in cryo-electron microscopy (cryo-EM) allowed structural biologists to obtain
structural data for DENV envelope proteins with resolution of 3.5 , including transmembrane
regions (PDB IDs 3J2P, 3J27). These data are a perfect starting point for molecular dynamics (MD)
simulations. We have constructed a homology model for TBEV envelope building block, consisting
of 2 E and 2 M subunits, and simulated 500 nanoseconds of molecular dynamics for two states with
neutral and protonated histidine residues, mimicking the states of the protein before and after fusion
initiation, using the explicit water and explicit POPC membrane.
The main idea of such extended simulations is to obtain a bit more 'realistic' representation of the
protein structure compared to the one provided by cryo-EM and homology modelling, which can
still be biased despite very high resolution of the structure. The structures are generally stable, but
certain differences between the protonation states are observed, including a tendency to the fusion
peptide release in the trajectory for the protonated state. The character of membrane-protein
interaction is described, pointing out protrusion of lipids surrounding the protein. The movement
pattern of outer surface of the envelope proteins allows construction of mechanistic hypotheses
regarding the interaction of the proteins with low-affinity cell receptors, being an important early
stage in virus-cell interaction.
The data from our simulation provide valuable insights into understanding of the early stages of
flaviviral membrane fusion.



66

EFFECT OF ATTRACTIVE INTERACTIONS ON THE WATER-LIKE ANOMALIES OF
A CORE-SOFTED MODEL POTENTIAL

Shashank Pant
1,2
, Tarun Gera
3
and Niharendu Chowdhury
4
1
Department of Chemical Sciences, Indian Institute of Science Education and Research-Kolkata,
Mohanpur-741252, India
2
Manchester Institute of Biotechnology, The University of Manchester, 131 Princess Street,
Manchester, M1 7DN and School of Chemistry, The University of Manchester, Oxford Road,
Manchester, M13 9PL
3
Department of Chemistry, Indian Institute of Technology-Delhi, New Delhi, 110016, India
4
Theoretical Chemistry Section, Bhabha Atomic Research Centre, Mumbai-400 085, India
shashankpant.lko@gmail.com

Understanding the origin of anomalous properties of water is of prime importance. A pertinent
question in this respect is: whether tetrahedral orientational interaction is a prerequisite for the
manifestation of water-like anomalies. Recently it is shown that spherically symmetric potentials
with two length scales, popularly known as core-softened potentials yield water-like anomalies. In
the present study, we investigate the effect of attractive interactions among the particles on the
existence and location of density anomaly in the temperature-density (T-) plane. Using a suitable
form of the core-softened potential, we employ extensive molecular dynamic simulations to
understand microscopic origin of the density anomaly in terms of local structure of the model fluid.
We observe that with the increase of the attractive interaction, the density anomaly region shifts
towards higher densities and higher temperatures. Many of the previous studies involving model
water and a class of core softened
1
potentials have concluded that the structural anomaly region
encloses the diffusion anomaly region, which in turn, encloses the density anomaly region, the same
pattern has also been observed in the present study for the systems with less depth of attractive well.
We found for the systems with deeper attractive well the diffusion anomaly shifts towards higher
density and not always enclosed by the structural anomaly region.

References
1. A. B. de Oliveira, P. A. Netz, T. Colla, and M. C. Barbosa, J. Chem. Phys., 125, 124503, 2006
2. Pant, S., Gera, T. and Chowdury, N. J. Chem. Phys., 139, 244505, 2013.




67

Probing the outer membrane of Pseudomonas aeruginosa using molecular
dynamics simulations

Jamie Parkin, Dr. Syma Khalid, Dr. Tim Carpenter

School of Chemistry, University of Southampton, Southampton, Hampshire, SO17 1BJ

J.Parkin@soton.ac.uk


Pseudomonas aeruginosa (PA) is a bacterium of particular interest due to increasing antibiotic
resistance. PA is a pathogen that can be fatal in persons with compromised immune systems.
Increasing resistance to current antibacterial agents has generated a requirement to develop not only
novel methods in drug design, but also a greater understanding of the drug delivery process. The
resistance of PA, in part, is due to a complex outer membrane (OM) arrangement, which has a low
permeability. Drug transport across the outer membrane is mediated by beta-barrel proteins, which
exhibit substrate specificity. It is this substrate specificity that provides the innate defence
mechanism. Two methods have been used to help understand the barriers that these substrates face.
The research of Eren et al. comments upon the specificity of the some Occ family proteins during
mutation studies, as well as a look at the OccD1 protein using molecular dynamics simulations [1].
We present here, using molecular dynamics (MD) and steered molecular dynamics (SMD), an
increased understanding of the arginine transport pathway through the PA outer membrane protein,
OccD1. Specific binding events thought to be key to transport have been observed; such as OccD1
loop movements, hydrogen bonding network interactions and transport substrate orientation.
Previously, Bemporad et al. have calculated the energetic profiles for small molecules across simple
symmetric lipid bilayer systems [2]. To further understand the barrier that membranes impose, we
have performed free energy calculations on a range of substrates across the complex OM to
determine the energy landscape that must be overcome for permeation. The results provide a key
step towards developing novel antibiotics for PA.
References

[1] Eren E. et al, Toward Understanding the Outer Membrane Uptake of Small Molecules by Pseudomonas
aeruginosa, J. Biol. Chem., (2013), 288, 12042-12053.
[2] Bemporad, D and Essex, J. W., Permeation of Small Molecules through a Lipid Bilayer: A Computer
Simulation Study, J. Phys. Chem. B, (2004), 15, 4875-4884.


68

Solvent Extension of MDmix methodology and its impact on the quality of the
predictions

Kevin Pinto Gil, Daniel lvarez Garcia, Xavier Barril

Departament de Fisicoqumica, Facultat de Farmcia, Universitat de Barcelona, Av. Joan XXIII s/n,
08028 Barcelona, Spain.

kpinto.gil@gmail.com

Binding hot spots are points on the protein surface displaying high propensity to interact with other
molecules. They can be found in protein-protein interfaces as well as on ligand binding sites, where
the hot spots inferred from the protein structure should overlap with pharmacophores derived from
the alignment of multiple ligands. Molecular Dynamics (MD) simulations with explicit
aqueous/organic solvent mixtures (MDmix) have been proposed as a means to identify binding hot
spots.(1)(2) This computational experiment is analogous to solvent mapping carried out by X-ray
crystallography and NMR, which are well known and accepted means to detect binding sites and
their corresponding hot spots.(2)

MDmix simulations can use a wide range of small organic molecules as co-solvents, hence it is
important to know if the information provided is redundant or can be complementary. In this work,
we first expand the set of organic molecules to be used in MDmix in order to cover a wide range of
chemical moieties. We then proceed to evaluate the information provided by MDmix simulations of
these solvents on beta-secretase 1 (BACE-1), comparing the results with the exhaustive structural
and pharmacological information available for the system. The most important pharmacophoric
points defined by the ligands coincide with the most intense binding hot spots predicted by MDmix,
and we find a large degree of overlap between different atom types (e.g. aromatic vs. aliphatic), but
also some notable differences. This allows us to propose a ranked list of solvent types to be used in
MDmix studies, to gain the maximal amount of information with the minimum number of
simulations.

Finally, we have identified some currently unexploited hot spots and selected compounds for in
vitro testing that should be able to interact with, and gain binding affinity from, those points.

References

1. MacKerell, Jr., A. D., Bashford, D., Bellott, M., Dunbrack Jr., R. L., Evanseck, J. D., Field,
M. J., Fischer, S., Gao, J., Guo, H., Ha, S., Joseph-McCarthy, D., Kuchnir, L., Kuczera, K.,
Lau, F. T. K., Mattos, C., Michnick, S., Ngo, T., Nguyen, D. T., Pr M. No Title. J Phys
Chem B. 1998;102:3586616.
2. Seco J, Luque FJ, Barril X. Binding Site Detection and Druggability Index from First
Principles. 2009;236371.


69

Accurate Prediction of the Thermodynamics and Kinetics of Drug
Binding

Giorgio Saladino, Francesco L. Gervasio

Institute of Structural and Molecular Biology and Department of Chemistry,
University College London, 20 Gordon Street, London WC1H 0AJ, UK

g.saladino@ucl.ac.uk

Nowadays, the complex machinery behind the discovery of new drugs is facing a setback,
as the ever increasing yearly investments in R&D fail to translate into the approval of new
drugs on the market [1]. As few as 7 new compounds were marketed by the largest
companies in the last 5 years, leading to the so-called curse of attrition [2]. One of the
reasons for this situation is that many compounds in the drug discovery pipeline fail in late
stages due to lack of efficacy and toxicity. In particular, the importance of drug bindi ng
kinetics is increasingly recognized as an important factor in drug development [3]. To
address this problem, many computational methods are being developed that try to predict
these parameters in early development stages [4]. We recently demonstrated that
Metadynamics [5] can be effectively applied in a pharmaceutical workflow to predict free
energies of binding and rank drug candidates [6][7]. Here, we move a step further and
apply our recently developed method, Transition State Partial Path Transition Interface
Sampling (TS-PPTIS) [8] to predict the thermodynamics and kinetics of binding for a new
drug targeting to the Heat Shock Protein 90 (HSP90) molecular chaperone.

References

1. Walker, S. M & Davies, B. J. Drug Discov. Today, 16, 467471, 2011.
2. Bunnage, M. E. Nat. Chem. Biol. 7, 3359, 2011.
3. Swinney, D. C & Anthony, J. Nat. Rev. Drug Discov., 10, 50719, 2011.
4. Van Drie, J. H. J. Comput. Aided Mol. Des. 21, 591601, 2007.
5. Laio, A. & Parrinello,M. Proc. Natl. Acad. Sci. USA 99, 12562, 2002.
6. Saladino, G., Gauthier, L., Bianciotto, M., & Gervasio, F.L. J. Chem. Theory Comput. 8, 1165-
1170, 2012.
7. Herbert, C. et al. Cancer Cell, 23, 489-501, 2013.
8. Juraszek, J., Saladino, G., van Erp, T.S. & Gervasio, F.L. Phys. Rev. Lett.,110, 108106, 2013.


70

Development of a New Series of Bis-Triazoles as Anti-Tumour Agents

Maysaa Saleh
1
, Christopher Moody
2
, Charles Laughton
1


School of Pharmacy and School of Chemistry, University of Nottingham, Nottingham NG7 2RD,
UK

paxms7@nottingham.ac.uk
maysao40@hotmail.com


The unlimited proliferation of cancer cells depends on a telomere maintainance mechanism which is
most commonly provided by the telomerase enzyme
1
. The telomeric ends form structures called G-
quadruplexes. Stabilization of these structures by small binding molecules called G4 ligands
inhibits telomere elongation, and targets telomere maintainance mechanisms, resulting ultimately in
delayed cell death and abrogation of tumourigenicy in vivo
2
. In this project, we are developing a
new series of triazoles which are designed to bind to and stabilize G-quadruplex structures
selectively, and which may therefore have potential as anti-cancer drugs. These triazoles are
synthesized via 1,3-dipolar cycloaddition reactions of organic azides with quinone derivatives
which afford two possible bis-triazole products: centrosymmetric and noncentrosymmetric
regioisomers. The different regioisomers are expected to have different DNA-binding
characteristics. Quantum mechanics calculations have been used to investigate these reactions, and
predict that the centrosymmetric regioisomer is the most stable geometry and favoured product. The
target ligands have been produced through multi-step synthesis in moderate to good yields, as the
QM-predicted regioisomers. Their selectivity for G-quadruplex DNA over duplex DNA was
investigated using a FRET-based assay. This showed four compounds to be moderately effective
G4 binders over the concentration rage examined, with particular affinity for the quadruplex formed
by the Hsp90a promoter sequence. Encouragingly, good selectivity for G-quadruplex DNA vs.
duplex DNA was displayed by all the ligands over the concentration range used. To rationalize the
affinity of the ligands in binding to G4-DNA, docking studies were performed using Glide
programme in which the ligands were docked to both 3- and 5-sides of the telomeric parallel G4
template, and cluster analysis was used to analising docking data and grouping similar docking
poses into clusters. The study showed good correlations between Glide docking scores and the
experimental FRET results for two clusters, and revealed the reason for the moderate binding is the
presence of two sp
3
methylene groups within the aromatic region that affects the - stacking of the
ligands on to the G4 scaffold. An optimised new series of bis-triazoles were designed and docked to
the same template of G4-DNA, then analised in the same way. The optimised structure for best
binding affinity toward telomeric G4-DNA was determined, by using the good correlations from the
x-study. MD simulation studies will be performed to explore the binding interction motion between
the telomeric parallel G4 and both synthesised and optimised bis-triazoles, and calculate their
binding energies.


References

1. Moorhouse, A. D.; Haider, S.; Gunaratnam, M.; Munnur, D.; Neidle, S.; Moses, J. E., Mol.
BioSyst., 4, 629642, 2008.

2. Campbell, N. H.; Patel, M.; Tofa, A. B.; Ghosh, R.; Parkinson, G. N.; Neidle, S.,
Biochemistry, 48, 16751680, 2009.


71

Investigating the reliability of the MM-GBSA method for
predicting protein-ligand binding free energies

Twana Salih, Weng Chan, & Charlie Laughton

Division of Medicinal Chemistry and Structural Biology, School of Pharmacy and Centre for Biomolecular
Sciences, University of Nottingham, NG7 2RD
Paxts2@nottingham.ac.uk

In the early stages of rational (structure-based) drug design and discovery projects, the Molecular
Mechanics Generalized Born Surface Area (MM-GBSA) method is extensively exploited to assist
researchers decide on optimal molecules for experimental validation. However, there is now much
evidence that correlations between predicted and experimental binding affinities are variable and
system dependent. We have recently embarked on a cancer drug discovery project based on the
design of inhibitors of key protein-protein interactions within telomeres. Specifically, we are
seeking, based on crystal structure data, peptide-like molecules that compete with the telomere
component TIN2 for its binding site on TRF1.
Standard MM-GBSA methodologies have been used to predict binding affinities of five TIN2-like
peptides for TRF1. Analysis of large numbers (up to 50) of replicate MD simulations of each
system reveal a wide spread of calculated G values. We present an analysis of this data, asking the
question as to whether this variability is an inevitable consequence of insufficient sampling, or
whether other factors are at work that may be corrected for. In addition, fluorescently labeled
versions of these peptides have been synthesized and experimental binding affinities determined;
correlations between the predicted and experimental binding affinities of the systems will be
presented.
References
1. Alder, M. & Beroza, P. J. Chem. Inf. Model, 53 (8), 20652072, 2013
2. Srivastava, HK. & Sastry, GN, J. Chem. Inf. Model, 52 (11), 3088-98, 2012


72

IMPROVING DRUG DELIVERY: COMPUTATIONAL STUDIES OF
PROTON DEPENDENT OLIGOPEPTIDE TRANSPORTERS

Firdaus Samsudin, Nicolae Solcan, Mark SP Sansom, Simon Newstead, Philip W
Fowler

Department of Biochemistry, University of Oxford, South Parks Road, Oxford OX1 3QU

mohd.samsudin@wolfson.ox.ac.uk


The human peptide transport system is a promising route for drug absorption. In addition to taking
up di- and tripeptides from protein digestion, the proton-coupled transporters PepT1 and PepT2 also
show affinity for a broad range of peptide-mimetic drugs like -lactam antibiotics and bestatin.
However, in the absence of structural data, rational drug design targeting these transporters has been
hitherto restricted to empirical approaches. The crystal structures of the bacterial homologs, PepT
So

[1] and PepT
St
[2], provide the first framework for understanding the key protein-ligand interactions
and building an accurate pharmacophore model.

Using molecular dynamics simulations, the binding of peptides on PepT
St
was investigated. Various
methods such as scoring functions (AutoDock), approximate free energy methods (Linear Interation
Energy (LIE) and Molecular Mechanics Poisson Boltzmann/Generalized Born Surface Area (MM-
PB/GBSA)) and rigorous free energy method (Thermodynamic Integration (TI)) have been used to
estimate binding free energy. Our results from LIE, MM-PB/GBSA and TI show good correlation
to experimental IC50 values from in vitro transport assays. To examine the binding preference of
this transporter, we predicted the binding energy of another 100 di- and tripeptides of various
residue combinations using LIE, and selected results were refined using TI. Overall, our studies
suggest that peptide transporters favor hydrophobic and polar ligands while the charged ones,
especially acidic residues, display much lower affinity. For dipeptides, the N-terminal side chain is
a crucial determinant towards overall binding affinity, while the C-terminal side chain is more
essential in tripeptides a stark contrast owing to the distinct binding poses. Free energy
decomposition points towards the importance of charged residues around the binding pocket.

References

1. Newstead, S. ; Drew, D. ; Cameron, A. D. ; Postis, V. L. G. ; Xia, X. ; Fowler, P. W. ; Ingram, J.
C. ; Carpenter, E. P. ; Sansom, M. S. P. ; McPhearson, M. J. ; Baldwin, S. A. ; Iwate, S ; Embo J.,
30(2) , 417-426, 2011

2. Solcan, N. ; Kwok, J. ; Fowler, P .W. ; Cameron, A. D. ; Drew, D. ; Iwata, S ; Newstead, S. ;
Embo J., 31(16) , 3411-3421, 2012




73

Scrutinizing Hybrid AA/CG Force Fields through Free Energy Calculations

Alexander B. Kuhn, Srinivasa M. Gopal, Marten Prie, Lars Schfer

Lehrstuhl fr Theoretische Chemie, Ruhr-University Bochum, Germany

Lars.Schaefer@ruhr-uni-bochum.de

Hybrid AA/CG models in which atomistic (AA) solutes are embedded in a coarse-grained (CG)
environment can substantially increase the computational efficiency of biomolecular simulations as
compared to fully atomistic representations. However, interfacing both levels of resolution is a
major challenge that requires a balanced description of all relevant interactions. We developed and
critically tested hybrid AA/CG schemes that combine different atomistic and coarse-grained force
fields in such a way that electrostatic interactions between the two regimes are explicitly taken into
account. This coupling is particularly important for polar solvents, such as water, that screen the
electrostatic interactions. To enact electrostatic coupling, various recently developed polarizable CG
water models were used. In addition to using a purely CG solvent, we also investigated the
influence of atomistic solvation shells of increasing size around the solutes. Potentials of mean
force (PMFs) between pairs of amino acid side chains and solvation free enthalpies of uncharged
amino acid side chains obtained from the hybrid AA/CG simulations are compared to the respective
data from fully AA and fully CG simulations. Finally, the structure and dynamics of small atomistic
proteins in CG water and of atomistic membrane proteins embedded in solvated CG lipid bilayers
are discussed, with a focus on the observed over-stabilisation of intra-protein interactions due to the
lacking hydrogen bonding capability of the CG solvent. Our work highlights some key challenges
and suggests possible solutions on the way toward hybrid AA/CG models that are both
computationally efficient and at the same time sufficiently accurate to address biomolecular
systems.

References

1. T.A. Wassenaar, H.I. Ingolfsson, M. Prie, S.J. Marrink, L.V. Schfer. J. Phys. Chem. B, 117
(13), 3516-3530, 2013.



74

HOW DYNAMIC TRANSMEMBRANE HELICES CAN INFLUENCE
PEPTIDE/LIPID INTERACTIONS: A MOLECULAR DYNAMICS STUDY

Christina Scharnagl
*
, Matthias Mrch
#
, Alexander Gtz
#
, Dieter Langosch
#


*
Technische Universitt Mnchen, Fakultt Physik, Freising, Germany
#
Technische Universitt Mnchen, Lehrstuhl Chemie der Biopolymere, Freising, Germany

christina.scharnagl@tum.de


Integral membrane proteins exhibit conformational flexibility at different structural levels and time
scale. Our objective is to systematically study the flexibility of transmembrane helices and its
impact on the structure of surrounding membranes. To this end, we have developed a set of low-
complexity peptides (LV-peptides) which are designed by mixing helix-stabilizing and helix-
destabilizing amino acids in the hydrophobic core and adding basic flanking residues. The backbone
dynamics of the helices is investigated by deuterium/hydrogen-exchange experiments that report
local and transient unfolding reactions of intrahelical hydrogen bonds.
1
Their impact on the
membrane is characterized by functional assays which report that these peptides induce lipid splay
where one acyl chain transiently protrudes to the surface of the bilayer, catalyse the translocation of
lipids from one bilayer leaflet to the other one
2
(lipid flip) and bind lipids selectively. These
activities increase with the flexibility of the helices.

To answer the question of how peptide/lipid interactions can modulate membrane structure and
catalyse lipid-specific flip and splay we use all-atom molecular dynamics (MD) simulations in
hydrated mixed bilayers and compare structure, dynamics and interaction of peptides, lipids and
water. The slow rate of lipid diffusion (~10
-7
cm
2
/s) implies insufficient lipid mixing within current
state-of-the art simulation times (several hundred ns for hydrated membrane patches with a
peptide/lipid ratio of 1/200, ~90 000 atoms). To sample different lipid environments around the
peptides and to improve statistical convergence we set up multi-copy simulations where the
peptides face initially different randomly composed lipid micro-environments. To analyse the
datasets across multiple simulations we use clustering techniques and similarity measures based in
information theory. Trajectories are produced with NAMD2.9 (CHARMM27 force field) running
on 800 Intel Sandy Bridge-EP cores with an output of ~25 ns/day at the SuperMUC HPC system
(Leibniz Computing Center, Garching, Germany).

The MD simulations reveal long-range lipid ordering in several shells, which is interestingly more
pronounced around the more dynamic peptides, where the surrounding lipids exhibit also a higher
lateral diffusion coefficient. The formation of hydrogen-bonds between lipid headgroup phosphates
and the basic flanking residues is responsible for the formation of the primary lipid shell as well as
for membrane deformations (bilayer thinning) around the peptide termini. In addition, the charge
state of the peptide termini defines the headgroup specific lipid affinity. Furthermore, lipids in the
primary shell around the more flexible peptides tend to splay their tails more strongly as compared
to the more rigid peptides. The results from the simulations support a model
2
where strong binding
of lipids with acidic headgroups in the first shell promotes lipid flip from a disordered second shell
of lipids around the peptides.

References

1. Quint, S.; Widmayer, S.; Minde, D.; Langosch, D.; Scharnagl, C. Biophys. J. 99, 2541-2549, 2010
2. Langer, M.; Sah, R.; Veser, A.; Gtlich, M.; Langosch, D. Chem. Biol. 20, 63-72, 2013

75

BREAKING DOWN PROTEIN CONFORMATIONAL TRANSITIONS WITH
COARSE-GRAINED MODELS

Pedro Sfriso, Agusti Emperador and Modesto Orozco

Institute for Research in Biomedicine (IRB Barcelona), Barcelona, Spain
Joint BSC-CRG-IRB Research Program in Computational Biology, Barcelona, Spain
pedro.sfriso@irbbarcelona.org


We present a computational method to trace conformational transitions in proteins. The method
uses discrete molecular dynamics as engine to sample protein conformational space. Protein
conformational transitions are treated as topological changes by means of structure-based potentials
from more than one structure. This algorithm is then coupled to up to two biasing methods: i)
metadynamics adapted to respect proteins easiest deformation pattern, and ii) a Metropolis test to
gain efficiency. The method shows an unprecedented computational efficiency, it is able to trace
with success rate >95% a wide range of known experimental transitions. Contrary to morphing
methods our strategy does not introduce distortions in the chemical structure of the protein and is
able to reproduce well very complex non-linear conformational transitions. The physical nature of
our approach is compatible with any perturbation analysis showing promising insights of protein
movements. In cases where only one structure is available, we combine sequence information, like
conservation or co-evolution of residues, to enrich the sampling along relevant conformational
states. When tested along known conformational transitions, this hybrid evolution-physical method
finds the target structure with reasonable resolution.

References

1. Sfriso, P. ; Hospital, A. ; Emperador, A.; Orozco, M. Bioinformatics, 29 , 1980 -1986, 2013


76

Towards faster-to-implement and extensible python-based tools for molecular
simulation data analysis
Ardita Shkurti
1
, Ramon Goi
2
, Modesto Orozco
2,3,4
, Charles Laughton
1*

1
School of Pharmacy, The University of Nottingham, Centre for Biomolecular Sciences, Clifton
Boulevard, University Park, Nottingham, NG7 2RD
2
Barcelona Supercomputing Center, Jordi Girona 31, Barcelona 08034
3
National Institute of Bioinformatics, Parc Cientfic de Barcelona, Josep Samitier 1-5,

4
Departament de Bioqumica, Facultat de Biologa, Avgda Diagonal 647, Barcelona
ardita.shkurti@nottingham.ac.uk
Principal component analysis (PCA) techniques emerge as a winning approach in two main
aspects of biomolecular simulation: (i) Insights into structural, dynamical behaviour and
conformational space of biomolecules; (ii) High compressed data storage in the context of the
potentially huge sizes of modern simulation trajectory files. For example, PCA-based compressed
files can preserve more than 90% of the variance of a typical data set, with a size less than one tenth
of the original trajectory file.
In addition, the tremendous improvements of hardware in terms of speed, capacity and decreasing
of storage and per-core costs have enabled the availability and accessibility of large amounts of data
especially in the field of molecular simulation. In this scenario, applying extensible and faster-to-
implement programming techniques to the PCA analysis of molecular simulation data becomes a
priority. In this context, the Python programming language has many attractions: (i) Python requires
less code to express the same concepts than high-level programming languages such as Fortran,
C/C++, Java etc.; (ii) Python provides a large and continuously increasing community; (iii) Python
provides a growing support for scientific packages that developers might easily load, integrate and
extended into their own codes.
In the present work, a Python-based PCA analysis tool (PyPCAZIP) has been developed,
amplifying the analysis functionalities of its Fortran and C-based predecessors [1], that include: (i)
Better handling of memory issues when dealing with very large data sets; (ii) On-the-fly selection
of subsets of atoms of interest for the PCA analysis from the available data sets (thus avoiding
unnecessary trajectory duplication); (iii) Flexible support for the simultaneous analysis of multi-
trajectory datasets that vary in their molecular topology and number of atoms.
Extending our previous work [2], we will discuss results from the application of this enhanced
analysis code to the investigation of ligand-induced changes in molecular flexibility in the Major
Urinary Protein (MUP). We have generated 100 independent 100 ns simulations of the apo-protein
and another 100 of the complex with the pheromone isobutylmethoxypyrazine, i.e. a total of 20
microseconds of data and two million snapshots. We will describe how pyPCAZIP has permitted
the rapid and quantitative analysis of this data, providing new approaches to the analysis of
equilibration and convergence. Ligand-induced changes in molecular flexibility, both at the macro-
level of configurational entropies and at the micro-level of residue-specific changes in dynamics,
will be presented and compared with pervious estimates, which relied on much smaller datasets.
References
1. Meyer, T. ; Ferrer-Costa C. ; Perez A. ; Rueda M. ; Bidon-Chanal A. ; F. Luque J. ; Laughton C. A. ;
Orozco M. J. Chem. Theory Comput., 2 (2), 251258, 2006
2. Roy, J; Laughton, CA; Biophys J. ; 99(1): 218-226, 2010


77

ABSTRACT WITHDRAWN



78

MOLECULAR DYNAMICS STUDY ON THE STRUCTURAL TRANSITION
OF DNA UNDER SUPERHELICAL STRESS

Thana Sutthibutpong
1
, Gabriel Gouvea-Slade
2
, Agnes Noy
1
, Charlie Laughton
3
, and Sarah Harris
1,4


1. School of Physics and Astronomy, University of Leeds, UK.
2. Sao Paulo State University, Sao Jose do Rio Preto, Brazil.
3. Centre for Biomolecular Sciences, School of Pharmacy, University of Nottingham, UK.
4. Astbury Centre for Structural and Molecular Biology, University of Leeds, UK.


s.a.harris@leeds.ac.uk


The study of the mechanical properties of DNA plays an integral part in our understanding of
genetic regulation [1][2]. We have performed a series of atomistic Molecular Dynamics (MD)
simulations of negatively supercoiled 108bp DNA minicircles to investigate how superhelical stress
affects DNA structure and dynamics. At a superhelical density of -0.10, the superhelical stress is
strong enough for structural transitions, such as kinks or denaturation bubbles, to occur within the
DNA [3]. Our study of the sequence dependence of supercoiling driven DNA denaturation led to an
unexpected result; DNA minicircles containing a totally random sequence (equal numbers of AT
and GC base pairs) can be denatured more easily those containing the AT-rich Far Upstream
Sequence Element (FUSE) [4]. We later compared the mechanical properties of these minicircles,
and found that the global shape of the molecule affects the propensity for the DNA to denature, as
well as the sensitivity of the sequence to the torsional stress.


References

1. Kouzine, F. ; Gupta, A.; Baranello, L. ; Wojtowicz, D. ; Benaissa, K. ; Liu, J. ; Przytycka, T. M. ;
Levens, D. Nat. Struct. Mol. Biol., 20(3), 396403, 2013

2. Dorman, C. J. ; Biochemical Society Transactions, 41(2), 2013

3. Du, Q. ; Kotlyar A. ; Vologodskii, A. Nucleic Acids Research, 36(4), 1120-1128, 2008

4. Kouzine, F. ; Liu, J. ; Sanford, S. ; Chung, H. J. ; Levens, D. Nat. Struct. Mol. Biol., 11(11),
10921100, 2004



79

Dynamics of retinal chromophore in rhodopsin: from cis/trans isomerisation to
activation

Siri van Keulen, Ursula Rothlisberger

SB - ISIC LCBC, cole Polytechnique Fdrale de Lausanne, BCH 4108, CH - 1015 Lausanne,
Switzerland

siri.vankeulen@epfl.ch



Rhodopsin is a G-Protein Coupled Receptor (GPCR) that is present in the human eye. This receptor
is part of the class A GPCRs and forms a template for all the 662 receptors that are included in class
A. Hence, if the process of rhodopsin activation can be understood, it could provide new insight
into the activation mechanism of also other GPCRs. Rhodopsin is a membrane protein that consists
of seven trans membrane domains, three intracellular loops, three extracellular loops, a C terminus,
an N terminal domain and a retinal molecule that is covalently bound to the receptor in the active
site. When rhodopsin is exposed to light, the covalently bound retinal is excited and isomerises
from a cis to a trans configuration. The cis-trans isomerisation induces a cascade of conformational
and configurational changes that lead to the active state of the receptor.
Over the last decades, the photo activation cycle of rhodopsin has been characterized. However, the
relationship between the relaxation of the retinal to the trans configuration and its effect on the
structure of the protein is not fully understood. Secondly, an atomic force microscopy study found
that rhodopsin proteins were highly ordered in rows of two monomers in native membrane. Because
of this high order the question arose whether rhodopsin transports the signal of its active state as a
monomer or if a rhodopsin dimer is more efficient.
The presented work will cover the relaxation process of a rhodopsin dimer after light exposure.
Only one out of the two retinal molecules present in the dimer was isomerised in order to monitor
not only activation, but also the effect of activation on an inactive rhodopsin protein. Quantum
mechanic/moleculer mechanics molecular dynamics, performed with CPMD, and classical
molecular dynamics, using Gromacs, were employed to monitor the occurring events. Currently,
also multiconfigurational time dependent hartree is performed on the rhodopsin protein in
combination with the local control technique. This combination of techniques provides the
opportunity to monitor the wave packet distribution after cis/trans isomerisation is induced by an
optimal pulse.



80

A Molecular Dynamics study of mechanisms of sequence recognition and DNA
binding in two telomeric proteins, TRF1 and TRF2

Milosz Wieczor, Pawel Wityk, Adrian Tobiszewski, Jacek Czub

Department of Physical Chemistry, Gdansk University of Technology
Narutowicza 11/12, 80-233 Gdansk

jacczub@pg.gda.pl

Telomeres are large nucleoprotein assemblies that form at the ends of linear chromosomes to
provide protection and stabilization of chromosomal termini. They were developped as a solution to
the end replication problem, which results in steady shortening of eukaryotic chromosomes. In
many species, including humans, telomeric DNA consists of long runs of repeating 6-bp tandem
sequence 5'-TTAGGG-3', so that it can be lost and restored over the course of cell divisions with no
loss of genetic information. A ribonucleoprotein called telomerase is capable of extending telomeric
DNA if necessary, and is a key inducible factor that allows for unlimited proliferation of a cell.
Besides that, telomeres are particularly sensitive to oxidative stress, and can presumably act as
sensors of environmental stress. For those reasons telomere maintenance is a crucial issue in both
cellular aging and carcinogenesis.
Structural maintenance of telomeres is regulated by a six-component protein complex called the
shelterin, which is recruited to telomeric DNA by means of sequence-specific binding of two
telomere repeat-binding factors, TRF1 and TRF2. These two homologues, although similar in
structure, have been found to govern different telomere-associated pathways. Both are, however,
indispensible for telomere protection due to their multifunctionality they promote the folding of
functional telomeric structure and act as accessory or inhibitory agents for other proteins. Since
their depletion leads to senescence or apoptosis in proliferating cells, they are often considered as
potential drug targets.
In the current study, we report on the molecular details of sequence recognition and binding
mechanism of DNA-binding domains of TRF1 and TRF2. We identify and quantify key
determinants of the DNA binding process by decomposing the free energy profile, and interpret the
obtained results on the structural level. With this approach, we draw conclusions regarding the
consecutive stages of sequence-specific association and propose a novel aspartate-dependent
mechanism of sequence recognition. By comparing our results with experimental data we also
highlight the applicability of free energy methods to studying protein-DNA interactions.



81

Lipase enzyme interactions with lipid bilayers
Nathalie Willems, Prof. Mark Sansom
Biochemistry Department, University of Oxford, South Parks Road, OX1 3QU
nathalie.willems@bioch.ox.ac.uk

The immobilisation of enzymes has played an instrumental role in the development of the
biotechnological industry. The advantages offered by immobilisation, such as increased enzyme
stability and up-scaling processes, have spurred much experimental investigation into optimising
the interactions between the enzyme and surfaces of interest. The lipase class of enzymes has been
the subject of much recent research due to its relevance in applications such as biodiesel and
polyester production. However, specific factors such as how immobilisation affects the structural
dynamics, stability, and activity of enzymes remain unclear. In order to develop a comprehensive
understanding of how different surfaces affect the conformational dynamics of the enzyme in
question, one needs access to high-resolution structural data. Here, we apply a multi-scale
molecular dynamics (MD) approach to investigate these types of questions.

Using coarse-grained molecular dynamics (CG-MD), we can simulate interactions within model
systems composed of different lipase enzymes and membrane surfaces of varying lipid composition
and size. From these systems, we are able to assess the residues implicated in interactions between
the enzyme and bilayer, and derive conclusions about possible binding modes involved depending
on the lipid composition of the membrane. Both enzymes display a common interfacial activation
mechanism observed in many lipase enzymes, and we can assess how interactions with a fluid
membrane impacts the dynamics of the regions implicated in this process. The CG simulations can
then be refined to atomistic representations to provide a higher level of detail regarding the forces
and residues that drive these interactions. This information can be used to provide predictive
insights that focus on optimising enzyme stability and binding, in the context of enzyme
immobilisation.


Exploring the Interaction of Agonists with the human alpha1 Glycine Receptor
Rilei Yu
1
, Timo Greiner
2
, Elliot Hurdiss
2
, Remigijus Lape
2
, Lucia G Sivilotti
2
, Philip C Biggin
1

1
Department of Biochemistry, University of Oxford, South Parks Road, OX1 3QU, Oxford, UK

2
Department of Neuroscience, Physiology & Pharmacology, University College London, WC1E
6BT, London, UK
The glycine receptors (GlyR)s are anion selective ligand-gated ion channels that are members of the
Cys-loop receptor family. The GlyRs play central roles in mediating inhibitory neurotransmission in
the spinal cord and brain stem and excitatory neurotransmission in embryonic neurons. They have
also been implicated in many disease states including tinnitus, spasticity and chronic pain. Thus,
they are considered important targets for drug design. Despite significant progress in recent years,
the determinants for efficacy of activation of different agonists are, at present, still not well
understood. In this study, computational modelling techniques including homology modelling,
docking, molecular dynamics simulations, and potential of mean force (PMF) calculations were
used to investigate the properties that confer efficacy on a series of compounds at the molecular
level in combination with electrophysiology experiments.
Homology models of the GlyR1 were built based on the GluCl channel from C. elegans, and the
binding modes of different agonists were investigated using docking and MD simulations. The
accuracy of our models was validated by site-directed mutagenesis studies in combination with
patch-clamp experiments. Our results suggest that the efficacy of the agonists is closely coupled to
the energetics of the C-loop of the receptor. With the presence of different ligands in the binding
sites, the C-loop closure energy profiles given by potential of mean force (PMF) calculation reveal
key differences that correlate with their overall efficacy. The results suggest that the combinat ion
of modelling and patch-clamp experiments for agonists can be a powerful approach to deciphering
the atomistic details of glycine receptor activation.