Predicting drug-target binding kinetics through multi-scale simulations

Lead Research Organisation: University College London
Department Name: Chemistry

Abstract

Drug-target binding kinetics has recently emerged as a critical parameter in determining the in vivo efficacy and toxicity of lead compounds. Unfortunately, the rational optimisation of this parameter to design more effective and less toxic drugs is extremely difficult as the features of small ligands and their protein targets that affect binding kinetics remain poorly understood.
The aim of this project is to systematically fill this knowledge gap by combining state of the art computational approaches. We propose to combine our state-of-the-art enhanced sampling and QM/MM methods to compute reliable association free energy profiles and rationalise the binding kinetics of receptor-ligand complexes in terms of the structures and energies of the transition state ensemble.
We will test and validate these methods, which have not been previously combined for this purpose. Together, they tackle the two central challenges of biomolecular simulation, i.e. conformational sampling and accuracy of potentials. We will apply the techniques we develop to well-characterised biomolecular systems of real biomedical importance, such as the influenza target neuraminidase, the anti-cancer targets p38a, Abl, Src and FGFr tyrosine-kinases, the chaperone HSP90, and other drug targets in collaboration with our pharmaceutical industrial partners (UCB, GSK and Sanofi). The results will shed new light on the kinetics of drug binding and their molecular origins. The methods we develop should find wide application, and we will make them available and accessible.

Planned Impact

The proposed project will 1) lead to new computational methods that can predict the binding kinetics of small molecules to protein targets, and 2) apply these methods on key pharmaceutical test-cases, such as anti-cancer targets. Although drug-target binding kinetics have recently emerged as a critical parameter for the efficacy and toxicity of drug leads, there are currently no effective techniques that can assist in rationally optimizing binding kinetics. By combining state-of-the-art biomolecular simulation approaches, we will develop computational tools that can efficiently and accurately map the binding free energy pathway (and thereby the binding kinetics) of small molecules to biomolecular targets. These tools are likely to have significant impact on the drug-design process in pharmaceutical companies, and we are indeed collaborating with major pharmaceutical companies to apply and validate the methods and tools we develop. In addition to informing drug-design, the methods can also be used to increase our understanding of biochemical networks in organisms and their regulation more generally, as binding kinetics plays an important role. The methods we develop are therefore expected to be taken up in industry as well as academia; to facilitate this, we will make easy-to-use computational tools and workflows freely available (and advertise them).

In the short term, the project will deliver methods that can accurately predict binding kinetics, and, by application of these tools, offer detailed insight into the small-molecule binding pathways to key targets of biomedical importance, such as established influenza and anti-cancer targets. This insight can directly inform the ongoing research and drug development in industry, through our collaboration with several major pharmaceutical companies.

In the medium term, the methods that will be developed are likely to form a key new step in the tool-chain for drug development. By systematically predicting the binding kinetics of lead compounds to their targets and relevant off-target proteins, the methods have the potential to significantly reduce failure rates in clinical trials, e.g. due to in vivo toxicity or side-effects. This has obvious direct economic benefits; clinical trials are the most costly part of drug development, and failure of drugs at this stage represents a major loss of investment for pharmaceutical companies. As indicated earlier, the computational assessment of binding kinetics will also help increase the understanding of biochemical networks in organisms, which in turn may open new avenues for the prevention and treatment of disease, as well as other applications, such as improving the efficiency of production of valuable chemicals by microorganisms (industrial biotechnology).

In the longer term, once the developed methods are established in the drug-development pipeline, the general public (as well as pharmaceutical industry) will benefit as drugs will be developed more quickly (and thus more cost-effectively), are likely to be more effective at lower dosage, and should have less side-effects. Other long-term effects that will benefit society could be the increased efficiency of biotechnological production of valuable chemicals, with the potential to make this production significantly more sustainable (cleaner, more energy efficient and with efficient use of sustainable starting materials).

Publications

10 25 50

publication icon
Comitani F (2018) Exploring Cryptic Pockets Formation in Targets of Pharmaceutical Interest with SWISH. in Journal of chemical theory and computation

publication icon
Haldar S (2018) A Multiscale Simulation Approach to Modeling Drug-Protein Binding Kinetics. in Journal of chemical theory and computation

publication icon
Hovan L (2019) Defining an Optimal Metric for the Path Collective Variables. in Journal of chemical theory and computation

publication icon
Huggins D (2019) Biomolecular simulations: From dynamics and mechanisms to computational assays of biological activity in Wiley Interdisciplinary Reviews: Computational Molecular Science

 
Description We have discovered that in most cases drug-target binding kinetics is associated with a complex free energy profile with many local minima. Thus to describe it accurately it is necessary to go beyond the simple transition state theory and use approaches such as transition interface sampling. What is more, by analyzing the interaction energies of structures in the transition state ensemble by QM/MM, we found that the polarizability is important and does not cancel out with that of the bound or unbound states. This has important consequences on the computational methods that should be used in predicting binding kinetics as the assumptions commonly used in most simulations-based approaches wont be able to provide accurate predictions.
Exploitation Route The computational platform that we are developing (based on our findings) will be made widely available by the end of the project. We believe it will have a very significant impact on rational drug discovery.
Sectors Healthcare,Pharmaceuticals and Medical Biotechnology

 
Description Our computational platform for binding kinetics is being used to help develop more effective drugs in collaboration with Evotec and UCB and Heptares. Binding kinetics plays a fundamental role in the efficacy and selectivity of drugs. Up to now has been very difficult to rational design drug-like compounds with a given binding kinetics. This is due to the fact that the kinetics is determined by the energy and nature of the transition state that, by its own nature, is very short lived and impossible to capture by high-resolution experiments. Thus accurate computer simulations can be very important in understanding what molecular features affect the binding kinetics. Our platform combines state-of-the-art enhanced-sampling atomistic simulations with ab-initio simulations to provide the most accurate description of the transition state ensemble.
First Year Of Impact 2016
Sector Healthcare,Pharmaceuticals and Medical Biotechnology
Impact Types Economic

 
Description AstraZeneca CASE studentship
Amount £29,500 (GBP)
Organisation AstraZeneca 
Sector Private
Country United Kingdom
Start 03/2017 
End 02/2021
 
Description Industrial PhD studentship
Amount £30,000 (GBP)
Organisation Heptares Therapeutics Ltd 
Sector Private
Country United Kingdom
Start 03/2017 
End 02/2020
 
Title SWISH a new Hamiltonian Replica Exchange-based computational algorithm 
Description We developed a novel and effective computational approach to predict cryptic binding sites on targets of pharmaceutical interest. 
Type Of Material Improvements to research infrastructure 
Year Produced 2016 
Provided To Others? Yes  
Impact The method has been described in an high-impact publication (JACS) and in a number of high-profile blogs in drug discovery. The PI has been invited by Pfizer and other pharmaceutical companies to give talks about the method. 
 
Title SWISH a new Hamiltonian Replica Exchange-based computational algorithm 
Description We developed a novel and effective computational approach to predict cryptic binding sites on targets of pharmaceutical interest. 
Type Of Material Computer model/algorithm 
Year Produced 2016 
Provided To Others? Yes  
Impact The method has been described in an high-impact publication (JACS) and in a number of high-profile blogs in drug discovery. The PI has been invited by Pfizer and other pharmaceutical companies to give talks about the method. 
 
Description Collaboration with Bristol university on predicting drug-target binding kinetics 
Organisation University of Bristol
Department School of Chemistry
Country United Kingdom 
Sector Academic/University 
PI Contribution We contributed our enhanced sampling simulation algorithms including TS-PPTIS. Our approach will be combined with Prof. Mulholland's QM/MM algorithms to accurately predict binding kinetics.
Collaborator Contribution Prof. Mulholland's contributed his QMMM algorithms as well as Waterswap to the combined computational platform.
Impact A combined computational platform to predict binding kinetics and model the transition state ensemble.
Start Year 2015
 
Description Collaboration with SANOFI and Evotec on the allosteric regulation of receptor Tyrosine kinases. 
Organisation Evotec
Country Germany 
Sector Private 
PI Contribution In this successful partnership, we helped designing a new class of allosteric anticancer drugs by using our novel computational algorithms to sample rare events and predict binding kinetics.
Collaborator Contribution The partners contributed to the success of the research by performing large scale ligand screening ad providing high resolution crystallographic structures and biophysical data on the ligand-target complex.
Impact A novel allosteric inhibitor is in re-clinical development for cancer.
Start Year 2015
 
Description Collaboration with SANOFI and Evotec on the allosteric regulation of receptor Tyrosine kinases. 
Organisation Sanofi
Department Aventis
Country France 
Sector Private 
PI Contribution In this successful partnership, we helped designing a new class of allosteric anticancer drugs by using our novel computational algorithms to sample rare events and predict binding kinetics.
Collaborator Contribution The partners contributed to the success of the research by performing large scale ligand screening ad providing high resolution crystallographic structures and biophysical data on the ligand-target complex.
Impact A novel allosteric inhibitor is in re-clinical development for cancer.
Start Year 2015
 
Description Developing new computational approaches to inhibit "undruggable" targets. 
Organisation UCB Pharma
Country United Kingdom 
Sector Private 
PI Contribution We shared with UCB early versions of a computational platform we are developing to target allosteric sites on otherwise "undruggable" targets, i.e. pharmaceutical targets that are difficult to target with traditional drug design approaches based on substrate competitive ligands.
Collaborator Contribution UCB contributed to the project with high quality structural and biological data and is hiring a dedicated PDRA to work on the collaboration for 3 years.
Impact The new tools have been used to design novel drugs.
Start Year 2016
 
Description Industrial collaboration with EVOTEC 
Organisation Evotec
Country Germany 
Sector Private 
PI Contribution We helped EVOTEC to rationalize the binding mode of a novel allosteric modulator of FGFR. By using our novel "SWISH" Hamiltonian Replica exchange algorithm, we predicted a previously unknown binding cavity in the D3 domain of FGFR3c, which was then validated by NMR spectroscopy.
Collaborator Contribution Evotec provided a plethora of unpublished experimental data on the binding mode and on the biological effect of the new tool compound in cells.
Impact The collaboration is multi-disciplinary involving Computational Chemistry, Chemical Biology, Structural Biology, Cellular Biology and Drug Discovery. A new manuscript is in preparation and will soon be submitted to a very prominent and high-impact journal. The PI (FLG) has been invited to a number of high-profile national international (ACS-meeting) conferences to discuss the results.
Start Year 2016
 
Title Plug-in and scripts for enhanced-sampling molecular simulations. 
Description We developed a new interoperable plug-in compatible with PLUMED and many widely-used MD codes (such as GROMACS) to run our TS-PPTIS approach for binding kinetics. The tool can be used to predict ligand and folding binding kinetics. 
Type Of Technology Software 
Year Produced 2016 
Open Source License? Yes  
Impact The tool is able to accurately predict the binding kinetics of drugs to their biological targets, paving the avenue to the rational design of new molecules with fine-tuned biomedical effects.