PhD student profiles

     
     

Tomasz Kosciolek
Wellcome Trust 4-year Interdisciplinary PhD Programme, beginning in Autumn 2011

Project title
Investigations of structural ensembles and disorder-to-order transitions in intrinsically disordered proteins.

Principal investigator: Prof. David Jones, ISMB, UCL
Co-investigator: Prof. Bonnie Wallace, ISMB, Birkbeck

   

Background
I graduated from the Jagiellonian University in Krakow, Poland with an MSc in Chemistry in 2010. My master’s thesis was undertaken in collaboration with Professor Andrzej Kolinski from the University of Warsaw and concentrated on novel approaches to protein threading. Between 2009 and 2011 I worked as a research assistant in the Department of Medicinal Chemistry, Institute of Pharmacology in Krakow investigating structure-based approaches in GPCR drug design.

In September 2011 I joined the Wellcome Trust PhD programme in Structural, Computational and Chemical Biology. I work in Professor David Jones’s group, where I investigate applications of predicted intra-protein contacts to de novo structure prediction, and on expanding our understanding of the dynamic properties of intrinsically disordered proteins and protein design.

 

Rotations
Rotation 1
Prof. Peter Rich, UCL (chemical biology) Mid-infrared spectroscopy as a diagnostic tool to distinguish healthy and pre-cancerous biopsy samples

Rotation 2
Prof. David T. Jones, UCL (computational biology) Developing new multiple sequence scoring functions for protein structure prediction (the project served as a basis for the paper: Kosciolek T, Jones DT (2014) PLOS One 9 (3), e92197)

Rotation 3
Prof. Bonnie Wallace, Birkbeck (structural biology) Investigating the structure of intrinsically disordered proteins using CD spectroscopy

 

PhD Project

For a long time it was thought that proteins were well-structured molecular machines, which could only undergo subtle structural changes. In the early 21st century it was realized that some proteins (as many as 1/3 of eukaryotes, including humans) were intrinsically disordered. This means that at least one part of such a disordered protein has no fixed structure and we cannot describe it in terms of a single set of Cartesian coordinates. Experimentally studying such proteins is therefore difficult – in X-ray crystallography we cannot observe disordered regions (they “disappear”) and only by using NMR can we observe disordered proteins (or IDPs – Intrinsically Disordered Proteins) in atomistic detail. This group of proteins is responsible for many important functions. Because they are disordered, IDPs can interact with multiple partners and form networks that are crucial in cellular signalling and regulation, or the binding of nucleic acids.

Computational techniques are indispensable for the study of proteins and other biological systems but similarly to experimental techniques, studying IDPs computationally is difficult. It is currently possible to computationally label an amino acid within a protein in order to tell whether it is going to be “ordered”, or “disordered”, but it is challenging to go beyond that. There are some computational techniques that can virtually replicate the effects of an NMR experiment, but we can use them only if we have a starting structure (a reference point) and they require massive computational resources; weeks of non-stop computations on a computer cluster.

In my PhD project I have tried to predict the ensembles of intrinsically disordered proteins from sequence. An ensemble is the set of structures which describes the dynamics of the protein. For an ordered protein a good ensemble consists of a single structure – it has all of the necessary information about the structure of the protein; for IDPs we need more structures to take “snapshots” of the disordered region. Our goal is to require only the protein sequence, because there are approximately 2 orders of magnitude more known protein sequences than there are structures; by limiting ourselves to learning the dynamics of only known structures we’d leave out most currently known proteins.

To achieve this goal, I use FRAGFOLD – a program developed to fold proteins from their sequence using fragments of known structures and physical potentials learned from observations of protein structures and physics. FRAGFOLD was originally used to successfully fold ordered proteins. I built a new framework for IDPs and validated my findings against known experimental (NMR) ensembles. From my research, we learned that FRAGFOLD can be used to study IDPs and that by combining FRAGFOLD predictions with clustering techniques (to generate ensembles of structures) it is possible to capture the disordered behaviour we observe from NMR experiments.

I also want to take the investigations further and show that we can make a disordered region ordered by mutating (exchanging) one or more amino acids in the disordered region. To do this, I will rely on already available experimental data. The ability to design disorder-to-order transitions could help to improve our understanding and facilitate further studies of the phenomenon of protein disorder.

 
 
Figure: An example of a disordered protein (ribosomal protein S6; PDB id: 2KJV). The structures show an NMR ensemble and the colours correspond to FRAGFOLD-IDP predictions (blue - rigid; red - highly disordered).

 

 

*




   
     
Institute of Structural and Molecular Biology, University of London
ismb-admin@ismb.lon.ac.uk
ucl logospacerbirkbeck logo