top title      

2016 ISMB Symposium Student Blog Competition

We organised a blogging competition for final-year Ph.D. students at this year’s ISMB symposium. Before the symposium Clare Sansom ran a workshop on science communication skills for these students, with a focus on blogging; the students were then invited to write a blog entry based on either one talk or one discipline-based session at the symposium. There were four judges: Clare focused on writing style, and three of the session chairs on accuracy and scientific content.

The workshop was very well received and stimulated some lively discussion, but there were disappointingly few competition entries: only four from the 20-odd students who had been invited to take part. However, these four were of such good quality that we decided, in the words of the Dodo from Alice in Wonderland: “Everybody has won, and all must have prizes”. We therefore publish all four blog posts and award first, second and two runner-up prizes. The first prize, of £100, went to Harshnira Patani from UCL’s Division of Biosciences for an accurate and accessibly written piece on the way that bacterial toxins ‘break and enter’ host cells.

Despite the disappointing turn-out for the competition itself, this was a useful and interesting exercise for which we have Clare to thank. I hope that we will be able to run similar competitions in future years, maybe involving students from earlier years, who will not be so busy writing up their PhD thesis.

Gabriel Waksman, Director of ISMB (based on text provided by Clare Sansom)

1st Prize: Harshnira Patani

Breaking and entering: Tc toxins in action

Pathogenic bacteria have evolved remarkable ways of invading and infecting host (including human) cells. They do this for a variety of reasons, such as to find food, or to destroy immune cells which might otherwise clear the ‘foreign’ bacterial cells from the host.

A common route by which some bacteria target host cells is by punching holes in the cell membrane using toxin proteins which cause the host cell to rupture (e.g. diphtheria and anthrax), enabling the bacteria to then scavenge for nutrients released by the dying host cell, or successfully evade immune clearance.

A class of toxins that use a unique mechanism of host cell invasion are the ABC-type toxins, or Tc toxins. They belong to a class of pore-forming (porin) toxins which form channels in the host cell, allowing bacteria to inject toxic enzymes into the host. Porin toxins are found in a range of lethal bacteria, such as the bioluminescent Photorhabdus luminescens and the plague pathogen Yersinia pestis, which form pores in (and therefore kill) insect and human cells respectively.

Recent studies in the Raunser lab have used a variety of cutting edge structural biology techniques to determine the precise mechanisms by which the Tc toxins work. They used techniques including x-ray crystallography and cryo-electron microscopy which provide information about the overall shape of the toxins, as well as the finer internal details. Using these techniques, they produce detailed high resolution snapshots of the P. luminescens Tc toxin in its pre-pore and pore states, with atomic resolution. The differences in the shapes of the pre-pore and pore atomic models revealed that the Tc toxin penetrates the host cell membrane by a syringe-like mechanism. The internal stalk-like channel component of the Tc toxin (yellow) passes through the surrounding ring structure and breaks and enters into the target cell, creating a pore through which more toxic proteins can pass into the host cell.

The Raunser lab believes that this mechanism of Tc toxin injection could be universal to the whole family of proteins. I think this can have many really important widespread implications. First, the atomic models could be used to help design bio-pesticides using insect porins to target agricultural insect pests. Second, they could be useful in the clinic for designing ‘medical shuttles’ for delivery of drugs by targeted ‘syringe injection’ to specific cells in a range of diseases including cancers. Finally, these experiments shed light on key steps of infection by bacteria that use Tc toxins, increasing our knowledge of these host-pathogen interactions and setting the stage for the design of new therapies against these lethal pathogens.

1) Gatsogiannis C, Lang A, Meusch D, Pfaumann V, Hofnagel O, Benz R et al. A syringe-like injection mechanism in Photorhabdus luminescens toxins. Nature. 2013;495(7442):520-523.

2) Meusch D, Gatsogiannis C, Efremov R, Lang A, Hofnagel O, Vetter I et al. Mechanism of Tc toxin action revealed in molecular detail. Nature. 2014;508(7494):61-65.


Above: The P. luminescens Tc toxin protein in its pre-pore (left) and pore (right) states, showing the internal stalk-like channel component (yellow) being injected into the host cell membrane by a syringe-like mechanism.
Credit: the Raunser lab, Max Planck Institute of Molecular Physiology, Dortmund, Germany.

2nd Prize: Pedro Tizei

Talk by Dario Neri – ETH Zurich

There are an estimated 2.5 million people in the UK currently living with a cancer diagnosis (Macmillan Cancer Support, 2015) and, as many of the readers will know from personal experience or that of somebody close, the treatments that can hold back or even cure cancers frequently have terrible side effects of their own. The main reason for this is the fact that many of the drugs used for chemotherapy are not targeted, i.e. they act throughout the body and will attack healthy cells in addition to the cells that form the tumour.

One strategy to reduce these side-effects is to make drugs which act only on the tumour and do not reach healthy parts of the body. In his talk at the ISMB Symposium, Professor Dario Neri from ETH Zurich in Switzerland, presented the work his group has been doing in this direction. The new drugs are made up of two separate parts: one that targets the tumour while avoiding the rest of the body and another part that kills or at least slows down the growth of the tumour.

The part of the drug that finds the tumour among all the healthy cells in the body is an antibody. These are molecules naturally produced by our bodies' immune system as part of the defence against microbes that can cause diseases. They work by attaching to specific shapes on the surface of the microbe, which are different from anything that is normally present in our bodies. Scientists can produce antibodies in a lab which can detect specific features that tumour cells have but healthy cells do not. Once these tumour-specific antibodies are attached to an active part, they can work as a drug.

The simplest way to make these targeted drugs is to simply attach a molecule already used for chemotherapy to the antibody that brings it to the tumour. However, Prof. Neri's group is engineering another part of the body's immune system to produce these drugs with even fewer side effects. Naturally, the immune system produces molecules called cytokines that tell cells to kill invading microbes that cause disease. When these molecules are attached to the targeting antibodies, they direct the immune system to attack the cancer cells and prevent damage to the healthy cells around them.

This approach has been so successful in laboratory animal studies that a company was founded to continue the development of these novel drugs. Clinical trials are being carried out to ensure that they are safe and effective for human cancer patients. If these results are confirmed in the trials and the drugs are given approval by regulatory agencies, doctors would then have these safer weapons in their battle against cancer.

References: Macmillan Cancer Support Statistics Fact Sheet, accessed on 14 July 2016.

Runner-up prize: Millie Pang

Biological Matrimony
Alfonso Valenica: Co-evolution based methods:new methods and applications

Romeo and Juliet, the ying and yang and tea and biscuits are all akin in their combination of discrete entities which act in concert to serve a common and more enriching purpose. It could also be said that throughout time, the certain features of one - being influenced by the counterpart - have been optimised and somewhat dependent on the other to perform the function in question.The biological equivalent of this is termed co-evolution where the inherent differences between biological entities, such as residue mutations within proteins, are exploited to govern functional complementarity.

One of the leading researchers within his field of bioinformatics, Alfonso Valenica, gave an inspiring talk regarding such phenomena in the protein world, concerning co-evolution based methods for predicting protein structure and identifying residues of functional importance, along with their corresponding interactors (1). 2 techniques were introduced in this talk - DCA and Mirror tree which use co-evolutionary information to predict protein structure/interfaces and interactors respectively.

Structure and conformations - DCA
Correlated mutations and inter-residue contacts impose constraints within the protein structure - and after years of research from different labs he concluded that “the current best method” for predicting protein structure using co-evolutionary information was developed by Sander et al (2). Along with structure, DCA was also used in exploring protein conformation by using the co-evolutionary residue contacts which derived from homologous protein sequences (3). By using 2 separate multiple sequence alignments corresponding to 2 different protein conformations, one can see that co-evolved residue pairs exhibit high correlation to the conformation produced. Specifically, this study showed 80% of the co-evolved residue contacts predicted were correlated to conserved motions within the protein. As well as using DCA in predicting functional conformations, it is also shows success in predicting regions of protein flexibility. Therefore such studies highlights complementarity in using sequence co-evolutionary information with coarse grained protein models of dynamics to guide the elucidation of functional protein conformations from static structures.

Protein interaction interface sites - DCA
An insight into how a protein binds it partners also serves as an imperative step into understanding its function. In addition to protein structures and conformations, DCA has also been used to predict the binding surfaces and inter-domain pairs of residues within structurally conserved interactions. by using co-evolutionary information. This method is called i2h and was developed by the Valenica group in 2002 (4). i2h identifies the most likely sequence regions implicated in the interactions between physical protein complexes by analysing the co-evolution of distant residue pairs within the multiple sequence alignments of the 2 proteins.

Mirror tree: finding protein interactors - The yang from the ying
As well as considering the protein interaction surfaces, co-evolutionary information can also be used to determine the binding partners of a protein query. As opposed to DCA, another method called MirrorTree (5) was developed to find interacting partners at the genome level, based on the assumption that interacting/functionally related proteins tend to show similar MSA - derived phylogenetic trees due to residue co-evolution. One of the unique features of this method is that each tree from 2 proteins is compared to a background of 58 random proteins, 10K times, in order to assess significance in partner identification. This all versus all tree comparison is to figure out which which sets of trees have the ‘highest correlation’ to the comparing tree, either protein 1 or 2. This is basically analogous to Juliet dating and deciphering from a subset of chaps, which one is my Romeo, here based on personality compatibility. MirrorTree was performed in e.coli which showed that in here there is 90% of choosing the right partner, providing evidence that co-evolutionary information is a useful feature for predicting protein partners. Because protein mutations can differ between different species and in diseases such as cancer (6), applications of the Mirrortree method can include characterising these different interaction partners in different evolutionary and clinical contexts - since interactors depend on structurally and sequence conserved interfaces. Last but not least, the methods of his work was inspired by an article in a contrasting field of neurology - highlighting the imperative bridging of approaches within different fields of science, which in turn act together to improve scientific research.


1. de Juan, D., Pazos, F. & Valencia, A. Emerging methods in protein co-evolution. Nat. Rev. Genet. 14, 249–61 (2013).
2. Marks, D. S. et al. Protein 3D structure computed from evolutionary sequence variation. PLoS One 6, (2011).
3. Sutto, L., Marsili, S., Valencia, A. & Gervasio, F. L. From residue coevolution to protein conformational ensembles and functional dynamics. Proc. Natl. Acad. Sci. U. S. A. 112, 13567–72 (2015).
4. 1. Pazos F, V. A. In silico two-hybrid system for the selection of physically interacting protein pairs. Proteins 47, 219–27 (2002).
5. 1. Pazos, F., Juan, D., Izarzugaza, J. M. G., Leon, E. & Valencia, A. Prediction of protein interaction based on similarity of phylogenetic trees. Methods Mol Biol 484, 523–535 (2008).
6. 1. Nishi, H. et al. Cancer missense mutations alter binding properties of proteins and their interaction networks. PLoS One 8, e66273 (2013).

Runner-Up prize: Tanya Prentice

A review of: The power of correlative super-resolution imaging by Dr Aleksandra Radenovic, EPFL, Lausanne, Switzerland

The dynamic processes of life rely heavily on proteins. Cell signalling, the replication of cells, defence against invading pathogens and damage, and subsequent repair of any damage all require protein and protein complexes. To understand how these fundamental processes work, the mechanical mechanisms of the nano-machines involved needs to be elucidated.

Atomic resolution structures of individual components of nano-machinery and increasingly, larger protein complexes can be achieved using well established techniques. These methods are often referred to as the traditional structural techniques and include X-Ray Crystallography and Electron Microscopy. These structures are not obtained under physiological conditions, and such techniques cannot track the dynamic nature of the protein complexes, integral to their function.

Atomic Force Microscopy (AFM) scans the surface of biological specimens to provide information regarding height and friction, with sub-molecular resolution and can be performed under physiological conditions in real time. However, there is no biomolecular specificity, so we do not always know what we are looking at.

Conversely, Single Molecule Localisation Microscopy (SMLM) can localise specific proteins. Two SMLM techniques, direct Stochastic Optical Reconstruction Microscopy (dSTORM) and Photo Activated Light Microscopy (PALM) use fluorescent dyes, probes or genetically engineered proteins to provide multicolour images, which can be compatible with live cell imaging. However, what one observes is the fluorophore and not the biomolecule itself, and therefore SMLM requires correlative imaging. This is often in the form of Transmission or Scanning Electron Microscopy techniques, which are not performed under physiologically relevant conditions.

To overcome the limitations of AFM and SMLM techniques, correlative methods have been developed. These methods augment inherent weaknesses by combining the strengths of each method. The specificity of SMLM imaging in conjunction with the high resolution of AFM allows for a much clearer interpretation of the data. In this manner super resolution imaging of live cells and protein complexes in action can be achieved. The integration of AFM and SMLM into a signal machine does not diminish either technique, and with continued development, will allow for simultaneous acquisitions.

The high spacial resolution of SMLM PALM techniques is however, at the expense of temporal resolution, thus dynamic processes of protein complexes can be difficult to study. Focal adhesions are dynamic fibrous bands of protein that connect cells with the Extra Cellular Matrix. They are large macromolecular assemblies of proteins and allow signalling between cells and the ECM as well as mechanical forces to be exerted. These assemblies are highly dynamic, moving at about 100 nm/min. Current SMLM requires imaging times of tens of minutes, but to get the temporal resolution necessary to follow the dynamic process of focal adhesions, imaging of less than 1 minute would be required.

Super Resolution Optical Fluctuation Imaging (SOFI) allows for better tolerance of overlapping emitter signals, improving the temporal resolution of recorded images. The use of both SMLM methods and SOFI provides enhanced insight into the structure of dynamic protein systems, with the enhanced spacial resolution of SMLM and better temporal resolution of SOFI. This allows for the imaging of moving adhesion complexes in living cells with a spacial resolution of about 100 nm and temporal resolution of 10s.

Continued advances in the integration of correlative methods will greatly enhance the resolution of live imaging techniques, essentially allowing us to watch life’s nanomachines in action.








Institute of Structural and Molecular Biology, University of London
UCL logo