Simulation of Growth Factor Receptor Clustering

Jacki Goldman (1,2), Bill Gullick (2), Dennis Bray (3), Colin Johnson (1)

Computing Laboratory (2) / Biosciences Department (1), University of Kent and Department of Zoology (3), University of Cambridge

Introduction

Growth factors are hormone-like substances which play an important role in the maintenance and development of normal tissues and in human cancer. These bind to receptor tyrosine kinases (RTKs) present on the cell surface. This causes a cascade of protein interactions within the cell which can ultimately result in cell growth, differentiation, migration or death. The erbB family of RTKs can be important in the genesis of several common solid tumour types. Because these proteins make ideal targets for anti-cancer drugs, a detailed understanding of the molecular interactions resulting in receptor activation and intracellular signalling is needed.

These systems are very complex. The system that we have been studying (the epidermal growth factor receptors (EGFR)) has many different growth factors. Each of these growth factor ligands differ in binding affinity, the degree which intracellular signalling is induced and tissue distribution, both during development and in adult tissues.

We have been developing a computer simulation, which uses animation to display the interactions between receptors. Click here to go directly to the simulation.. If you want to know more about the system before looking at the active model it is described in detail below.

(Fig 1) The type 1 family of growth factors and their receptors. The ten ligand genes are shown at the top of the diagram. The twenty eight possible homo and heterodimers formed between the four receptors and their splice variants are shown at the bottom. CLICK ON THE PICTURE FOR A LARGER VERSION.

Details of the process

Binding of growth factor ligands to receptors induces dimerisation and tyrosine phosphorylation of the receptors themselves, thereby initiating the signalling cascade. There are two possible explanations for this. The traditional explanation is stabilisation of dimers by ligand binding, causing each member of the dimer to phosphorylate the other (autophosphorylation), however recent experimental work has suggested an alternative model where substrates of an active dimer are actually other receptor dimers. (Sherrill, J (1997) Self-phosphorylation of EGF receptor is an intermolecular reaction. Biochemistry 26 1443-1451).

(Fig 2) Model of cell showing the sequence of events following ligand binding to growth factor receptors. We are currently exploring this process by GFP tagging of members of the Type 1 growth factor receptors. cDNAs have been microinjected into cells and their distribution in the presence and absence of ligand determined by low light digital microscopy.

Experimental work.

Figure 3. Clustering in Growth Factor Receptors. The epidermal growth factor receptor has been tagged with fluorescent protein. The figure shows microscope images from two stages in the process.

• A. Prior to ligand addition. The surface of the cell looks uniform, as the receptors are distributed evenly across the cell surface and are too small to show up individually on the picture.
• B. 5 minutes after ligand addition. The receptors have largely clustered together, and these clusters can be clearly seen as small dots on the cell surface.

We have also tagged the p85 subunit of PI3 kinase and filmed this clustering in response to ligand.

Computational Simulation

We have been developing a computer simulation, which uses animation to display the interactions between receptors. The simulation allows the addition of ligand to between 5 and 100% of receptor and follows their subsequent aggregation into dimer and multimers. The process can be paused and the composition of the components shown as a graphical output. The simulation can be resumed and will run to 5000 iterations and then stop.

The simulation is based on a model of interactions between unliganded monomers, liganded monomers and higher aggregates.

Figure 4 illustrates the interactions between different species of molecules in the simulation. Forward arrows show probability of binding. Reverse arrows show the probability of dissociation, the value of which determines the stability of the complex formed.

Figure 4:

Figure 4(a). Two un-liganded monomers (or one un-liganded and one with ligand bound) have a very low probability of binding to each other to form dimers which are very unstable.

Figure 4(b). Two liganded monomers have a very high probability of binding to each other to form dimers which are very stable.

Figure 4(c). Two liganded dimers have a very high probability of binding to each other to form stable tetramers. Also dimers bind to these clusters with high probability to create clusters of four or more subunits. The resulting aggregate is quite stable.

Figure 4(d). Higher order clusters (4+subunits) have a moderate probability of binding to each other to form even larger aggregates. A cluster formed in this way has a very low probability of dissociation. Regardless of the way in which the cluster was formed (i.e. aggregation of two large clusters vs. addition of a dimer to a cluster), dissociation always involves release of a dimer. Thus, the size of clusters tends to increase over time, up to an equilibrium point.

Figure 4(e). Key to above diagrams.

Active model

To run the model click the button.

This will bring up another window, which initially gives the simulation in unliganded state. To simulate the adding of ligand press the "add ligand" button at the bottom of the page, which will bring up another window where you can select how much ligand to add. A good value to see the effects quickly is 80%. At any point in the simulation you can pause the process and look at graphs of the progress so far.

Here is a static picture of the model:

---

Page created by:

Colin G. Johnson.

Computing Laboratory, University of Kent

C.G.Johnson@ukc.ac.uk