PI3K-Like Protein Kinases

Mathematical modelling can be used in cancer therapy

Dr Arne Lindqvist - Karolinska Institutet

I’m interested in the cell cycle and stress to the cell cycle. So in particular this goes into DNA damage signalling and how it affects the cell cycle and how it affects decisions that the cell is going to make, whether it’s going to stop and resume growth or if it’s going to die or if it’s going to just sit there forever. How are the interactions between cell cycle and DNA damage signalling bringing about these kinds of decisions?

How do you look at these interactions?

We look at single cells most of the time. We have developed assays where we can quantitatively , assess the signal transduction status of certain players in both the cell cycle and the DNA damage response. Then we quantify over time the changes in, for example, phosphorylation status of certain proteins. Then in the second step we use mathematical modelling and we use the modelling to make predictions on novel mechanisms and pathways that should be there or should not be there. We also use it to explain fairly complex problems. When you have a lot of pathways interacting then mathematical modelling can help to really make sense of what are the consequences of these problems.

Could you give an example of a pathway that you have looked at using this method?

We looked at lot at what is starting cell division and how is the decision taken for the cell to. The human cell sits for 23 hours and then, boof, it goes in and divides suddenly whereas this is a gradual activation, it’s a gradual process of activation coming up that leads to sudden total activity that is very dramatic for the cell. For this, I’ve been modelling it, it’s essential to understand how can you transform a gradual increase in protein levels to a complete switch in activity that’s going to bring about.

What implications would this have for cancer where a cell has lost control?

These kinds of models can help us to get a very good basal understanding. In the very long term it’s my hope that it could be used in personalised medicine because if we get the model working and we can take patient samples and assess what are the levels of the individual proteins, modifications we can feed this data into the model and make predictions on what would be the best combination of drugs that we could use to treat these very patients with.

Is there anything like this that is already active?

Not in this sense. There are things being done stepwise towards this goal where personalised medicine is coming when you are testing certain things from the patient samples but not in the way of having a full mechanistic model, making predictions on what are the best ways for treatment are going to be, not to the level of where I would like to bring this.