My talk is going to focus on three aspects: first of all I’m going to show how we use fly genetics plus medicinal chemistry plus computational chemistry to build drugs step by step that hit multiple targets. The complexity that we build into the drug is designed to deal with the complexity of the tumour, that’s the first topic. We’re even using that approach to take currently approved drugs and create what we call super versions of them that correct their deficiencies.
The second thing I’m going to talk about is how we build complex fly models. We can build up to 12 or 15 hit fly models that are designed to capture the genetic complexity of individual cancer patients. What I’m going to show is that as the complexity of the model goes up the number of drugs that will work against it goes down and that you really have to go to cocktails of drugs or polypharmacology, which has really been motivating our drug design.
Then I’m going to finish by talking about our efforts at the Center for Personalised Cancer Therapeutics, which I opened relatively recently. We currently have 25 patients in the centre and we sequence their tumours, we build personalised fly avatars, we do high throughput drug screening using robotics to create recommendations for treatments to the patient and their oncologist.
Is what you do unique?
Yes, what we do is a little unusual in that sense. There has been a debate in the field about what’s called targeted therapy which really has come to mean hitting a single target with the idea that the cleaner you hit it, the less toxic the drug will be. That hasn’t always panned out. Also, issues of resistance, tumour heterogeneity, it’s had some difficulties; it’s had successes but also difficulties. The problem with polypharmacology, I don’t think most people have a conceptual problem with it, the problem with it has always been that if you create what we would call a dirty drug that hits many targets, which we like, chemists don’t really know what to do with it. So if you want to take a drug, for example, from our flies and improve it so it has better druggable properties to a patient and you don’t know all the activities that are important, you don’t know what you can mess with and what you can’t. So that motivated us to create a pipeline where we actually develop the drug stepwise, we know all the targets that matter, that we’ve engineered some of them in, and that we can go to the chemists and say, “Yes, you can get active properties, PKPD, all that stuff that you need to, optimised to humans but don’t touch these activities. And here’s the assays for monitoring them.” So we’re basically giving this type of approach, polypharmacology, a rational way forward.
Also, what we find when we do this screening in a whole animal system, it’s not just that we find drugs that hit many targets, we find… using our fly genetics we can show that many of those targets are actually not required in the tumour. They’re required in the microenvironment, they’re required distantly in the skeletal muscle, in the fly liver and so on, that then feed back to help choke the tumour out. Those sorts of activities, you can’t capture those in what I would call tumour focussed screening and they’re very difficult to keep track of for a chemist. This is why it’s important for us to be able to map all of these things out and explain to the chemists what they can and can’t do to the drug. So, for those reasons, we’ve had to build that structure to allow us to take that approach.
What success have you had so far?
We did help identify vandetanib which has now been approved as standard of care for medullary thyroid carcinoma. We have developed other drugs which we’ve published that have now been licensed by biotech companies with the intent of bringing them into clinical trials. And, as I mentioned, we’ve also used a modified version of this approach to consider treating patients.
Where does your funding come from?
First of all, one of the beauties of working in flies is that they’re very cheap. So I don’t envy my mouse colleagues who have mouse costs which are pretty insane. Actually cost is mainly manpower. We do get funding both from the US government, from Mount Sinai, who has provided funding for our centre, and of course from private foundations. So people are curious about this approach and so it has… and people have been willing to basically give us some room to play in this arena.
How do you get a fly avatar from a person?
First of all I have to say I stole the name avatar from Pier Paolo Pandolfi who has made mouse avatars with a somewhat similar intent. So we sequence the patient’s tumour and then the difficult part comes as the patient will have mutations in many genes. So we’ll sit down with Eric Schadt’s bioinformatics group and go through them gene by gene and figure out which ones are actually altering protein function. That will generally bring the number down to between five and fifteen. If we can’t bring it down further than that then we have access to the fly genetic tools, we’ll actually test the candidates against what we call a base model and functionally figure out which ones matter. The key point is at the end of the day we’re not looking for what’s common between patients – “Oh, I recognise this as a driver of cancer, therefore I will add that to the model,” – we try to take everything that we have evidence is functional and if we’re not sure it goes in.
