Guilt by association

We’re starting to discover some of the genetic faults and variations that are associated with dementia, but we don’t know much about what they actually do. Two people who are trying to find out are Raffaele Ferrari, also at the UCL Institute of Neurology, and Claudia Manzoni from the University of Reading. Kat Arney spoke to them to find out more.

Raffaele - The one thing you really want to do is to identify genes as a first thing because there is a direct relation between genes and proteins which is important because we immediately know what the effect could be.

Kat - So this is finding out what genes makes stuff that’s in the nerve cells that could be going wrong.

Raffaele - Right. We would be able to name a protein which is affected by the genetic variability. There is a direct relationship between a gene and the protein. Normally, we know about the function of a protein so we may be able at first sight to understand a little bit more about what is molecularly happening within either brain cell.

Kat - So you can say, “Well, this gene looks like it’s quite badly broken. It’s probably going to make a badly broken protein, so it’s probably not going to work.”

Raffaele - That’s a great thing that we can do but this is literally just a tiny piece of information that is not still able to give us a broad view of what's going wrong in the brain cells.

Kat - So the basic problem is that sequencing studies, genome-wide association studies, that family studies have thrown up all these genes or these areas of DNA that you say, “Yeah, they're doing something in these kind of diseases.” But we still don’t know what – so I guess it’s like, if you look at an office, you can say, “Okay, this person works in an office, but we don’t know what their job is and what they're doing.”

So Claudia, how do we try and take these genes, what we know about the genes, and work out what the flip are these things actually doing because it’s alright to have a list I guess of the genes that you find. But it’s nothing if we don’t know what they do.

Claudia - This is exactly the problem that we are facing, that everyone is facing. Genetics from a certain point of view is simpler than functional biology because you identify genes in isolation. But then when you try to understand what that gene does, you need to go into the cell environment and the gene in the cell environment, works in cooperation with other genes and proteins. So you need really to look at something that is more complex.

Functional biology normally takes a lot more time than genetics to then analyse and identify the function associated with a certain gene. Even when we know the function that is associated with a certain gene, maybe we don’t know the function in disease because the function of a gene in the normal condition is one thing, but then we need to understand how the function changes when there is a mutation.

Kat - So normally, we do this with all those pesky experiments in the lab with cells or with animal models taking years and years, and years. I know people can spend an entire career just studying how one gene and the product of that gene goes wrong in a certain disease. We haven't really got that kind of time. What are you doing to speed that up?

Claudia - And also the money that you need for doing all those experiments. So, this is why Raf and I started talking a few years ago now about how can we think of doing something different and find a way to not solve the problem, but to find a way to ease the passage of information from genetics to the functional biology.

Kat - So to give you some kind of clues about where to start looking at the function.

Claudia - Yeah, exactly. So, we decided that the way to go was to stop looking at one gene at a time, but try to have a more broad look, a more general overview of what's going on, evaluating all the genes that we know are associated with a certain trait, and look at all the genes together to see whether they will point us in a certain direction. For doing that, we decided to start using databases that are already available, already generated, and are freely available in the public domain.

Kat - To go back to the office analogy, you're trying to use data that's already out there to work out – does this person work in the finance team or the human resources team? – what sort of data are you looking at?

Claudia - Yeah. We look at a protein-protein interaction type of data at the moment. We are planning to move to other data sets but at the moment, what we are really focusing on is protein-protein interactions. The idea is that proteins that work together, they interact. There is a principle which is called Guilt by Association principle.

If we know the function of protein A and we know that protein A interacts with protein B, but we don’t know the function of protein B, well, just by knowing that protein B is able to interact with protein A, we can infer the functional protein B. So we are using this principle to build networks of proteins, knowing that they interact together and to see for known proteins, to see in which pathways, in which functions they are associated based on the network of other proteins that they interact with.

Kat - To go back to our office analogy, if you know that two people are always going to meetings together, they're probably working together.

Claudia - Yeah, exactly. That is the perfect analogy, yeah, definitely.

Kat - So through this work – through understanding the genes, getting really good clues about what they do to direct the lab research at the right way, to find targets for drugs, what do you want to see as the key outputs for this? What do you see as coming out to benefit patients and how long, how far away is that going to be?

Raffaele - Ideally, I think we’re talking about two outcomes. One is that of understanding the molecular mechanism – what is impacted in the brain cells that leads to their death? This is critical because if we don’t know that, we don’t know how to handle it and how to fix it in a way, if it works out, it’s rewarding, just the beauty of understanding something. On the other hand, the understanding can lead to identifying either biomarkers in both preventive medicine or as a monitor of the disease progression, or as well as for developing therapeutic measures. We might be able to identify elements that fit all of these needs – so preventive measures, monitoring measures, or therapeutic measures.

Claudia - And the other point is that if we know the mechanism, we can actually think of a drug that is able to impact the mechanism because at the moment, we don’t have drugs. We don’t have therapies for all these disorders unfortunately. But we have some drugs, that we can prescribe to patients but these are symptomatic drugs. So they don’t cure the neurodegeneration. They just work on the symptoms.

For example in Parkinson’s, we can reduce the tremors, but we don’t stop the progression of the disorder. But if we really know the molecular reason why the cells in the brain are dying, we can think of an intervention that is actually directed to the problem and then we can prevent or stop the neurodegeneration. So, this is why it’s very, very important to understand the molecular mechanism at the base of the degeneration.

Kat - Raffaele Ferrari from UCL and Claudia Manzoni from the University of Reading.