An interview with Patrick Barth

What is your research background?

I was trained as a physicist/chemist in France. I went through a classic University and École Normale Superieure path, going on to a PhD at the Commissariat à l’Energie Atomique in Saclay, France. During that time, I really wanted to understand how energy was captured, used and transduced in living cells – a fundamental process that is still very poorly understood.

We were working on very complex membrane protein systems called photosystems: these membrane proteins capture light and then turn light into electron transfer from one side of the membrane to the other. And then somehow an enzyme has to capture the electron from that membrane protein. We were trying to understand how these enzymes interact with the large photosystems through protein engineering, and we had plenty of mutational data but very little structural information to interpret them.

This led me to refocus my research to developing software that can predict the structure of these proteins, and can rationally re-engineer their function. This took me to UC Berkeley and Stanford, for my postdoc, and then later on to the University of Washington where I developed software to model and predict how proteins recognize each other and how membrane proteins, with their architecture, move when they begin to function. These proteins have huge potential for synthetic biology and cell therapies, so I also developed software to design membrane receptors with novel functions.

So I had done a lot of experimental work during my PhD, while my postdoc focused on computation. I wanted to combine the two, so I went to Baylor College of Medicine to start my career there.

Have you spent much time looking at transmembrane receptors, like G-protein-coupled receptors (GPCRs)?

Yes, one main focus of the lab is on GPCRs – almost all what we do in the wet lab is on them. We typically track and predict how they look like, their architecture, and then based on that, we try to predict how molecules, or lipids, or peptides interact with these receptors, and lead them to signal or shut them off.

There are a lot of molecules in the cell that can activate these receptors. And, somehow, when these small molecules bind the receptors, they trigger some change in 3D conformation – they change their shape or how they move. On the other side of the membrane, in the cell, there are proteins that recognize these changes, and interact with the GPCR and start to activate different signaling pathways.

We are trying to decode how these GPCRs interact with multiple molecules, how a single receptor can respond to a large number of molecules that trigger different signaling pathways. If a given molecule can trigger a different output starting from the same receptor, then the receptor can end up with a different shape, different architecture – all depending on the stimuli. We are trying to understand this molecular code, and it’s actually a fascinating biological computation problem to solve.

A key exciting aspect of our work is that we try also to engineer receptors with novel sensing and signaling properties for manipulating cellular functions. So GPCRs provide us with a great starting versatile template to achieve this goal.

What are some examples of these proteins?

There are many examples because GPCRs perform many different functions, but one example is the chemokine receptors found for example on the cell membranes of T lymphocytes. These GPCRs are involved in immune responses and they are actually hijacked by HIV, so you have HIV proteins that are going to interact with these proteins, and use these proteins to enter the cell.

Another example is dopamine receptors, which activate very important neuronal pathways upon dopamine release. We have also many other GPCRs that respond to neurotransmitters like serotonin, which control mood, and receptors that control lipid uptake, which has to do with obesity and many others that respond to hormones.

When we talk about the biophysics of membrane-bound receptors, what are the big challenges?

One challenge is basically trying to predict, at the atomic level, what are the different states that are put in motion, depending on how they bind and which molecules they bind to. And the question is: if a receptor binds to a given molecule, how long will it stay in the particular architecture that allows for the recruitment of cytoplasmic effector and the activation of a signaling pathway?

The receptor is always going to move between different conformations, each of which can trigger different signals inside the cell. When the ligand binds, the receptor will eventually stabilize in one particular conformation. The question is for how long, and if can we measure that using experimental biophysical techniques, using single-molecule spectroscopy for example, and visualize it using electron microscopy.

On the modeling side, the challenge is to predict how the atoms of the receptor (including the drug) interact with each other, with which strength, and how this modulates the shape and trigger or constrain the movements of the receptor.

How important is the lipid bilayer to the conformations of a membrane-bound receptor?

It is phenomenal – which is why we tend to think about membrane receptors not as a single entity but as a molecular machine that functions properly under a particular environment. So this will have drastic effect on its stability. Then if you actually change the composition of lipid membrane, you can actually have a very important functional and mechanistic effect on the receptor itself.

That's why it's been so difficult to model membrane proteins and study them. Compared to soluble proteins that function most of the time reasonably well in buffer with water molecules and ions, the lipid membrane is actually complex and plays a very important role in stabilizing the conformation of the receptor. Step by step, we include this level of complexity in our computer modeling. On the experimental side, much progress has been made in recent years to control lipid environment in vitro so it is now more straightforward to work with membrane receptors.

How do you see your work affecting pharmacology?

That's a key question. We believe our work can really impact receptor pharmacology. In terms of drug development, there is this big, central question of how you can design a drug that gives us the desired response. And when you have a receptor, which is such a flexible molecule and can actually trigger different responses upon sensing molecules that are not that different from each other – this is a huge challenge.

What is key is to better understand the molecular mechanisms behind signaling pathways, which would allow us to better design drugs. And the fact that we can now more accurately model receptor conformations means we may link them to the signaling path of a ligand.

And then we are also developing new tools to model how peptides and mini proteins bind and regulate the function of membrane receptors. These have been very challenging because they are much larger molecules than drugs, so they can adopt many different shapes themselves. But because they involve many more atoms than small drugs, we think that we can achieve much higher cell activity and binding affinity and selectivity than small drug molecules. So this is going to be also a very important application for pharmacology.

What are your immediate research plans at EPFL?

I aim at two directions. First, to set up our dry lab and continue developing modeling software and tools. This is a long-term project, we want also to distribute software tools that are accessible to a large community of scientists. So we need to continue doing it.

The second direction is to show that we can actually engineer as proof-of-concept. We want to demonstrate that we understand how receptors function, and actually engineer a receptor to perform a particular function. If this engineered receptor works the way we want it to, it will directly demonstrate that we are right; if not, then that would mean that our hypothesis, or starting model, was not correct. And so that's the bridge to the test. We already achieved a few membrane protein design proof-of-concept milestones in the lab but this is just the beginning.

And then there are plenty of exciting applications in synthetic biology, engineered cell-based therapies that would benefit from smart protein machines that we design in our lab.

Overall, at EPFL, I want to continue to develop fundamental tools to predict and design structure and function of receptors, but also to bring all these tools now to more interesting applications that pertain to fundamental biology and even clinical therapy.