May 11, 2020 - by Simone Ulmer
Mr. Dal Peraro, your area of expertise is computational biology and the modeling of biological systems. You are interested in multiscale modeling of large molecular systems, especially membrane proteins systems. You have now successfully submitted a project to investigate the envelope protein E of the corona virus SARS-CoV-2 using the “Piz Daint” supercomputer.
This is correct, yes. The primary expertise I am using in our proposal for researching COVID-19 is based on our longstanding interest in membrane proteins. Since my PhD and postdoctoral training, I have been working on membrane proteins. When I established my lab at EPFL, I continued working on membrane proteins. Now I have several lines of research on membrane-protein interactions, which we tackle using mainly computational chemistry for molecular simulation and modelling, integrated with experimental data. One key aspect is to study their evolution in a realistic environment, or as realistically as we can reproduce it with a computer. To do this, we must run molecular simulations on supercomputers. This is the background of our research and the foundation of the proposal.
Why the E protein?
Compared with what other people in our community are studying, protein E is not in the spotlight. Most groups are studying the spike protein or the nucleocapsid protein; this is a natural choice, as they are the main targets to develop vaccines or drugs, for instance. There are a lot of people already working on these targets, and we did not want to duplicate their efforts. We wanted instead to best use our specific expertise on membrane proteins. We want to look at the E protein, a pentameric channel that allows the transfer of ions into membrane compartments. These functional properties can be relevant for blocking the occurrence or the progress of COVID-19. This has a potentially therapeutic outcome, although, at the moment, the biology of the virus cycle is not yet well understood, nor is the role of protein E.
Specifically, what are you studying?
When we look at the genome of this virus, some proteins are heavily post-translationally modified, like the spike protein as well as the membrane proteins. Protein M and E, for instance, are palmitoylated. This means that some of their cysteine residues are acylated to provide an anchor point to the membrane. This property is something that we are already investigating for other proteins in collaboration with experimentalists at EPFL, trying to understand the role of this palmitoylation and lipidation in general for protein membrane association.
What makes this protein E so important, and what do you want to find out in your simulations?
Membrane protein E of the coronavirus has three sites of palmitoylation, and one of the goals of the research is to study their role for stability and assembly of the channel formed by protein E. We know that the two membrane proteins of SARS-CoV-2, M and E, are actually very important for assembly in the virus cycle — for the transport and reassembly of the virus in the host cells. However, they are not yet well structurally characterized. The molecular biology of the virus, and protein E in particular, is still poorly understood, and work is in progress to advance in any front. For other viral proteins, like the spike protein, scientists have recently published structures that people were able to use to perform molecular modeling and simulations. For protein E, we currently have some characterization based on nuclear magnetic resonance spectroscopy, but it is, however, obtained in a non-physiological environment. One of the first goals of our proposal is therefore to refine the current structural information that we have for the protein E.
How will you go about it?
As I mentioned earlier, protein E is expected to form a pentameric channel. Using molecular modeling and simulation, we plan to probe and see if this conformation is stable in the membrane environment; in the simulations, we model it to be as close as possible to the composition of the membrane compartments where protein E is trafficking, the endoplasmic reticulum and the Golgi apparatus. Protein E is known to conduct cations, but its channel lumen is highly hydrophobic. Molecular simulations will thus be helpful to better understand its functional mechanism.
For the study, you wrote in your proposal, you would like to draw on your experience in researching the so-called influenza A M2 channel, as you see functional similarities between the two viruses.
Yes, that is another motivation for picking the E protein to study. In influenza A virus, the M2 protein forms a channel that is one of the main targets for antiviral drug development. In SARS-CoV2, the E protein, being also able to form ion channels, could have similar functions. Inhibiting it could provide an additional therapeutic strategy, as we know that viruses not expressing E are much less virulent.
The main thing for you is to find out the real molecular structure of this E protein envelope?
Yes, this is one of the goals. Our simulations will test and refine the structural information available so far and in the future, as more structures might appear soon. We have to characterize states of this channel that can be targeted by small molecules, which can be later developed into effective antivirals. Of course, during the six-month period of this project, we hope to provide some basic additional information about this protein that is mostly unknown at the moment. In the long term, my hope is that our contribution can foster new investigations and collaborations around this protein. One of the things I am planning is to get in contact with experimentalists and other computational scientists that are working on this in order to establish exchanges and collaborations.
Intensive research is ongoing on SARS-CoV-2. Has anything changed in the initial situation since you wrote your research proposal in March? Have certain questions perhaps already been clarified?
Not that I know of about E protein. But I am aware that other labs are trying to do similar work, and we are trying to coordinate in order to not replicate the efforts and advance faster and more efficiently.
When do you expect your first results that could influence the further course of SARS-CoV-2 research?
We started with running some tests on our clusters at EPFL. Now the systems we want to study are ready, and my students have started to launch the simulations on “Piz Daint”. Most of the development will depend on the first results of our simulations that will allow us to better understand the structure of the E protein in a realistic membrane environment. Within in a few weeks, we may already have information for these specific aspects. We will continue the work based on these first results. Regarding the impact on fighting COVID-19, our research remains at the fundamental level and now focuses on the molecular mechanism of one specific viral protein; thus, our contribution will hopefully foster more research efforts, but it can have an impact on curing the disease only in the long run.
Is there a lot of exchange with the different research groups?
My idea for this project is to be completely open, to share the information and not compete with anybody, and try to collaborate with other people. Even within my group, we are trying to do this together – as a PI, I normally assign a project to one person, but for this project, many people are contributing. We will likely collaborate with other groups from the experimental and computational side as well. This is an unusual type of approach, motivated by the fact that in these days, there is a necessity for us to make results available very quickly, even before publication, to most people in the scientific community. My feeling is that everything is much more open. This also seems to be the case for other research groups working on COVID-19 programs — now there is more collegiality. People are sharing and are more open to exchange of information so that we can proceed faster than usual. I think this is a great opportunity for the scientific community to explore and get used to different ways to work together.
Apart from the fact that theoretical calculations sent from home to the supercomputer "Piz Daint" are currently easier to perform than laboratory experiments, what is the advantage of simulations over laboratory experiments in this case?
Simulations can give you information that experiments sometimes cannot provide. Of course, there needs to be an interplay between experimental and computational work. The type of simulations we do need to start from experimental data, but they can then provide new, richer information that is not visible in the experiment. From experiments, you usually gain a static snapshot of these proteins in one particular state, which might be, in our case, not the most physiologically relevant one. With the simulations, we are able to expand these static snapshots and add dynamic information. Our protein during simulation can be explored in other states based on temperature effects and different environmental conditions, like a specific membrane compartment, for instance. In this case, we take the structure of E protein stabilized in micelles and explore its state in a phospholipid bilayer that is modelled in a way to mimic the endoplasmic reticulum or Golgi apparatus. In the same spirit, palmitoylation is something that is not captured structurally in experiments, at least the ones we have available, so this can instead be modelled in the simulation. We can probe to find the effect of these post-translational modifications for the stability and conduction of the channel. This is something that is extremely difficult to do experimentally, but computationally, we have theoretical models and techniques that are very robust and reliable to study protein function.
And beyond that?
What the simulations provide on top is a realistic molecular movie of these proteins. You can observe all the atoms of the system. This gives you a very detailed understanding of the molecular mechanism that proteins use to interact with each other or with the membrane. You can really probe molecular mechanisms atom by atom. This is highly informative, especially when you want to develop interacting entities like a drug. Even for vaccine design, people are using this kind of approach, i.e. structure-based, to engineer proteins that mimic the epitope of some of these viral proteins. Now, we have the next six months to develop some basic understanding of this protein, which is much less characterized than others in the SARS-CoV-2 virus; based on our findings, we can think about the next steps.
We are conducting this interview through Zoom, because, like many others, we are both currently forced to work from our home offices due to the very virus that you are studying. Can you carry out your research in your home office in the same way as in your office at EPF Lausanne?
There are some things you can easily do at home, for instance writing papers and sending emails, but what is missing is the personal contact with my students. We meet on Zoom so we can have interactions, but it is not actually the same. When I am at EPFL, I stop by any time at the lab and offices to discuss matters with my people, but now I lose a bit that personal relationship. The computational work is not heavily affected, as the lab is still running and most of the work can be done remotely, but some people doing experiments are now completely on hold, because everything is closed at EPFL. Everything is a bit slower, because at home we also have kids to keep up with and other distractions, and everything is not as efficient as in the lab. However, we are getting used to handling this new and unexpected situation.
(Image above: Shutterstock)
About Matteo Dal Peraro:
Matteo Dal Peraro graduated in physics at the University of Padua in 2000. He obtained his Ph.D. in biophysics at the International School for Advanced Studies (SISSA, Trieste) in 2004 and received postdoctoral training at the University of Pennsylvania. Since 2007, he has been at EPFL School of Life Sciences, where he is Associate Professor and director of the doctoral program in Computational and Quantitative Biology. His research at the Laboratory for Biomolecular Modeling (LBM), within the Interfaculty Institute of Bioengineering (IBI), focuses on the multiscale modeling of large molecular systems.