Phase Space Invaders (ψ)

Episode 9 - Michele Vendruscolo: Preventing protein misfolding, fostering public engagement of scientists, and AI tools in drug design

• Miłosz Wieczór • Season 2 • Episode 9

Share episode

Send us a text

In the ninth episode, Michele Vendruscolo and I discuss the current state of research on misfolding diseases, typically associated with excessive protein aggregation and formation of insoluble amyloids. Michele presents an optimistic perspective in which the convergence of recent clinical and software developments opens up new avenues for efficient treatment of such debilitating conditions as Alzheimer's and Parkinson's disease. This interconnection between the everyday work of scientists and the experience and outlook of affected individuals inspires us to reflect on the societal responsibilities of a scientist, from disseminating our results to inspiring newcomers to engage with urgent global crises to mentoring the next generation of problem-solvers and revolutionary thinkers.

milosz_1_05-02-2024_171619:

Welcome to the Phase Space Invaders podcast, where we explore the future of computational biology and biophysics by interviewing researchers working on exciting transformative ideas. We're kicking off season two of the podcast with a conversation with Michele Vendruscolo, a professor of biophysics at the University of Cambridge, known for his outstanding contributions to the computational study of protein aggregation and amyloid formation. These processes are a leading cause of misfolding diseases. such as Alzheimer's and Parkinson's. And recent years saw a huge effort from the simulation community to understand and address the molecular origin of these debilitating conditions that wreak havoc on so many families worldwide. So I asked Michele to tell us where we stand in our fight against misfolding diseases and how his group is making use of the new AI based tools. to find potential new treatments. We discuss how much of a breakthrough these AI tools have been and what kind of progress we can expect within the next decade. Then we move on to the broader question of the role of a scientist or public intellectual in society. How the discoveries we communicate find their way into people's lives or don't and whether we're doing a good enough job at making young people excited about the often slow, solid, and humble science in the era of instant news and bold influencers. I hope we inspire you to think about it more. So let's go. So, Michele Vendruscolo, welcome to the podcast.

squadcaster-05aj_1_05-02-2024_161619:

Well, thank you. My pleasure being here.

milosz_1_05-02-2024_171619:

So I remember a moment when, misfolding diseases and amyloid formation was the kind of go to topic for quite a number of groups in computational biophysics to the point where, you know, you could go to any major conference and every other poster would be on either beta amyloid or alpha synuclein, but you were there way before it was cool, right? So I wonder if you can tell us more about what your original motivation was what have we learned as a community through simulations? And, um, you know, what is the outlook? Are we on the cusp of finding solutions, treatments, or rather we're still uncovering the next layers of complexity there?

squadcaster-05aj_1_05-02-2024_161619:

Okay. Well, perhaps a bit of personal history because, uh, you know, I, I started as a PhD student in condensed matter physics, but, I, I felt that, uh, you know, my interests were going towards molecular biology and biophysics. I looking for a problem in that area and this was in the mid 1990s. So, you know, protein folding at the time, uh, was, uh, you know a big deal. And uh, so I started working in that area. Uh, that was the final year of my PhD. And then, you know, at a very, uh, supervisor, Amos Maritan in Italy. And, you know, we started working on, problem. folding protein design as well. So the problem of finding a sequence that fold in a given, uh, confirmation in a given native state. And then, uh, you know, I was looking for, a mentor for a postdoctoral, uh, training and, uh, you know, I met Eitan Domani at the Weizmann Institute in Israel. And, uh, you know, he was working on the department of complex systems. So he was doing of different problems in, biophysics and, but he was very keen and in, in, in protein folding. And so I went there for a couple of years and then I really enjoyed the, the, field. And so, but I thought, uh, you know, but still doing it from a physics department that was not completely fulfilling because I was missing the, all the biology part. And so decided to fill that gap as I moved to, to the uk, where, uh, I was, fortunate enough to end up in the lab of Chris Dobson, who was one the pioneers in, in the protein folding field. And the, but the, the lab was, uh, fully experimental. So I was, uh, effectively the first, uh, computational person. But I was lucky at the same time. Well, first of all, because it was a really great lab I worked to be, but also because just happened that Martin Karpus at the time was in sabbatical there in, in with Chris Thompson. And so I started working with Martin and, uh, of course he was, you know, truly a computational person. Uh, so we got along very well. And, uh, we, we really made it, what I think was, you know, very happy period in my career scientifically. and that was about the time in which people started looking at protein misfolding and aggregation. So I had this sort of, uh, particular through the early work of Chris Dobson. So I, I was in the right place at the right time. start early. of course, initially it was extremely difficult to do anything by computational methods because the timescale for the misfolding aggregation processes and the, spatial and and timescales are incredibly long and large So, you know, anything through computation looked daunting, but as history has shown, then people may start making progress. And especially, now, more recently with artificial intelligence method, really the, the field is moving incredibly fast. I guess you're. Partially right when you say there was a lot of interest a few years ago, there will be a lot of interest the near future through the advent of machine learning, in particular, in terms of drug discovery. So, in other words, ways to intervene. in the process for therapeutic purposes. So we are at the sort of time in history in which we can design drugs in the computer that actually modify the and aggression process in ways that are useful in drug discovery, so they can turn these compounds designed computationally can be taken forward into, drug discovery pipelines. So, you know, I started perhaps, uh, earlier than others for, you know, Lucky circumstances, but I'm very much looking forward to the next few years in which, uh, all these efforts, uh, cumulatively will, uh, hopefully make an impact, to the life of people, especially, know, patients and their families.

milosz_1_05-02-2024_171619:

Right. And what is the landscape let's say, the molecular landscape of those misfolding diseases, right? Because with different therapeutic modalities, people try to impact different aspects of, of the disease from the production of the, protein part that forms the amyloid to the actual disaggregation to maybe Addressing downstream issues that are caused by aggregation. how do you see? There was even this question of whether aggregation is the actual cause of the disease or is it just a symptom, right? So, uh,

squadcaster-05aj_1_05-02-2024_161619:

uh, that is

milosz_1_05-02-2024_171619:

what are the lessons learned?

squadcaster-05aj_1_05-02-2024_161619:

very broad debate that is going on since, uh, you know, probably 30 years now since the amyloid was first put forward in the early 1990s. think now, especially with the approval of the first, drugs for Alzheimer. Which target the, the amyloids one can be fairly confident that the amyloids are causative agent in some form. You know, they're

milosz_1_05-02-2024_171619:

Mm hmm.

squadcaster-05aj_1_05-02-2024_161619:

of, uh, of not just Alzheimer, but all, uh, neuro degenerative diseases in some form or in another. And so, but having said that, the protein aggregation, aggregation processes. are extremely complex. So I don't want even to talking about other therapeutic approaches, you know, targeting, for example, inflammation or targeting mitochondrial function, you know this is a, be too broad as a discussion. But even if one restricts the attention To the amyloid state as a target one realizes that, you know, there is a variety of targets within this broad area. So, one can target the fibers, for example, antibodies can target. The, the amyloid plaques to the effector function. So effectively they can target the amyloids towards degradation they can do some things which are perhaps a bit more sophisticated, because Amy Amyloids. Catalyze the formation of other amyloids. So, you know, one can

milosz_1_05-02-2024_171619:

Mm hmm.

squadcaster-05aj_1_05-02-2024_161619:

or other compounds that the catalytic sites on the fibers and therefore block the auto catalytic, growth of the aggregate. Or one can target intermediate species, you know, because of the most toxic, uh. aggregates are not the large fibers in most cases. In some cases they are, but in, in most other cases, they aren't, in particularly Alzheimer and Parkinson most likely the most toxic are small aggregates called oligomers. 10 to 20 nanometers, maybe 10

milosz_1_05-02-2024_171619:

Mm hmm.

squadcaster-05aj_1_05-02-2024_161619:

molecules.

milosz_1_05-02-2024_171619:

Do we know what they do exactly? Like, what are the reasons for their toxicity?

squadcaster-05aj_1_05-02-2024_161619:

is that they do lots of things and, uh, because they should not be there. So they are orthogonal to biology, so to speak, Not being designed out by biology, they tend to interact in an abnormal manner with lots of cellular components. For example, they can interact with the cell receptors and trigger downstream pathways or can even interact directly with the, with the cell membranes and perturb their integrity. So lots of things that are particularly unhealthy for, for a cell. So one can try and remove those oligomers or, uh, one can try also more radical things like reducing the overall expression levels. of the aggregating proteins. After all, proteins aggregate when they are at high concentration, so where the concentration exceeds the super saturation threshold, right So the idea is that if you block the production, the synthesis the cell, or even you can block already at the mRNA level. Um, you reduce the overall amount

milosz_1_05-02-2024_171619:

Yeah. So we have many points of possible intervention.

squadcaster-05aj_1_05-02-2024_161619:

yes, there are pros and cons in all this, mechanism of actions. For example, if you reduce the, the level of expression of a given protein, you prevent aggregation, but you also prevent the function. So you have a loss of functional. Activity, which,

milosz_1_05-02-2024_171619:

Yes.

squadcaster-05aj_1_05-02-2024_161619:

the long term, I mean, these are slowly progressive long term diseases. So you, you cannot really the functionality over decades lots, uh, lots of different, uh, sub targets when, when you want to, go against, protein aggregation. Of course, we, we have explored a few of them and more traditionally we've been trying to develop compounds that inhibit the aggregation process. Say for example, the block, this, uh, autocatalytic activity that I mentioned a minute ago. More recently, I quite, excited by methods in which, one uses more molecules to stabilize the native states, because in you know, to me that the best of, uh, both worlds. So you prevent the toxic gain of function to aggregation, and you also avoid the loss of functional activity of the native state. So you. you. know, stabilize the native form of the protein. So prevent aggregation and promote functionality.

milosz_1_05-02-2024_171619:

even before the nucleation starts, yes.

squadcaster-05aj_1_05-02-2024_161619:

Exactly. So these compounds are called

milosz_1_05-02-2024_171619:

Okay.

squadcaster-05aj_1_05-02-2024_161619:

pharmacological chaperones. And there are a few examples already approved. And these examples are for proteins that are folded so what we're trying to do is to develop, uh, pharmacological chaperones for disordered proteins. So this is a challenge because, you know, disordered proteins don't have binding pockets. so, you know, we have to, um, understand, uh, how we can design compounds that have a high affinity for the disordered proteins and also high specificity. So we are working in the, in this area mostly now. And again, here is where artificial intelligence is helping tremendously because we've been, uh,

milosz_1_05-02-2024_171619:

Right. I was about to ask,

squadcaster-05aj_1_05-02-2024_161619:

Yes.

milosz_1_05-02-2024_171619:

was about to ask how did this whole workflow change now in the last, let's say two, three years that we have all those tools for predicting changes due to mutations for predicting binding sites in order of magnitude more accurate way than before.

squadcaster-05aj_1_05-02-2024_161619:

Yeah, I wouldn't want to the case, but for us here is being a game changer, right? Because this idea of pharmacological chaperone, we had already probably 10 years ago, but we couldn't really find any way. of systematically identifying candidate small molecule that could act as pharmacological chaperone. We had one or two examples that we, you know, really investigated in depth, but the discovery was almost by chance. We were lucky to find, so no systematic way. Until the advent of machine learning methods. Now we, we have really Uh, systematic ways of screening chemical libraries to discover compounds that have predicted the binding affinity for disorder of proteins. And these compounds are, uh, know, then we can test them in, in the lab. The, the predictions are really quite accurate. So if we predict, uh, let's say we take a library of, uh, million compounds, we predict the top 20 and at least 10 of these, uh, will actually bind. So there is a very high hit rate. and so there is essentially a much reduced need if at all of, high throughput, experimental screening programs.

milosz_1_05-02-2024_171619:

you would say, this is, you would say specifically due to the inclusion of AI tools, or that was the case even before,

squadcaster-05aj_1_05-02-2024_161619:

No, this,

milosz_1_05-02-2024_171619:

let's docking.

squadcaster-05aj_1_05-02-2024_161619:

you know, specifically due to the introduction of the AI tools, because I'm talking about disorder proteins, because if you want to use docking programs, you know, docking is a very well established field uh, it is quite effective. So you can do chemical library screening with docking methods and find, uh promising. If you target folder process. For disordered proteins, it was, uh,

milosz_1_05-02-2024_171619:

Right.

squadcaster-05aj_1_05-02-2024_161619:

I think it's still not possible. but

milosz_1_05-02-2024_171619:

So are we talking,

squadcaster-05aj_1_05-02-2024_161619:

yes,

milosz_1_05-02-2024_171619:

are we talking about some sort of joint generation of. Protein structure condition on a small molecule binder. Or what is the

squadcaster-05aj_1_05-02-2024_161619:

mean, I

milosz_1_05-02-2024_171619:

the avenue to, to the prediction?

squadcaster-05aj_1_05-02-2024_161619:

many ways, but, uh, for example, you know, uh, you know, the introduction of the transformer architectures with AlphaFold, you know, that can can be, uh, sort of exploited, to predict binding affinities as well. So, you know, it essentially is a way of predicting intermolecular contacts or intramolecular contacts. So it's an alpha fold, but you know, you have alpha fold multimers, for example, you,

milosz_1_05-02-2024_171619:

Right. Yes.

squadcaster-05aj_1_05-02-2024_161619:

know, you can predict inter, molecular interactions. On those grounds, you can predict protein small molecule interaction or protein,, RNA interaction, not yet with the high accuracy, but yeah,

milosz_1_05-02-2024_171619:

There were, yes, definitely efforts in this direction. Yeah. Yeah.

squadcaster-05aj_1_05-02-2024_161619:

towards that, uh, so yes, AI has been absolutely crucial in transforming our ability to predict, intermolecular interactions for disordered proteins and also for ordered ones. But I, I'm mostly focusing on disordered ones because these are the most common ones in neurodegenerative diseases.

milosz_1_05-02-2024_171619:

So let's say five or 10 years down the line. We might see things popping out In clinic applications or, that will be very optimistic.

squadcaster-05aj_1_05-02-2024_161619:

No, no, it wouldn't be at all, in fact, that is, more or less the timescale that I would expect.

milosz_1_05-02-2024_171619:

I see. And what what is the current situation of the approved drugs? What, what do they do? How do they achieve that and how effective are they?

squadcaster-05aj_1_05-02-2024_161619:

well, I

milosz_1_05-02-2024_171619:

Mm-Hmm.

squadcaster-05aj_1_05-02-2024_161619:

as I said, I mean, have been four or 500 failures in clinical trials, in the last 10 years. 30 years for drugs that could change the course of the disease. Of course, had drugs since quite a long time for Alzheimer's, Parkinson's, and other neurodegenerative conditions that could treat the symptoms. until very recently, we didn't have any way of, curing the disease or slowing it down for that matter, until, uh, three years ago because in 2021, uh, the for and Drug Administration, the FDA in the US approved the first, anti-amyloid drug for Alzheimer's Aducanumab, uh, by Biogen. And, uh, this has really been a huge historical moment. because it is, you know, the first time, you know, patient could have some hope in fact also the amyloid hypothesis was validated. So huge boost of confidence, both for the patients and for the people working in drug discovery. And soon after in, you 2023, the second drug with the same mechanism of action targeting amyloid plaques, Zirconumab. by EZA and Biogen as well, was also approved by the FDA. And on top of that, now this year, perhaps a third drug, again, an antibody targeting um, amyloid plaques, uh, Dunanima Eli Lilly. is expected to be approved. So, you know, there is a really a very strong validation of the target and antibodies of increasing efficacy. Now, I wouldn't want to sound too optimistic. These antibodies are not widely used in the clinic yet because they have, they've quite serious side effects. So people, you know, clinicians and patients are worried about prescribing them. They're also very expensive. this is

milosz_1_05-02-2024_171619:

Mm-Hmm.

squadcaster-05aj_1_05-02-2024_161619:

different discussion, but, you know. It is not all positive, but it is really a transition moment in which we went from no idea, you know, shooting in the dark, to having very concrete evidence that there are therapeutic strategies that can work. In fact, one of the things that we and others are now is to, find other compounds in particular, small molecules, but the mechanism of action of these antibodies that they've been approved, but, uh, they have probably less, side effects. so I, I think perhaps I should stress this, again. So we went from. A situation in which, uh, uh,

milosz_1_05-02-2024_171619:

Mm-Hmm.

squadcaster-05aj_1_05-02-2024_161619:

only guesses about what could be a good target for, uh, for, for the disease, to having a validated one. I'm not saying this is the only possible one. I'm, I'm saying that, you know, if you want to do other targets, you're back in the situation that you are uncertain. even if you develop a compound that engages with the target, binds and changes the behavior, you don't know whether it will change the course of the disease. Uh, with amyloids instead, you know, so

milosz_1_05-02-2024_171619:

Right.

squadcaster-05aj_1_05-02-2024_161619:

it's a much simpler problem.

milosz_1_05-02-2024_171619:

Yeah. I think this is very important to state because you know, with the media landscape, you get sometimes such a confusing picture of what is going into the clinic and, what is just a noise around incremental scientific discoveries, right? Because people mentioned this, Oh, there's a new drug. There's a new, this new that, and then in five years, you don't see any, effect. So that's just, to, to switch gears to the other question you wanted to touch upon, which is what is the role of public engagement for a scientist? Or what is the role of public individual, when you're a scientist, right? So,, how do you see this question of, say, communicating with the general public? Especially in topics such as Alzheimer's, which goes on to leave a mark on many people's lives in a very profound way.

squadcaster-05aj_1_05-02-2024_161619:

Yes, I, I think, uh, scientists have almost a duty to communicating their research and and their results, uh, to the general public. know, generally speaking, because are funded by the public, and so they have to, you know, give something back. But specifically in, in fields where there is a real impact, at least in principle, of the research on the life of people, like in drug discovery, there is a particular need of getting the public, not just to know what are the efforts, but also to support these efforts. And so somehow there is a, synergy between, what the society needs, and what the scientists, working in, uh, in drug discovery can give back. So there is a you know, if I understand your point is that, you're suggesting that there isn't enough engagement, many times from the scientists and I, I, would agree with that. I think, uh, that, we should mentor, especially the younger generations, make them aware that this public engagement is an important aspect of a scientific career. Also the problem of taking, routes in scientific research that perhaps, uh are not having too much impact, you know, because tend to think in a detached way. From society.

milosz_1_05-02-2024_171619:

Yes, I see that, you know, one issue we face is that science is almost always mediated by the media layer, right? And, that doesn't always serve best. Um, the interest off attracting people to science. maybe that's why it's hard to tell for me. But it certainly skips many aspects off what science is. What are the uncertainties? It very much tries to get to the point off. Oh, this is now solve the scientists work on this for two decades. And now we know, right? While the reality I'm thinking, what is the role in conveying also, the uncertainties. That was, I think, very much on everyone's minds during the COVID pandemic, right? How do we convey the uncertainty without, making the point that we don't know anything, right? Is there even

squadcaster-05aj_1_05-02-2024_161619:

right. Perhaps there is a, you know, there is a expectation from the public that the scientists could solve problem you throw at them. And this has been true, of course, in some cases in history and COVID is definitely one example. But no, what I meant is that, you know, in our time with. Really facing global challenges we have an impact on our generation, but especially in the future generation. So, you know, climate change, uh, energy materials, uh, intelligence, uh, these are challenges that be solved by a segment of the society. They are really involving everyone. And so I think there should be a cooperation between different components of the society, the public, the governments, the industry, and the scientists. So they should cooperate to identify. Priorities setting up effective, strategies and programs, address these, these challenges,

milosz_1_05-02-2024_171619:

Right,

squadcaster-05aj_1_05-02-2024_161619:

increasingly. urgent. So curious, that several of these challenges are a direct consequence of the success of science, so to speak, right?

milosz_1_05-02-2024_171619:

true, yes.

squadcaster-05aj_1_05-02-2024_161619:

um, are coming from there, right? This is science that created success. And now somehow science should also define other ways of dealing with problems. So, um,

milosz_1_05-02-2024_171619:

It's always going to be iterative, yes.

squadcaster-05aj_1_05-02-2024_161619:

Yes, and, you

milosz_1_05-02-2024_171619:

But do you think we have, we have good platforms for this engagement with the public? I can think of a few science communicators that have YouTube channels or, you know, just reach the general audience with higher efficiency. But I think we, in our community, in the computational community, maybe lack this popular level of public outreach that would really have famous popular communicators that, achieve maybe not commercial success, but deep reach within the general public.

squadcaster-05aj_1_05-02-2024_161619:

yes, and I wish your podcast could, uh, to fill this gap. It is true, I think, historically, computational science has not been, say, on the front pages of newspapers. But, uh, as you said, the, you know, internet and social media give us, it's like a a continuous experiment. So people can try what works and what doesn't work. And there are many, many attempts, some, some of which hopefully will succeed. so, but it's certainly true that, people, you know, should try more things like. You're trying yourself, right? You know, extending the reach of what we're doing. Now, I mean, again, with the of artificial intelligence, and the impact that it is having on the computational community, this task may become easier, because almost anybody now would know what artificial intelligence is, or at least that happening, would be interested in learning more. And the type of problems that computational science can start and solve are becoming real ones. So, in the sense, real impact on the lives of people. So, it's much easier to explain person that there is a new drug for Parkinson's rather than, I don't know, a new free energy calculation method. You see what I

milosz_1_05-02-2024_171619:

Definitely, yes, yes, yes but to that point, if we get to attract people, uh, into science, Especially the computational science, what would be your advice to people who want to start, who are, you know, at the beginning of their scientific journey, and want to do something impactful, want to, you know, change the world, want to contribute towards a scientific question

squadcaster-05aj_1_05-02-2024_161619:

yes, that's a really deep question. I think there is something in the human nature that, biases people, you know, forces people to look for answers. There is a sort of, uh, human desire to understanding how nature works. So it's almost, you know innate. So that's a driving force, I think, for most people. Now, can result in efforts that can be detached from reality can have a real impact on our society. So My inclination would be to advise younger people that, you know, if they feel this urge of pursuing scientific inquiries, they should direct their efforts, uh, towards, fields, uh, where could scientific contributions are really needed. You know, I was mentioning global challenges, so it's not something that we can, postpone. So we have to act now, right, in the coming. Months and years. And so, you know, it's important for younger people to understand that they should direct their, their intellectual curiosity, towards having a real impact on this, on this problem. It is certainly possible

milosz_1_05-02-2024_171619:

right, so, you know, looking at the list of your lab's alumni. It's hard not to notice that you're one of the few people who actually start a sort of scientific family. So that's, if that can be a testament to the fact that you're actually working on some of those really deep and foundational questions

squadcaster-05aj_1_05-02-2024_161619:

well,

milosz_1_05-02-2024_171619:

I think that proves that very well.

squadcaster-05aj_1_05-02-2024_161619:

you know, I'm, I'm flattered that you're saying that. I didn't think think of myself like in this way, but it is certainly true that this is my personal experience. You know, I was fortunate to work with scientific mentors, uh, that really made. You know, a huge impact the way I do science and the way I think about it. And so, again, my advice would be to actively look for mentors that can really represent a good match, right? You know, you can work together with them. You can learn from them and you can raise to their level. You know, the, the mentorship is, uh, is absolutely crucial in a scientific career, probably in all aspects of life, but is extremely hard to learn to do science by yourself., If there is a very experienced person or people that are teaching you, you, you can fly much, uh, much more easily.

milosz_1_05-02-2024_171619:

Now that you say that, given your credentials. uh, we have to trust you, Okay. Uh, thank you Michele. Uh, I think that was a lovely conversation. Thank you for your time and, uh, background story and, uh, all the developments that you told us about. Um, yeah, thank you for contributing to the podcast

squadcaster-05aj_1_05-02-2024_161619:

and, uh, it's been a privilege being here

milosz_1_05-02-2024_171619:

and wishing you a great day ahead.

squadcaster-05aj_1_05-02-2024_161619:

and to you.

Thank you for listening. See you in the next episode of Face Space Invaders.