Subscribe
Share
Share
Embed
Welcome to UCL Brain Stories, the monthly podcast series from the UCL Neuroscience Domain presented by Caswell Barry (UCL Division of Biosciences), Steve Fleming (UCL Division of Psychology & Language Sciences) and Selina Wray (UCL Queen Square Institute of Neurology). UCL Brain Stories aims to showcase the best of UCL Neuroscience, highlighting the wide range of cutting-edge research going on within the Neuroscience Domain as well as bringing you the people behind the research to share their journey of how they ended up here. Each month we’ll be joined by a leading neuroscientist to offer their perspective on the big questions and challenges in Neuroscience research, to find out what stimulated their fascination with the brain and hear how they ended up becoming part of the UCL Neuroscience community.
For more information and to access the transcript: https://www.ucl.ac.uk/research/domains/neuroscience/brain-stories-podcast
Sonia Gandhi
Wed, Aug 14, 2024 11:51AM • 47:53
SUMMARY KEYWORDS
disease, cell, parkinson, protein, brain, patients, neuron, aggregates, work, form, dish, mechanism, trial, called, protein aggregation, image, model, lab, imaging, generate
SPEAKERS
Sonia Gandhi, Selina Wray, Steve Flemming
Steve Flemming 00:00
Steve, hello and welcome to brain stories. I'm Steve Fleming, and I'm here with my co host, Selena Ray
Selina Wray 00:09
on brain stories. We aim to provide a behind the scenes profile of the latest and greatest work in neuroscience, highlighting the stories and the scientists who are making this field tips.
Steve Flemming 00:22
We don't just ask about the science. We ask how the scientists got to where they are today and where they think their field is going in the future
Selina Wray 00:30
and today. It's a huge pleasure to be joined by my colleague and friend, Professor, Sonia Gandhi. Sonia is a Professor of Neurology at the UCL Institute of neurology and also has a lab at the Crick Institute. Sonia, welcome and thank you for joining us today. Thank you for having me. So maybe we could start just by hearing for a few minutes in your own words what your research is focused on.
Sonia Gandhi 00:58
Okay, so I'm a clinician scientist, so the disease I study and work on and also see and treat patients is Parkinson's disease. Parkinson's disease is the second commonest neurodegenerative disease. It's the fastest growing neurological disease globally at the moment, and it's a disease that affects several parts of the brain. So patients first have a problem with movement, so they'll have difficulty with initiating movement and slowing of movement, and they may also have stiffness and tremor. But in actual fact, Parkinson's affects many parts of the brain as well, so they may also develop some sleep disturbances, problems with cognition and memory loss, as well as autonomic problems and psychiatric problems. So it's a systemic problem and a problem that affects many parts of the brain, and as well as seeing and treating patients with this condition, my lab is really focused on trying to understand the origins of the condition, what triggers Parkinson's in the brain, and how it progresses over time, and trying to understand the kind of molecular and cellular basis of why somebody will develop Parkinson's disease, With the ultimate aim that if we could understand the mechanisms, we could then, hopefully someday, find treatments and targets super and
Selina Wray 02:29
I think you know, one of the reasons we wanted to get you on the podcast, and one of the reasons I think your research is so powerful, is because it really does span the full remit of bench to bedside. So I wonder if we can maybe start by, first of all, talking a little bit about the bench aspect and your mechanistic work that you mentioned, and then, and then we'll move on to the clinical research. So you said that your lab tries to focus on the origins of Parkinson's in the brain. We've had guests on the show talk about Alzheimer's and Huntington's, and in both those diseases, you have these specific protein aggregates. Is this also the same in Parkinson's? And can you tell us a little bit about the pathology of the disease? Yeah,
Sonia Gandhi 03:09
so I mean, a lot of the neurogenerative diseases are very linked together by these common hallmarks what we see under a standard microscope in the brain tissue, and that is the switch of a protein in its normally soluble and often unfolded state into becoming insoluble, clumping together and forming what we call a protein aggregate. And in the case of Parkinson's, but not just Parkinson's, Parkinson's and a set of disorders that include dementia with Lewy bodies and multiple system atrophy. We see a protein called Alpha synuclein, which is highly abundant and and has a range of physiological functions. We see that protein convert into becoming a pathological clumped or aggregated form, and that happens inside the nerve cells and neurons of the brain, what we are very good at is studying the end part of that process. So we're very good at seeing these very large aggregates inside brain cells, and that's using conventional microscopy methods that are able to see that. But in actual fact, the process of protein misfolding starts when a protein starts to come together and self assemble. At that stage, it's probably only a few nanometers in size, and then it will form a series of intermediate protein assemblies that may be 50 nanometers, 100 nanometers, 150 nanometers, and then eventually it will form these very large structures which are at the very end stage. And what we're trying to do is understand the process all at the very earliest stages, which means that we have to see those protein assemblies. And the problem is when obviously seeing is believing, but you can only see. See what what physics allows us to see, and because the diffraction limit of light, the wavelength of light is about 202 50 nanometers, it means it's hard to see things smaller than that using normal microscopy methods. So instead, what we do in our lab is use a series of methods that are inspired by super resolution microscopy that basically allow us to see things that are smaller in the human brain or in cells, so that we can capture those really early events. And the second thing to say about protein aggregation, which I'm sure you've all heard of from other podcasts, has been that when when a protein starts to misfold, it forms lots of different types of of misfolded aggregates, not just one type. And sometimes the one that's causing disease might be rare, so you might have only 1% of the protein actually forming the abnormal disease causing form. So how do we find 1% in a sea of 100% of you know protein so we have to look for rare events as well. So as well as looking for small ones, we look for rare ones. And the way we look at rare events is to use a set of technologies called single molecule technologies, which allows you to look at protein assemblies one by one, not the whole population in one go, but look at them one by one and find those really difficult needle in a haystack style events in the brain.
Steve Flemming 06:26
I mean, it's absolutely fascinating. And this is a completely different area of research to what I'm engaged in, which is more in the looking at cognition and the connections to the brain. And I'm just wondering what an experiment on a kind of day to day level looks like here, and what do you start with? So you mentioned studying both the human brain, but also potentially animal models and where, what are the cells that you're starting with that you're using to then try and look at these very early protein dysfunctions?
Sonia Gandhi 07:04
Yeah, it's a good question. You could start either from the bottom up or top down. And when we first, when I first started in this field, what we would do? Many of these methods have come and been inspired by physical chemistry, and I was partly based at the Department of Chemistry in the University of Cambridge, and there, the laboratories of Chris Dobson and David kanaman have basically worked on in vitro systems, so in buffer systems, where it's one protein, and under simplified background, and what they would do is Shake the protein, make the protein in in the lab, and then shake it inside a test tube, and then take the solution and pass it through single molecule confocal microscopes and label it, and then start to look at all those different species. And what we would do would be to start up in Cambridge, bring those protein aggregates down to London, and then put them on our cell systems and ask, what's it also doing inside a biological system? But we've got a lot better now at using those technologies directly in human brain or human cells. So now what we do is we start from the other end. We might take a really complex system, a human cell, or even a human bio fluid like cerebrospinal fluid, and cerebrospinal fluid, for example, contains 4000 other proteins. But now what we can do is use the super resolution microscopy, or the single molecule microscopy, and look in a complex system directly for those protein assemblies. So now our experiment might start in a cell and search for the proteins and so, yeah, so we do the experiment both ways around,
Selina Wray 08:46
looking in these complex systems. Can you tell us a little bit about the cell models that you use? Because these are kind of next generation cell models, right? They're really at the forefront of what can be done for modeling human disease. Yeah, I think
Sonia Gandhi 09:03
when we think about why we can't find cures and drugs very easily for these diseases, the biggest problem is that we do not have a model of the human brain, obviously, but we also don't really have models of the constituents of the human brain, the human cells, the brain cells that we can study in detail. But of course, since stem cell technology has changed the field radically, and since the discovery of being able to take a adult skin cell, for example, reprogramming it and turn it back into a stem cell, and then being able to give that stem cell a set of cues, or postcodes, if you like, to turn that stem cell into a cell that will become the nervous system, and then in the nervous system, the cell that gets affected in disease that that technology has allowed us now to make a model of the human. And brain in a dish, and in its most simplified way, what we're trying to do is we know the cell types that are affected. In Parkinson's. It's a cell type called the dopaminergic neuron. It secretes and makes a specific neurotransmitter called dopamine, and it's in the midbrain, so in our dish, then what we're doing is taking cells from patients, turning them into the stem cells, and then making the cell type that they have in the brain that's affected in Parkinson's in a dish. And we also make cells from the front of the brain, the cortex, because we know that they're involved in the memory loss that can happen in the disease as well. And more lately, we've been working on making the other cell type in the brain, the glia, which is consists of sort of astrocytes, microglia and oligodendrocytes, three really crucial cell types in the brain that work with the neurons. And in fact, we're increasingly understanding that they're really important in disease as well. So now that's what we do in a dish, is we make those cell types, and then we try to understand the mechanisms in those those cells. And
Steve Flemming 11:05
I this is probably a naive question, but is there anything understood about why dopamine cells in particular suffer from this aggregation and cell death? Yeah. I The
Sonia Gandhi 11:23
the reason why certain cell populations are vulnerable in disease has been the subject of huge, huge research, and we know that there are a number of things about the dopamine secreting cells that are affected in Parkinson's that might make them vulnerable. So, for example, a projection, a neuron that goes from the substantia nigra to the striatum, which is what we think one of the early neurons that gets affected has a massive dendritic arborization, and it's a hugely complex architecture. It's unmyelinated, and it has huge energy demands because of the architecture and the and the footprint of the neuron. So it's already quite vulnerable to any aging stress, for example, anything where you know you have less energy or you have more oxidative stress in the brain, then the dopamine neurotransmitter itself, the chemical itself is metabolism generates a lot of free radicals and agents that might also damage the cell. So the cells that dopaminergic have a lot of stresses to deal with, and it's thought to make them vulnerable. But they're not the only vulnerable cell type in Parkinson's, but they certainly are one of the earliest ones to be affected.
Selina Wray 12:39
And so what do you think, or kind of where is the feel that in our understanding of what these aggregates are doing to the cell, then how, how we know we talk about aggregates potentially having toxicity, but what are the different mechanisms by which That toxicity arises? So
Sonia Gandhi 12:58
we think that the that the aggregates are forming in the cell, and at the earliest stages, once they have a certain structure that can interact with the lipids of the cell and the membranes of the cell, what they can do is they can disrupt a lot of the critical organelles in the cell. So a lot of the work that we've done has been to show that when alpha synuclein starts to aggregate, it can aggregate in one of the critical organelles, the mitochondria of the cell, and it can disrupt the way in which that organelle generates energy, or ATP, and it can also disrupt a lot of the signaling of death pathways from the mitochondria to the rest of the cell, and it can even disrupt mitochondrial calcium handling. So once you have now a physiological problem with the way in which the cellular homeostasis is maintained that puts an extra stress on the cell. But then on top of that, the aggregate needs to be cleared by the cell. So there's a whole system in place that we call proteostasis, which is how to refold misfolded proteins, and if you can't refold them, how do you degrade them? And that involves another organelle in the cell called the lysosome. And we understand that when aggregates form, they also put the lysosome under stress as it tries to clear these misfolded proteins. And if the lysome can't function properly, that causes a backup or an accumulation of these misfolded and toxic proteins in the cell to continue doing damage to other parts of the cell. So you can imagine a cellular cell trying to handle this insult, this this protein misfolding insult is now engaging all of its pathways to try to deal with that at the same time and suffering from sort of injury in different compartments within the cell. So it's a very difficult place to be in. And then you have the added problem, which is then when an aggregate is secreted and goes into the extracellular space outside the. Neuron, you have the reaction of all the other cells around it. So the glial cells are the inflammatory cells of the central nervous system, and those aggregates can be seen as damaged molecules, and they can activate the glial cells, and then they will generate an inflammatory response, which then goes on to cause more neuronal injury. So you get into this vicious environment and vicious cycle, if you want, in the if you like, in the cellular, micro environment of the aggregates, inducing toxicity within the cell, coming out of the cell, inducing more toxicity in the environment as well.
Steve Flemming 15:35
I mean, just going up a bit further back up the causal chain. So you mentioned you take the stem cells from the patients and then use these in your lab to generate dopaminergic neurons that you can then use to as a model system for understanding this misfolding and so I'm wondering, how does that work in practice, in terms of do you then need to Wait a certain amount of time before the misfolding starts to happen. And does that? Does the kind of temporal aspect of this give you insight into the causal factors earlier on in the cell life that then lead to the misfolding process? We
Sonia Gandhi 16:19
We are constrained with the model in the dish by it being a model of of an image or neuron, which essentially means that when we make our models from the stem cells, we're generally making making models that map more to the fetal versions of the neuron. And because of that, it's really a developmental system. So we have to wait in culture for the developmental system to take its its natural course, if you like, until it until its identity is stabilized. But even when the identity of the neuron is stable, it's still sitting as a fetal dopaminergic neuron. That means that our experiments are very long. So we often wait a long time in culture to be able to use the neuron for our experiments in the when we make the midbrain dopamine neuron, that can be six to eight weeks in culture. But actually, when we use some of the neurons from the cortical areas. We can wait, you know, up to 100 120 days. And the other problem with that is that, since Parkinson's disease is a disease of aging as well, people who develop the condition will have the aging stress that may have contributed to the initial triggers and drivers. Recapitulating that in a dish is very, very difficult. It's a disease that happens over a period of 20 to 30 years with very slow toxicity and slow neuronal loss. So how do you do that in a dish in a time frame that we can study? So one way is to apply external stresses or aging factors in the dish, and again, try to make an aging model on the background of a fetal model. As you can see, it gets more and more complicated, and the other way is to take ourselves from patients who have specific forms of Parkinson's disease that is driven by a genetic mutation. And the reason for taking those is that their disease comes on much earlier in life. The mutation is driving a specific pathway that we will already understand, because we know where the genetic mutation is, and when we take cells from patients that have these familial forms, they're rare forms of Parkinson's, but they're driven by a mutation. Then actually in the dish, we are able to see the protein misfolding at much earlier stages. So using the fetal model is is sort of offset by using a disease model of a more extreme form of the condition, if you like, or a more severe form of the condition, I must
Selina Wray 18:59
follow up and ask you about this, because one of for me, one of the most exciting potentials of this technology is this idea that we can maybe build personalized disease models. And I know in some recent work, you've been combining your stem cell models with AI to really look at how cellular phenotypes might link to the clinical picture of the patient presentation. I wonder if you could tell us a little bit about that.
Sonia Gandhi 19:29
Yes, this was a fun piece of work we did during the pandemic when we couldn't do any culture, but we realized we had very, very large data sets of imaging cells from from patients with Parkinson's. And what we were trying to do was to say that we know that the disease is heterogeneous. By that mean, by that, we just mean that it's clinically variable. So we have patients who will have disease that we know is spreading through the brain and gives them a lot of. Cognitive loss. And when we look at post mortem we we know that that's because they've got a lot of protein aggregation in their brain. But then we have patients as well who have a very benign form where the disease stays mainly in the midbrain. They have what we call a negropathy, and the neuronal loss is really confined and doesn't affect other brain regions. And when we look at the familial forms, we know that that form is usually driven by a problem with the mitochondria in the cell. So we have, then, on the one side, rapid disease with protein aggregation. On the other side, slow disease with a mitochondrial problem. And then, of course, everyone in between is going to be have a slightly different mechanism happening in their brain, and that's why they are having these different clinical phenotypes. So the premise was that if mechanistically, the mechanism determines your clinical presentation, then can we get to those different mechanisms in the dish? And that hasn't really been done before, because, you know, we don't even have biomarkers for the disease, let alone biomarkers for different mechanistic subtypes. So what we did in this program of work was say, well, we can artificially generate different subtypes of disease. We can generate a mitochondrial form by giving a mitochondrial toxin. We can generate a proteinopathy form by giving this folded protein so we can sort of engineer chemically these different subtypes. And then we can go one step further by taking patients with different mutations in those two pathways, a mutation in the mitochondrial form and a mutation in the protein aggregate form. So now we have these subtypes mechanistically in the dish, and we can image them, because that's what we do. We're an imaging laboratory, and we can image multiple things about the cells. We can image their mitochondria. We can image the lysosome, the pretty stasis system. And then we can say, well, as humans, what we do is we analyze specific features from that data set. But what can a computer do in an unbiased way? So what we did is we asked the computer, first, the algorithm, to first train on all the features that come out of the machine where you just do a straightforward imaging experiment. So out of the confocal microscope and the processing software will pop out about 250 features, many of those we ignore, actually. And we asked the algorithm that, if you look at those features for lots and lots and lots of cells, can you predict which mechanism was occurring in the in the cells? And that did with quite a lot of accuracy. Actually, it was sort of 80, 85% accuracy. But then we decided to go a step further and say, Actually images hold much more information than quantitative data, and all that information is hidden, so we see it as humans, but we don't really, really process that when we when we do our experiments. So instead, what we did is we trained these algorithms just on the images, so we broke down our images into single cell images, so we had, I don't know, 600,000 images, and we asked the algorithm, if you look at an image from a healthy person's neuron versus an image from a mitochondrial form versus an image from a proteinopathy form of Parkinson's, can you tell the difference? And actually, on that basis. So it's machine learning on the basis of computer vision, the accuracy was really high because we were obviously capturing in our images using cell painting, the right parts of the cell that had all the information, and then the algorithm was capable of being able to distinguish different mechanisms. So it's just the beginning, and I wouldn't want to overstate what we found. It was just our ability to see that machine learning and deep learning is is a powerful tool that actually it's a powerful tool on imaging, and we could potentially use, then our patient, example of their own brain in a dish to be able to understand what might be happening inside their brain in a way that standard imaging, MRI scans, for example, can't do with. They can't get to the mechanism of what's happening inside the cell, but the model of the patient in the dish just might be able to
Steve Flemming 24:20
this is super cool stuff. And I'm just it makes me wonder whether there was there anything in the computer vision model in terms of the features that it picked out to predict the disease mechanism that surprised you, that was not part of the standard way of doing this kind of science. Yeah,
Sonia Gandhi 24:38
it's a great question. So the reason we took the two approaches of using features that are quantitative or quantitative already is that there's really high explainability, so you can then rank the features and work out which is actually giving you the prediction. And that was very intuitive. So what happened was it was using the mitochondria of the cell and. The lysosome of the cell in the top 10 features, and the combination of the two to tell us which mechanism. Now in computer vision, explainability is poor generally, and that's one of the problems, right? So we don't understand how it's working, but what you can do, and what we did in this in this experiment, was you can drop out different parts of the image and see if the prediction still holds. So we were able to understand which part of the image is being used most. And there was a bit of a surprise, because when we use the mitochondrial channel alone, that could predict 80, 85% of the overall prediction. But So could the lysosomes. But in fact, using the nucleus, which is, you know, headquarters of the cell, and not where we thought a lot of the signal was, was also contributing to the prediction. So for us, that was new biology, because what is different about the nucleus, the nuclear morphology in disease cells, that was allowing the algorithm to make the prediction. So, yes, new biology can, I think, emerge from these methods? What they mean is a little bit difficult for us to say right now.
Selina Wray 26:11
It's really timely for us, because we are doing our first live recording this evening, and for any listeners, our next episode will be entirely on neuro AI. So that was a nice little plug. And please join us to listen more for more. Sam, you. I mean, that's a very productive way to spend your pandemic, and I want to come back in a few minutes and ask you about other things that you were doing in the pandemic. But before we do that, we've really focused only on your lab research, but you're also still very active clinically, and particularly in the area of experimental medicine, and thinking about new ways that we can design clinical trials so that we improve patient outcomes. So before we do move on, maybe you could just tell us a little bit about the more clinical research that you're part of.
Sonia Gandhi 26:59
Okay, so we are really, really fortunate at UCL, because there are just so many clinicians and healthcare professionals as well as scientists studying Parkinson's disease or unseen patients and and about five years ago, we all came together and set up a movement disorder Center, which is sort of translational. It's a translational research center that essentially brings together our patients on a common platform, tries to understand how we might one day deliver precision medicine, right treatment at the right stage of their condition and to the right individual. And to do that, we've understood that we really do need to see that as very much a team, a team science approach. There's no way that that sort of endeavor can be delivered by any any individual or any laboratory alone within the center. We also conduct quite a lot of clinical trials for different stages of the disease, so symptomatic treatments at all stages and also disease modifying therapies, which, of course, is is the thing that we're all trying to work towards, is finding better disease modifying therapies. And what we've learned is that the current clinical trial designs are inefficient and very lengthy. So one of the difficulties with treating neurodegenerative diseases, but not just treating them, bringing new drugs to the clinic, is that it takes about 10 to 15 years to go from the beginning of a phase two trial to the end of a phase three trial. And not just that, all of the phase three trials for disease modifying therapies for the past 20 years have ended in failure, but failure at the end of the trial. So at the end of that 10 to 15 year journey, we find out that something hasn't worked. And it's not just a massive financial cost, it's actually a hugely intensive labor cost for the people conducting the trials. You dismantle the entire trial machinery, you reassemble it a few years later to do the next phase of the trial, and only to find it doesn't work, and then it's a massive cost to patients. I mean, I think the patients who invest their time and effort and the hope in these trials just can't be, you know, you can't put a put a value on that and and for them to be either in the control arm or a treatment arm of a trial that then fails after many, many years is just devastating. So there's been this big movement, then from patients and from trialists that actually we could design the trials better, and we could make this journey much, much more efficient, which would be ultimately better for patients. And so the trial design we're looking at and we're adopting, and in fact, starting the first trial next year is something called a multi arm, multi state. Each trial. You may have heard of them in the context of covid. So the covid recovery trial was this style of trial, and it's been used before, very successfully for prostate cancer. It went out of UCL. And in this trial design, you have one control arm and multiple drug arms, and they run simultaneously. So patients like it, because there's only one control arm and not one control per drug arm. And then the second part of it is you do an interim analysis at 18 months, which means that if there's no signal, you can drop that drug arm out and immediately take on a new one into the trial, so you have an early signal of futility. And then the third part of it is the phase two runs straight into the phase three, so no big gaps. The trial machinery keeps running until a drug is found. So in five years, you can actually evaluate far more drugs than you could in the original way in which we conduct trials. So it's really a, you know, it's an obvious thing to try to do. It has huge challenges, because doing this in neurodegenerative disease is not straightforward, and our trial, called the Edmund J Safra accelerating clinical trials in Parkinson's Disease Initiative, is set up, really, to operate across the UK, the people doing the trial, and it's led by Tom Fauci and Camille Carroll and but it involves approximately 90 people. It'll be 40 sites across the UK, and we're really very, very excited about it. So that's the other major initiative that I've been involved in.
Steve Flemming 31:39
Could you say a little bit about what the therapies are, that the kind of cutting edge therapies are that you're looking to test within this trial? We,
Sonia Gandhi 31:49
in the first iteration of this trial, we are actually going to use repurposed drugs, and the reason for that is because they're available. We the cost of the trial infrastructure, of course, is huge, and we really want to show that this trial design can work and can deliver a result. The process of selecting drugs for trials like this is not straightforward. So actually, we've gone through a very, very intense process where we've joined up globally with a global consortium that's looked at a lot of the drugs that are currently in the pipeline or being studied, we've looked at all of their pre clinical data, their epidemiological data, any genetic data, come up with a short list, and then that shortlist has gone through another few rounds. So we have a very final shortlist of what those drugs do. And we're trying to also target different pathways developing new therapies, and the cutting edge therapies, for example, the genetic therapies, antisense oligonucleotides, are also being trialed, but not within this specific trial platform at the moment, but we hope one day we'll
Selina Wray 32:54
have to get you back on when we know the results then, so we can discuss Those Absolutely So Sonia, you told us about this wonderful work on on AI and looking at your cell images that you'd done during the pandemic. But actually, although you couldn't do cell culture, I think we don't want our listeners to think that you just down tools in the pandemic. Because I think, unlike a lot of us, your lab at the crick kept running. But in a very different way. Can you tell us a little bit about what your initiative started in the covid period?
Sonia Gandhi 33:27
Wow, the covid was. The covid pandemic was a really, it's sometimes hard to capture back what it was like at the start of the pandemic. But in March 2020, I was on call in the hospital at at UCLH. And I think it's, it's, it was definitely one of the first times as a doctor that we experienced seeing all patients with one condition, all the staff with the same condition. You know, seeing, for me, young patients on the intensive care unit with covid. And that time was just, just an extraordinary and very, very painful time as a doctor. And at that time, as we were going into the first lockdown, the leadership at the Crip Hall nurse had sat me down. I was on my way to clinic, and I remember this very clearly. He sat me down on the sofa in the collaborative spaces and said, What can we do to help the NHS? And I was there in conversation. Then I said, Well, we, you know, let's talk to them, and we work very, very closely with UCLH. And UCLH were, you know, a phenomenal hospital that time. I think they were really forward thinking, and they and they said, you know, the biggest thing we need right now today is testing. Because at the moment, we have no we do not know who has the disease, who has our. Covid, two infection, and we just are operating blindly, really, and so what we did was, at the time when we didn't close, what we did was repurpose the entire Crick into a testing facility. And the reason we did that was because we have a very, very large, advanced sequencing facility, so we have all of the robotics and the automation and the and the sequencing to do that. But at the time, the biggest constraints were, could we set up a testing pipeline that could interface with the NHS? So we had to interface interfacing with NHS digital it was quite a complicated problem, so we we had to put in place a pipeline that allowed staff to put in swabs, then we would collect them, test them, barcode them, track them, and report out a result, eventually to a portal or and then to their their phones. And setting up that whole pipeline took some time, but there were other constraints we'd run out of, like the whole country had run out of RNA extraction kits, and we were waiting for the last shipments to come for that from China. I remember speaking to a pilot flying out just before the first lockdown about whether we would have that on time. And so what we did in those early days was ask, actually across the Crick, what other technologies can we use? And there was a lot of innovative technologies coming through, but one group had been working on how to extract ancestral DNA from from materials that were, you know, millions of years old, and they'd been working on this method. So we said, well, let's try that. So we tried this method where we could synthesize every chemical in house to do that. And it worked. And I remember the day it worked. And and then we we turned to this method of extracting the the extracting the genetic material, and then making all of our chemicals in house. And so the entire pipeline was completely self sufficient from start to finish. Was not dependent on any external provider. And then we also provided our chemicals throughout the UK for diagnostic testing. And we did the whole thing. I think we went from beginning to end in about two to three weeks for the smallest number of samples. I remember it because it's my birthday when we launched the first formal testing on April the first and then, of course, we started to scale up. So at one point, we were doing, I think, about 4000 tests a day, and we were testing for the 1010 hospitals and about 150 care homes around us as well, where staff just couldn't otherwise test, and then we did also our own institute and some other institutes around us. So it was about trying to keep people safe through testing. And I really enjoyed it. I guess it was helpful having the scientific background and being able to understand both how diagnostic testing needs to work, but how, how laboratory innovation can change testing? So that was the beginning. And then once the vaccination program had come underway, we then decided to use the crick to set up a vaccination center for our local population, and that was staffed by many, many staff from UCLH as well and from the National Hospital for neurology, and we delivered just a very large number of vaccinations. And then the final thing we did was try to, at one point we recognized that we had maybe a million swabs as well from staff and from staff within Crick and staff across UCLH was to say, Well, look, you know, actually, there's quite a lot of research we can do here. So we applied very early on for the ethics to be able to use the samples to understand viral evolution and to understand immunity. And the other thing that the crick has, which was very useful was the worldwide influenza Center, which we're very used to, having neutralization assays, where you can test a patient's serum and to see whether it has the right antibodies that would neutralize the SARS covid Two virus. So we set up a live neutralization assay in high content that's now very, very widely used to be able to test the serum of people who are part of our large research cohort now to see whether they would be immune to these different variants that kept evolving and that resulted in a in really, we would always share our results back with the government and all the and all the bodies that were deciding on public policy. So we played a big role there as well in in understanding immunity. So a lot of work we did in the end. But I'm not a virologist or an immunologist. I mean,
Steve Flemming 39:50
it's such an A such an amazing story. And I think it's, I mean, more broadly, so many stories from covid about the under. Unpredictable relevance of basic science, and the importance of having a robust infrastructure for basic science. I mean, the story you just told about the ancestral DNA extraction, who would have thought that that would then become relevant for, you know, applying a test in the middle of a global pandemic? I mean, it's just fantastic to hear these stories. And
Sonia Gandhi 40:18
I think in some ways it taught us a lot about The juxtapositioning, the fact that you have a research institute that is multidisciplinary right next to the hospital, in the same way that the universities could, meant that you could really translate, and we were translating in the time of covid really, very, fast. And I think that is a bit of a blueprint to show that translation is possible, to go from bench to bedside very quickly, if we have the right infrastructure and obviously the mission and motivation to do that. So it taught me a lot about translation.
Selina Wray 41:01
So maybe we can, we can just switch gears a little bit now Sonia and I think one of the things that we must discuss, there will be a lot of people listening to this, thinking, wow, how do you end up with a job that is so diverse and has such wide range? So maybe you could just spend a few minutes telling us about your scientific journey through university and your training, perhaps, what were the kind of key transition points and choices that you made that allowed you to have the role that you have now? Okay,
Sonia Gandhi 41:31
the first thing I'm just going to say is it's a very long journey to become a clinician scientist, but I would say, also a very rewarding one. So I started out by doing a medical degree. And the undergraduate part of my medical degree was in Cambridge, and as part of my third year, where you do an intercalated a bachelor's degree was in neuroscience. So I think very early on, I'd understood and was really fascinated by the science that underpins medicine, of course, but also neuroscience. And I came into contact with some phenomenal teachers and people who inspired me there as well. And I met, I met Huxley, who had essentially described and discovered the action potential, and he had been in the same college as I had, and he's he talked to me over it, over a dinner, about how he made this discovery of the action potential using the squid axon and his sort of humility around what was going to be one of the greatest discoveries, how he made it, and how kind of serendipitous science is really struck me. And I think I was already hooked into understanding the brain and the way in which the circuitry and the networks of the brain operate that I think I'd already engaged very strongly into the idea that I wanted to do science and I wanted to neuroscience. And then I did my clinical training and the rest of my clinical degree in Oxford, and there I came across a number of neurologists who, again, really inspired me into what I would have to decide which would be My clinical specialty. And from there, I left Oxford and did my clinical training as a junior doctor, as a house officer and senior house officer, as they were called, then in in Oxford bath and then London. And so for many years, straight after my medical degree, I was really training as a junior doctor, doing very, very long hours in renal medicine, in hematology, cardiology, and eventually came to the National Hospital for neurology at Queen Square, and did my neurology training as a junior Doctor. And so I think there I learned that neurology was going to be the specialty for me, and that was good, because that was the science that I'd wanted to do as well. But then, you know, it's, it's a phenomenal place to practice neurology. It's, you know, the brain I got, the advice I was given very early on is you just have to choose what will be the most interesting thing in life for you, and therefore for me, for all parts of the body, that was the brain, the mysteries of the brain, but also the way in which we practice neurology is a little bit different to the way other specialties of practice. And the advice I was given was to go away read Sherlock Holmes, and if that's what you find interesting, that's what you need to do, and you feel like you are a detective when you do neurology that you're piecing together using the art of observation deductive reasoning a very complex puzzle, and what you end up with is is a very incomplete solution to the puzzle, as well as a neurologist. And but it's utterly fascinating, and for me, has really held my my interest and my passion. So that's how I got into neurology. And then I had to really do decide whether I was going to be an academic physician or a clinician scientist, and I opted to do a PhD, and the PhD was in partly genetics. It was shortly after the discovery of one of the genetic forms of Parkinson's disease, the discovery that pink one mutations in the pink one gene can cause a mitochondrial form of Parkinson's disease. And so I worked on how you go from gene discovery to understanding the function of the protein, and then the fun the malfunction of the cells. And that was just really good training for me. So I did my PhD at UCL, and then took on neurology training as a specialist. And I went to East London, to the Royal London, and spent some time in Essex as well doing my neurology training. And then came back to academia to be a clinical lecturer at Imperial and did another four years of training where I did 50% research as a postdoc and 50% clinical, but I actually had all my well, two of my children during that time, so I spent a lot of time on career breaks from the clinical world. And after that period of training that took me, you know, eight years where it would have otherwise taken a neurologist, maybe four years at that time, I then became a consultant. And at the same time, I got a fellowship from the Wellcome Trust to set up my own laboratory. So in 20, end of 2013 beginning of 2014 I established my own laboratory at UCL, and then I seconded to the Francis Crick Institute just two years into that, the move. We actually moved in 2017 and I've been there ever since. So does that sum it up? Yeah. I
Steve Flemming 46:58
mean, it's an amazing journey, and as you say, a long road to combine both a clinical career and an academic one.
Selina Wray 47:06
Well, thank you so much. Sonia, that was a really fascinating discussion. I think we could have gone on for another hour. Thank you so much for joining us on this episode of brain stories. Thank you everyone for listening and see you next time.
Sonia Gandhi 47:19
Thank you for having me.
Steve Flemming 47:20
We'd like to thank Matt Wakelin, my super and Trevi smart for their roles in taking brain stories from an idea to a fully fledged podcast. Patrick Robinson and UCL digital education for editing and mixing and please follow us on pecs at UCL brain stories for updates and information about forthcoming episodes. You