AudioHelicase Podcast: Olivia Corradin on a new way to probe disease genetics

On this episode of AudioHelicase podcast, Whitehead Fellow Olivia Corradin talked about investigating the genetic underpinnings of diseases through a new technique she developed, the outside variant approach. Applying the method to study the autoimmune disease multiple sclerosis (MS), Corradin and colleagues identified a role for a cell type in the brain in MS, offering a new way of understanding the disease. She also discussed running a lab during the COVID-19 pandemic, and evaluating scientific information in a time of uncertainty.

Subscribe to AudioHelicase on iTunes or SoundCloud and never miss an episode!

EDITED TRANSCRIPT

CONOR GEARIN, HOST: Welcome to AudioHelicase, the podcast from Whitehead Institute where we unwind the science and the people behind some of the Institute’s most exciting discoveries. We think of some diseases as genetic and some diseases as transmitted through a pathogen like a virus or bacteria. But it’s not always so black and white. There are many conditions in which your DNA sequence can raise or lower your risk of developing a disease on a sliding scale. Whitehead Fellow Olivia Corradin has developed a new approach to investigating disease genetics that can help reveal previously hidden mechanisms for how diseases operate in specific locations in the body.

I’m Conor Gearin, digital media specialist at Whitehead Institute. In this episode, we’ll talk to Corradin about her new technique, called the outside variant approach, and how it can reveal the genetic underpinnings of conditions that have remained a mystery to scientists—including multiple sclerosis, an autoimmune and neurodegenerative disease. We’ll also hear her thoughts on running a lab during the COVID-19 pandemic and her suggestions on how to evaluate new scientific studies and knowing which sources to trust. Corradin earned her PhD at Case Western Reserve University in the lab of Peter Scacheri. She’s the Scott Cook and Signe Ostby Fellow at Whitehead Institute. As a Whitehead Fellow, she runs her own lab and conducts independent research while still an early-career scientist. 

Olivia, welcome to AudioHelicase. Can you start us off by talking about what your lab is focused on right now?

OLIVIA CORRADIN: My lab is a human genetics lab, so what we’re most interested in is understanding the genetic components behind human diseases. What we particularly focus on are those genetic variants that alter the genome but don’t directly affect proteins. These are DNA variants that occur outside of genes—which make up only 2% of your genome–so we’re really interested in how those can still work to have an effect on disease, even though they’re not directly affecting a protein’s function.

GEARIN: Right, which brings up something that might be a source of confusion, which is that when we say a disease is genetic, some people might think of that as a binary—that a disease either is or isn’t genetic. Is there a better way of thinking about genetics and disease risk?

CORRADIN: People tend to think that when you say something is genetic, that means it’s fully determined by your genome and nothing else. But actually most of the diseases we talk about are going to be a complex relationship between your genetics and your environment, your lifestyle choices, and even things like what happened to your mother when you were in utero. So it’s really a complex relationship between these variables. What we think about most often are genetic risk factors. So what genetic variants you have that make it more likely to develop a trait. But it doesn’t mean that developing that disease or trait is predetermined. It just puts you at a greater risk of it happening. The one that is easiest for most people to understand is the relationship between genetics, lifestyle, and obesity. Yes, there are genetic factors that put you at greater risk, but there are also lifestyle factors that can put you at lower risk. It’s really a combination of all these contributors that will, in the end, define your status.

GEARIN: How did you first get interested in studying how genetics affect disease risk?

CORRADIN: It started back in undergrad in a fruit fly lab, which I think is a way a lot of geneticists get started. It’s a really great way to see genetics in action. When I went to graduate school, I found a surprising new love, which was understanding the whole genome—really thinking about how one DNA template can make for a huge amount of diversity in the human population. How one DNA template can make your brain cells function, as well as your blood cells function. 

That led me into a big data problem. If you’re talking about trying to understand the whole genome, you need to develop a whole new set of skills. That involves interpreting and analyzing really large datasets. It shifted me a bit from molecular genetics into studying genomics and more computational biology. I just ended up really loving that. I joined a lab that I never expected to in graduate school, studying the epigenetics of cancer, which is understanding how the shape of the genome changes in different individuals. That led us to this problem of—we just have so little ability to understand how changes outside of genes contribute to disease. That’s where I really found my greatest passion and really started pushing forward to understand how changes in that area of the genome can really have an effect.

GEARIN: When scientists try to uncover the genetic components of a disease, there’s a particular type of study they can do called a genome-wide association study, or GWAS. Could you describe what this approach entails and what kind of data it provides?

CORRADIN: Genome-wide association studies have been around for about a decade. They’re by far the most common approach that we use to understand genetics underlying complex traits. By that, we mean things like schizophrenia, multiple sclerosis—even height we would consider a complex trait. So we try to understand what genetic variants or changes might contribute to these different traits. And the way this is done is relatively simple on the surface. You just take a large group of people that have a particular trait, and a large group of people who don’t. And then you just start looking at genetic variants for differences. This gets complicated really fast when you start talking about the fact that these people have different lifestyle choices, they have grown up in different environments, they may come from different ancestries. So we have more complex models that allow us to account for some of these variables. But in the end it’s really the sheer number of people—we just collect huge numbers in order to see what variables, what different genetic variants have an effect beyond all of these different confounding variables. A lot of the variants that we talk about are pretty common in the human population. They don’t necessarily cause a negative effect, they’re just what makes us different. So the general variation that’s possible and inherited from your parents is what we talk about.

GEARIN: You developed a new approach to look at something called outside variants. Can you describe the method and what it adds to GWAS?

CORRADIN: What’s been so successful about GWAS is that they’ve found so many genetic variants that associate with disease. We’re talking about millions of different genetic variants that have been associated with thousands of different human traits and diseases. The field has been incredibly good at finding these variants. But that only takes us so far. We’ve been trying to figure out what these variants actually are doing that makes them more likely to increase your risk of developing a disease. My favorite analogy is a genetic game of Clue. So you have a suspect, but you have no idea what the “weapon” is, and you have no idea where in the body this effect is happening or what “room” this is happening in. What the outside variant approach does is it really focuses on trying to figure out what part of the body is affected by this genetic variant in a way that contributes to the disease pathogenesis.

 So a little bit more detail about how the outside variant approach works: it’s really looking at three-dimensional interactions in your DNA. Your genome isn’t just a line. It’s compacted into the nucleus. We take advantage of that 3D shape to help us study a particular genetic variant. We look at not just what the genetic variant is, but also all the genetic variants that touch and interact physically with that particular variant. And then we ask if those additional DNA changes make you more or less likely to develop the disease. So instead of having just one data point, we’ll end up with somewhere from ten to one hundred data points that can tell us about the particular region of the DNA. So that gives us much more deep information that we can compare to different cell types and say, well maybe one variant is important across all of the cell types we’re interested in, but if you look at all twenty of them, all of them act in the brain, but only one of them acts in the blood. So we can start taking advantage of these 3D interactions to get this extra information—all by looking at small changes. You start with maybe a 50% increase in your risk of developing a disease, and then we ask if that can be modified by a small amount by a particular DNA variant.

For example, when you’re studying a disease like multiple sclerosis—this is an autoimmune disease, but it’s also a neurodegenerative disease, so it affects both your immune system and your neuronal system. What we want to do is that if we know a particular DNA variant is contributing to disease, we don’t know if it’s because it’s altering your T cells or if it’s altering a neuron. What this approach does is focuses in on that problem and identifies some additional DNA variants that contribute to the effect. By looking at some of these more nuanced, small-modifier changes, we can start looking at what cell type is most important. And then we can start moving some of the hundreds of DNA variants that we’ve associated with a trait into categories—figure out which ones are affecting the immune system, which ones are affecting the brain—and that helps us reveal patterns. If we look at just the patterns that are affecting the brain, we reveal things that we might have missed if we looked at all the variants together.

GEARIN: You’ve been able to apply the outside variant approach to learn more about multiple sclerosis, or MS. So first of all, what is MS and why is it that parts of its biology remain a mystery?

CORRADIN: MS is an autoimmune disease, so it involves a patient’s immune system mistakenly recognizing self for bad. It actually leads to your immune system attacking a part of your brain called myelin, which is this protective layer that wraps around your neurons, and that helps the signals in the brain to move very quickly—which is really important for a complex system like the human body. What happens in MS patients is this myelin layer, this protective layer starts to degrade over time, and that leads to these neurological symptoms. What’s interesting about MS is that the patients’ symptoms tend to occur and then remit. They’ll have periods of time of severe symptoms and then periods of remission. That’s a complex interplay between the immune system attacking myelin and the myelin being repaired. A lot of the difficulty in studying a disease like this is what day you look at the patient is going to affect what kind of data you’re collecting. So understanding how two systems, two complex systems, the immune system and the nervous system interplay to cause this disease is what led to the challenges to furthering our understanding it.

GEARIN: And what did the outside variant approach reveal about MS?

CORRADIN: Using the outside variant approach, we really focused in on the cell types. When you think about MS genetics, if we look at all of the results from genetics studies, we would conclude that MS is mostly an autoimmune disease. If we look at the hundreds of DNA variants that have been shown to increase your risk of developing MS, we would predict most of them to affect the immune system, and so many that we couldn’t find any that affect the nervous system. So knowing that the pathology of the disorder involves this interplay between the immune system attacking and the brain responding, but none of the genetic variants seem to be affecting the brain, that makes it really hard for us to understand what are the factors in the brain that we need to improve upon in order to help patients. The outside variant approach looked at each genetic variant associated with disease independently. So we just took one of these sites at a time and analyzed them at a great depth, and evaluated which of those were more likely to affect the brain.

When we did that, we found a couple of genetic changes that appeared to be altering the function of a special cell type called the oligodendrocytes. So these cells are responsible for generating new myelin in the brain. Those became really interesting to us, because we knew myelin was really important for the disease. We went and looked at the function of the genes in these regions, and found that the ones we think are acting in oligodendrocytes, if you alter them, if you decrease how much gene product is made, then the cells no longer produce new myelin. So we were really interested in understanding how some people might be born with some genetic variants that make them not as strong in response to an immune attack. We’re hypothesizing that these variants are affecting the efficiency in which your body can generate new myelin.

GEARIN: And does that open up any new ways of understanding how MS works, and how to approach treating it?

CORRADIN: That’s always the hope, that you’re actually getting to insights that will affect how we can actually treat or diagnose patients. For MS, the majority of treatments are focused on blocking your immune system. So what they do is they prevent the immune system from interacting with the brain. That has been a really successful treatment strategy. It slows progression of the disorder, but it doesn’t halt it. What that’s treating is further development of symptoms. It doesn’t treat the symptoms the patients have already developed. If you could understand what aspects of the brain could be modulated in a way that’s beneficial, you might imagine a combination therapy where you’re both blocking the immune system and repairing the nervous system. So this is opening a door for a strategy like that to pursue making patients more efficient at generating myelin as well as preventing its damage in the first place.

GEARIN: Are there other disease mysteries that you think are amenable to the outside variant method? And is your lab doing any of this work right now?

CORRADIN: There’s a lot of diseases where we tend to only think about one part of the body at a time. And when we think really deeply about how a disease works, most often it’s going to involve multiple cell types and multiple parts of the body. There’s a number of easy examples. When you think about a cancer, for example, you might expect that a lot of the genetic variants that cause your tumor are in the cell type that becomes the tumor. If it’s in your colon, you might expect that it’s the colon cell that’s affected. But more and more, we’ve learned about the importance of the immune system. You can imagine a scenario where some people are born with less ability to detect cancer cells, and their immune system is less efficient. That would put them at risk of developing a tumor. So trying to distinguish genetic variants that are affecting the cell type of origin of a cancer versus the immune system would be a really good next step. We’re starting to pursue that with breast cancer. Other diseases we’re looking at include substance abuse disorders, especially opioid use. That is really a wide open question of what areas of the brain and what cell types in the brain are most critical to defining one’s susceptibility to addiction. So we’re starting to look at genetic risk factors and whether they affect glia, astrocytes, neurons, or if it’s more about the amygdala, more about the nucleus accumbens. We can start to make some of these distinctions by using our approach.

GEARIN: When working with datasets of this scope, I’m imagining it could be hard to know where to start or even how to sort through all the data.

CORRADIN: Yeah, it’s always a question for us. The datasets are huge. We can try to study all of the genetic loci of interest, we can just take a hammer approach and just try to do everything, or we can be a little more precise and look at loci that we have hypotheses for, so more of a chisel approach. The way we do that is we look at a lot of publicly available data, so we’ll look to see what genes are nearby, if there’s any information about what those genes do, and if those are related to some published literature on a trait. Sometimes we learn a lot by not trying too hard to make a hypothesis, just being really open-minded. It’s a lot of time on the computer. The compute time is quite long. But we do get more novel results that way, where we can find things we weren’t necessarily expecting. One of the hardest parts for us is that we can apply this approach to any trait that we can get data for. So choosing the traits that are most valuable for our method is important. We’ve really been focusing on traits that we really think more parts of the body are involved, but haven’t been indicated yet by current studies. The classic example is cancer and the immune system. But others as well, where you might think that maybe more glia cells are involved, support cells in the brain that aren’t just neurons, and trying to find some of those nuances as well.

GEARIN: Shifting gears a bit, how has COVID-19 affected work in your lab? Is there anything that has helped you continue pursuing your research?

CORRADIN: COVID has caused everyone unique challenges for sure. We’re in somewhat of a lucky position in that everbody, even our experimentalists in the lab, do some computational biology, so we’re able to keep pursuing new science from the computer. We use a lot of data that’s been published, so it’s a matter of downloading and asking and answering new hypotheses. It led to some new interesting and creative ways to pursue projects that were stalled by to the inability to do new experiments. There have been some really funny Zoom calls where we’ve gone back and forth about different projects and sharing screens and using the whiteboard function on these Zoom calls to draw what we’re talking about. Of course we miss the in-person interaction. We’re usually very ad-hoc, you can sort of pull someone into a conversation really easily. So to help us recreate that, we’ve adapted this two-hour, we call it our highly focused time where everyone’s available. At the drop of a dime, between 1 and 3 pm every day, everyone is ready to jump on Zoom or any other communication for immediate response. So it helps us be like, quick—let’s pull someone into this conversation and get their advice. It helps a little bit with the feeling we’re missing where we’re all just in the same room.

GEARIN: It’s great to hear that. During the COVID-19 pandemic — as well as before — there are always new studies coming out and papers posted to preprint servers ahead of formal peer review. Sometimes it can feel like an onslaught of new data. If you’re not a specialist in a particular subfield that a study is coming from, it can be hard to know whether you can trust the conclusions or not. As a scientist, are there criteria you use to evaluate and sort through new research?

CORRADIN: When you’re looking specifically at scientific studies, really pay attention to sample size. If the number of people they’re looking at seems really low compared to the number of people who are affected by a trait, affected by COVID, then you know it’s something you want to be cautious about interpreting. Another big warning sign is big conclusions. Science tends to happen in baby steps, so you’ll see a lot of “this may suggest.” And that’s actually a good sign. That means the researchers are recognizing that their data is supportive of a fact but doesn’t prove it in one study. So when you see warning signs—the title of the paper is really extreme, or if they say “this proves”—those can lead you to question what you’re looking at. With preprints, you'll find Twitter to have lots of scientists commenting on the paper. You’ll see comments at the bottom of the preprint servers. A lot of times, Twitter—it has its pros and cons—but if you’re listening to scientists discuss a paper, if they’re talking about it as a scholar, then you can start to use that as a test as well. But when in doubt, bide your time and read the peer-reviewed version.

GEARIN: Sometimes, motivated patients or patient advocates dive into the primary literature to learn more about a condition—especially if it’s a rare disease. I’m wondering, for a person looking at info that’s outside of their specialized training, what kind of tools should they use to evaluate that information?

CORRADIN: I do it myself. When I broke my foot recently, I googled scientific papers about whether surgery is the right option or not. Having the ability to distinguish is important. Always google the journal name. If you’re reading a paper somewhere that you think is a peer-reviewed journal, google the journal name—make sure it is something that is a recognized journal, because there are these sort of predatory journals out there that are publishing work but are not as rigorous about reviewing its content.

I think it’s all about trustworthiness of the source. So is this a website you recognize? What information is on that website that maybe you have more confidence on? You can use that as a litmus test. Do they claim the dinosaurs aren’t real? You can see if their other news stories have a high degree of validity as a starting point. Then, look at the study itself, and if you find something questionable, google it. You’re going to get a mixed bag. Some people will agree with a particular statement, some won’t, but you’ll start to see trends. You’ll see the reputability of the news source will be a really good indicator of whether or not that’s a particularly high-powered study. So a really easy example, I had someone on one of the social media websites that I know post that COVID-19 meant that it was the 19th virus, and they posted an article that said so. I Googled COVID-19, I found the CDC hit that said that, no, it’s not the 19th COVID, it’s the COVID virus that was discovered in 2019. So cdc.gov, that’s a highly reputable source, and so you’re able to verify a confusing fact like that immediately. Every hit on the Google search underneath says 2019, 2019. So sometimes you have to use quantity to help you. If it’s the only source claiming a fact, then it’s something you really want to dig deeper into. 

GEARIN: You can learn more about Olivia Corradin’s research on our website at wi.mit.edu. Find past episodes of AudioHelicase and stay tuned for new ones by subscribing on iTunes and SoundCloud. Thanks for listening.