Pushkin. I'm Jacob Goldstein and this is What's Your Problem, the show where I talk to people who are trying to make technological progress. My guest today is Manola's. Kellis Manola's is a professor of computer science at MIT, and he works in computational biology. It's a field where researchers take giant data sets relating to things like genetics and health outcomes and try and understand basically what's going on, things like what are the cellular mechanisms of disease and
how can we intervene to keep people healthy. In particular, Minola's research focuses on genomics and a related field called epigenomics. Here's how Manola's explains.
What that means. What's extraordinary with genomics is that we
can see beyond the limits of human imagination. We're talking about millions of cells across hundreds of people, across thousands of genes, and now we can now look at how the single genome manifests in every cell type of the human body in a slightly different way to create this extraordinary symphony that is the human life, that is human thought, that is human understanding, cognition, and every biological process that ability to now start understanding the building blocks of how
this human genome manifests into all of these myriad of cell types and their interactions and their combinations and their coordination and their communication is what we can do for the first time. They're also giving us the entry points for understanding the basis of human variation, the basis of human disease, and the basis for reversing human disease.
So that is the very big picture view of what Manola's does. In our conversation, we got into a lot more detail. For one thing, Manola's talked about his work on obesity, and that work is based on epigenomics, which is basically the way in which different genes are turned on and off, and this turns out to be a really big deal. Manola's and I also talked about his
work on Alzheimer's disease. In that part of the conversation, he talked about how he and his colleagues are trying to find these key biological pathways that contribute to lots of different diseases, and how they're trying to come up with drugs to target those pathways. We started our conversation by talking about Manola's early work on the human genome, which led to the work he's doing now.
So the human genome was mapped by K ninety nine or two thousand and three, depending on how you count. And then we had all of the nucleotides, all of the letters through into billion letters. Then the hard part begins, how do you make sense of that book? So that was the Book of Life. So we had all of the letters, how do you make sense of the book? My own PhD was developing evolutionary signatures for understanding systematically the human genome. So how do you recognize where are
the protein coding parts? What are the parts that code for protein? We didn't even know.
And just to be clear, sort of non intuitively, most of the human genome is not protein coding, right, Like there's this very basic idea that like, oh, sure the genome, that's what codes for proteins, but in fact, most of the genome is not doing that.
Ninety eight percent of the human genome does not code for protein.
It's wild. That is so nonintuitive, correct.
So in that mysterious ninety eight percent of the genome lie control regents that are responsible for turning genes on and off. And that's where ninety three percent of the disease associated genetic variants are sitting.
Huh, it's not the genes that actually code for proteins, it's the genes that control when are proteins made, when are they not made, how much are they made.
That's exactly right.
Okay, so I get that in a broad sense. That's sort of the state of affairs when you're coming into the.
Field's exactly right. So I wrote a series of papers, both as a student and as a faculty member that sought to then uncover how to even parse the genome, how to even start understanding reading that book of life. So that's one part. The second part is where the regulatory motifs are. What are regulatory motifs. They are the short words of the language of DNA that are bound
by regulators to turn genes on and off. So there's these regulatory regions, and within these regions lie these words which are the regulatory mode.
And just to be clear, the regulatory motifs are part of what determine sort of when and how much different genes express different proteins.
That's exactly right, that's exactly right. And that's where the human epigenome comes in. So what we needed to now understand is how that genome turns to life. So you can think of the epigenome as the living genome, as the genome. There's the genome itself is static. It's just the book the tablets, if you wish that Moses brought down from the mountain, and then the epigenome is the
music that gets played from the orchestra. The epigenome tells you which parts are active in the brain and the liver, and the heart and the muscle and so and so forth.
So your work on the epigenome is really interesting to me. And I know you've done some work on obesity, and the epigenome tell me about that.
The strongest genetic association with obesity sits in one gene called FTO, and FTO was renamed fat and obesity associated after that discovery, and it remained mysterious for seven years. People had no idea how that gene works.
You just saw correlate.
There was a correlation.
There was a correlation.
Just the problem of genetics and the beauty of genetics. The beauty of genetics is that it tells you what region is responsible for disease. Regardless of how it functions. The downside is that it after he tells.
You it's the same thing.
After it tells you that he has a role, you have no idea how it functions. And what we showed in our work is that that region doesn't affect the FTO gene at all.
So like in the middle of a gene, there is this whatever series of nucleotides, but those those nucleotides are just randomly in the middle of that gene and actually have nothing to do with that gene. I didn't even know you could do that.
Fairly, you can't. So there are eighty nine differences, eighty nine common variants, common genetic variants that are all coinherited. If you get a here, you get all of the other you know, actage, you get that passage. If you get that package, it spans fifty thousand letters. But there are only eighty nine differences in these fifty thousand letters. Wow, and these will increase your body weight by one standard deviation, which is like how much it's like nine pounds, Like
it's a lot, okay. So so basically what that does is that it functions somehow to increase your risk for a basits, it's like the strongest genetic association before. And what we reason is, how could it be acting. It could be acting in your brain to decide whether you like sweets or salting. It could be acting your muscle to make you more fit or less fit. It could be asking in your digestives. So we basically said, okay,
well that's speculation. Let's look at the data. And we looked at the data and we found that there was this massive control region that was active in mesenchymal stem cells what are mesimo cells and sells. They are the progenitors of brown fat and white fat. Now, white fat is white because it's full of lipids. That's where all the calories are stored. Brown fat is brown because of all of the iron in the mitochondria. That's where the
calories are burned. So it turns out that our fat cells make a developmental decision in their first three days of differentiation to go down the white path lineage or the brown path lineage. And what the white fat does is it stores energies and brown burns energies. So it turns out that I'm actually homozygous risk for the store calories position, which is the obesity risk.
So you have the obesity.
I have two copies of the obesity risk. My wife has zero. I can tell you, you know, we look very different. Fair So we basically realize that it sits in the progenitor cells of white and brown flat and then we could show that the true target was not the ftogene at all. It was instead two other genes that are sitting one point two million letters away from this region and six hundred thousand letters away, and those genes turned out to be master controllers of thermogenesis. They
are basically switching your metabolic state. So my cells are stuck on the store position and my wife cells are stuck on the burn position.
And so what is the relationship between the genes that are acting here and this this you know, package variant that is far away from them.
It comes back to the epigena. So our DNA is stored inside a tiny little space. The way that gene regulation works is that you have these control regions that are scattered throughout the region of every gene that are linked together to that gene in three dimensions. So they do around and.
So it's it's far away. If you think of it as a strand but in three dimensional space, right there, three dimension pats right, Ah, that's satisfying.
And when we took these genes and we modulated them, we show that you can turn off one gene in mouse, in specifically the adipocytes of mouse with a dominant negative cus of fat cells with a dominant negative construct, and that turned the mouse fifty percent leaner. They eat the same amount, they exercise the same amount, but they burn calories when they're awake and they burn calories when they're sleeping.
And what's really fascinated with that story is that the variant associated with obesity is at two percent frequency in Africa, but forty two percent frequency in Europe and forty four percent frequency in Southeast Asia. So it rose from two percent to forty four percent maybe because of positive selection. Maybe it was beneficial to be able to store every kind of.
Places where food is, where you have food is scarce in moments of famine, exactly.
In the out of Africa event, this may have been selected for. Or in the you know, ice ages, it may have been selected for. And it's only after World War two that this variant became associated with obesity.
Because food became so abundant.
And we stopped exercising as much. So it's fascinating to see how the environmental shift led to a new genetic association which is now plaguing our society, and of course the hope that by understanding the circuit systematically, we can now solve so many different circuits and ultimately so many different pathways and ultimately so many different disorders.
In a minute, Manola's describes how he and his colleagues are trying to turn their genomic research into new medicines. That's the end of the ads.
Now we're going back to the show.
Another area where Manola's and his colleagues have done a lot of work is on Alzheimer's disease. They looked at a common genetic variant called apo E four. People with two copies of this variant have a much much higher risk of getting Alzheimer's, and Manola's and his colleagues were trying to figure out why. They found that having this Apoe four variant was linked to problems with moving cholesterol
around in the brain, a process called cholesterol transport. Then they did experiments and mice that found that drugs that restore cholesterol transport actually restored cognition in the mice. Now that's in mice, and Alzheimer's is a notoriously difficult disease to treat in humans. So I asked Minolas what it will take to move his research from mice to humans,
and his answer was really interesting. It pointed not only two ideas about treating Alzheimer's, but to bigger ideas about treating human disease more generally.
The way that I'm thinking about this, the way that our team is thinking about these, is how do we enable personalized medicine and precision medicine. Namely, Alzheimer's is not going to be only about transport. It's going to be a combination. Every person has some combination of these regulations. A point four is the strongest genetic risk, but there
are many others. And the question is how do we now systematically take a person with Alzheimer's, or take a family with risk, develop treatments that are either directly addressing the root causes rather than treating the symptoms, and are not only preventative but adapted to every family and every person.
And just to be clear, like having you know, two copies of the APO four lil is neither necessary nor sufficient to get Alzheimer's. Right, that's exactly both of them and not get it. You can have neither of them and get it. So it's exactly so complicated hard.
So, as with everything with human disease, genetics is not destiny. Genetics is a predisposition, and there are environmental factors. There are behavioral factors, there are nutritional exercise factors, there are socio economic factors. There's so many other factors that are affecting how your genetics will manifest ultimately into disease. But now the question is for every person, how do we create a drug? And it's not going to be feasible economically or in any other way to create one pill
for each person. The way that we're going to enable personalized medicine is by understanding what are the hallmarks of disease, what are the hallmarks of Alzheimer's, the wholemarks of obesity, the whole moods of diabetes, the hallmarks of cardiac disorders, and develop therapeutics for every one of those hallmarks. So think of it as an arsenal of twelve or twenty different drugs for Alzheimer's that you're going to be taking a combination of it.
Seems like oncology is already some way down that road, right, I mean, you know her two positive breast cancers have certain drugs that target them that sort of thing, right, is that the model?
That's exactly the model. So the hallmarks of cancer have been the way of thinking about cancer for twenty plus years now. And the difference in cancer is the following. Cancer is subject to positive selection. What does that mean? That means that because it's a replicative disorder where the cell, the cancer cells make more of themselves. If a cell acquires a mutation that allows it to replicate faster, you
will have more of that cell. So you are subject to positive selection where the bad mutations are increasing in frequency in every generation of the cancer. By contrast, most complex disorders are subject to purifying selection, where the mutations that are responsible for them are maintained at low frequency by evolution.
Huh.
So it's working at the opposite ends of the evolutionary spectrum. So cancer has a small number of genes that drive the disorder. Complex traits have thousands of genes that are maintained at low frequency or at weak effects.
Except that sounds much harder. It's harder to figure out what's going on harder.
But the saving grace is that even though you have extreme heterogeneity in the number of drivers, for every one of these disorders, they coalesce, they cluster, they converge in a small number of recurrent pathways, and these are the hallmarks.
Huh.
So you can find multiple genes associated with lipid transport, you can find multiple genes associated with new inflammation with DNA damage, so.
You target the sort of pathways where they converge.
That's exactly right. So we're not going to make a drug for Alzheimer's that we might make a drug for DNA damage, a drug for lipid metabolism, a drug for cholesterol transport, et cetera. And that's what we're working.
That's satisfying. That's a satisfying explanation.
It basically says that it is a limited number. There's a billion people in the planet. We're not going to have a billion drugs. What we're going to have it's a small number of drugs, one for each pathway, and these are sometimes going to be actually reused between different disorders. So we work on cardie disorders, we're finding the same genes underlying Alzheimer's, and specifically the lipid and cholesterol component are in fact reused in the heart disease. And again
it's about lipids. It's about saturation of the fat stores of an individual and now the lipid escaping into the blacks into the bloodstream, forming these plaques that will then cause heart you know, failure and heart damage and so and so forth. So that's where we're at.
So is there. I mean, the dream is that there is some dysfunction that is common to all these different diseases that you could target, right, Like, I mean, the naive dream is find the cure for everything, or not everything, but find the cure for a lot of things, or at least find a drug that will reduce risks of many different bad things, right, I mean, is that plausible or am I just naive in going there? From what you're saying.
So you're right that some of the time these pathways that we're finding are going to be helping in multiple frauds, And then that's absolutely the dream. We should basically start not with what is the worst disease, but maybe what is the best pathway that if we fix that one, we're going to have an impact on most diseases.
Right, like the highest return on investments for example.
Like, Yeah, that's a great way to think about it. But the way that I would say is that for each person, this might be a different molecule.
So now I'm not hopeful.
But that with a small number of these molecules, say one hundred, one hundred and fifty two hundred molecules.
When you say molecule, you mean drug.
I mean trust, might I mean drust. Yeah, Basically that there's going to be a small number of pathways and a small number of these modulators, and that those are going to be mixed and matched in each person to then target a communatorially large number of people.
Yeah, it just got hard. I know, I know biology is hard, but I got up to for a second.
There's not going to be a single silver bullet for all of those. In fact, for any one of these diseases, there's no silver bullet. But the moment you build your panelbly of fifty silver bullets, then you're going to be hitting two hundred diseases. That's the beauty of it.
Fifty bronze bo there's no silver bullet, but maybe.
You can find it for hearts exactly right.
We'll be back in a minute with the lightning round. Now, let's get back to the show. I read that you have been an author on more than two hundred and thirty papers, which is a lot. Which one was the most fun?
Oh? You know what, don't I tell you about my very first one?
Sure?
And the very first paper was published in c graph and it now has like two thousand citations, And it was about how do we reconstruct the surface of an object from a cloud of points? So you can basically use laser scanning to sort of figure out points in three D and then the question is what is the surface that goes between them. I've always been fascinated with three D space, so it was very fun for me to just like you know, as a kid, basically as as a freshman at to work on such a project
and then showing up at the conference. He was in Disneyland, so it was my first time in Disneyland as an author of a vapor.
Sounds relevant for motion capture, not knowing anything about it. When I think of, like, you know, people, the way they make movies now exactly as they put a bunch of censors on people and they move around and then you can turn them into a dragon or whatever you want.
Yeah, that's exactly right. So you know that paper has been quite influential and used for a lot of a lot of different things.
What's the most overrated Greek island?
Oh my god, I can tell you about the most underrated Santorini. Definitely not overrated tons of people, but worth every time. I can tell you about my first day in Santorini, which is I walked out on this balcony and I asked the owner of the restaurant if I can take a look at the view and I'm not order anything. He said, please be my guest, and I walked out, and ten minutes later, I'm like, I can't leave. I'm gonna have to order. He tells me, ten years ago, I came here to look at the view.
I want you to throw a little bit of shade. I want you to get in a little bit of drug.
Can't.
What's one place in Greece I should not cannot.
It's not possible. I mean, you know, if you keep insisting, I'll give you another twenty amazing places to visit.
Well, that's fair, that's fair. I did what I could do. If everything goes well, what problem will you be trying to solve in five years?
I think what I'm trying to solve now of actually creating these drugs in such a modular, AI driven, personalized, reusable way, centered on pathways. That's going to keep me busy for a long time. And I hope that in five years we have actually sold a big chunk of the platform and that we have a few drugs in clinical trials. So you know, my dream needs to take all of these circuits that we have uncovered and make a difference for humanity, make a difference for you know,
my fellow beings. That's my big goal.
Great, it's fun to talk to you.
I learned a lot, such a pleasure, thank you, and I love that you're fearless. You're like, well, we're gonna jump into this new topic and find it all about it.
Man nola's Kellis is a professor of computer science at MIT. Today's show was produced by Edith Russelo, edited by Karen Chakerji, and engineered by Sarah Bruguer. You can email us at problem at pushkin dot FM. I'm Jacob Goldstein. One last thing we are going to be taking a break for a couple of weeks, but we'll be back with new shows in early twenty twenty four. Thanks for listening, Happy New Year, that t