Andrew Huberman interviews geneticist Dr Oded Rechavi on inherited memories and transgenerational epigenetics
Andrew Huberman speaks with Dr Oded Rechavi about whether traits and experiences acquired during a lifetime can be passed to offspring through RNA.
Summary
Andrew Huberman hosts Dr Oded Rechavi, a geneticist whose laboratory studies transgenerational inheritance — specifically whether acquired traits, behaviors, or experiences can be passed from parent to offspring through molecular mechanisms beyond DNA. The conversation covers the foundational biology of DNA, RNA, and proteins before moving to the two classical barriers thought to prevent inheritance of acquired traits: the Weismann barrier (separation of somatic and germ cells) and epigenetic reprogramming. Dr. Rechavi then presents work from his laboratory using the nematode worm C. elegans demonstrating that small RNA molecules produced in the brain of a parent worm can alter the behavior of offspring for multiple generations — without any direct manipulation of the offspring's nervous system. He argues this constitutes clear, replicated proof of transgenerational inheritance of acquired traits in worms, and discusses what this may eventually mean for human health, diagnostics, and intervention.
Key Takeaways
FULL TRANSCRIPT
Introduction to DNA, RNA, and the Basic Machinery of Inheritance
Andrew Huberman: Today, what I mainly want to talk about is the incredible questions that you probe in your lab, which are incredibly significant for each and all of our lives. I think most people have a general understanding of what genes are, what RNA is, and so on. But maybe you could explain to people in very basic terms. I'll just preface all this by saying that I think most people understand that if they have two blue-eyed parents, there's a higher probability that their offspring will have blue eyes than brown eyes. But most people generally understand and accept that if they spend part of their life studying architecture, and if they have children, there's no real genetic reason to assume that their children would somehow be better at architecture because they contain the knowledge through the DNA of their parents. They might be exposed to it in the home — so-called nurture in that case — but they wouldn't inherit the knowledge. Today I'm hoping you can explain to us why eye color, but not knowledge, is thought to be inherited, and the huge landscape of interesting questions that this opens up, including some evidence that — contrary to what we might think — certain types of knowledge at the level of cells and systems can be inherited.
Dr Oded Rechavi: DNA is the material, the genetic instructions, that is contained in every one of our cells. We have a set of genes — the entire set is called the genome — and this is present in every cell of our body. The same set of instructions. Genes are made of DNA, and chromosomes contain the DNA. Chromosomes are the DNA and the proteins that condense the DNA, because we have a huge amount of DNA in every cell that you need to condense.
Huberman: Sort of like thread on a spool?
Rechavi: Right — huge amounts that you have to condense. We have the same genome, the same DNA, in every cell in our body. It's good to have an analogy to understand how it works. This is like the IKEA catalog that you have in every cell in your body — the instructions to make everything that you need in your house: the chairs, the kitchen, the pictures. But in every room, you want something else. In the kitchen, you want things that fit the kitchen, and in the bathroom you want things that fit the bathroom. So you only remove one particular page of instructions — the instruction for how to build a chair — and you place that in the living room. The genome is the instruction to make everything. This is the IKEA catalog. And in every cell we take just the instructions for making one particular piece of furniture. This is the RNA. And in the end you build a chair — the chair is the protein.
This is true for one particular type of RNA, which is messenger RNA. And in fact this is just a small percent of the RNA in the cell. We have a very big genome and less than 2% of it encodes for this messenger RNA. However, a lot of the genome is transcribed to make RNA that does other things. Some of these RNAs we understand, and many of them we don't.
Huberman: I think it's a beautiful description — and IKEA is not a sponsor of the podcast, so it's totally fair game to use the IKEA catalog as the analogy for DNA. The specific instructions for specific pieces of furniture is the RNA, and the furniture pieces are the proteins that are essentially made from RNA using messenger RNA.
Rechavi: Right.
The Weismann Barrier: Why Somatic Cells Can't Pass Information to Offspring
Huberman: Despite the fact that the same genes are contained in all the cells of the body, is it fair to say that there is one very important exception — somatic cells versus germ cells? Would you mind sharing with us what that distinction is?
Rechavi: Yes. Every cell type is different. We have cells in the legs, cells in the brain — and in the brain we have cells that produce dopamine, cells that produce serotonin, and so on. But we can make one very important distinction between the somatic cells and the germ cells. The germ cells are supposed to be the only cells that contribute to the next generation — the cells out of which the next generation will be made. Each of us is made from just a combination of a sperm and an egg. These are two types of germ cells. They fuse, you get one fertilized egg, and out of this one cell all the rest of the body will develop. What happens in the soma — which is all the cells that are not the germ cells — should stay in the soma and should not be able to contribute to the next generation.
This is very important, and is thought to be one of the main barriers for the inheritance of acquired traits, the inheritance of memory, and so on. Because, for example, with the example you gave of learning architecture — if I learn about architecture, the information is encoded in my brain. Since my brain cells can't transfer information to the sperm and the egg, because the information is supposed to reside in synaptic connections between different neurons in particular circuits that developed, what happens in the brain shouldn't be able to transfer to the next generation.
Even simpler: if you go to the gym and you build up muscles, you know that your kids will have to work out on their own. This is something that we know intuitively even if we don't have any background in biology. This is connected to the fact that, as we said at the beginning, every cell in the body has its own genome, and the next generation will only form from the combination of the genomes in the sperm and the egg. Even if you somehow acquire a mutation or a change in your DNA in one particular brain cell, it wouldn't matter, because there's no way to transfer that mutation to the DNA of the germ cells that will contribute to the next generation.
The man who first thought about this barrier is called August Weismann — this was in the 19th century — and it is called today the Weismann barrier: the separation of the soma from the germ line, with only the germ line transmitting information to the next generation. This is also called the second law of biology. It is very, very fundamental.
The other main barrier is called epigenetic reprogramming. The genetic material in our cells acquires all kinds of chemical changes, but these modifications are largely erased in the transition between generations. In the germ line — in the sperm and the egg — and also in the early embryo, most of the modifications are removed. So we can start on a blank slate based on the genetic instructions. This is crucial, because otherwise — and I'll note that in some organisms this doesn't really happen — we would not develop according to the species-typical genetic instructions. To preserve this, we erase all these modifications and start anew. In mammals and in humans, this is largely true. Most of the modifications in the sperm and in the egg are removed — about 90% of them.
Lamarck vs. Darwin: Two Theories of How Traits Are Transmitted
Huberman: There is this idea — and I'll say it so that you don't have to — that dates back to Lamarck: Lamarckian evolution, very controversial, and maybe not even just controversial but offensive to certain people. This idea of inheritance of acquired traits — the idea that one could change themselves through some activity. Using the example of going to the gym, we could also use the example of somebody who becomes an endurance runner, then decides to have children, and has in mind the idea that because of all the running they actually did — not just because they were biased towards running in the first place — their offspring would somehow be fabulous runners. This Lamarckian concept we believe is wrong. So how do we talk about inheritance of acquired traits? What's the proper language for us to frame this discussion?
Rechavi: Lamarck believed this is how evolution progressed. Later, Darwin showed that it's really natural selection — the selecting of organisms that already contain particular qualities, selected based on whether they survive or not in particular environments. Therefore their evolution progresses, they become more common, and take over. These are two very different explanations.
The most common way this is contrasted is the neck of the giraffe — the classic example. According to Lamarck, giraffes had to stretch their necks towards the trees to eat when the trees were high, and because of that they transmitted the trait of long necks to their children, who also had long necks. According to Darwin, a giraffe happened to be born with a long neck, survived because it ate, and so its heritable material — Darwin didn't know about genetics, but the concept holds — took over, while the rest of the giraffes with different heritable material died. So this is natural selection versus inheritance of acquired traits.
When we go back to studies about inheritance of acquired traits, there were also theoretical problems — barriers that have to be breached for this to happen — and you can narrow it down to those two main barriers. The Weismann barrier is the first. The second is epigenetic reprogramming. These are so important to how our bodies work that natural selection is considered the first law of biology and the Weismann barrier the second.
There is one additional thing to mention: there are also other mechanisms that might transmit information between generations, including transmission of RNA. There are different types of RNA — not just messenger RNA, which encodes the information for making proteins, but also other RNAs that regulate gene expression. In recent years, also in the mammalian field, RNA as the molecule with the potential to transmit information between generations has taken center stage. This is the cutting edge — a lot more to understand — but RNA has a lot of potential for doing that, as we'll explain soon. But we have to go to worms first.
Why C. elegans Is an Ideal Model Organism
Huberman: Many if not most of our listeners are focused on humans and human biology and health. But I cannot emphasize enough the importance of model organisms and the incredible degree to which they've informed us about human health, especially when it comes to very basic functions in cells. Before we go into the description of worms per se, could you explain to a general audience what a model organism is and why you've elected to work on a particular type of worm to study these fascinating topics?
Rechavi: Model organisms are organisms around which a huge community of researchers combines resources to create all the tools and understanding that accumulates. We have learned about every aspect of biology through them, including many important diseases. The reason that we can learn a lot about humans by studying these animals is that we all evolved from the same ancestor. We share a lot of our functions with them, and also a lot of our genes. They sometimes have things that are much more apparent in them that we can study. Another important reason for studying them is, of course, that you can actually experiment on them. We can't do this to humans. The things that we do to these animals — we can change their genes, do all kinds of things — would not be permissible in humans.
The community of people that study C. elegans has literally numbered and named each neuron, so that two laboratories on opposite sides of the world can publish papers on the same neuron knowing that it's the same neuron in both laboratories. Something that is extremely hard to do in any mammalian model — mouse or certainly in humans — and this poses huge challenges that give great advantages to studies of things like C. elegans.
C. elegans nematodes always have 959 cells, out of which 302 are neurons. We have had a connectome since the 1980s — like a subway map that tells us which neuron talks with which other neurons — and it is the same across individuals. Not only that, the worms are transparent, so we can actually see the neurons fire using particular tools, and we can activate genes and silence genes using optogenetics. On top of that we have great understanding of the genetics of the worm, of the genome. C. elegans is the first animal to have its genome sequenced — before humans. Each mother produces about 250 babies which are almost genetically identical. We know where we grow them; the environment is very controlled. We grow them on a plate with just bacteria, so we can easily separate between nature and nurture. The generation time in C. elegans is three days. So you can do hundreds of worm generations in one PhD. Every worm will produce hundreds of progeny that are genetically identical, so you will have great statistics for your experiments.
RNA Interference and the Nobel Prize-Winning Discovery
Rechavi: In the worm, we now have very obvious and clear-cut proof that there is inheritance of acquired traits — so much so that I don't think that anyone in the epigenetics field argues against it.
Huberman: What was the first experiment that you did on C. elegans that confirmed for you that inheritance of acquired traits is real?
Rechavi: We set out to test whether worms can produce transgenerational resistance to viruses. These worms don't have dedicated immune cells like we do — they don't have T cells or B cells. They defend themselves from viruses using RNA that destroys viruses. These are called small RNAs.
In 2006, two researchers who were studying C. elegans — Andrew Fire and Craig Mello — got the Nobel Prize for showing that there is a mechanism that regulates genes through small RNA. What they showed is that if you inject the worms with RNA molecules that are double-stranded, they shut off the genes that match in sequence to this RNA. It's sort of like taking the specific instructions for the coffee table from your IKEA handbook, inserting a copy of that into the book, and in doing so preventing the expression of — essentially erasing — the original page.
Huberman: Perfect explanation.
Rechavi: They found that double-stranded RNA — RNA that has two strands — is what starts the response leading to the production of small RNA molecules, which are the ones that actually find the messenger RNA and lead to its destruction. You silence it, so you don't get proteins in the end. For that they got the Nobel Prize after people found that this is conserved in many organisms including humans. There are now drugs that use this mechanism. It is called RNA interference — RNA interferes in the expression of a gene, in the function of a gene. It's also called gene silencing, because these RNAs enforce the silencing of genes instead of the genes being expressed.
They showed two very important things. You don't only see the action in the cell that you injected, or in the tissue that you injected — you see it all over the worm's body. It spreads. This includes also the germ cells. So if you inject the double-stranded RNA just to somatic cells, even to the head, you will get the effect also in the germ cells and in the next generation. Later they showed that you can just take worms and feed them on bacteria that produce this double-stranded RNA, and that the silencing would move from the gut — where the bacteria are eaten — to the rest of the body and also to the next generation. This is not controversial at all. This is done routinely every day by any C. elegans biologist in the world. It has been replicated a million times.
Demonstrating Transgenerational Viral Immunity in Worms
Rechavi: When I started my work, I wanted to see whether, in addition to artificial double-stranded RNA, some natural traits can also transmit across generations because of small RNAs.
Huberman: Right — because injecting RNAi, or short interfering RNAs, or putting worms into an environment with an abundance of inhibitory RNAs as an experiment, is very different from worms experiencing something and then passing on that acquired trait to their offspring. It's a world apart in my opinion, because one is an extreme manipulation that illustrates an underlying principle. The other is something that in theory occurs in the passage of generations just naturally.
Rechavi: We're going from the more artificial to the less artificial. The advantage of the more artificial approach — just like with model organisms — is that it's easier to know exactly what you did: you introduced one factor and you can follow the result. This is always the tradeoff. In fact, this is probably the reason that these small RNAs evolved in the first place — to get rid of viruses and other parasitic genomic elements. This is a mechanism to fight them.
We demonstrated this very clearly using a fluorescent virus. If the virus replicates successfully, the worm turns green. If the virus is destroyed, the worm stays dark. This is very simple and clear-cut. We took worms, infected them with a fluorescent virus, and they destroyed it — this had been done in the past. But then what we did is we neutralized the machinery that makes small RNAs in the descendants of the worms. So the descendants cannot make small RNAs from the start on their own, because they just don't have the genes needed to make them. Then we asked: what will happen when we infect these worms with the virus? Will they be green or dark? They can't make their own small RNAs, so they can't protect themselves on their own. The only way for them to stay dark — for the virus not to replicate — is if they inherit the small RNA from their parents. And this is exactly what happens. All the worm progeny, although they don't have the gene needed for making the small RNAs, stay dark. They silence the virus, and this continues for additional generations.
Huberman: So the parent worms effectively put something into the genetic instructions of the offspring that would afford them an advantage if they were to be confronted with the same thing that the parents were.
Rechavi: Right. And we know exactly what this advantage is. The advantages are small RNAs that match the viral genome — they just chop up the virus in the next generation. We can identify these small RNAs in the descendants, although they don't have the machinery to make them, simply because they inherited them. We can identify them by RNA sequencing — which is like DNA sequencing, where you get the actual sequence of the RNA molecules — and we can see that they correspond to the virus. The descendants have inherited small RNAs only if their parents were infected with the virus.
Can the Brain Transmit Heritable Information to Offspring?
Rechavi: It is true that also in mammals, RNAs and small RNAs are a leading candidate for something that could mediate the transmission of stress protection, or also of harmful effects, between generations. Perhaps RNA does it, and it's very interesting to think about when we talk about inheritance of memories. Can brain activity of some sort transmit — at least in worms? In mammals, I've said this disclaimer multiple times: we don't know. Time will tell. In worms, we know a lot.
Can worms transmit brain activity? Do they have the specificity to do so? I think that any tissue that can transfer RNA to the next generation and affect the next generation is interesting — the gut, muscles, everything. But the brain can synthesize information about the environment and about internal states, and can also think ahead. The most provocative thing you can say is that you could somehow plan the fate of your offspring using your brain, after taking many things into account.
Huberman: Without talking to them.
Rechavi: Right — without talking. We have to understand that the brain uses a different language than the language of inheritance. It keeps information in synapses — in the connections between different neurons. When you learn something, you make some connections stronger and other connections weaker, and you wire the nervous system in a different way. On the other hand, heritable information of any sort has to go through a bottleneck of one cell — the fertilized egg — because we all start from just one cell. So the question is: can you translate the information encoded in this three-dimensional structure of synapses and the architecture of the brain into heritable information, into a molecular form?
You can teach worms things about the world, even though they have just 302 neurons. For example, you can take an odor that the worms like — the worms have thousands of odorant receptors and can recognize many many molecules — and pair it to something bad like starvation. Then the worms will learn to dislike this odor. We don't know that this learning necessarily involves changes in the strength of synapses. It's a possibility, but it doesn't have to be the case. It could be that the receptor for this particular odor is simply removed, and this is how they learn. Now they won't have the receptor, they won't smell it, they won't like the odor.
This type of thing you could perhaps transmit to the next generation — not that anyone has shown it convincingly — because all it would take is an RNA that controls this particular receptor. People have shown things like this in mammals: that you learn a certain thing, and then in the next generation a particular receptor would be methylated or would change, and this would transmit the response. On the one hand it could be true. On the other hand, you need to prove — and this hasn't been done convincingly enough yet — how exactly the information transfers from the brain to the germ cells, and then in the next generation from the germ cells back to the brain, to where the receptor needs to operate. This is the challenge. This is the current state of the field.
The 2019 Cell Paper: Brain-to-Germline RNA Signaling Across Generations
Rechavi: What we did in C. elegans is we showed that the brain can communicate with the next generations using small RNAs, and that this can change behavior — and it doesn't require any translating between languages. It is very simple. What we showed is that if you take a worm and you change the production of small RNAs just in its brain, in the next generations their behavior will be different, even though you don't touch their brains. This is a paper that we published in 2019 in Cell.
We showed that you just manipulate the production of endogenous, natural RNAs in the worm's brain — RNAs that are always made, but you change their amount — and this changes the capacity of the worms in the next generation to find food, not only in one generation but three generations down the road. The way it works is that perturbing the production of these small RNAs in the brain affects, in the end, the expression of a gene in the germ line. One gene is called sago-2. We can do all kinds of controls where we manipulate the activity of the gene and see that this also affects behavior. And this gene works in the germ cells. The information needs to go from the brain to the germ cells — it doesn't need to go back from the germ cells to the brain to affect behavior.
We know that this is a true epigenetic effect because it goes on for multiple generations, and also because it requires the machinery that transfers RNA between generations. If you don't have the protein that physically carries the RNA between generations, it doesn't happen.
Huberman: So it has to be RNA.
Rechavi: It has to be RNA. We can also find the RNAs in the next generation that change — we sequence the actual RNAs that change in the next generation.
It sounds strange that you change germ cells — sperm and egg — and it changes behavior. But if you think about it, the germ cells affect the soma, including the brain, in many ways: by secreting certain chemicals, and also because the other cells develop from the germ cells. So some information could be transmitted over the course of development, or the course of development could be altered because of changes that occur in the germ cells.
For example, in mammals, one of the explanations for how heritable information transmits is that it just affects something very early in development — when you just have a few cells, or even in the placenta that develops during pregnancy — and this later throws everything off. Because of that you have many problems with metabolism and so on. This is the idea of the developmental origins of health and disease: that many functions are determined early on in development.
Future Applications: Diagnostics, IVF, and Intervention
Huberman: In terms of the work in either C. elegans or in other model organisms — but in particular in C. elegans — where do you see this going next?
Rechavi: Assuming that we will discover similar things in humans — which we don't know to be the case, but let's say we find it — I think there are many things you can do. You could change parental inheritance by having the parent exercise, for example. Some things like this have been done. There are experiments in rodents where they show that overfeeding the rodent creates problems for the next generations, for the children. However, if you let the rodent exercise, it corrects the parental inheritance. So this is one possibility.
You can also manipulate it at the source. If it's RNAs, in the future — if we understand how it works — you could actually change the composition of the heritable RNAs. If you do IVF, you could perhaps change the composition of the RNAs in what you introduce. But way before that, what you could do — perhaps even in the not-so-far future — is use this for diagnostics. DNA-based diagnostics for every couple that wants to have a child are already done in Israel for most couples. You can look at the DNA and look for genetic disease. But no one is looking at the RNA at the moment. If we understand how it works better, we'll have another level — a whole new world to look at. And perhaps there will be some RNAs that correlate with disease.
The beauty is that, unlike DNA, this is plastic. With DNA, this is your DNA — perhaps we can choose another embryo. But here you could say — again, in the future, this is science fiction, it doesn't happen now — if we understand this and it's true, we can say: maybe you should run on the treadmill a little bit. This will change the profile of your RNAs, and then we will use it for IVF because it correlates with healthy profiles of RNAs. This is a level that no one looks at now, and it holds great potential — again with the disclaimer that we don't know how it works in humans at all.
Huberman: Yet. But of course this is why it's so interesting today. You've taken us on an amazing journey through the genome, RNA in particular, the work in your laboratory — which is just incredible — and also this introduction of model organisms. So thank you so much.
Rechavi: Thank you.
Huberman: It's been a real pleasure.
Rechavi: The pleasure was all mine. Thanks a lot.