Don't Call it a Brain in a Dish!

with @PascaStanford, @VijayPande, and @omnivorousread Our understanding of the human brain and its disorders has always been limited by our lack of access to living, human, developing brain tissue. For the first time, that's changing. In this episode, Sergiu Pasca, Professor of Behavioral Science at Stanford, talks with a16'z General Partner Vijay Pande and Hanne Tidnam about the wild new tech that's pioneering a whole new approach to understanding the brain: brain organoids. This conversation covers what these organoids can and can’t do; what they’re good for understanding and where that understanding becomes limited; why calling these “brains in a dish” or “mini-brains” isn’t the right terminology at all; and finally, how far can this new tool and model be taken now and in the future, leading us closer towards understanding psychology itself on a molecular level.

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Hanne Tidnam?

  Hi and welcome to the A16Z Podcast. I'm Hannah and in this episode, general partner Vijay Pandey and I talk with Sergio Pashka, professor of behavioral science at Stanford, all about a new technology we have for understanding brain disorders, the wild and very sci-fi new frontier of brain organoids. So what are brain organoids? How were they developed and how can we use them? The conversation starts with the essential problem that we've never had real access to the tissue and actions of the developing brain, or even a living, normal brain, and the problems with all of our existing models for understanding it, from genetic studies to autopsies to primates. We look at those models we've relied on in the past and what this new model of brain

• 0:44Hanne Tidnam?

  organoids now brings us, allowing us to study the human brain, both how it develops and what goes wrong in certain disorders with living, human brain cells in a dish for the very first time. We talk about what these organoids can do and can't, what they're good for understanding and where that understanding becomes limited, and why calling these mini-brains or brains in a dish isn't the right terminology at all. And finally, how far this new tool and model might be taken now and in the future, and how it will lead us closer towards one day, even perhaps, understanding psychology itself on a molecular level.

• 1:23Hanne Tidnam?

  We're here today to talk about understanding brain disorders and some of the new tools we're developing for how to do so. So let's start where we actually are in that. Are we actually anywhere significantly more advanced than we were in the days of hysteria? You know, thinking about things like labeling these sort of societal conditions that we had no clue, like where are we actually right now?

• 1:45Speaker 01

  Psychiatric disorders are still behaviorally defined and there are very few biomarkers that are considered reliable for diagnosis. The truth is that our understanding of psychiatric disorders is actually quite limited. I often like to joke that I suffer from an oncology envy syndrome,

• 2:02Hanne Tidnam?

  Oh, right.

• 2:03Speaker 01

  which is essentially this deep frustration that you feel as you see just like how fast cancer research Has uh has gone in the last few decades. From really like no treatments whatsoever to almost completely curing certain forms of cancers. And if you look carefully at it, you realize that one of the reasons for this incredible progress is that oncology has really made use of the revolution molecular biology. And it has done so because it actually has access to tissue, to the tissue of interest. We know almost nothing about how the human brain develops because it's it's completely inaccessible.

• 2:43Speaker 01

  And so, again, we are defining psychiatric disorders based on combinations of behaviors, presence or absence, or certain patterns of behavior. We've made, I guess, a lot of progress into classifying these disorders and reclassifying them. But the truth is that our molecular understanding of psychiatric disorders or brain disorders more broadly is very limited. And probably behind any other branch of medicine, which I think is reflected in the therapeutics that we have.

• 3:15Vijay Pande?

  Well, and the complexity, I mean, it's fun to think about, like, you know, in the eighties molecular biology was this hot new term. I mean, you're talking about something almost like molecular psychology. Taking this big sort of emergent phenotype that is, you know, a behavioral and then trying to connect it. Not just at the cell tissue level, not just at the cellular level, but all the way to molecular level. That is a hard thing to do. It's hard to imagine like if someone has schizophrenia or severe depression, what's the target to hit?

• 3:41Speaker 01

  Right.

• 3:42Hanne Tidnam?

  You said something really interesting about just never being held back by not really having the tissue.

• 3:46Speaker 01

  Right.

• 3:46Hanne Tidnam?

  And you mean by that, you know, that we, the first time we get to look at the tissue is after somebody who has suffered from a psychiatric disorder has died, right? That is our primary tool at the moment.

• 3:56Speaker 01

  Yes. And there are a number of challenges associated with studying postmortem tissue from patients. Of course, the obvious one is the fact that the tissue is not alive.

• 4:06Hanne Tidnam?

  Yes.

• 4:06Speaker 01

  For me, as a neuroscientist, it is really important to be able to record electrical signals from cells. To really look at how they're communicating with each other. But at the same time, another limitation is actually the availability of tissue. I mean, if you were To just think, for instance, about autism spectrum disorders.

• 4:23Hanne Tidnam?

  Yeah.

• 4:23Speaker 01

  Which is very common, one in sixty or so individuals. And there is even an autism brain bank.

• 4:30Hanne Tidnam?

  Mm hmm.

• 4:30Speaker 01

  But the number of brains that we have in a brain bank is really in the hundreds, not in the thousands for a disorder that is like so common.

• 4:37Vijay Pande?

  And it's probably for adults too, right?

• 4:39Speaker 01

  And it's, right. Another limitation is actually the age of these individuals, but very often also the cause of death. Because in most of these cases, it's actually traumatic death. And most of psychiatric patients will take many, many medications and undergo various therapeutic interventions across their lifespan. We don't know, for instance, what is, how is that influencing what we're seeing in postmortem tissue?

• 5:03Hanne Tidnam?

  So you're getting a very small amount of information that may not even be accurate.

• 5:08Vijay Pande?

  Or very anecdotal.

• 5:09Hanne Tidnam?

  Or very anecdotal, yes. And that's the only real tool that you have at the moment besides behavior?

• 5:15Speaker 01

  Well, I think, of course, there are imaging studies that you could use. For instance, MRI and functional MRIs. The problem with those studies is that you don't really get the molecular resolution. You don't get to really study the tissue. An alternative, which has been, you know, used in the last decade or so, has been to model many of the disorders with animals. And that has been quite an exciting field. That was primarily accelerated by identifying genes associated with psychiatric disorders. But I, I think we always have to be aware of the differences between, between species, right? Even in how the brain, the structure of the brain, the fact that there are millions of years that separates us in evolution. The behavioral repertoire is very different across species.

• 5:59Speaker 01

  Now, of course, there, the behavioral repertoire is much closer to that of humans. But as you can imagine, again, the limitation there is how scalable is that? How many primates can we really use for this type of studies? And who can afford to do this experiment on a large scale? The truth is that most of the psychetics have a very complex genetics. It is very rare, like one single gene.

• 6:24Vijay Pande?

  Yes.

• 6:24Speaker 01

  Or one single variant. But very often, a combination of this, and it's not just a. Obviously about the genes, but what are the cells and the circuits that are affected by this? And I think that only once we start to understand some of the molecular machinery behind the psychiatric disorders. Can we, as it happened in, I guess, in the cancer field, start thinking about therapies? That have been designed for specific disorders rather than identified by chance. Because many of the drugs that we have for psychiatric disorders today have actually been identified by chance.

• 6:56Vijay Pande?

  Yeah, most, right? And not only identified by chance, but the method of action even after the fact is poorly understood.

• 7:02Speaker 01

  Absolutely.

• 7:03Hanne Tidnam?

  Yeah, we're using them and even we don't quite understand why. So what are some of the major shifts that we're beginning to see? Where we are beginning to engineer a better model? Or a better mapping of what's actually going on behind the scenes?

• 7:18Speaker 01

  There are a number of important discoveries that have happened over the last few decades. That I think brought us in this unique position right now to start asking questions about human brain disorders with human patient cells. And the first one is probably even the fact that we were able to maintain some stem cells in a dish. Which was done in the eighties. For a long time, we were unable to really maintain pluripotent stem cells in a dish.

• 7:46Hanne Tidnam?

  So living cells.

• 7:47Speaker 01

  Living cells, embryonic stem cells maintained in a dish. Another paradigm shift has happened, you know, 12 to 13 years ago. When it was shown that development is really not a one way street. Cells don't just differentiate and never can go back. But there are ways of actually pushing them back in time to resemble those pluripotent stem cells from which every cell that we have is made. Essentially, you can take skin cells from any individual, put them in a dish. And then just over express a series of genes that are important for a cell to be pluripotent. And that is sufficient to actually push that cell in time to look like a pluripotent stem cell. So this is.

• 8:31Hanne Tidnam?

  It's like time travel for cells.

• 8:33Speaker 01

  But I think what that.

• 8:34Vijay Pande?

  Or fountain of youth.

• 8:36Speaker 01

  But I think what that allows us to do is essentially be able to obtain in a non invasive way. Right? Because this is not invasive by getting a skin biopsy or a blood sample. Obtain pluripotent stem cells from an individual. And by pluripotent, as the name already implies, that means that they can become multiple cell types. Including brain tissue.

• 8:57Hanne Tidnam?

  And so that is where it becomes about studying the brain.

• 9:00Speaker 01

  Right. So I think that the third series of breakthroughs has to do with us slowly learning how we can instruct this pluripotent stem cells to become brain-like tissue in a dish.

• 9:15Vijay Pande?

  So it comes from cells from an adult.

• 9:17Speaker 01

  Right.

• 9:18Vijay Pande?

  That then become under differentiated.

• 9:20Speaker 01

  Right.

• 9:21Vijay Pande?

  And then the sort of the fun begins. It's kind of a mind-blowing, so to speak, to take like a skin cell or a blood cell, turn it into a brain cell, neuron cell with others.

• 9:31Speaker 01

  Right.

• 9:32Vijay Pande?

  And then developmental starts again.

• 9:34Speaker 01

  Starts again, absolutely. The human cerebrocortex has a diversity of cell types, which are arranged in layers. But those layers are not born all at the same time. They're born in a specific sequence. And initially you get layers of deep layers, like layer five and six, and slowly you get upper layers. Now, if you start to recapitulate that process in a dish, then you start seeing that you don't have to provide all the cues all the time. You just have to provide the initial cues. And once the process starts, the progenitor cells start making first deep layer neurons.

• 10:08Hanne Tidnam?

  Wow.

• 10:08Speaker 01

  And then slowly start to make upper layer neurons.

• 10:10Hanne Tidnam?

  But doesn't it take years, as long as it does in the human brain?

• 10:14Speaker 01

  Exactly. And it's, well, the human, if you think about the human cerebrocortex, all the neurons in our cerebrocortex are made by 27 weeks of gestation.

• 10:23Hanne Tidnam?

  Okay.

• 10:23Speaker 01

  So that's second part of gestation. But that's still a very long time.

• 10:27Hanne Tidnam?

  That's long, yeah.

• 10:28Speaker 01

  And essentially.

• 10:29Vijay Pande?

  Long for research.

• 10:30Speaker 01

  Precisely. And this is exactly what motivated us to even develop new methods. Because initial methods were keeping cells at the, on a flat surface, on a dish, on a plastic, flat surface. And so what happens when you start to keep cultures, these cultures, the cells, for like, you know, 10 weeks, 12 weeks.

• 10:49Hanne Tidnam?

  It doesn't seem like they'd be very happy.

• 10:51Speaker 01

  They're not very happy. And essentially what they do is they start peeling off. And my frustration initially was that we couldn't go farther in develop.

• 10:58Hanne Tidnam?

  Yeah.

• 10:59Speaker 01

  You could see that they were progressing, but they were just like peeling off.

• 11:03Hanne Tidnam?

  They have nowhere to go. Yeah.

• 11:04Vijay Pande?

  So what was the magic? What was the secret?

• 11:05Speaker 01

  Essentially, I mean, if I don't know if we can call this innovation, but around, I think, the summer of 2011 or so, I thought that maybe it would be easier to just move the cells to a plate that doesn't allow the cells to attach at all. It's essentially a plastic dish that is coated with a substance that doesn't allow the cells to sit down. So because they cannot sit down, then essentially they form a ball of cells, a sphere.

• 11:32Hanne Tidnam?

  To each other.

• 11:33Speaker 01

  Exactly.

• 11:34Hanne Tidnam?

  Oh, my gosh.

• 11:35Speaker 01

  And it's interesting enough that then they don't, they don't really.

• 11:38Hanne Tidnam?

  They're not bothered.

• 11:39Speaker 01

  Yeah, they're not bothered. So you can keep them for very long periods of time. And in fact, we've kept some of the longest, if not the longest cultures ever maintained, which went on for 800 days and beyond. Initially, we just thought, oh, let's just like do that so we can keep them a little bit longer. But then we realized that that enabled the cells to self-organize because they're in a three-dimensional culture. They start to interact very differently, one with each other, and they just can just develop and progress in development for much longer.

• 12:06Hanne Tidnam?

  So you see more. They're doing more.

• 12:08Speaker 01

  You can recapitulate. Exactly. You allow them actually to, you know, essentially develop as they would.

• 12:15Vijay Pande?

  Kind of getting out of their way.

• 12:16Speaker 01

  Right. Exactly. Exactly.

• 12:18Vijay Pande?

  You know, what's interesting about this is that we're, in many ways, the science here is just replicating the natural aspect of developmental biology. That developmental biology starts with, you know, originally these stem cells, and these stem cells have to then differentiate into becoming all different types of tissue. The fact that actually we can go backwards is interesting. And also the fact that actually, in some ways, the experiments will recapitulate. That developmental biology is very intriguing.

• 12:45Speaker 01

  Yes. I mean, if we really work to think like how the human brain comes together, right, there are a group of stem cells, which are cells that are capable of turning into other cell types under appropriate conditions. And those cells start organizing and forming the brain over a very long period of time. The human brain literally takes hundreds and hundreds of days and then years afterwards to mature and come together. And that, I think, is one of the most exciting possibilities for the cellular reprogramming technologies is that they allow us to essentially recapitulate in a dish some of the cellular processes in a way that allows us to study them and, of course, to do this in the context

• 13:30Speaker 01

  of patients.

• 13:31Hanne Tidnam?

  It really does make me think about when you say, well, the original black box is the human brain, right? It is truly the, you know, that we see nothing inside it, except for studying the behaviors that are as a result of that.

• 13:42Vijay Pande?

  And actually, you know, the amateur developmental biologist in me likes to point out that all this organization that goes on kind of recapitulates evolution to some degree, too. And so, you know, we have, the brain has all this complexity of lower brain and then other parts and eventually cerebral cortex gets added through evolution. That's seems to be what you're talking about in terms of it being developed and recapitulate into mental biology. So now the issue is like, since if we care about these disorders and probably most of these disorders are not going to be in the lower sort of animal brains.

• 14:13Speaker 01

  Absolutely.

• 14:14Vijay Pande?

  You know, we'd have to be able to build models to have them.

• 14:16Speaker 01

  Right.

• 14:16Hanne Tidnam?

  So now, this is what some media have started to refer to as brain in a dish or organoids, or we've heard mini-brains as well. Those are all attempts to try to describe this model of how we can begin to understand more from live human brain cells. Can we talk about, are all of those labels, do they all get at something different?

• 14:37Speaker 01

  There's still a little bit of confusion in the field about the terminology, but some of these terms are definitely inaccurate and I think they do not reflect the science. Probably even the term organoid is not really the most appropriate.

• 14:49Vijay Pande?

  And that's a more generic term because you can have liver organoids or cardiac.

• 14:54Speaker 01

  Absolutely.

• 14:54Vijay Pande?

  Yeah.

• 14:55Speaker 01

  An organoid broadly speaking is a three dimensional cell culture that self organizes and recapitulate some aspects of organ function, but not all.

• 15:07Vijay Pande?

  Yeah.

• 15:08Speaker 01

  But actually, the majority of the organoid field is about cells that are derived directly from patients. So for instance, you take a biopsy, a gut biopsy or a liver biopsy.

• 15:17Vijay Pande?

  Yeah.

• 15:18Speaker 01

  And you just want to maintain those cells. So you keep them in a three dimensional culture. That's also called an organoid.

• 15:23Vijay Pande?

  Yes.

• 15:23Speaker 01

  I mean, organoid, of course, as the terms implies is organ like.

• 15:27Vijay Pande?

  Because they don't have organ functionality.

• 15:29Speaker 01

  No, they don't. And, and most of them don't also recapitulate many of the structures. And I, I, I think there is a little bit of confusion, but it.

• 15:35Vijay Pande?

  But in the organoid field, like there are, I think are lung and liver organoids that have some recapitulation of function.

• 15:40Speaker 01

  Right.

• 15:40Vijay Pande?

  So in that sense, maybe that's appropriate. But you're saying on the brain side, I mean, what aspect of function?

• 15:45Speaker 01

  All of the organoids recapitulate some features of the organ. Of course, none of them recapitulates all the features of the organ.

• 15:52Vijay Pande?

  Yes.

• 15:52Speaker 01

  Like neurons in brain organoids, fire action potentials.

• 15:56Vijay Pande?

  Hmm.

• 15:56Speaker 01

  They make synapses with each other, so they do communicate. But of course, it is not like the brain. Most of these brain organoids resemble very specific brain region.

• 16:07Vijay Pande?

  Yeah.

• 16:07Speaker 01

  The, the, the brain is not a homogeneous organ, not like the liver. It, it does matter which part you actually probe.

• 16:13Vijay Pande?

  I didn't realize that you could make a little visual cortex or a little hippocampus or a little.

• 16:17Speaker 01

  Yes, I, I will actually, paradoxically, the easiest structure to make is the cerebral cortex. It turns out is kind of like the default is the one that has the least a number of instructive signals.

• 16:28Vijay Pande?

  Yes, yes.

• 16:29Speaker 01

  That you provide. But then you can, you know, and we've shown this, you can derive multiple brain regions. You can make a midbrain, you can make a spinal cord, uh, you can make a striatum, you can make a hypothalamus. And guide their differentiation with essentially small molecules.

• 16:42Hanne Tidnam?

  These brain organoids where you're using them to try and understand more about these brain disorders. How are we getting more information out of them? What is that beginning to teach us?

• 16:52Speaker 01

  I think one of the first things is that they recapitulate many aspects of human brain function. Again, by maintaining them for very long periods of time, we show, for instance, that cell types are generated in a sequence, in the same sequence as they would in the brain. So for instance, you don't just get neurons in the brain, you also get glial cells, which for a very long time were thought to be like unimportant cells, they just like glue the brain together. But they come late in development, especially in humans. For instance, astrocytes are essential for neurons to even form synapses with each other. But we know very little about them. Now, if you keep this organized for long enough, and it does take 20 to 30 weeks, but once you do that, you start seeing, for instance, astrocytes also appear at the right time, and they actually mature.

• 17:35Speaker 01

  In fact, one of the most surprising things that we discovered by maintaining these cultures for a very long time and comparing them to primary tissue, to tissue obtained from actual patients, is that the cells do, they know the maturation time point. And for instance, they know when birth should happen.

• 17:52Hanne Tidnam?

  Oh my god, that's crazy. Is there a moment, I'm imagining all of you in the lab, at this point, you're seeing these developments and hoping that you might start to see these different types of cells form?

• 18:05Speaker 01

  Yeah.

• 18:06Hanne Tidnam?

  And then suddenly, one day, they did? These, is that like, was it like a birth?

• 18:09Speaker 01

  No, I think it was like a slow process, like for us even, and for me personally, it was like difficult to accept. As a physician, I always kind of like thought as like birth as being like this dramatic event to the brain that triggers a series of processes. But, you know, birth in humans is actually not even happening when it should happen. It should happen three months later. It should happen much, much later. Exactly. But there are limitations because of head size. And so it's not really about the birth itself, it's just about the maturation of the brain.

• 18:35Hanne Tidnam?

  But you must have been amazed to see the maturation reach that level.

• 18:38Speaker 01

  If you think about glial cells, astrocytes in particular are very different before birth and after birth. And astrocytes, as they approach nine to 10 months of keeping them in a dish, they slowly, not suddenly, slowly transition to a postnatal signature, as if there is some sort of program that once they start it, it just progresses. And we know that not just at a gene expression level, but at the chromatin level, epigenetic level, and even functional level, many of these cellular processes are, and again, Is it surprising? Sure, it's surprising for us, but if you think from a point of view of like developing a brain, the mechanisms for making a brain must be must be very robust. Because you have to make the same brain over and over again.

• 19:20Speaker 01

  And of course, there are a lot of differences between our brains, but to a large extent, they're quite similar, right, structurally and functionally.

• 19:26Vijay Pande?

  Well, yeah, and it's a funny thing about biology, because at some level, it's amazing anything works. And then another level, like, it doesn't take much, like, if we're talking about a very different context. We're talking about plants or something like that. We're just planting seeds and letting them go.

• 19:38Speaker 01

  Yes.

• 19:38Vijay Pande?

  I mean, that's amazing in its own right.

• 19:40Speaker 01

  Right.

• 19:40Vijay Pande?

  And in a sense, these IPS cells or these stem cells are seeds of sorts.

• 19:44Speaker 01

  Absolutely.

• 19:45Vijay Pande?

  And and then we put them together, they interact with each other, the cell-to-cell communication creates this developmental profile.

• 19:52Speaker 01

  Yes.

• 19:53Vijay Pande?

  There's probably a cellular clock, and it's just it's just it's just a machine going tic, tic, tic, tic, tic, rolling along, as long as you just don't get in its way and create an environment that's kind of close enough to what it needs to be.

• 20:07Hanne Tidnam?

  But we're talking about development of, you know, sort of normal development in a way. How does this begin to connect to abnormal development and our understanding of when things are going wrong?

• 20:17Speaker 01

  Well, I think the fact that we can even model in a dish in a non-invasive way, right? Because it doesn't involve like taking anybody's tissue, brain tissue, right? I think the greatest opportunity is in accessing these stages of brain development that were previously inaccessible, that are likely related to disease. Just to give you an example, a very specific example about modeling disease and asking questions about disease. The human cortex doesn't have just cells that excite other cells. It also has about 20% of the cells that inhibit other cells, other neurons. So there is a very clear balance.

• 20:58Hanne Tidnam?

  A dance.

• 20:58Speaker 01

  Yes, precisely, between excitation and inhibition in the cortex. And that is a very important, because if you think, for instance, about epilepsy, right? In epilepsy, there's more excitation, or there's not enough inhibition, right? This balance just goes awry. It's also thought that, to a large extent, this balance also goes awry in autism. So very often, autism is thought to be a disease of excitation-to-inhibition imbalance. But here's an interesting fact about the developing of the human cortex. All of this inhibitory cortical neurons, all of these neurons that put a brake, are actually not born in the cortex in the brain. But they're born literally inches away from the cortex in another part of the brain.

• 21:41Speaker 01

  And sometimes around mid gestation, start all of them to migrate one by one. And they go up to reach the cortex and populate that region.

• 21:51Vijay Pande?

  So which which region do they originate from?

• 21:53Speaker 01

  So they they come from the ventral forebrain.

• 21:55Vijay Pande?

  Okay.

• 21:56Speaker 01

  In a region called MGE. And they have to literally, again, one by one migrate and reach the cortex. They don't just crawl on a surface. What they do is they have a very peculiar way of movement, the cells. And they have a very long process that they point towards the direction which they want to move. And at one point, their cell body almost breaks in two.

• 22:19Hanne Tidnam?

  No.

• 22:20Speaker 01

  The nucleus stays behind. And then almost like a muscle contraction, the nucleus is pulled up about thirty microns.

• 22:26Vijay Pande?

  And so that's how they move?

• 22:28Speaker 01

  That's how they move. They move in a very peculiar way.

• 22:30Vijay Pande?

  Yeah, wow.

• 22:31Speaker 01

  And they do this every, you know, three hours or so.

• 22:34Hanne Tidnam?

  It's like an inchworm, almost.

• 22:35Speaker 01

  Kind of, yeah, yeah, exactly. And so we started watching to see what, and again, it was unclear also how it really happens in humans. Because this is one of those stages which is completely inaccessible.

• 22:47Vijay Pande?

  Yes, yes. Do you think that recapitulates evolution in in some sense? Or is it, I mean, because it's odd, but that, I mean, we had to tack on this brain on top of another brain.

• 22:56Speaker 01

  Yes. Well, I I think it has to do with the fact that these cells, because they're so different, require different cues when they're like developed. It's interesting, like, one of the most surprising things for for students is just to learn that most of the cells in the brain are not born in the place in which they reside.

• 23:14Vijay Pande?

  Yeah, that's really interesting.

• 23:15Speaker 01

  You just assume that they're like there.

• 23:16Hanne Tidnam?

  The movement.

• 23:17Speaker 01

  Precisely. If anything, there a rule of development is that cells have to move from where they were born.

• 23:23Vijay Pande?

  And inches, inches.

• 23:24Speaker 01

  Inches.

• 23:24Vijay Pande?

  Yeah, yeah, yeah.

• 23:25Speaker 01

  And and this happens, actually, there is evidence in in rodents. This migration stops before birth.

• 23:31Hanne Tidnam?

  I was just gonna say, does it continue as the brain grows?

• 23:34Speaker 01

  Yes, in humans, interestingly, there is like recent evidence that shows that in humans, in primates, but in humans particularly, this continues up to the second year of life in humans. So towards the prefrontal cortex, there is a population of interons that continues to migrate.

• 23:48Hanne Tidnam?

  In motion.

• 23:49Speaker 01

  Right, so we know very little about this. So a number of years ago, we wanted to really model this complicated process in a dish. And we did so with a new approach, which we often call the second generation organoids or assembloids, which essentially involve deriving these two brain regions separately in a dish from pluripotent stem cells. So for instance, making the cortex that has all the excitatory cells, and then making this ventral forebrain regions that has all the interons, developing separately, we put different cocktails of small molecules. And at one point, at the right point, we essentially put them together in a tube, essentially at the bottom of a tube. We leave them to sit close to each other overnight. And next day, they essentially are fused to each other.

• 24:34Hanne Tidnam?

  I mean, were there lightning bolts? Mary Shelley must be like rolling over in her grave.

• 24:39Speaker 01

  I think what was really surprising, because we really thought it's gonna be really difficult to actually do this experiment, that we're gonna have to develop all kinds of like engineering tools to like stick them to each other until somebody in the lab came and said, look, it's very simple. You put them at the bottom of the tube, you leave them overnight, the next day, they're essentially fused. But I think what is even more surprising is what they do after. Because you can color the cells differently, and you can watch them under a microscope. And you start seeing over the coming weeks that those inhibitory cells that are migratory start to all move towards the other side. And the migration is actually quite specific, because the other cells don't care. And even if you fuse different combinations, even if you put two spheres, two organoids, a ventral,

• 25:24Speaker 01

  they don't, they don't move. So this, there's something that are attracted to. And so they move on to the other side, and once they arrive, they even change their shapes. They become very complex, and they start making synapses with excitatory cells.

• 25:36Hanne Tidnam?

  So that's the moments where this, where energy starts being transmitted?

• 25:40Speaker 01

  So the, the cells, like initially, they're not connected to each other, they're not synaptically connected. But once they arrive onto the other side, they start connecting to the other cells. Now, one of the things that we discovered, and I think this brings us back to how can we use this model, is that we've been studying for a while a form of autism and epilepsy called Thymathys syndrome. And this disease is very, very rare. There are only probably a couple dozen patients all around the world. What makes them really unique is that they only have a point mutation. That means one single letter in their entire genome is changed in a calcium channel. And it's been known that this calcium channel is important for cells to migrate.

• 26:24Speaker 01

  And so we thought, could it be that in patients, this mutation affects this cellular process? We, again, recruited patients that have this disease, which are very rare. We brought them to Stanford. We got skin cells from these patients. We took those skin cells and we turned them back in time to make them look and behave like pluripotent stem cells. And then we took those pluripotent stem cells and guided them to become either cortex or ventral forebrain, and then put them together. And what we noticed is that the patient cells had the cells jump much more often. So they would just engage in these jumps prematurely, so to speak. And every single time they would jump, they would actually jump a shorter distance.

• 27:09Speaker 01

  So they wouldn't do a very good job. They don't move very well. Now, the good thing is that there are drugs that can modulate this channel. And once you add them and you block the activity of the channel, we showed that within essentially a few hours, you can completely restore this abnormal cell behavior. And the reason why I think some of these technologies are so exciting is because they offer us access to certain cell states, cell behaviors, that would be very difficult to access otherwise, and so understand some of these molecular, subtle molecular mechanisms in that context.

• 27:42Hanne Tidnam?

  So one of the big limitations seems like that you're kind of constrained to the time of human development. Is that for the foreseeable future, you're stuck with like this pacing? Or would the next kind of iteration of engineering this process be speeding that up somehow?

• 27:58Speaker 01

  No, I agree. One of the limitations is obviously how long these cultures really take. But I think that's both a limitation and I think an opportunity. Because now that we know that in a dish, some of this timing is recapitulated, it would allow us to really try to understand the molecular machinery. What is this clock? And I'm hopeful that once we're going to identify what that clock really is, it would allow us to maybe accelerate or decelerate it.

• 28:21Hanne Tidnam?

  I love that idea that there's almost like an invisible clock going on here that

• 28:25Speaker 01

  Right, and

• 28:26Hanne Tidnam?

  We don't even know what the mechanisms are.

• 28:27Speaker 01

  We don't know what the molecular mechanism is.

• 28:29Vijay Pande?

  And especially in the context of development of an organism, you might want certain things to time out with other things. You have to like have the bodies being developed while the brain's being developed. But in the context of just the brain, maybe those constraints go away and you can imagine speeding it up.

• 28:43Speaker 01

  Sure, right, right. It's also very clear that the human brain has slowed down in general, like even versus other primates, slowed down its development. So many parts of the brain are actually developing at a much slower rate through a process that is called neotony, essentially just putting a brake on brain development.

• 29:00Vijay Pande?

  Yeah, what would be the, what's the selective pressure towards?

• 29:03Speaker 01

  The main, I think, hypothesis is that this allows the human brain, we have enough brain to do the basic functions, but we also have enough brain that we can allow to slow down its development and allow for social learning. So if you just think about the prefrontal cortex, that is the last one to myelinate, to mature. And it's thought that because we're allowing social learning to happen more than any other species, more than any other primate. And so there are very clear mechanisms, and we know this is true. I mean, this is very well conserved across species.

• 29:33Vijay Pande?

  Yeah, it's probably one of the parenting books I've read talked about, actually took this analogy of development mirroring evolution one step further, which they talk about after they get out of the womb, the babies are first in a little caveman stage and then in a little villager stage.

• 29:46Hanne Tidnam?

  I can totally relate to that.

• 29:48Vijay Pande?

  And it's, I mean, caveman to villager is the social development of that frontal cortex that has to happen and maybe probably needs the stimulus for it to be done right.

• 29:57Speaker 01

  Right, but because of our social group, we can allow to just like have our children develop for much longer. And this is species specific, just to make it, I mean, we, for instance, you know, if you take chimp chimpanzee derived stem cells and you compare them side by side, the brain organites, they do develop at different paces. So the chimp finishes earlier development and the human continues. So this is recapitulated in vitro as it is in vivo. So the question is like, what is the molecular mechanism? How can we figure out? Because I think as you were mentioning, of course, that is a challenge because today we have to keep hundreds of days of culture.

• 30:31Hanne Tidnam?

  What are some of the other big challenges that you think would unlock a whole new level of understanding for us?

• 30:37Speaker 01

  One mistake very often in the field is to really think that this this model is really the answer to anything. I mean, I think as George Box famously said, all models are wrong, but some are useful. And so I think the idea is like, what is it useful? I think very often people confuse, oh, are they fully recapitulating the brain? Of course not. The question is, what is the question that you're trying to answer? What process, what disease are you really trying to answer? And I think, you know, there are a lot of things that are missing. There's like, for instance, no vascularization. Many of the immune cells are missing. For instance, all the microglia are not. They can be added, but I think it's fair to look at this as an incomplete model that can be tuned in a way and engineered so that you can ask questions.

• 31:18Hanne Tidnam?

  So if you were to translate that into brain disorders, what types of brain disorders can we then study usefully with this and what not?

• 31:26Speaker 01

  I think the neurodevelopmental disorders, so the ones that arise because of the human brain developing in some unusual way or in an abnormal way, I think those are the primary disorders that we can study. The ones that, for instance, are have a later onset, such as Alzheimer's disease, neurodegenerative disorders, I think those are a little bit more challenging to study right now. And of course, the ones that have an immune component as well as as the immune system has to be brought into. I mean, there are going to be ways, I think.

• 31:55Vijay Pande?

  What about things like schizophrenia?

• 31:56Speaker 01

  Schizophrenia, I think there are multiple genetic forms of schizophrenia right now that I think could be very useful. I think if we say broadly schizophrenia, I think that's going to be difficult because schizophrenia is again, right. And we, that's why we never really say, oh, we're modeling schizophrenia, because schizophrenia is a combination of behaviors. We don't have behaviors in the dish.

• 32:19Vijay Pande?

  Yes, yes, yes.

• 32:19Speaker 01

  What we're modeling is the molecular biology behind some of this patient's genetic makeup.

• 32:24Vijay Pande?

  Yeah. Well, you know, something that I've seen in other areas is that it's common outside of sort of neuroscience that just even a cellular phenotype can recapitulate disease enough that at least you can see how that cell responds to some molecules.

• 32:37Speaker 01

  Absolutely.

• 32:38Vijay Pande?

  And that is predictive of whether it's going to work in animals or humans. So if cells are enough to predict phenotypes in some areas, you can imagine this is one step beyond.

• 32:47Speaker 01

  Precisely.

• 32:47Vijay Pande?

  It's a group of cells. It's more complex. It's not all the way to a full organ or a full organism.

• 32:52Speaker 01

  Yes.

• 32:52Vijay Pande?

  But that, you know, it's not, it seems very reasonable that when you go up a level of sophistication, you can get to a lot more.

• 32:58Speaker 01

  Right. And our approach has been precisely to try to identify genetic forms of disease. Some of them like actually very common. So for instance, there is a large deletion of chromosome 22, which is present in 1% of all patients with schizophrenia. These patients have a 40% chance of getting schizophrenia lifelong versus 1% in the general population. So studying some of these patients that have a high susceptibility to developing this disease would probably, you know, represent at the end a window into maybe other forms. Because I think it's also important, and I think we started the conversation with this. All psychiatric disorders are behaviorally defined. We are ultimately going to have to define them in a different way biologically. The way we've discovered that the fact that cancer is in the pancreas,

• 33:43Speaker 01

  it doesn't necessarily mean that it's going to have the same treatment, right?

• 33:47Vijay Pande?

  Yeah, well, cancer is moving from tissue of origin to like a pathway.

• 33:51Speaker 01

  The pathway, right. So I think the same thing is probably going to be for brain disorders. It's probably going to be more circuit based, maybe even cellular based in some of these disorders. But we're going to have to accept that autism is not one disease, but it's a group of disorders. That schizophrenia is not one disease, but a group of disorders.

• 34:07Hanne Tidnam?

  So the DSM will look totally different.

• 34:10Vijay Pande?

  The molecular DSM.

• 34:13Vijay Pande?

  You know, part of what really intrigues me about this too is that that you can basically nudge the systems to build these things. And that it's repeatable and you can build upon your past successes. I mean, these are the hallmarks of an engineering process. I'm curious how you see that bearing out, what is the future looks like? How far do you think you can push the engineering of these things?

• 34:34Speaker 01

  I mean, I actually think a lot about this like these days because I, you know, I never really wanted to become a tool developer. I kind of like became one by chance. And I just wanted to understand disease and turned out to start using this, this, these new methods. And initially I also thought that we really need an engineering approach. Like, but the engineering approach actually is not working that well here. If you think about it, because an engineer, and I see this actually when I have students that have an engineering background. They want to understand. Every single part of the system before they would put it together. The beauty, I think, about biology is very often that a lot of this information is encoded in the genome. So, once you start the process, the cells actually find each other in a meaningful way.

• 35:14Speaker 01

  And actually, very often, we do reverse engineering in the sense that we make parts and we put them together. And we see that they connect in a specific way. And then we start asking, how did they actually connect? Because we had no idea before.

• 35:26Vijay Pande?

  It's almost which type of engineering, because you could have an electric engineer get hung up on the fundamental physics. And we're going to get stuck there. Or they can sort of accept that these are working blocks. And they're somewhat black box working blocks.

• 35:39Speaker 01

  Absolutely.

• 35:40Vijay Pande?

  But then, let's see what you can do to put them together.

• 35:41Speaker 01

  Exactly.

• 35:42Vijay Pande?

  And reverse engineering is, I think, exactly the right discipline. It sounds like for where we are in terms of the time of this.

• 35:47Speaker 01

  And very often, also, our disease modeling has actually moved in that direction. When initially, very often, we would say, okay, here's a disease, a psychiatric disease, let's try to model it. But now, very often, we say, what is an aspect of human brain function that nobody has had access to? Let's try to model it in a dish and then say, okay, here are interneurons moving into the other side. What are the molecular machineries behind it? And very often, we would identify disease-related genes. And so, in that case, we can say, oh, now we have the perfect assay in the first system. But if you would have tried to do that a priori, it would have been very difficult. Because you wouldn't have had the tool. Now, we can make more defined brain regions. And when we assemble them, we can actually get cells projecting at a long distance,

• 36:31Speaker 01

  connecting and forming small circuits.

• 36:33Vijay Pande?

  Well, and that is what I would expect from that type of engineering approach. Because if you had to do this purely as rolling dice and seeing what comes out, it's going to be really hard to start building these more complex structures versus sort of getting better year after year.

• 36:46Speaker 01

  Precisely, yes.

• 36:47Hanne Tidnam?

  So, I have to ask, if now we're in this new world of sort of beginning to grow essentially small human Brains or versions of brains that have some functionality to them. They're firing, they're doing some stuff. How do we begin to think about sort of some of the ethical and legal issues around that? I mean, is there a line at which you can say clearly like, no, this is a thousand miles away from consciousness? Or if you let that organoid grow in a dish for, I don't know, five years? Like, would you begin to be approaching? Like, how do you start thinking about what's actually happening there in terms of consciousness?

• 37:23Speaker 01

  Of course, there are a lot of discussions in the field. And what is clear is that this structures that we're building in a dish are very different than the actual human brain. First of all, because they're not a miniature version of the human brain. That's why we don't like the term mini-brain, because it's inaccurate.

• 37:40Hanne Tidnam?

  Yeah.

• 37:40Speaker 01

  But they're like-

• 37:41Hanne Tidnam?

  It's a slice, it's a chunk.

• 37:41Speaker 01

  But there are parts, there are incomplete parts of the human brain. And there are another very important aspect of human brain that is also missing here, which is actually sensory experience. We know that the sensory experience, everything that comes through our senses during development. Absolutely, that shapes the circuits. We know that that's critical periods.

• 38:01Hanne Tidnam?

  But is there no sensory input at all coming in to these cells?

• 38:04Speaker 01

  I mean, we can build all kinds of ways of actually stimulating, but there are no sensory inputs. So that is the reason why, overall, we don't really think that there are like major ethical concerns with the cultures maintained in the dish. Now, there are some discussions, some ethical discussions around the transplantations of some of these cultures into animals. So, for instance, one can take them and transplant them into a rodent and integrate them into the circuit of a rodent. In which case, for instance, they may receive some sensory input, they may participate to some of the circuit functions. So I think many of the discussions are around that. And of course, what is really considered not acceptable at this point is, for instance, transplantations into primates.

• 38:47Speaker 01

  We also, I think, have to put this into the broader context of how are we gonna model psychiatric disorders? And we call this broadly surrogates, brain surrogates, because they're all kind of like trying to mimic the human brain. And I think it's very clear that we need better model for understanding psychiatric disorders. We need models that resemble more and more the human brain. And obviously, the closer they resemble the human brain, the more uncomfortable we also feel with anything resembling the human brain. So I think- That's why I think ethicists and lawyers are in this because of course there are also legal implications, right? For instance, the cells come from human individuals. What are the limitations to what are really the applications for which they can be used for? And so I think

• 39:31Speaker 01

  those are discussions that are active in the field at this point.

• 39:34Vijay Pande?

  Yeah, and I think the the long term implications of this is kind of amazing because we're talking about this very much in the context of scientific discovery and disease. And that's one direction. Just as a technology, it's also just kind of amazing. The fact that we can engineer silicon allows us to have chips. The transistor is cute, but sort of having that engineering aspect of it is what allows us to have microprocessors. So this reminds me of the early days of that in some ways where you're putting these things together and you're seeing what they can do. I can imagine that they could start, especially, you know, since on the silicon side, we're spending so much energy on the silicon side doing neural nets.

• 40:14Speaker 01

  Right.

• 40:15Vijay Pande?

  You know, and running machine learning. It's interesting to think what you could do with this. Then that gets really kind of very science fictiony, like, could you create a little visual cortex and use that to do computation? Could you do this and that? And there the it's going to be a really interesting challenge on the bioethics side to figure out, you know, what makes sense.

• 40:33Speaker 01

  Right, right.

• 40:34Vijay Pande?

  You can see a path.

• 40:35Speaker 01

  Absolutely. And and one can think that once we put the cells together and they connect to each other, they form small networks.

• 40:40Vijay Pande?

  Yes.

• 40:41Speaker 01

  Of sorts, there is some sort of computation.

• 40:43Vijay Pande?

  Yes.

• 40:44Speaker 01

  There's like some information that is transmitted among the cells. So I guess one question is like, how complex that is? How similar is that to the human brain?

• 40:50Vijay Pande?

  Yes.

• 40:51Speaker 01

  But we do know that sensory input and many other external informations do shape that.

• 40:57Vijay Pande?

  Yes.

• 40:58Speaker 01

  Developing. So I think that addresses some of the ethical concerns when we're maintaining them purely in vitro. But again, one can really see it as some sort of computing device, so to speak.

• 41:09Vijay Pande?

  Yes.

• 41:10Speaker 01

  Which is biological.

• 41:11Hanne Tidnam?

  So I want to wrap up by just asking kind of where what the next near term exciting steps are here and then what the wildest ones that you can imagine is when we have this new tool, this new way of understanding the brain, new model, appropriate for some things, not for others.

• 41:24Speaker 01

  Exactly.

• 41:25Hanne Tidnam?

  What is the sort of like the the next major evolution that you see?

• 41:28Speaker 01

  I think the, I mean, the next step is really applying it broadly, right? I think it's still early days in developing some of these technologies and applying it. But we really need to see it at work.

• 41:39Hanne Tidnam?

  Yeah.

• 41:40Speaker 01

  More broadly and maybe bringing psychiatry into a New era, like a molecular era into what we would often refer to as molecular psychiatry.

• 41:49Hanne Tidnam?

  Yeah.

• 41:49Speaker 01

  So I think on the long term, that's what I see the most exciting avenue.

• 41:53Hanne Tidnam?

  Yeah.

• 41:54Speaker 01

  But of course, the models are gonna get, you know, more complex in many ways. We're likely gonna be able to build ever more complex circuits.

• 42:01Hanne Tidnam?

  Hmm.

• 42:02Speaker 01

  And I think some of the transplantation studies are also gonna be critical in that regard. And then, of course, there the implications about, you know, just understanding how the human brain function. I think how and what makes it so unique. There's a lot of interesting biology. We know very little about the actual human brain, not just the biology, but also the computing. How does it come together and compute?

• 42:24Hanne Tidnam?

  Do you think that it could bring us closer to understanding the nature of consciousness?

• 42:28Speaker 01

  I would be satisfied with just understanding the biology of one single neuron. I think that would be exciting for me.

• 42:34Vijay Pande?

  It's really intriguing time right now. You think about like, one analogy I often make is that birds and fighter jets both have wings, but they fly very differently. And maybe one was, the, the human version was inspired by the natural one, but that we can go beyond.

• 42:49Speaker 01

  Right.

• 42:50Vijay Pande?

  But to go beyond, it's useful for us to really understand and to be able to probe deeply and then figure out what we wanna do there. And the go beyond might be using neurons as an analogy as it is in machine learning right now and neural networks. Or it could be actually as literal neurons. And that back and forth is really interesting. And one of the most exciting things for me about the field of neuroscience is where it sits right now in that sort of hub between understanding the brain and understanding intelligence, understanding learning and both at this sort of macro scale, which we talked about, and the micro. Those connections, I think, will be useful sort of in many disciplines for years to come.

• 43:26Hanne Tidnam?

  Thank you so much for joining us on the A16Z podcast.

• 43:28Speaker 01

  Thank you so much for having me.