Hello. My name is Arnold Kriegstein. I'm a neurologist and a neuroscientist at UCSF, and I study human brain development and diseases and a little bit of evolution. And in this section, we're gonna talk mostly about human-specific features of brain development, how some of them can be modeled in animal systems and model systems like organoids, and a little bit about what they're teaching us that may inform us about diseases and also about human-specific features of development and brain evolution. So, I thought I'd begin just by mentioning... in my previous talk, we highlighted how we've used single cell RNA sequencing in my lab, and also in other labs, to start looking at genes that are uniquely expressed by particular cell types in the adult, and also in the developing human brain. We've particularly been interested in a special cell type that we described a number of years ago. We call it an outer radial glial cell, or oRG cell. These oRG cells are neural stem cells. They're highly enriched in the human brain. And our single cell transcriptome has shown that these cells in human brain are highly enriched for genes that are involved in these three pathways: extracellular matrix production; epithelial to mesenchymal transition, which is, we think, how these cells originate from the ventricular zone, through neuroepithelial cells turning into oRG cells; and also they have self-renewal or stem cell maintenance gene pathways. But each one of these enriched gene pathways that we found in these developing fetal cells in the human brain have been described in the literature not in developing human brain but, rather, in tumors, as shown here, known as glioblastoma, glioblastoma multiforme. These are terrible, untreatable diseases. They are deadly. They usually strike patients in adulthood. They're not developmental diseases that occur in childhood. And yet we were very intrigued that they seemed to contain gene expression signaling pathways that we find, in development, are unique to these outer radial glial cells. The outer radial glial cells are present only during brain development. They disappear, we think, before the patients are even born. So, the fact that these enriched genes are found in a cell type... in the tumor cells was particularly intriguing. And so, we were able to look at single cell gene expression patterns in tumors that were resected from patients at UCSF. And shown here is a heat map of gene expression in these tumors. And what I want to highlight is that the genes we found in oRG cells are shared by the most aggressive ones of these human glioblastoma brain tumors. That's highlighted in this red rectangle. So, based on the fact that we see expression for the cell type -- genetically, or molecularly, in the tumor samples -- we got some fresh samples from the OR, we sectioned them, stained them the way we do our normally developing fetal tissue, and then did time-lapse imaging, as shown here on the right. And what I want to highlight... if you pay attention to this cell in particular, it undergoes a mitotic behavior that we found previously was unique to the fetal outer radial glial cells. So, if I start this movie and you focus on the behavior of that cell, it undergoes what we call mitotic somal translocation. The nucleus translocated up along that fiber just prior to cell division. That behavior -- mitotic somal translocation -- as far as we know is unique to these fetal cells, these cells that are present in the developing human brain. So, we were very surprised to see the same cell type with many of the same dynamic behaviors I just showed you in adult tumors. So, we've been pursuing that unexpected finding, and we have evidence that these cells are present in many, as I mentioned, of the most aggressive glioblastomas. And we have a hypothesis that this behavior -- the mitotic somal translocation... the mitotic behavior -- may contribute, or may even be responsible, for how very aggressive and invasive these tumors may be. These are very early days, and we're still exploring that hypothesis. But I want to mention it, because it was an unexpected direction that our research has taken. And we've now been collaborating with experts in cancer biology at UCSF to try to pursue these questions further. The other way we've been using the information we've gained on molecular expression of individual cell types is to calibrate, or to understand a little bit better, models of human brain developments that are becoming very popular. And I want to spend much of the rest of the time talking about organoids in particular, and especially what we call cerebral organoids, which are models of human brain development in a dish. Just to outline briefly how these are created, they start with a pluripotent stem cell. What I show here on the left is a skin cell, which can be reprogrammed using those Yamanaka factors that were first introduced by Shinya Yamanaka, or similar reprogramming factors, that can convert an adult cell, like a skin cell, into a pluripotent cell, like an embryonic stem cell. From those induced pluripotent stem cells, you can then derive neural progenitor cells, using a variety of different protocols that seem to all have common features that allow you to take this undifferentiated cell that can become any cell in the body and direct its differentiation down a pathway that leads to a neural stem cell. Beyond that point, you can allow these cells to aggregate or clump together, and that's shown in this "induction molecules" step called the neural rosette, in the middle, where, as they aggregate, they start to signal to each other and they organize themselves. And in a way, they self-organize in a way that kind of reproduces what would happen during normal, in this case, brain development. And they form little spheres, or little clumps of tissue, shown here on the right, that as they become more developed, more mature, start producing neurons. And those neurons resemble the cells in the normally developing fetal brain, and they're referred to as cerebral or brain organoids. And those are... the brief outlines of the protocols that I want to mention. I'm not gonna go into great depth in terms of how we actually make these cells; there are many papers have been published on these protocols. And I also want to mention that in addition to just growing them in these aggregate wells or cultures, people have now begun to apply a variety of different engineering techniques in order to try to get them to grow bigger, more mature, try to better recreate or better reproduce human features of development. And much of this has involved the use of scaffolds or bioreactors, which are symbolized in this diagram here. Just to show you how these cells look, at least in our hands, we've taken iPS colonies, shown here on the far left, created these aggregates known as embryoid bodies, driven them down that neural pathway, as I mentioned, to create neuroectodermal or neuroepithelium... they're like radial glial cells at one stage. They form rosettes. And then, as they get more and more mature, they start resembling a more complicated cerebral or brain-like structure. Just to show you a little bit about the complexity of human brain development, I would refer some of you to the talk I gave a little bit earlier describing some of the complexities of the composition of the human brain, as well as its structural organization. Some of that is schematically shown here. And the question is really, how much of the features of normal brain development are reproduced in these organoids? And that's the question I want to focus on in the next few slides. First of all, at the early stages of brain development, the organoids seem to do a reasonably good job of not only capturing the cell type diversity but also the organization. So, shown on the left are these neural rosette-like structures, which form like a neuroepithelium around a lumen, an area that's highlighted here in blue. And the cells have apicobasal and apicolateral polarity. They resemble in some ways the human cortex shown here on the right. And then, in contrast, I'm showing you a similarly aged organoid at the bottom. And you can see the resemblance. They are organized across a kind of pseudostratified epithelium. And so, this early stage of neuroepithelial development seems to be well-produced, or at least reasonably well produced... reproduced in organoids. If we look at additional structures as the cortex normally develops, it becomes beautifully and very complicatedly organized. That's shown in these panels on the top, which are sections of the developing human brain at different stages. Below, I've shown you cross-sections of typical organoids at comparable stages of development. And we've used the same markers to stain cell types, so the colors of the cell types in the panels above, which highlight the cells in normal fetal development, are the same colors that are shown in the panels below, to highlight those cell types as they appear in organoids. And what I want to emphasize here is while the rosettes shown at the beginning of the panel... the panel... whoop... the panel way over there on the left are not too different than the progenitor zones in normally developing fetal brain, what's obviously missing is the highly structured organization of the rest of the brain. So, the layers of the cortex, which are so exquisite in the top panels and change over time as the neurons mature and migrate and develop, are not well represented in the organoids. The organoids are much more chaotic, as shown below. They don't have the same laminar, or layer-like, organization, even though they do have at least some examples of the major cell types, which is reflected by the red dots and the green dots and the yellow dots. The cell types are there, but they're not distributed in the right proportions, and certainly not in the right kind of cortical, or layered, organization. That's obvious just looking at these organs... the organoids. So, they're missing other cell types that weren't highlighted in the previous slide, and those are mentioned here in the schematic: endothelial cells; blood vessels, which these organoids don't have, because they grow from endothelial cells that come from a different source that we don't find in these neural stem cells alone; they're also missing microglia, which are the immune cells that migrate into the brain from external sources; and most protocols don't produce non-neural cells, the glial cells, like astrocytes or oligodendrocytes, the way they normally do in maturing normal brain tissue. The other thing that's missing is the extracerebral tissue. Of course, you're just mostly restricted to cerebral cortex. You don't have all the other brain regions together in the same dish, and so you don't have the connections that normally form, for example, from the thalamus. You don't have target structures that the neurons can project to, which include the thalamus and areas of the most basal ganglia, like the striatum and so forth. So, we have a very restricted or a reductionist view of brain development. Nonetheless, there are some... there's some good news. And the good news is that these cerebral organoids do show preservation of broad cell types, and this is true across many different protocols. So, what I'm showing you is data from our lab, where we've used more than one protocol, in fact, and multiple individuals to create cell lines and organoids, and compared them to individual primary tissue. What I call primary tissue is donated human samples from multiple individuals. For example, we've had 48 individuals contributing samples of normal developing brain that we've used, along with 10 individuals from which we've divide... devised... derived multiple cell lines. From those multiple cell lines, we've made 31 organoids. And we've looked across all the organoids and across all the normal human tissues at the cell types, the composition, essentially, of different stages of development, and put them all together to see how well they compare. And that's shown bioinformatically in this panel, which shows clusters of cell types that are distinguished according to color, in this example, that represent the major cell types that you'd expect to see in developing human cortex: excitatory neurons at different stages of maturation; interneurons or inhibitory cells at different stages of maturation; the progenitor cells, including of course the radial glial cells; and neuroepithelial cells; and intermediate progenitors. All the different cell types that have been described in developing human brain. Shown on the right, we've color-coded the contributions to each of these cell clusters that come from organoids versus primary human fetal tissue. And as you can see, the pink and the brown dots are more or less intermingled in a kind of salt-and-pepper distribution, suggesting that there are cell types, of all the major classes, that we find both from primary human tissue and also from organoids. So, at least we can say that the cell types, in broad strokes, are relatively similar. So, we've used our data set, our single cell RNA sequencing data set, to look more carefully at the gene expression patterns across individual cell types from both primary tissue as well as organoids. And what I'm showing now are the gene clusters that we've collected from five individuals across multiple brain regions and developmental time points. And the cell types, highlighted here, going from left to right on the top row include progenitors that express SOX2, which are the neuroepithelial radial glial-like cells; the outer radial glia, in the middle panel, that express HOPX; and intermediate progenitors that are expressing a gene known as EOMES or TBR2. And there are cell types of all of those progenitors found across all of our donated samples. In the bottom are lineages involving neurons, including newborn neurons at the left that express NEUROD, maturing neurons that express SATB2, and inhibitory neurons on the right. And you can see from those blue dots that we find those different cell types across multiple individuals -- in fact, all of our individuals -- that have been examined for their cortical cell types. So, these are the principal cell types that we can see maturing in normally developing cortex. If we do the same kind of analysis, based on our single cell gene expression data from organoids -- and this, again, involves multiple cell lines... four cell lines, three different protocols... we put all the cells together -- we find that the progenitors and the early-appearing newborn neurons shown on the left... those cell types are highly represented in the organoids. But the outer radial glia, the intermediate progenitors, the maturing neurons, and inhibitory neurons shown to the right are represented in very, very small numbers. So, they're present, but they're not represented in the same proportions that we'd see in normal developing brain. This highlights one of the differences between our primary tissue, or fetal tissue, and brain organoids. It's that the cell types across broad classes are represented, but the relative numbers are not the same. More importantly, the gene expression of each of the cell types is not normal. What are shown here in these Venn diagrams are just the genes that are uniquely expressed in individual cell types or classes. And shown on the large yellow circle are the cell type defining genes from primary donated human tissue. And there are about 600 of them. This shows that there are highly specific gene expression patterns across multiple cell types in the normally developing human brain. In the blue circle are similar genes that we can find that are selectively enriched in different cell types in the organoids. And you can see that there's a far fewer number -- only 46 compared to around 600. And if we look at the genes that are the same -- cell defining types in both fetal tissue as well as organoids -- it's a much smaller number: only five. So, there's a very significant and impoverished number of specific gene differences from cell type to cell type in organoids compared to what we see in the diversity of normal cell types in developing human brain. This also extends to the individual complexity of each cell type. So, what's shown in these bar graphs is the heterogeneity of cell types, in fetal tissue compared to organoids, across different progenitors as well as excitatory or inhibitory neurons. And at the far left, shown in the tallest differences, right here, is the fact that the fetal cells in the excitatory lineage are highly diverse. Much of those differences really reflect areal position. So, we see the same cell type in the front of the brain, the frontal region. It shows different gene expression patterns than that cell type in the posterior part of the brain, or the occipital region. Those areal differences, which account for some of the diversity of cell type enrichment in the normally developing brain, are not present or not reflected in the organoids. The organoids have far fewer gene differences from within a cell type. They don't have those regional identity gene differences that we'd normally find. That's not only true for excitatory cells; it's true, as shown in these other bar... bar graphs for other cell types, like the intermediate progenitors, as well as young neurons. So, there's a reduced diversity of gene expression across cell types. And then, finally, there's perhaps the biggest problem... is that there's a blending of gene identity in organoids that you don't see in primary tissue. And that's shown in this graph. So, in orange are the gene differences between radial glial and neurons in primary tissue. And those gene expression differences are very distinct. You see genes expressed in radial glia that are not found in neurons, and genes in neurons that are never found in radial glia. But in blue are those same genes in organoids. And what that shows is that these radial glial cells are expressing neural genes and that the neurons are expressing radial glial genes. So, there's a kind of blended or ambiguous gene identity that you don't see in primary tissue. So, how these gene differences are going to impact the ability of the organoid to model specific diseases, for example, is unknown, but I think it's something we have to be aware of, and we may have to take into account when trying to calibrate organoids against primary tissue. And one of the more dramatic differences that we see across all organoids compared to primary tissue is that the organoids are highly enriched in cell stress genes. What's shown here are glycolysis network genes, or glycolysis stress genes. And the violin plots to the right highlight that in primary tissue... and I'm just using two examples, which are far... far here to the left... show very little expression of this gene, PGK1, or this gene, ALDOA, which are both genes that are highly expressed when there's glycolytic stress. In the organoids, by comparison... this is across multiple samples of organoids... they all show highly enriched glycolysis network genes. In this case, these two genes, as an example, are highly enriched in the organoids compared to primary tissue. This is also true for genes that are associated with endoplasmic reticulum stress, metabolic stress. And once again, these two genes, which are just chosen as examples, are very lowly expressed in primary tissue, which are the first two violin plots, but all of them are highly enriched across protocols in the organoids. The organoids for some reason under high levels of metabolic stress that are not seen in normal primary tissue. And then we looked at published datasets to see if our data in my lab happens to be different than what's been published in other protocols by other investigators. And I've highlighted here multiple published lists of gene expression in organoids using different protocols. They're all relatively similar, but they are different, and they're in different laboratories. All of them, across all these protocols, show enhanced glycolytic stress and endoplasmic reticulum stress pathways, just like the ones in our own hands. So, there's something sort of fundamentally wrong or different in the protocols that people are using to generate organoids that put them under stress. And what the consequences of this stress enhancement might be in the use of these organoids, either for studying normal development or disease, I think needs to be considered. So, the caution here is that one should be careful when interpreting disease phenotypes in organoids that might be influenced by these metabolic stress pathways. And that's particularly important when studying metabolic disorders. However, we and others have been using organoids to study diseases, and I think if you do it properly, you can learn quite a bit about disease mechanisms that you probably can't understand or can't appreciate any other way. And one example I want to give you is for this disease, known as lissencephaly. A normal MRI scan on the left shows a highly folded, normal cortex. The one on the right shows a patient who has a form of Miller-Dieker lissencephaly, which is a very severe form of lissencephaly, which means the brain is smooth, and it's also a little bit smaller -- microcephalic. And that's shown on the right. It's a horrible disease. It's clearly a developmental disorder. The cell doesn't... the brain doesn't develop normally, and the cortical folds are entirely absent. So, we were able to get skin fibroblasts from both normal patients and patients who had this form of genetic lissencephaly called Miller-Dieker syndrome. And this is work that was done by Marina Bershteyn, a very talented postdoc, when she was in my lab. And from these patients, and also normal patients or wild type, normal individuals, she was able to derive organoids. Those are shown below. First by... the induced pluripotent stem cell stage, where she had pluripotent cell lines generated from multiple individuals, both normal and patients with Miller-Dieker syndrome. Those were then driven down a neural stem cell pathway, allowed to aggregate and form the kind of organoids that we've been discussing so far. And these are cross-sections of those organoids at different stages of maturation. The top row shows the normal, wild type. The bottom row shows the Miller-Dieker patients. And I hope you can appreciate that at each of these stages, things were quite similar. She was able to derive pluripotent cells, derive neural stem cells, allow them to form these rosettes, which are shown on the right. They're very beautiful, and they're very similar, although slightly different between Miller-Dieker and normal. The slight differences are that the Miller-Dieker patients... the cleavage angles when the cells divide are slightly different than in normal patients. But generally, they looked quite similar. And they were able to mature to the point where they showed genes, as shown here in the bottom, after 10 or 15 weeks that we'd previously associated with the development of these outer radial glial cells, which is a very specific cell type that we find in the human brain, that we don't find in mouse. And those gene expression differences suggested that we had outer radial glial cells in the organoids. And to confirm that, we took sections of the organoids, stained them and time-lapse imaged them. And here's that behavior. This is a looped film, so you see them dividing over and over again. But this is that mitotic somal translocation behavior that's so characteristic of these cell types, outer radial glia. So, we had outer radial glial cells in reasonable abundance in our wild type as well as in our lissencephaly patient organoids. But when we looked more carefully, we found that the outer radial glia weren't dividing normally in the organoids... in the patient organoids. And that's highlighted here. When the cells jumped, they jumped much longer distances than normal. But then, having jumped that far, they failed to divide normally, either not dividing at all or taking hours and hours to divide, as shown here, instead of minutes, which is the normal case. That abnormality in the behavior of the outer radial glial cell is a feature that you couldn't appreciate, and hadn't been appreciated, in studies of these mutations in animal models like the mouse, for example. But it highlights the role of the outer radial glial cell in this disease, and maybe links this disease to features like cortical expansion and cortical folding, which are clearly absent in this disease. And so, again, this highlights the use of a human organoid in a laboratory to try to get insight into a disease mechanism that was previously unknown, and that might be human-specific, or at least a feature that you couldn't model in animals that had been studied for many, many years before. The other feature that we've been using organoids for is... to look at human-specific features of brain development and evolution. This is something that we have trouble doing in primary tissue, simply because we don't have, for example, fetal brain from multiple species like, for example, chimpanzee. And to study human-specific features of brain evolution, you really need to look at differences between human and our nearest or closest living relatives, which are the chimpanzees. Humans and chimpanzee diverged 6-8 million years ago. The problem is that we don't have access to fetal developing monkey, or in this case, especially, chimpanzee tissue. So, to overcome that, a very talented postdoc who joined my lab a number of years ago, Alex Pollen, who is an evolutionary biologist, decided he would use organoids. And so, it took him a number of years to develop a protocol that worked equally well across great apes and humans to make induced pluripotent stem cells from skin cells, essentially. So, starting with skin cells from, for example, chimpanzees and humans, he was able -- with the same protocol -- to make pluripotent stem cells, which are shown in the first panels to the left, and then derive those into organoids. So, we had chimp organoids, human organoids, and a variety of other non-human primate organoids. And as shown on the right, they looked quite similar, at least grossly. They matured according to the same protocol over time. So, in the top row are some human organoids at different stages of development, and at the bottom are the chimp organoids from different stages of development. And I want to highlight that in order to do these evolutionary studies, it's very important to have more than one or two individuals, because you don't want to look at individual variation; you want to look at species differences. And so, for example, we've used ten human individuals to grow our organoids and eight individual chimpanzees to grow the chimpanzee organoids. And the differences that we've seen, and that I'm gonna highlight some... some of which I'm gonna highlight next, were seen across all of these individuals, and that's a very important feature. So, let me focus again on the progenitor cell types, namely the radial glia. And we've been able to see, as shown here, human-specific genes in the radial glial cells both in human primary tissue, which is fetal tissue, and organoids at the same stage of development that are not found in chimpanzee or macaque. These are two non-human primates. And what's shown if you look at the red dots are highlighted genes that are either found in human but not in non-human primates, or found in non-human primates but not in humans. So, these are genes that were either gained or lost in evolution, when humans and chimpanzees diverged. And they're specific to this cell type, namely radial glia. And I want to highlight some of them in particular that are unique to outer radial glia, which is a particular cell type of interest to me. So, what's shown here are two genes that are expressed in radial glia in both primary human tissue, which is shown in pink, and in human organoids, which is shown in brown, but not in either chimpanzee or in macaque monkey outer radial glia. So, these are genes that were expressed in radial glia in humans during evolution that have been essentially gained that were not present in non-human primates. Looking at the other side of the equation, we can see genes that were actually lost in the evolution between chimpanzees and humans in this particular cell type. And that's an example that's shown here on the right, KPNA4. This is a gene that's found in chimpanzee and gorilla radial glia, but was lost in human radial glia, as shown both in primary tissue and in organoids. So, this now allows us to look at human-specific changes in gene expression in individual cell types. And I want to go back to one of the differences we highlighted in my first talk: the mTOR signaling pathway, which as I showed earlier was something that was enriched uniquely in outer radial glia during human brain development. Shown here on the left in a different sort of landscape map are a variety of different genes that are part of the mTOR signaling pathway, all of which are enriched in the outer radial glial cells in developing human brain. This was confirmed in tissue sections, shown here on the right, by the expression of a protein called pS6, which is essentially a readout of enhanced or activated mTOR signaling. And you can see in yellow that this expression level is highly outlined in the radial glial fibers of oRG cells, outer radial glia, confirming that the outer radial glia do in fact have enhanced enrichment of this mTOR signaling pathway, as our gene data would suggest. Now, if we look at the genes from our organoids across species, which are shown here... and what I'm highlighting are just the mTOR signaling pathway genes in this heat map. And in primary human tissue -- which is, you know, the fetal tissue that I described, for example, in my first talk -- as we'd expect, there's enhancement of these genes, especially those that are activated in mTOR signaling. So, that confirms what we'd seen before. The human organoids, highlighted here, also show enhancement of those same gene pathways, which is very reassuring. It shows us that the human organoids are matching the expression in the outer radial glia of the primary tissue. So, the same genes in the same cell type are enhanced in organoids as in primary tissue. But if we look at the chimpanzee and the macaque tissue, they don't have enriched expression of these mTOR signaling pathways, suggesting that the enhancement I mentioned earlier in this cell type might be human-specific. So, to confirm that, we went to primary tissue. In this case, we had fetal tissue from the macaque monkey. And we were able to look at pS6, that readout of mTOR signaling. And as highlighted in the quantification down below, this confirmed that the outer radial glial cells in... human outer radial glial cells show enhanced mTOR signaling pathway expression but not the same cell type in monkeys, in this case the macaque. So, this has given us new insight into a difference that we previously knew was found in these outer radial glial cells. The outer radial glial cells had enhancement of this mTOR signaling pathway, but what this study tells us now is that this is a human-specific disorder, a human-specific feature of these outer radial glia. And the reason that's important -- or we think it's important... potentially important -- is because the mTOR signaling pathway has been implicated in diseases, including autism and tuberous sclerosis and macrocephaly. And so, these results suggest that the study of these diseases might require human models. It might not... not... not only not be modeled properly in mice, but it might not even be appropriate to use non-human primates, because of this enrichment of mTOR signaling that's a human-specific feature. So, that's an example of how one could use organoids not only to get insight into diseases but also to look across species at cell type-specific differences that might be evolutionarily important and also may have some disease implications. So, the conclusions of this talk are several, but I just want to highlight these three. Cellular diversity of radial glial neural stem cell... stem cell subtypes is much greater in humans than in mouse, and some of this diversity is reflected in human organoids as models of brain development. These organoids can reveal disease mechanisms as well as -- I've shown you more recently -- evolutionary changes. But, the cautions are that the organoids are under stress. They do not reproduce all the features of developing human cortex, and they have some differences of gene expression that might be important both in normal brain development and especially when analyzing the phenotypes of diseases. So, for the time being, I think they need more work, more improvement, and they need to be mod... they need to be standardized against normal developing brain tissue to really understand the differences, and try to improve the technology to make organoids that are better representative of what goes on during normal development. I want to end just by thanking all the people who did the work on this study, of which there were many. These are people who are currently in my lab, who have contributed to one or another of the things that I've mentioned. I want to highlight, for the organoids in particular, the contributions of Aparna Bhaduri, who is a terrific bioinformatician as well as a developmental biologist, and Madeline Andrews, who's been terrific at growing and studying these organoids that I focused a lot of my discussion on. And I also want to thank the funding agencies, and in particular, The BRAIN Initiative, which has really funded our single cell analysis of brain development and given us some of those insights into outer radial glial cells and other human-specific features. So, I thank you all for your attention, and I thank all these people who've contributed to the work. Thank you.
