Top

iBiology

Bringing the World's Best Biology to You

Asymmetric Cell Division and Brain Development

Part 1: Asymmetric Cell Division; From Drosophila to Humans

00:00:13.12 Hello.
00:00:14.12 My name is Juergen Knoblich.
00:00:15.12 I'm a scientist at the Institute of Molecular Biotechnology of the Austrian Academy of Sciences
00:00:19.19 in Vienna.
00:00:20.19 And, today, I would like to tell you about asymmetric cell division, one of the most
00:00:24.07 fundamental processes in developmental biology.
00:00:28.14 My lab is interested in the development of the human brain, one of the most complex but
00:00:35.10 also fascinating structures that nature has generated.
00:00:38.11 And, on this slide, I've summarized what we know about the development of mammalian brains.
00:00:45.24 The mammalian cortex develops from a neuroepithelium that lines a liquid-filled cavity, which is
00:00:52.16 called the lateral ventricle.
00:00:55.03 It's a polarized epithelium, where basal is on the outside and apical is on the inside.
00:01:01.04 And it's a very nice model system to study, because all the different cell types are nicely
00:01:05.04 arranged along the apical-basal axis of the epithelium in a layered-type fashion.
00:01:09.23 So, here, on the more basal side, is the so called cortical plate, and this is where the
00:01:15.13 neurons reside and where they send out their axons.
00:01:18.07 Here, on the more apical side, is the so called ventricular zone, and this is where the stem
00:01:23.11 and progenitor cells reside.
00:01:25.01 And these progenitor cells can undergo either one of three different types of divisions.
00:01:30.05 So, early during the development, they divide symmetrically and this leads to an initial
00:01:35.04 amplification of the progenitor pool.
00:01:37.20 And, later, they start to divide asymmetrically, so that one progenitor generates one progenitor
00:01:43.12 cell and one cell that migrates out into the cortical plate to undergo
00:01:48.01 terminal neuronal differentiation.
00:01:49.19 Or, alternatively, the second cell can be a so-called intermediate progenitor, which
00:01:54.00 divides once more into two terminally differentiated neurons.
00:01:58.00 And this is called direct or indirect neurogenesis.
00:02:02.19 Now, asymmetric cell division is a conserved process that happens during development of
00:02:09.03 many brains.
00:02:10.03 So, this is in the development of an insect brain, in this case, the fruit fly Drosophila.
00:02:15.18 During Drosophila development, the neurons are generated during the larval stages of
00:02:20.17 development and they arise from neural stem cells that are called neuroblasts.
00:02:25.12 And these neuroblasts divide asymmetrically into a large and a small daughter cell.
00:02:31.20 And, after division, the small daughter cell will divide once more into two terminally
00:02:37.00 differentiated neurons, whereas the large daughter cell will continue to divide asymmetrically,
00:02:42.07 in a stem cell-like fashion.
00:02:45.22 So, asymmetric cell division happens during the development of many brains and, if you
00:02:50.11 think about it, there's actually three different ways of how cells could divide asymmetrically.
00:02:57.00 Either a cell could generate just a bunch of daughter cells, where, then, some of them
00:03:02.10 stochastically assume fate A, whereas others assume fate B. Alternatively, the process
00:03:08.12 can be a bit more coordinated, for example, if the mother cell is under the control of
00:03:15.13 a polarized environment and depends on certain factors that are secreted from there, but,
00:03:20.21 after division, only one of the two daughter cells continues to receive those factors and
00:03:25.08 assumes fate A, whereas the other cells assume fate B. And in its most sophisticated way,
00:03:31.14 asymmetric cell division can involve the polarized segre... segregation of cell fate determinants
00:03:37.01 into one of the two daughter cells, which then initiate a particular developmental pathway
00:03:41.06 in this cell but not in its sister cell.
00:03:44.09 And it's this way of intrinsically asymmetric cell division that I would like to talk with
00:03:49.05 you about today.
00:03:50.23 So, asymmetrically segregating determinants are not really all that a new idea.
00:03:55.20 Already, at the end of the 19th century, a developmental biologist called Weismann formulated
00:04:01.08 what is now thought to be the first theory of a developmental cell fate specification,
00:04:06.05 and what Weismann thought was that cells contain a set of nuclear determinants, which are then
00:04:13.02 distributed during division among the daughter cells so that, at the end of development,
00:04:18.17 each cell has one of these determinants, and this is what specifies their fates.
00:04:25.11 Weismann's theory was very successful, because a few years later a developmental biologist
00:04:30.04 named Ed Conklin found the molecular equivalent of the segregating determinant.
00:04:36.09 Conklin studied the development of ascidian eggs and he found that ascidian eggs contained
00:04:42.08 a region of yellow cytoplasm that he called the yellow crescent.
00:04:47.21 And what he found was that this yellow pigmented area, during development, always segregates
00:04:54.09 with a particular cell lineage and, in the end, ends up in the muscles.
00:04:58.22 And so he thought that the yellow crescent contains a substance
00:05:02.23 that specifies muscle fate.
00:05:05.06 Now, although the protein that is responsible for this has been identified and, although
00:05:10.10 it is really a muscle fate determinant, this pathway turned out not to be conserved in
00:05:16.00 evolution and not really tell us about how asymmetric cell division actually works.
00:05:22.21 But this changed when, in the lab of Yuh Nung, Tadashi Uemura identified a gene that is called
00:05:31.11 numb, and numb is the very first asymmetrically segregating determinant that was identified.
00:05:37.21 numb was originally identified as a Drosophila mutant that affects the development of the
00:05:43.09 so-called external sensory organs.
00:05:45.14 Now, every one of you knows that the back of a fly is full with these hairs and these
00:05:50.06 hairs are not just two cuticular structures, but they're actually sensory organ.
00:05:55.07 And each of these hairs contains four cells: it contains a hair cell, a socket cell, and
00:06:00.07 then two internal cells that form a neuron, and a sheath cell that surrounds the neuron.
00:06:07.00 External sensory organs are a wonderful system to study asymmetric cell division, because
00:06:11.09 all the different cells in an external sensory organ are generated from one sensory organ
00:06:17.04 precursor cell in a series of asymmetric cell divisions.
00:06:21.11 So, the SOP cell will first divide into two intermediate progenitors, which then generate
00:06:27.05 neuron and sheath and hair and socket.
00:06:30.16 Normally, all these divisions are asymmetric, but in a numb mutant all of these divisions
00:06:36.13 become symmetric, so that all of the daughter cells in the end become socket cells.
00:06:41.16 And, conversely, when Numb is overexpressed, the opposite cell fate transformation is observed
00:06:47.24 and four neurons are formed.
00:06:49.15 So, the levels of Numb seem to be responsible for specifying cell fates in
00:06:54.10 external sensory organs.
00:06:57.01 How this works was found by Michelle Rhyu, in the lab of Yuh Nung Jan,
00:07:01.19 a couple of years later.
00:07:03.11 Because what she found was that numb encodes for a protein that localizes asymmetrically
00:07:10.14 and segregates into one of the two daughter cells during mitosis.
00:07:16.10 This movie shows you the segregation of Numb during mitosis.
00:07:20.05 So, initially, Numb, which is shown in green here, is uniformly distributed around the
00:07:25.05 circumference of the cell.
00:07:26.22 But, as the cell goes into mitosis, you will see that Numb accumulates on one side of the
00:07:33.07 cell cortex.
00:07:34.07 Then, the mitotic spindle orients, the sister chromatids separate, and the protein is specifically
00:07:40.16 segregated into only one of the two daughter cells.
00:07:45.18 So, from this, it was thought that the asymmetric segregation of Numb specifies the sister cell
00:07:52.16 fates in the external sensory organ lineage.
00:07:55.11 But it was later shown that in other lineages, like muscle lineages or the central nervous
00:08:00.00 system, Numb actually has a very similar function.
00:08:02.21 So, when I saw this subcellular localization for the first time when I was a graduate student,
00:08:08.09 I was truly blown away by this.
00:08:11.13 And so I applied as a postdoc to the lab of Yuh Nung Jan and wrote a proposal where
00:08:16.03 I said, hey, I'm going to do a few experiments to uncov... to uncover the mechanism of asymmetric
00:08:22.01 Numb segregation.
00:08:24.07 I failed miserably, but I did find a few things about Numb localization,
00:08:29.02 and these are shown here.
00:08:31.05 The first thing we did was we simply analyzed the localization of Numb and its relation
00:08:37.15 to other cellular compartments.
00:08:39.01 And what we found was that, in interphase, Numb is uniformly distributed around the circumference
00:08:44.13 of the cell, but the localization is specific to mitosis, and that Numb always localizes
00:08:50.24 over one of the two spindle poles.
00:08:53.07 So, we thought there must be a connection between spindle orientation and Numb localization.
00:08:58.08 However, when we disrupted the mitotic spindle, Numb was still asymmetrically localized.
00:09:04.14 And, conversely, in a numb mutant, the mitotic spindle still orients perfectly fine.
00:09:11.01 And so we came up with this conceptual model for how asymmetric cell division could actually
00:09:16.01 be coordinated.
00:09:18.20 We thought there must be an axis of polarity that forms before a cell wants to undergo
00:09:25.02 an asymmetric cell division.
00:09:27.08 And then, in mitosis, this axis of polarity is used for two independent cellular events.
00:09:33.20 It's used to orient the mitotic spindle and it is used to localize cell fate determinants
00:09:39.22 asymmetrically.
00:09:40.22 And, together, the coordinated activity of those two events will lead to the establishment
00:09:46.04 of two different cell fates in the two daughter cells.
00:09:51.04 So, this is an interesting conceptual model, but what are the molecules behind this
00:09:58.05 molecular machinery?
00:10:00.20 The first molecular entry point into the machinery of asymmetric cell division came from the
00:10:05.14 discovery of a molecule that is called Inscuteable.
00:10:09.22 Inscuteable is interesting because of its expression pattern.
00:10:12.15 So, this slide shows you an early Drosophila embryo, and Inscuteable is shown in blue,
00:10:18.03 and you can see that it's specifically expressed in this region in the head.
00:10:21.23 Now, what is special about this region becomes clear if we look at this beautiful map of
00:10:27.24 the pattern of mitosis during Drosophila embryogenesis that was generated by Victoria Foe.
00:10:34.15 The Inscuteable expression area corresponds to what Victoria Foe called Domain 9.
00:10:40.00 And Domain 9 is special in the orientation of its division pattern; while all other domains
00:10:47.16 divide symmetrically with a mitotic spindle that is oriented parallel to the surface of
00:10:54.04 the epithelium, Domain 9 reorients its mitotic spindle, the mitotic spindle assumes an apical-basal
00:11:02.03 orientation, and the cells divide asymmetrically to generate the very first neurons that are
00:11:06.24 formed during Drosophila development.
00:11:09.06 So, the expression of Inscuteable correlates with asymmetric cell division and with vertical
00:11:16.13 spindle orientation.
00:11:17.22 And this is interesting and, so, together with Rachel Kraut in the lab of Bill Chia,
00:11:23.20 who actually identified inscuteable, we set out to analyze its mutant phenotype, and this
00:11:29.07 is summarized here.
00:11:32.12 Normally, Domain 9 divides with a vertical mitotic spindle.
00:11:38.01 So, this shows you a top view onto the epidermis of a Drosophila embryo and you can see the
00:11:44.18 most... the mitotic spindles, that the centrosome is pointing up and the microtubules are going
00:11:50.20 towards the inside of the embryo.
00:11:52.17 In an inscuteable mutant, however, this is different.
00:11:55.14 Now, the mitotic spindles are all parallel to the surface.
00:11:59.15 So, Inscuteable is required for the reorientation of the mitotic spindle in mitotic Domain 9.
00:12:06.19 But Inscuteable is not only required, it's also sufficient for a spindle reorientation.
00:12:11.12 This slide... this image shows you mitotic spindles outside Domain 9 and what you can
00:12:17.11 see is that they're all parallel to the surface.
00:12:20.00 But, if we ectopic express the inscuteable gene in this area, now this mitotic spindles
00:12:26.12 reorient into an apical-basal direction.
00:12:29.24 So, Inscuteable is not only required, it's also sufficient for introducing an apical-basal
00:12:36.15 mitotic spindle orientation.
00:12:39.10 So, Inscuteable provided a molecular entry point into identifying the machinery for asymmetric
00:12:46.15 cell division.
00:12:47.20 And so we started to search for its binding partners.
00:12:51.20 And we found one key binding partner, which is illustrated on this slide.
00:12:57.13 We did this through biochemistry, and we did a domain analysis of the Inscuteable protein
00:13:03.08 and expressed several fragments, some of which retain the functions and others
00:13:08.22 do not retain the function.
00:13:10.15 And then, we pulled on the tag... with which those transgenes were tagged, and found that
00:13:17.15 the functional but not the non-functional form brought down two protein bands: one around
00:13:24.23 40 kiloDaltons and one around 70 kiloDaltons.
00:13:29.22 So, we then used mass spectrometry to identify the molecular identity of those protein bands
00:13:36.23 and found that the 40 kiloDalton band corresponds to a heterotrimeric G protein alpha subunit,
00:13:44.06 whereas the 70 kiloDalton band was a protein that was called partner of inscuteable,
00:13:50.23 or Pins.
00:13:52.09 And this protein was independently identified in the laboratory of Bill Chia in a two-hybrid
00:13:57.06 screen for Inscuteable binding proteins.
00:13:59.06 And the function of these proteins is shown on the next slide.
00:14:03.11 It turns out that Pins acts as a receptor-independent activator of heterotrimeric G proteins.
00:14:12.08 Now, Pins belongs to a family of proteins that are characterized by a so-called
00:14:17.17 GoLoco domain.
00:14:21.02 Heterotrimeric G proteins normally are activated when ligands bind to
00:14:26.23 a seven-transmembrane receptor.
00:14:29.06 This leads to the exchange of GDP for GTP, the dissociation of the beta-gamma subunit,
00:14:35.11 and then either the GTP-bound alpha subunit or the free beta-gamma subunit can transduce
00:14:41.04 the signal.
00:14:42.10 In the presence of Pins, or its mammalian homolog AGS, however, without any nucleotide
00:14:49.04 exchange, the beta gamma-subunit is separated and the G proteins are activated.
00:14:56.13 So, Pins and... in... and the heterotrimeric G protein alpha subunits are functional binding partners
00:15:05.06 of Inscuteable.
00:15:06.16 And the mutant analysis showed that both of them are required for Inscuteable to perform
00:15:12.09 its activity during asymmetric cell division.
00:15:15.22 So, how can those three proteins act together to orient mitotic spindles?
00:15:21.03 A hint is... as to how this could work came from the identification of yet another binding
00:15:26.07 partner, which is called Mud in Drosophila or NuMA in mammals.
00:15:31.22 NuMA and Mud are microtubule binding proteins that are also part of the Inscuteable/Pins/G alpha
00:15:41.12 protein complex.
00:15:42.23 And the way we think they act is that, normally, Mud and NuMA are on the lateral side of an
00:15:51.14 epithelial cell, and this leads to a horizontal orientation of the mitotic spindle.
00:15:58.10 But, when Inscuteable is expressed, now Mud and P... and Pins and G alpha are recruited
00:16:05.09 to the apical cell cortex.
00:16:07.15 This generates a microtubule attachment site on the apical cell cortex and lead leads to
00:16:13.24 an apical-basal mitotic spindle orientation.
00:16:17.21 So, taken together, Inscuteable, Pins, G alpha, and Mud/NuMA provide a microtubule attachment
00:16:25.17 site that is sufficient to orient the mitotic spindle in a particular direction.
00:16:32.20 So, how does Inscuteable know where to localize?
00:16:38.03 Why is it localizing on the apical side and not on the basal side?
00:16:44.03 A hint as to how this could happen came from a comparison of other model systems that are
00:16:50.18 also modeling asymmetric cell division.
00:16:53.09 One of the most famous model systems for asymmetric cell division is the worm C elegans.
00:16:59.01 Beautiful work from the laboratory of Tony...
00:17:00.17 Tony Hyman has actually shown that, during early C elegans development, the mitotic spindle
00:17:06.22 is pulled towards the posterior side of the zygote, and this leads to an unequal cleavage
00:17:13.23 into a small posterior and a larger anterior cell.
00:17:18.19 And work from the laboratory of Ken Kemphues has used this as a screening tool and identified
00:17:25.14 a set of proteins that are commonly referred to as the Par proteins.
00:17:30.11 These Par proteins -- Par-1, Par-2, 3, 4, 5, and 6 -- occupy opposite domains on the
00:17:39.03 early C elegans embryo.
00:17:41.20 And in their absence both mitotic spindle and cell fate become equal.
00:17:49.04 Evolutionary considerations have shown that the Par proteins are essentially involved
00:17:53.07 in almost any event that involves cell polarization.
00:17:57.14 And this goes from asymmetric cell division, to epithelial polarity, to the distinc...
00:18:04.07 the distinction between axons and dendrites, and many others... the polarization of an
00:18:09.16 oocyte and many other polarity events.
00:18:12.21 And they're also involved in asymmetric cell division.
00:18:16.01 The Par proteins are conserved in evolution and the most important Par proteins are Par-3,
00:18:22.05 or Bazooka in Drosophila, Par-6, and an atypical protein kinase C called aPKC.
00:18:31.19 And these three proteins form an evolutionarily conserved protein complex.
00:18:37.07 In Drosophila, this protein complex is always located at the apical side of epithelial cells
00:18:44.06 and of asymmetrically dividing cells.
00:18:47.19 In the absence of Inscuteable, however, the Par proteins are not linked to the spindle
00:18:54.00 orientation Pins/G alpha/Mud complex.
00:18:57.06 And so the Par proteins are on the apical side, the spindle orientation complex is on
00:19:02.01 the lateral side, and for this reason mitotic spindles are horizontal and the cells do not
00:19:07.14 divide asymmetrically.
00:19:09.13 When we turn on Inscuteable, however, Inscuteable binds to the Par protein complex, it goes
00:19:15.13 to the apical cell cortex, it will recruit Pins/G alpha/Mud to the apical cortex, and
00:19:21.03 this will lead to a reorientation of the mitotic spindle into an apical-basal direction.
00:19:27.21 So, the Par proteins determine the asymmetric localization of Inscuteable.
00:19:34.01 So, from this, we have a pretty good molecular understanding of the orientation of mitotic
00:19:42.00 spindles, but how do the determinants segregate asymmetrically?
00:19:46.12 A hint as to how this could happen came from beautiful work that was done in the laboratory
00:19:53.01 of Francois Schweisguth, who showed that one of the Par proteins, the atypical protein
00:19:59.00 kinase C, can phosphorylate the Numb protein.
00:20:03.09 Numb is recruited to the plasma membrane through a set of positively charged amino acids at
00:20:08.22 its N-terminus, and what aPKC will do is it will phosphorylate the Numb protein at this
00:20:15.05 N-terminus.
00:20:16.05 This will neutralize the charges and the Numb protein will form... fall into the cytoplasm.
00:20:22.18 So, during interphase, the Numb protein is uniformly distributed around the circumference
00:20:30.22 of a cell.
00:20:31.22 And this is because aPKC does not phosphorylate Numb.
00:20:36.24 In order for aPKC to phosphorylate Numb, a kinase, a mitotic kinase called Aurora A,
00:20:43.06 needs to become active, it will phosphorylate the Par-6 subunit of the Par protein complex,
00:20:51.19 and this will trigger aPKC to phosphorylate Numb.
00:20:55.13 And, as aPKC is only present on the apical side, Numb will be phosphorylated on the apical
00:21:01.12 side, on the apical side it will fall off the cell cortex, and for this reason it will
00:21:07.07 be concentrated on the basal side.
00:21:10.15 So, asymmetric phosphorylation of Numb is the principle of its asymmetric localization.
00:21:17.16 So, with this, we have assigned molecules to each and every step
00:21:23.23 of the asymmetric cell division machinery.
00:21:26.19 I told you that polarity in cells, before mitosis, is set up by the asymmetric localization
00:21:35.02 of the Par proteins to the apical side of the pre-divisional cell.
00:21:39.19 And the cell goes into mitosis, the Par proteins will then recruit, via the adapter protein
00:21:46.06 Inscuteable, Pins, G alpha, and Mud to the apical side, and this will direct the orientation
00:21:52.08 of the mitotic spindle.
00:21:53.19 At the same time, the kinase aPKC in the Par protein complex, which will phosphorylate
00:21:59.07 Numb on one side but not on the other side of the cell, and this will introduce its
00:22:04.21 asymmetric localization.
00:22:06.11 And, together, these events lead to the asymmetric segregation of the Numb protein into only
00:22:11.19 one of the two daughter cells.
00:22:14.02 So, with this, we have been able to assign molecules to each and every step of this postulated
00:22:19.14 mechanism for asymmetric cell division.
00:22:22.03 More recently, however, there have been some modifications of this model, for example,
00:22:26.07 it is now known there is a second spindle orientation complex, which is called the telophase
00:22:33.00 rescue pathway, that involves a kinesin and the protein Discs-large, that acts redundantly
00:22:39.03 with Inscuteable/Pins/G alpha.
00:22:41.03 And we also know that the crystal structure of Inscuteable suggests that there is a complex
00:22:48.11 exchange between these proteins during microtubule cortical attachment.
00:22:53.02 But, by and large, this model explains how cells divide asymmetrically.
00:22:59.15 Now, the key question of course is, how much of this is actually conserved when we move
00:23:05.17 from Drosophila to vertebrates?
00:23:07.15 Let me remind you, again, that in the mammalian cortex the progenitor cells initially divide
00:23:14.10 symmetrically to generate more progenitors, and only later they assume an asymmetric division
00:23:20.09 mode, where they generate one progenitor and one neuron, or one progenitor and one intermediate
00:23:26.16 progenitor, which then will generate two neurons.
00:23:30.16 Now, as it turns out, the orientation of these divisions is slightly different; while the
00:23:37.05 early, symmetric divisions occur with a mitotic spindle that's precisely parallel to the ventricular
00:23:43.11 surface, these later, asymmetric divisions occur with a mitotic spindle
00:23:48.02 that is slightly tilted.
00:23:49.19 And so, early hypotheses for how asymmetric cell division could be guided assumed that,
00:23:56.19 because the mitotic spindle is no longer horizontally oriented, this would lead to an asymmetric
00:24:01.01 distribution of cell fate determinants and would trigger asymmetric cell division.
00:24:05.22 However, beautiful work, particularly from the lab of Fumio Matsuzaki, has shown that
00:24:11.03 the number of cells that have an oblique mitotic spindle is far too low to explain the high
00:24:16.23 fraction of asymmetric cell division.
00:24:19.11 And to put this to a final test, my lab has generated a knockout mouse that contains
00:24:26.11 a mutation in the single homologue of the Inscuteable gene,
00:24:31.08 and the phenotype that we saw is shown here.
00:24:36.08 Normally, during the late stages of development, mitotic spindles are oblique.
00:24:43.16 But, when we knock out Inscuteable, as expected, now all the mitotic spindles become flat.
00:24:50.04 But, instead of a reversion of asymmetric to more symmetric divisions, we saw a different
00:24:56.02 phenotype and this is illustrated here.
00:24:58.22 In a normal mouse, the progenitor cells can undergo either direct or indirect neurogenesis.
00:25:05.12 During indirect neurogenesis, the intermediate progenitors would generate
00:25:08.23 two differentiating neurons.
00:25:11.13 But what you see when you turn the mitotic spindle, and make it more flat, is a bias
00:25:16.07 towards direct neurogenesis.
00:25:18.10 So, unlike in Drosophila, where Inscuteable decides between symmetric and asymmetric cell
00:25:23.18 division, in the mouse it decides between direct or indirect neurogenesis.
00:25:30.14 But the most interesting aspect of the Inscuteable phenotype comes from an overexpression experiment,
00:25:36.23 and this is illustrated here.
00:25:38.12 So, we also generated a mouse that we can use to overexpress the inscuteable gene in
00:25:43.10 the developing mouse cortex.
00:25:45.18 And what we then find is that the number of neurons that are generated is vastly exceeded.
00:25:52.05 So, there's many more neurons that are generated.
00:25:56.02 And this is through a mechanism that is very interesting from an evolutionary point of
00:26:01.09 view and is illustrated here.
00:26:04.19 The overexpression of Inscuteable leads to a vast enhancement of oblique or even
00:26:11.07 vertical mitotic spindles.
00:26:13.04 And, as a result, some of the daughter cells no longer reside in the ventricular zone.
00:26:18.16 Instead, they become progenitor cells which now reside outside the ventricular zone.
00:26:26.12 And such progenitor cells are called outer radial glia cells or, by other labs,
00:26:30.24 basal radial glia cells.
00:26:33.03 Normally, this type of progenitors does not occur in the mouse, or only in very low numbers.
00:26:39.12 Instead outer radial glia of basal radial glia cells are typical for the primate cortex,
00:26:46.03 and this is shown here.
00:26:48.00 So, in rodents, the cortex contains the ventricular zone, there's an area that is called the subventricular
00:26:55.11 zone, which contains the intermediate progenitors, and then there is the cortical plate.
00:27:00.17 But, in the human cortex, there is an additional layer,
00:27:05.01 which is called the outer subventricular zone.
00:27:08.19 And this layer contains a cell type that is called the outer radial glia cells.
00:27:13.16 And these outer radial glia cells act as a transit amplifying population.
00:27:18.04 So, in mice, per progenitor division, either one neuron or two neurons are generated.
00:27:26.05 But, in humans, the progenitors can generate these outer radial glia cells, which then
00:27:33.11 continue to divide asymmetrically, generating hundreds of intermediate progenitors and hundreds
00:27:39.11 of neurons.
00:27:40.11 Well, this is good news, because it explains why we have so many more neurons than rodents,
00:27:46.17 but it's also bad news because it means that we can no longer use the mouse as a model
00:27:52.17 system to study this aspect of asymmetric cell division.
00:27:57.23 And, of course, the Inscuteable overexpression phenotype suggests a very provocative possibility,
00:28:04.00 which would be that enhanced levels of Inscuteable would be responsible in part I generated...
00:28:10.23 for generating these outer radial glia cells.
00:28:14.06 Now, in order to do... to test this hypothesis, we would actually have to remove Inscuteable
00:28:19.15 from a human brain, but of course this experiment is not possible.
00:28:23.02 So, how can we address the development of a cortex in a human or a primate setting?
00:28:31.05 In order to do so, my lab has more recently generated three-dimensional culture methods,
00:28:37.06 which we now can use to model human cortical development in the lab.
00:28:42.00 So, this is one example of what we call a cerebral organoid.
00:28:46.07 There is a cortical tissue, here.
00:28:47.24 Here, you actually see the development of a human eye.
00:28:51.22 Here is another example.
00:28:55.04 This is cortical tissue, you can see the development of a ventricular zone, here.
00:28:59.11 Here is a lateral ventricle.
00:29:01.08 And here you see differentiating neurons, which are beginning to form a cortical plate.
00:29:06.16 Here's other areas of the human brain that formed different brain regions, and that have
00:29:14.16 slightly different morphologies.
00:29:16.21 And we can generate these organoids from pluripotent stem cells that we can generate from essentially
00:29:23.05 any patient, either healthy or suffering from a neurodevelopmental disorder.
00:29:28.24 And we've already used them to model some of the major... some of... some of the more
00:29:33.23 severe neurodevelopmental disorders, like microcephaly.
00:29:38.16 And we can now use them to actually analyze brain development and the role of asymmetric
00:29:45.14 cell division in humans.
00:29:47.14 And, in the second lecture, I will tell you more about how we generate these organoids
00:29:53.23 and about the results that we have obtained with them.
00:29:58.11 So, in the end, I would just like to summarize that I've shown you that asymmetric cell division
00:30:05.01 can occur through the asymmetric segregation of protein determinants.
00:30:10.08 I've also shown you that Inscuteable is a molecular trigger for the orientation of the
00:30:15.18 mitotic spindle.
00:30:17.15 I've shown you that Inscuteable acts by binding to its interaction partner, Pins, and,
00:30:24.09 with that, through a heterotrimeric G protein alpha subunit.
00:30:27.13 I've shown you that the Par proteins are important for setting up the axis of polarity during
00:30:32.02 asymmetric cell division.
00:30:33.12 And, finally, I've shown you that Inscuteable regulates spindle orientation in the mouse
00:30:38.21 cortex, but in the mouse this only regulates the balance between direct and indirect neurogenesis.
00:30:45.02 And, finally, I've shown you that the human cortex contains progenitors that you do not
00:30:51.16 run into... in very small numbers, find in rodents, but that we can generate organoids
00:30:56.24 from human pluripotent stem cells that we can use to study human cortical development
00:31:02.15 in cell culture.
00:31:03.20 So, in the end, I would just like to acknowledge people who actually contributed to this.
00:31:08.14 I would like to thank all the members of my lab, whether in past or present, who have
00:31:14.22 contributed to these various results that elucidated mechanisms of asymmetric cell division.
00:31:20.24 I mentioned particularly the work of Matthias Schaefer, who identified Pins and G alpha,
00:31:27.11 Markus Schober, who found the role of the Par proteins,
00:31:31.01 Mark Petronczki, who found the role of Par-6,
00:31:34.24 Sarah Bowman found the Mud protein,
00:31:37.08 and Frederik Wirtz-Peitz and Takashi Nishimura identified the mechanism
00:31:42.03 for Numb localization.
00:31:43.15 The inscuteable knockout was generate by Maria-Pia Postiglione,
00:31:47.08 and the cerebral organoids were generated by Madeline Lancaster.
00:31:51.12 I would like to thank the Austrian Academy of Sciences, the European Research Council,
00:31:56.08 the Austrian Research Fund, and EMBO for funding my research.
00:31:59.19 And my entire lab for being a great group that is fun to work with.

Part 2: Asymmetric Cell Division; From Drosophila to Humans

00:00:13.04 Hello.
00:00:14.04 My name is Juergen Knoblich.
00:00:15.04 I am a scientist at the Institute of Molecular Biotechnology of the Austrian Academy of Sciences
00:00:21.00 in Vienna.
00:00:22.00 And, in this second part of my lecture, I would like to tell you about how we can recapitulate
00:00:28.06 human brain development, in the lab, starting from pluripotent stem cells, using three-dimensional
00:00:35.05 culture methods.
00:00:36.05 Now, I'm sure every one of you would agree with me that the human brain is the most complex
00:00:43.06 but also the most fascinating structure that nature has generated.
00:00:47.08 It contains about 87 billion neurons that have to be born at the right time, migrate
00:00:52.21 to the right position, and wire up in the right way in order to allow us to perform
00:00:56.24 the cognitive processes that we're able to do.
00:01:00.06 But despite its complexity, the brain develops from a limited number of stem and progenitor cells
00:01:07.19 in a set of predefined rules.
00:01:10.14 And essentially everything that we know about those rules comes from experiments that were
00:01:14.13 done in rodent model systems, particularly in the mouse.
00:01:18.16 So, this shows you a very simplified view of mouse cortical development.
00:01:24.24 The mouse neocortex develops from a neural epithelium that lines a liquid-filled cavity,
00:01:31.08 which is called the lateral ventricle.
00:01:33.21 It's a polarized epithelium with basal to the outside, apical to the inside, and all
00:01:39.06 the different cell types in this epithelium are nicely arranged in a layer-type fashion
00:01:43.19 along the apical-basal axis of the epithelium.
00:01:46.17 So, here, on the more basal side, is the so-called cortical plate, and this is where the neurons
00:01:51.17 are and where they send out their axons.
00:01:53.16 And, here, on the more [apical] side, is the so called ventricular zone, and this is where
00:01:58.16 the stem and progenitor cells reside.
00:02:01.05 And these progenitor cells undergo either one of three different types of division.
00:02:05.16 Early during development, they divide symmetrically and this leads to an initial amplification
00:02:10.23 of the progenitor pool.
00:02:12.11 Later, they divide asymmetrically, where one progenitor generates one progenitor cell and
00:02:17.24 one cell that migrates out to become a neuron.
00:02:21.00 Or, the other cell forms a so-called intermediate progenitor, which divides once more into two
00:02:27.03 terminally differentiated neurons.
00:02:29.01 And this is called direct or indirect neurogenesis.
00:02:33.15 Now, when we move from rodents to humans, there is two characteristic differences.
00:02:40.01 First of all, the initial set of symmetric divisions lasts a lot longer
00:02:47.02 in humans and in primates.
00:02:49.16 This is a rodent cortex in... a rodent brain and a primate brain at about... at a comparable
00:02:57.17 stage of development, magnified to a comparable size.
00:03:02.01 And what you can see is that, while the more posterior brain parts are actually very similar,
00:03:07.17 the cortex is vastly expanded in the primate brain.
00:03:11.16 And this is because these cells had a lot more time to divide symmetrically and to amplify.
00:03:18.14 The second big difference is that, in addition to the ventricular zone and the cortical plate,
00:03:26.15 the primate brain contains an additional layer that is called the outer subventricular zone.
00:03:33.16 The outer subventricular zone is characteristic for primate brains and it contains a cell
00:03:38.24 type that is called outer radial glia cells or basal radial glia cells.
00:03:44.05 These cells are not, or in only very small numbers, present in rodent brains.
00:03:49.19 But, in primates, they act as a transit amplifying population during neurogenesis.
00:03:55.21 So, while in the rodents, per progenitor division, either one neuron or two neurons are generated,
00:04:04.02 in the primates, the progenitors generate these outer radial glia cells, which continue
00:04:09.12 to divide asymmetrically, generating hundreds of intermediate progenitors
00:04:14.03 and hundreds of neurons.
00:04:16.05 So, this is good news because it explains why we have so many more neurons than a mouse,
00:04:21.14 but it's also bad news because it means that there are certain brain developmental processes
00:04:25.19 that cannot be modeled in a rodent model system.
00:04:30.02 And this is particularly emphasized... empha... emphasized on this slide.
00:04:35.17 This shows you an MRI scan of a patient who suffers from a very severe neurodevelopmental
00:04:41.03 disorder that is called microcephaly.
00:04:43.19 What you can see is that the patient's brain is much smaller than that of a healthy patient.
00:04:49.11 This patient carries a mutation in a gene that is called Nde1.
00:04:53.03 Nde1 is conserved all the way through evolution, from yeast to humans, but when you make a
00:05:00.06 mutation in a mouse in Nde1 you see no, or only a very weak, phenotype.
00:05:06.15 So, certain neurodevelopmental disorders cannot properly be modeled in rodent models.
00:05:13.23 And, for this reason, there have been many attempts to actually model brain development
00:05:19.12 in a human setting.
00:05:20.15 So, how can we actually do this?
00:05:24.11 The easiest way, of course, is to use human tissue.
00:05:28.12 Human cortical tissue can be obtained from aborted fetuses, it can be fixed,
00:05:33.14 it can be stained.
00:05:35.07 There... one can even do live imaging and cell tracing on this.
00:05:41.20 But, of course, such experiments place a huge experimental burden and ethical burden, and
00:05:48.08 are very complicated and usually suffer from low N numbers.
00:05:51.14 And, for this reason, people have been trying to model brain development starting from
00:05:56.12 pluripotent stem cells.
00:05:58.03 The easiest way of doing this is to turn the pluripotent stem cells into neural stem cells,
00:06:03.17 which can then be forced to undergo neuronal differentiation, and one can study the phenotype
00:06:10.01 in those patient-derived neurons.
00:06:12.07 But, of course, neurons like to be in a three-dimensional environment and so there's many limitations
00:06:19.18 to those two-dimensional experiments.
00:06:21.03 And so, there have been many attempts to actually model the development of human tissues in
00:06:27.05 three-dimensional culture.
00:06:28.16 And, really, the pioneer of this has been a Japanese scientist named Yoshiki Sasai.
00:06:36.19 Yoshiki Sasai, around 2012, was able to recapitulate the development of a human eye in culture.
00:06:49.16 And this was clearly a pioneering experiment in the field of
00:06:55.12 generating human tissues in culture.
00:06:58.12 Now, I think Yoshiki Sasai clearly was one of the leading developmental biologists of
00:07:04.19 our times and it's very sad that he's no longer with us.
00:07:08.15 And so, based on his work, but also adding other experimental approaches to it,
00:07:13.24 Madeline Lancaster, a couple of years ago, in my lab, has decided to develop an in vitro three-dimensional
00:07:22.18 culture system that we can use to model the development of a human cortex in culture.
00:07:27.24 And here's the method that she came up with.
00:07:30.14 We start with pluripotent human stem cells, which we dissociate and then rapidly reaggregate
00:07:36.23 on the bottom of a 96-well plate.
00:07:39.09 We then allow those cells to develop into embryoid bodies, giving them just enough time
00:07:45.18 to form the three germ layers.
00:07:47.15 We then replace the medium with a neural induction medium, so that only the neural ectoderm survives.
00:07:53.18 And these balls of neural ectoderm are then placed into droplets of Matrigel gel, which
00:07:58.19 is a collagenous 3-dimensional support matrix that supports the development of
00:08:06.00 stem cell-derived tissue.
00:08:08.06 We then culture these Matrigel droplets, first in floating culture and later either in a
00:08:13.10 spinning bioreactor or, more recently, in an orbital shaker.
00:08:17.18 And, over time, Madeline saw the development of fairly complex tissues.
00:08:22.19 So, these are two examples of what we call cerebral organoids.
00:08:27.07 Here is one example and you can see a developing human cortex, here.
00:08:32.07 Here is a lateral ventricle.
00:08:34.17 Here is another piece of cortical tissue.
00:08:37.06 And here is another one.
00:08:39.01 This is another example.
00:08:40.21 There is cortical tissue, here, and over here.
00:08:45.06 And down here you see the development of a human eye.
00:08:50.20 This is a cross-section through a cerebral organoid.
00:08:54.03 There's cortical tissue, here.
00:08:55.20 You see a ventricular zone, here.
00:08:58.04 Here is the lateral ventricle.
00:09:00.14 And here are the differentiating neurons.
00:09:03.24 Down here is what we call the choroid plexus, which is the area that generates the
00:09:09.02 cerebral spinal fluid.
00:09:10.19 And here are other brain areas -- in the absence of markers, I do not know what they are, but
00:09:15.24 I want you to note that they have a different histology, indicating that they are different
00:09:23.01 areas of the developing human brain.
00:09:25.14 So, cerebral organoids can be used to model the development of
00:09:30.09 various parts of the human brain.
00:09:33.08 But, for our analysis, we focused on the developing cortex, which is the most complex but also
00:09:39.05 the most fascinating part of our brain, and also the one that is most different between
00:09:43.16 us and rodents.
00:09:45.15 This is a cross-section through a cerebral organoid.
00:09:48.18 This is the cortical part.
00:09:51.02 In red are progenitors, in green are the differentiating neurons, and what you can see is that, at
00:09:54.23 this stage, the organoid histology is essentially indistinguishable from that of a developing
00:10:01.09 mouse cortex.
00:10:02.09 So, cerebral organoids can recapitulate the three-dimensional organization of a developing
00:10:08.07 human cortex.
00:10:10.03 We also analyzed neuronal differentiation.
00:10:14.04 The neurons in the cerebral organoids send out long axons that contain growth cones,
00:10:21.01 they often branch out, they can bundle together, and the neurons also send out
00:10:27.12 large dendritic trees.
00:10:29.06 But, most importantly, the neurons in our organoids are electrically active.
00:10:34.06 This is a calcium imaging experiment where you can see that the neurons spike action
00:10:40.01 potentials, and they communicate with each other, and the pattern of these electrical firings
00:10:46.08 is far from random.
00:10:48.01 And... and... and there are certain correlated neurons and anti-correlated neurons.
00:10:54.08 So, cerebral organoids recapitulate both the three-dimensional organization of the developing
00:11:01.02 human cortex, and proper neuronal differentiation and function.
00:11:06.14 Now, the neurons in the developing cortex actually come in different flavors.
00:11:11.15 They are arranged in a layer-type fashion and these layers are formed in an inside-out
00:11:17.17 manner, where the deep layers are formed first and then newly formed neurons have to migrate
00:11:23.23 through these deep layers, adding additional layers to the outside.
00:11:27.09 There are various markers for these layers -- SatB2 is a layer for the out... is a marker
00:11:31.09 for the outside neurons, Ctip2 for the inside.
00:11:35.03 When we stain early organoids, essentially all of the neurons are positive for the deep
00:11:40.05 layer marker, Ctip2.
00:11:42.06 A bit later, SatB2-positive neurons appear, initially they are intermingled, and then
00:11:46.23 they show some kind of a sorting out, although we do not really see the formation of proper
00:11:53.11 neuronal layers.
00:11:54.11 But, the temporal specification of various different neuronal subtypes can be recapitulated
00:12:00.21 in the cerebral organoids.
00:12:03.10 So, we've generated a three-dimensional cultural system that we can use to recapitulate the
00:12:09.06 development of the human cortex in culture.
00:12:11.14 So, what can we do with it?
00:12:13.13 In my view, we are currently experiencing a complete revolution in the way of how we
00:12:18.24 do biomedical research.
00:12:20.13 And this is because there are currently four technological developments that are
00:12:25.02 coming together.
00:12:26.11 The first one is the availability of more and more complete human genome sequences,
00:12:32.10 many of them associated with complete patient records.
00:12:36.00 The second one is our ability to generate pluripotent stem cells
00:12:40.07 from each of these patients.
00:12:42.21 And the third one is our ability to generate, in the lab, more and more different tissues
00:12:48.06 as organoids, that recapitulate particular organs in those patients.
00:12:52.23 And, finally, we can edit the genome and introduce or remove mutations from any of these
00:12:59.14 pluripotent stem cells.
00:13:00.21 And how we can combine these technologies to analyze human neurodevelopmental disorders
00:13:05.23 will be shown on the next couple of slides.
00:13:08.18 For this, we teamed up with Andrew Jackson, a pediatric neurologist at the University
00:13:13.12 of Edinburgh, who works with a patient who suffers from a severe form of microcephaly.
00:13:18.02 This is a MRI scan of the patient; you can see the brain is much smaller than that of
00:13:22.06 an age-matched healthy patient.
00:13:25.10 We obtained a biopsy from the patient, reprogrammed the cells into iPS cells, and then generated
00:13:31.05 organoids from them.
00:13:32.17 These are healthy, control organoids and patient-derived organoids.
00:13:37.06 And what you can see is that the control organoids contain a nice differentiated ventricular
00:13:44.15 zone, large ventricles, and many neurons, but the patient-derived organoids are much
00:13:49.17 smaller, they contain a lower number of neurons, and only a very tiny ventricular zone.
00:13:54.21 So, using our organoid system and patient-derived iPS cells, we can recapitulate the small-brain
00:14:01.07 phenotype of a microcephaly patient.
00:14:03.17 And we can now go back in history and ask, why are there so fewer neurons in these
00:14:10.02 microcephaly-derived organoids?
00:14:13.10 This is a control organoid and a patient-derived organoid at a much earlier stage of development.
00:14:19.18 At this early stage, in the control organoid, all the progenitor cells are still undergoing
00:14:25.16 the symmetric divisions, and neurogenesis has not started.
00:14:29.19 This is very different in the patient-derived organoids, where we can already see the appearance
00:14:34.02 of individual neurons.
00:14:35.14 And we conclude from this experiment, and many others that I don't have the time to
00:14:38.24 show you, that the cells have switched to an asymmetric division pattern in a premature
00:14:45.01 state, this leads initially to the formation of too many neurons at a too early stage,
00:14:51.02 but at the same time the progenitors are not sufficiently amplified.
00:14:54.15 And we believe that this is the reason for why those organoids are so much smaller.
00:14:59.13 So, premature neurogenesis and incomplete progenitor amplification are correlated with
00:15:05.18 the appearance of a microcephaly phenotype.
00:15:07.18 So, how can we explain this?
00:15:10.14 For this, we performed whole exome sequencing on the patient and we found that the patient
00:15:16.01 carries a compound heterozygous mutation in a gene that is called CDK5Rap2.
00:15:23.23 Both of these mutations introduce premature stop codons and, consistent with this, we
00:15:28.13 do not detect the CDK5Rap2 protein in the patient-derived cells.
00:15:33.19 So, what is CDK5Rap2?
00:15:36.19 CDK5Rap2 is a protein that originally has been identified in the fruit fly Drosophila,
00:15:41.09 where it was called centrosomin.
00:15:43.01 It's a centrosomal protein that is absolutely required for the correct orientation of the
00:15:48.15 mitotic spindle.
00:15:49.15 And, consistent with this, while in the control organoids, at an early stage, all the mitotic
00:15:56.01 spindles are precisely aligned with the ventricular surface, in the patient-derived organoids
00:16:01.01 they assume a more or less random orientation.
00:16:04.07 And so we believe that, during the early stages of cortical development, the orientation of
00:16:09.04 the mitotic spindle is very important to ensure the equal distribution of the apical and basal
00:16:15.23 lateral plasma membrane domains, and to make sure that both daughter cells can maintain
00:16:20.24 the progenitor fate.
00:16:23.09 In the patient-derived organoids, however, a tilt of the mitotic spindle no longer allows
00:16:28.12 the symmetric inheritance, so that some cells start undergoing neuronal differentiation
00:16:33.01 and only a few maintain the progenitor status.
00:16:37.04 So, a defect in spindle orientation, leading to a lineage defect, might be responsible
00:16:44.00 for the microcephaly phenotype.
00:16:46.11 So, this is where a normal human genetics analysis would stop; we've recapitulated the
00:16:51.21 disease, we've found a plausible gene, and a plausible mechanism.
00:16:56.15 But, now, genome editing tools allow us to unambiguously ask, was it really this mutation
00:17:04.04 that was responsible for the disease phenotype?
00:17:06.13 And for this, we used genome editing to repair one of the two premature stop codons.
00:17:12.13 This is a control organoid, a patient-derived organoid, and an organoid that's derived from
00:17:18.17 cells where we have repaired one of the two mutations.
00:17:22.15 And what you can see is that both the size defect, as well as the premature neuronal
00:17:27.08 differentiation, can be rescued.
00:17:30.00 So, cerebral organoids can be used to recapitulate neurodevelopmental disorders.
00:17:35.20 And they can also be used to unambiguously associate
00:17:39.20 particular mutations with those disorders.
00:17:43.18 Now, microcephaly is a very severe disorder and, ultimately, we would like to recapitulate
00:17:51.19 more subtle disorders like epilepsy or autism.
00:17:56.16 And the holy grail of neurodevelopmental disease modeling is a type of neurons that have a
00:18:03.05 very complicated developmental origin, and these are the GABAergic interneurons.
00:18:11.05 This is a schematic view of the developing cortex.
00:18:15.03 The cortex develops from the dorsal neuroepithelium and all the neurons that arise in the dorsal
00:18:22.21 neuroepithelium express an excitatory neurotransmitter, glutamate, and become excitatory neurons.
00:18:30.06 In addition to those neurons, there are inhibitory GABAergic interneurons, which arise from the
00:18:36.07 ventral part of the developing cortex.
00:18:38.21 They then migrate tangentially to go into the dorsal areas and integrate into the circuits
00:18:44.21 of the dorsal cortex.
00:18:46.08 And it is thought that migrating interneurons is what actually is absolutely essential for
00:18:55.20 creating functional neuronal circuits.
00:18:58.22 And the types of mutations that are associated with diseases like epilepsy or autism suggests
00:19:06.07 that defects in interneuron formation and migration
00:19:10.06 may be associated with many of these diseases.
00:19:13.13 And so we asked, can we actually model this very complex developmental event in our
00:19:20.04 cerebral organoids?
00:19:21.15 The first thing we wanted to know is, do we actually have ventral and dorsal cortex in
00:19:27.00 the organoids?
00:19:28.05 This is shown here.
00:19:30.07 The dorsal cortex can be identified, because the neurons and the intermediate progenitors
00:19:35.19 express a gene that is called Tbr2, whereas the ventral cortex, the lateral ganglionic eminence,
00:19:41.15 in this case, can be identified by the expression of Gsx2.
00:19:45.03 When we section an organoid, we find, typically, that the organoids contain both Tbr2-positive
00:19:52.11 dorsal regions and Gsx2-positive ventral regions.
00:19:56.20 So, both ventral and dorsal areas are actually present in our cerebral organoids.
00:20:01.08 So, can we use them to model interneuron migration?
00:20:05.05 This slide shows you a dorsal cortical area where the intermediate progenitors express
00:20:10.13 the dorsal marker Tbr2, and when we stain the adjacent section with a vGad, an interneuron
00:20:18.06 marker, and somatostatin, a marker of particular interneuron subtypes, we find that interneurons
00:20:24.05 are actually present in the dorsal cortical areas of the organoid.
00:20:30.05 So, interneuron migration from the ventral into the dorsal part can actually be modeled
00:20:35.21 in cerebral organoids.
00:20:37.13 But there's one key problem with this.
00:20:39.20 And this is illustrated here.
00:20:41.20 In the current protocol, organoids are a little bit like a car where the wheels are up, the
00:20:46.22 engine is up, the seat is in the back, the windshield is in the front... so, all the
00:20:51.05 individual parts are present, but they are in a random arrangement.
00:20:54.17 And so, if we make a section through such an organoid, we have to be very lucky to hit
00:21:00.00 both a dorsal and a ventral part, and to be able to model interneuron migration.
00:21:07.05 And so, we asked, can we actually introduce a polarity axis in organoids, for example,
00:21:14.03 a dorsal-ventral axis?
00:21:16.01 And so, in order to do this, Josh Bagley, a postdoc in my lab, set out to develop a
00:21:21.08 protocol where he separately generates dorsal organoids and ventral organoids.
00:21:28.06 And then he places these two organoids together into one droplet of Matrigel.
00:21:36.03 One organoid is labeled with RFP expression; another organoid is labeled with GXP...
00:21:41.00 GFP expression.
00:21:42.00 Initially, when the two organoids are co-embedded, they lie next to each other, but over time
00:21:48.10 they fuse with each other and, if I would not show you the red and green color, you
00:21:52.19 would not be able to distinguish the boundary between the two organoids.
00:21:57.05 So, we have developed a co-culture protocol that we can use to generate organoids where
00:22:03.02 one part is dorsal and the other part is ventral.
00:22:06.19 So, can we use this to model interneuron migration?
00:22:12.04 This slide shows you a typical dorsal-ventral fusion organoid, and what you can see is that,
00:22:17.16 within the dorsal area, you see these green spots.
00:22:22.05 When we focus in on this, we find that these green areas are actually full of cells that
00:22:28.10 have the migratory appearance of interneurons.
00:22:31.21 When we fuse dorsal with dorsal organoids, we no longer see this.
00:22:36.15 And so, from this experiment and many others that I don't have the time to show you, we
00:22:40.19 conclude that, when we fuse ventral with dorsal organoids, there are cells that are migrating
00:22:47.04 from the ventral into the dorsal part of the organoid.
00:22:50.23 So, what are those cells?
00:22:53.07 This slide actually summarizes a very large number of experiments and I only show you
00:22:57.20 the key experiments.
00:22:59.12 Here, I show you that the migratory cells express GFP and all of them also express GAD1,
00:23:08.02 which is a marker for developing interneurons.
00:23:11.24 This is a high magnification view and you can actually see the elaborate cell shapes
00:23:17.01 that these develop... these migrating interneurons actually have.
00:23:20.15 We also use many other markers of migrating interneurons and various interneuron subtypes,
00:23:27.11 like parvalbumin, somatostatin, neuropeptide Y and also calbindin and calretinin, and we
00:23:34.05 find all of these subtypes in our migrating fused organoids.
00:23:39.12 So, taken together, this tells us that we can use fused organoids, where we have regenerated
00:23:46.12 a dorsal-ventral axis, to model the long-range interactions of various parts of the human
00:23:52.22 brain, in particular the migration of cells from one part of the brain into the other.
00:23:59.19 Now, of course, ideally, we would like to image this process of interneuron migration
00:24:05.07 in real time.
00:24:06.11 And, in order to do so, Josh developed a method that we can use to do long-term imaging of
00:24:13.01 migrating interneurons and this is shown here.
00:24:15.20 For this, Josh makes very thick slices from fused organoids and those slice... slices
00:24:23.17 are then imaged on a spinning disk time-lapse microscope for about three days.
00:24:29.07 This is a still of one of these migratory movies and what you can see is that we can
00:24:34.24 actually see migrating cells that go from one part of the organoid into the other.
00:24:41.17 This slide shows you a movie of migrating interneuron.
00:24:46.19 The interneuron can be seen here; here is the cell body and here are two processes.
00:24:52.15 I would like you to focus on this one cell and you will see that it migrates across the
00:24:57.09 entire field of the movie.
00:25:00.10 And interneuron migration is one of the most complex but, in my view, also both beautiful
00:25:06.00 processes that occur during development.
00:25:09.03 You can see the cell extends many processes, then some of them are retracted, and it migrates
00:25:14.11 along other processes.
00:25:17.10 The second movie shows you a more realistic view.
00:25:20.23 It shows you how crowded the field is.
00:25:23.06 There are many different migrating interneurons.
00:25:27.11 I want you to focus on this one.
00:25:31.00 And you will be able to see that this cell actually migrates across the entire field
00:25:35.20 of the image field.
00:25:38.01 So, here, you can see that the cell extends different processes, some of the processes
00:25:44.05 are then retracted.
00:25:45.15 It makes decisions to actually follow processes which initially are weaker.
00:25:52.12 And so, with this, we can now use organoids to model long-range interneuron migration,
00:26:00.22 and we're in the process of deriving iPS cells from patients that actually suffer from neurodevelopmental
00:26:07.06 disorders to see whether interneuron migration is affected.
00:26:10.22 Now, it is known that the migration of interneurons is in response to a chemokine that is called
00:26:21.01 CXCL12.
00:26:23.18 And the receptor for this chemo... for this chemokine is CXCR4.
00:26:29.03 And there is a drug called AMD3100, or Plerixafor, which actually is an inhibitor of
00:26:36.20 this chemokine receptor.
00:26:38.06 And so we asked, is interneuron migration that we actually see in our cultures dependent
00:26:44.13 on this receptor?
00:26:46.00 This is a normal organoid, where you can see a lot of interneurons migrating from one into
00:26:50.23 the other part, and this is an organoid that has been cultured in the presence of this
00:26:55.12 CXCR4 inhibitor.
00:26:57.18 And what you can see is that the migration of the interneurons is
00:27:01.19 actually almost completely suppressed.
00:27:04.08 I want you to note that the growth of axons from one part of the organoid into the other
00:27:09.08 is actually not affected, indicating that the other part of the organoid is actually
00:27:14.00 still alive.
00:27:15.04 So, we can use fused organoids to model interneuron migration and also to test chemical compounds
00:27:22.17 for their effect on interneuron migration.
00:27:25.14 And we are currently generating various models for neurodevelopmental disorders that we can
00:27:33.16 use to then test whether interneuron migration is affected.
00:27:38.24 So, in the end, I would like to summarize what I've told you.
00:27:42.24 I've told you that we've generated a three-dimensional culture system that we can use to recapitulate
00:27:49.20 early human brain development in culture.
00:27:52.19 Our cerebral organoids generate various parts of the human brain that can actually functionally
00:28:00.07 interact with each other.
00:28:01.12 We can use the organoids to model neurodevelopmental disorders and we can generate various parts
00:28:08.10 of the human brain separately in culture, then fuse them together and the interactions
00:28:14.05 between these various parts are maintained.
00:28:16.23 And, particularly, the last part of this of course has huge potential.
00:28:22.02 We will be able to generate any pairs of brain regions and watch the migration of cells from
00:28:30.01 one into the other, or a migration of axons, and hopefully eventually even be able to visualize
00:28:36.08 the formation of some of the major axon tracts.
00:28:39.15 In the end, I would like to acknowledge people who actually contributed to this work.
00:28:43.22 The hero of the organoids is Madeline Lancaster.
00:28:47.12 She was a postdoc in my lab, now has her own lab at the LMB in Cambridge.
00:28:53.01 For looking at dorsal-ventral patterning, she worked together with a grad student,
00:28:56.04 Magdalena Renner.
00:28:57.23 The fusion organoid system was generated by Josh Bagley, another postdoc in my lab.
00:29:03.07 These are the names of other people who work together in these various projects.
00:29:08.09 The reprogramming of the cells was a collaboration with the lab of Joseph Penninger.
00:29:12.07 I would particularly like to thank the patient and their family for allow... for allowing
00:29:16.11 us to use the cells in our experiments.
00:29:21.21 And our collaborator at the University of Edinburgh is Andrew Jackson.
00:29:25.21 I would like to thank the Austrian Academy of Sciences for funding our work.
00:29:30.17 In addition, we obtained funding from the European Research Council, the Austrian Science
00:29:35.16 Fund, and from EMBO.
00:29:36.24 And I would like to thank my entire lab for being such a fantastic group.

Videos in this Talk
  • Part 1: Asymmetric Cell Division; From Drosophila to Humans
    Part 1: Asymmetric Cell Division; From Drosophila to Humans
    Audience:
    • Researcher
    • Educators of Adv. Undergrad / Grad
  • Part 2: Asymmetric Cell Division; From Drosophila to Humans
    Part 2: Asymmetric Cell Division; From Drosophila to Humans
    Audience:
    • Researcher
    • Educators of Adv. Undergrad / Grad
Speaker: Jürgen Knoblich
Total Duration: 1:02:22
Recorded: May 2017
All Talks in Development and Stem Cells

Talk Overview

Dr. Knoblich begins his talk by explaining the key role that asymmetric cell division plays in development of the human brain.  During mammalian brain development, neuronal progenitor cells initially divide symmetrically to increase their numbers.  Later they divide asymmetrically to produce one progenitor cell and one (or two) cells which will terminally differentiate to become neurons.  What determines which daughter cell will become which? Working in Drosophila, Knoblich and others elucidated a signaling pathway in which Par proteins are asymmetrically localized before cell division. This recruits a complex of proteins which defines the orientation of the mitotic spindle and causes the localization of Numb protein at one pole of the cell. Upon cell division, only one daughter cell will inherit Numb protein and this ultimately will determine the fate of the daughter cells.  Interestingly, this signaling pathway is conserved from insects to mammals, however, Knoblich found an important difference that may explain why humans have many more cortical neurons than mice.

In his second talk, Knoblich describes experiments in his lab to develop 3 dimensional brain organoids from human pluripotent stem cells. While studying the development of rodent brains has proved extremely useful, there are some important developmental differences that require human tissue for investigation.  In addition, some diseases such as microcephaly cannot be modeled in mice.  Knoblich and his colleagues have developed cerebral organoids that mimic early human brain development and can be used to model neurodevelopmental disorders.  They have also been able to generate separate organoids from various regions of the human brain and then fuse them and follow the migration of live neurons between these parts, opening the path to many more studies of neuronal development.

Speaker Bio

Jürgen Knoblich

Jürgen Knoblich

Jürgen Knoblich is a senior scientist and deputy director of the Institute of Molecular Biotechnology of the Austrian Academy of Sciences (IMBA) and an Adjunct Professor at the Medical University of Vienna.  Knoblich’s lab is interested in understanding how the complexity of the human brain is generated from progenitor and stem cells during development. To… Continue Reading

Playlist: Cell Division

Related Resources

Knoblich JA, Jan LY, Jan YN. (1995) Asymmetric segregation of Numb and Prospero during cell division. Nature. 377(6550):624-7.

Chenn A, McConnell SK. (1995) Cleavage orientation and the asymmetric inheritance of Notch1 immunoreactivity in mammalian neurogenesis. Cell. 82(4):631-41.

Kraut R, Chia W, Jan LY, Jan YN, Knoblich JA.(1996) Role of inscuteable in orienting asymmetric cell divisions in Drosophila. Nature. 383(6595):50-5.

Knoblich JA, Jan LY, Jan YN. (1997)  The N terminus of the Drosophila Numb protein directs membrane association and actin-dependent asymmetric localization. Proc Natl Acad Sci U S A. 94(24):13005-10.

Schaefer M, Shevchenko A, Shevchenko A, Knoblich JA. (2000) A protein complex containing Inscuteable and the Galpha-binding protein Pins orients asymmetric cell divisions in Drosophila. Curr Biol. 10(7):353-62.

Bowman SK, Neumüller RA, Novatchkova M, Du Q, Knoblich JA. (2006) The Drosophila NuMA Homolog Mud regulates spindle orientation in asymmetric cell division. Dev Cell.10(6):731-42.

Wirtz-Peitz F, Nishimura T, Knoblich JA. (2008) Linking cell cycle to asymmetric division: Aurora-A phosphorylates the Par complex to regulate Numb localization. Cell. 135(1):161-73.

Postiglione MP, Jüschke C, Xie Y, Haas GA, Charalambous C, Knoblich JA. (2011) Mouse inscuteable induces apical-basal spindle orientation to facilitate intermediate progenitor generation in the developing neocortex. Neuron. 72(2):269-84.

Part 2

Lancaster MA, Renner M, Martin CA, Wenzel D, Bicknell LS, Hurles ME, Homfray T, Penninger JM, Jackson AP, Knoblich JA. (2013) Cerebral organoids model human brain development and microcephaly.  Nature. 501(7467):373-9.

Lancaster MA, Knoblich JA. (2014) Generation of cerebral organoids from human pluripotent stem cells. Nat Protoc. 9(10):2329-40.

Bagley JA, Reumann D, Bian S, Lévi-Strauss J, Knoblich JA. (2017) Fused cerebral organoids model interactions between brain regions. Nat Methods. 14(7):743-751.

Renner M, Lancaster MA, Bian S, Choi H, Ku T, Peer A, Chung K, Knoblich JA. (2017)

Self-organized developmental patterning and differentiation in cerebral organoids. EMBO J. 36(10):1316-1329.

Reader Interactions

Leave a Reply

Your email address will not be published. Required fields are marked *