Vertebrate Organ Development: The Zebrafish Heart

Part 1: Vertebrate Organ Development: The Zebrafish Heart

00:00:13.05 Hello.
00:00:14.05 My name is Didier Stainier.
00:00:15.05 I'm a Director at the Max Planck Institute for Heart and Lung Research in Bad Nauheim,
00:00:20.00 and my lab works on the formation and function of a number of vertebrate organs, including
00:00:24.00 the cardiovascular system.
00:00:26.07 Today, in this first part, we're going to be talking about vertebrate development and
00:00:31.04 specifically about the use of a model system, a zebrafish, a tropical fish, in trying to
00:00:37.18 understand how vertebrate organs, including the heart, form and function.
00:00:43.07 And so, after a few introductory slides about the general points and general questions about
00:00:48.03 development, we'll go into the various models that have been used to study development,
00:00:52.21 and then look to see how and why the zebrafish is helping us understand the fundamental...
00:00:58.04 fundamental questions about vertebrate biology, including developmental questions.
00:01:03.10 So, the question, simply put, is shown here.
00:01:07.02 So, essentially, imagine a fully-grown organism, as shown by Elvis Presley, who actually was
00:01:14.11 stationed in Bad Nauheim.
00:01:16.20 You can see that we have an organism of about 180 pounds, which consists of thousands of
00:01:22.05 different cell types.
00:01:23.08 And, 22 years, before, essentially, this organism consisted of a fertilized embryo, fertilized
00:01:29.20 egg, which weighed about one nanogram and, of course, was consisting of... of only a
00:01:35.08 single cell type.
00:01:36.14 And so, the question is, really, how does a fertilized egg become, over a period of
00:01:44.12 gestation of nine months and then the maturation period of many years, a fully grown organism,
00:01:50.17 such as the one shown on the right?
00:01:53.08 And, just to put... and identify some of the questions involved in vertebrate development
00:01:58.07 or in developmental... in terms of processes in general, we can see a number of questions,
00:02:04.14 here, including cell differentiation, cell proliferation, cell death, cell motility,
00:02:10.08 how the various cell types come together to form the various organs, tissues, and then,
00:02:15.17 finally, how the organs function and maintain during... during homeostasis.
00:02:22.08 And, essentially, in the human genome we have about 21,000 genes and these genes are expressed
00:02:30.07 in different combinations, different times, different places, and there are several questions
00:02:34.23 including, which of these genes, which of these 21,000 genes are required for tissue
00:02:39.15 and organ development?
00:02:41.07 And what are their functions?
00:02:42.09 And also, how are these genes uhh... essentially expressed in different cells or groups of
00:02:49.00 cells?
00:02:50.00 And so, how do we essentially address or answer these questions?
00:02:55.05 And so, the field has been using a number of model systems -- some of them are shown
00:02:58.18 here -- including the fly, the sea urchin, the chick, you can see the Xenopus right here,
00:03:06.12 the worm C. elegans, and Danio rerio, which is right here, the fish that will be the focus
00:03:13.00 of this talk.
00:03:14.05 And, of course, let's not forget the mouse.
00:03:16.19 And so, essentially, these model systems have been used for many years, decades, in fact,
00:03:22.18 centuries, some of them, and there had been mainly two approaches to study development:
00:03:27.10 an embryological approach and a genetic approach.
00:03:31.08 Let's just say a few words about the embryological approach, which was, again, the main approach
00:03:36.01 used for many years, starting in the late 18th century.
00:03:40.13 And so, a classical experiment, done here in Germany by Mangold and Spemann in the early
00:03:46.14 '20s, consisting of essentially taking part of the dorsal tissue, as you can see here,
00:03:53.14 and putting it on the ventral part of an early-stage embryo, and that, once the embryo was raised,
00:04:00.09 essentially, you can see at the bottom, that uhh... these... there was a second axis formed
00:04:05.20 on the ventral side of this embryo.
00:04:08.24 And this led to the concept of embryonic organizers, whereby tissues or groups of cells were secreting
00:04:16.03 growth factors that were inducing adjacent tissues to form specific structures.
00:04:24.08 And much work has been done since then to identify not only the genes but also the mechanisms
00:04:29.03 involved in these tissue inductions, not only from this dorsal organizer the so-called Spemann
00:04:34.23 organizer, but also other organizers.
00:04:39.08 The other major approach that has been used, essentially, to study development is a genetic
00:04:45.12 approach.
00:04:46.21 And so, forward genetics and reverse genetics have been used traditionally.
00:04:51.14 And just to first define these terms as they'll be used throughout these talks, forward genetics
00:04:58.10 is the means by which, essentially, mutations are introduced... induced randomly throughout
00:05:03.11 the genome.
00:05:04.23 These mutations are homozygosed and then the organisms are screened for interesting phenotypes.
00:05:11.11 In contrast, reverse genetics consists of specifically focusing on the single gene or
00:05:17.02 several genes.
00:05:18.21 These genes are specifically mutated and then the homozygous mutant organisms are analyzed
00:05:23.04 for interesting phenotypes.
00:05:25.17 So, let's focus a little more on forward genetics, and, to use a phrase used by many geneticists,
00:05:34.18 including in the late Ira Herskowitz, the awesome power for genetics has really led
00:05:39.18 to many insights about key biological processes, including, for example, from the yeast, shown
00:05:48.15 on your left, we have gained tremendous insight into the control of the cell cycle; forward
00:05:54.13 genetics in C. elegans has given us insights into mechanisms of apoptosis or cell death,
00:06:01.00 as well as the discovery of microRNAs; and, of course, the logic of embryonic patterning
00:06:06.08 and developmental processes has come from for genetics in the fly.
00:06:11.21 And what about vertebrate organisms?
00:06:13.15 So many vertebrates have organs that are not seen in these invertebrate model organisms
00:06:18.01 and also many genes that are specific to vertebrate species.
00:06:22.16 And so the idea, then, is, can we identify a vertebrate organism in which to do such
00:06:27.23 forward genetic screens?
00:06:30.05 And certainly there's been the mouse.
00:06:32.21 Many screens done in the mouse, again, for... since... many years, and these screens have
00:06:38.22 certainly led to many insights including, for example, about the signaling function
00:06:43.06 of these specialized structures called cilia.
00:06:46.06 Now, doing forward genetic screens in the mouse is a fairly complex and expensive process
00:06:52.10 for a number of reasons, and, of course, the litters, the number of pups that one can get
00:06:57.11 from single-pair matings of mice, is fairly limited.
00:07:02.02 And so, in the '80s, this tropical fish Danio rerio, which originates from the Ganges in
00:07:08.13 India, became a... an intriguing model system that was possibly amenable to forward genetic
00:07:15.18 screens.
00:07:16.18 And the main reasons that Danio rerio seemed to be a good model are shown here.
00:07:22.24 And so, from single-pair matings, you can get a large number of embryos, which, of course,
00:07:27.04 for a genetic screen is key.
00:07:28.20 And, also, in terms of development, you can see that fertilization is external, which
00:07:34.01 means essentially, then, that one can visualize and manipulate these embryos from the onset
00:07:39.17 of development.
00:07:41.00 And this, of course, is very important in terms of trying to observe development as
00:07:45.05 it takes place.
00:07:47.09 And so, a typical fish room is shown here.
00:07:49.24 You have, of course, many tanks, and each housing a different genetic strain.
00:07:55.16 You can also see, right here, the breeding boxes that are used, and you can separate
00:08:01.13 the males and the females from their eggs.
00:08:06.05 And essentially, this brings us to not only, are these embryos fertilized externally?
00:08:14.11 But the development of these embryos is also very rapid, so within 24 hours the heart has
00:08:20.01 started beating and by 48 hours most of the organ systems are in fact functional.
00:08:26.06 So, not only that, but these embryos are also transparent or translucent, as well the larvae.
00:08:33.15 And so these and other attributes of the system really allow us to do not only forward genetics,
00:08:38.15 but also... and take.... use embryological approaches and really try and get to a cell
00:08:43.12 biological understanding at single-cell resolution of development or processes.
00:08:48.01 I should also mention that uhh... the zebrafish community has become a very large community,
00:08:53.08 more than 1,500 labs and, so far, more than 30,000 papers, and, as I said, started around...
00:08:59.15 in the early '80s.
00:09:01.14 So, I'll now show you I... a movie of the... about the first 24 hours of zebrafish development
00:09:09.07 and, before that, let me just show you a few key developmental stages.
00:09:13.12 So, on the panel A, you can see that, in fact, we have the single... the one-cell stage embryo,
00:09:22.08 which consists of a cell sitting on top of a yolk mass -- the yolk, of course, is being
00:09:26.04 used for nutrition prior to the start of feeding.
00:09:29.24 You have then... have a series of rapid cleavages to give you the blastula stage, and then gastrulation
00:09:34.18 starts, and this is when the germ layers -- the ectoderm, mesoderm, and endoderm -- are being
00:09:39.14 formed.
00:09:41.01 Within essentially 12 hours of development, you know, gastrulation completes, and then
00:09:46.15 you have the formation of the somites, as shown right here.
00:09:50.01 By 24 hours, as I showed you, the heart starts beating, and then by 48 hours most of the
00:09:55.22 organ systems are functional.
00:09:58.01 So, this is the movie again, starting from the one-cell stage.
00:10:01.09 As I said, a series of rapid cleavages forms the blastula, and then the gastrulation starts,
00:10:09.00 during a process called epiboly, by which the cells surround the yolk mass.
00:10:13.09 You have somitogenesis and then by 24 hours the heart starts beating.
00:10:18.16 So, just to focus on a few of the organ systems that are clearly visible in the fish, you
00:10:25.15 can see on the top panel right here the heart, which is very prominent and will be the focus
00:10:30.16 of this first and second part.
00:10:32.20 You will see the vascular system; not only the main vessels of the trunk, but also the
00:10:39.03 smaller vessels are formed through angiogenesis.
00:10:42.13 And then, finally, and this is now using a transgenic line, and we're looking at the
00:10:47.08 liver and the pancreas and, again, I'll show you a confocal stack rotated to give you a
00:10:53.22 sense of the resolution one can achieve using the simple imaging techniques in this model
00:11:00.12 system.
00:11:01.12 So, again, we have the liver, right here, and the pancreas, right here, and then this
00:11:06.18 is the part of the pancreatic duct connecting the two.
00:11:09.14 And, again, I'll just rotate this movie.
00:11:11.19 Behind these is the swim bladder, which is a buoyancy organ for the fish.
00:11:17.07 And, again, you can get a sense of the single-cell resolution that one can achieve for these
00:11:22.20 vertebrate organs.
00:11:23.22 There are two main advantages to the zebrafish model, as listed here.
00:11:28.13 The first one is the live imaging, as I've shown you before, and the second one is the
00:11:33.15 ability to do large-scale screens, taking advantage of the fact that one can get large
00:11:37.04 numbers of embryos from single-pair matings, and this is both, now, for genetic screens
00:11:41.09 as well as small molecule screens, or chemical screens.
00:11:44.17 So, let's have a first look at the live imaging by focusing on the heart and take a historical
00:11:52.00 perspective.
00:11:53.10 And so, this is the way the heart was looked at in the early '90s, when... which was when
00:11:58.07 the initial large-scale screens were done.
00:12:00.21 There were two large-scale screens done, one here in Germany by Christiane Nüsslein-Volhard's
00:12:05.22 lab in Tübingen and the other one by a former student of hers, Wolfgang Driever, who then...
00:12:12.09 was leading a lab in... at MGH in Boston.
00:12:16.23 And, essentially, simple white light microscopy was looked... was used to be examining and
00:12:22.13 screening these embryos, as shown here, and I will come back to this organ system later
00:12:28.13 on.
00:12:29.13 So, in 2004, people started taking advantage of fluorescent transgenic proteins including
00:12:35.21 GFP.
00:12:37.01 For example, here, GFP is labeling the myocardium, or the muscle cells in the heart.
00:12:42.17 And it took... a few years later, and a special microscopy technique was revived called selective
00:12:50.03 plane illumination microscopy, or SPIM.
00:12:53.12 And you can see, here, again, that we're looking at the heart, and looking at the cardiomyocytes
00:12:59.04 in the muscle cells in the heart.
00:13:01.03 And you can see the blood flowing, essentially, from the yolk mass -- around the yolk mass
00:13:06.17 -- into the atrium and then into the ventricle -- the atrium being the collecting chamber
00:13:12.12 of the heart and the ventricle being the pumping chamber of the heart.
00:13:15.20 Here, we're looking at just a single cell layer, but of course the heart consists of
00:13:19.23 multiple cell layers, as shown here.
00:13:21.20 So, now, we're labeling in dark red the myocardium, in bright red blood, and in green the endocardium,
00:13:30.15 which is the endothelial layer inside the heart.
00:13:33.10 And, again, we're looking at single plane images using SPIM imaging, but of course the
00:13:38.06 heart is a three-dimensional organ, and so, now, taking advantage of spinning disk microscopy
00:13:45.17 followed by image reconstruction, you can see... imagine you're standing inside the
00:13:49.19 ventricle looking out, and what you're seeing are these muscular ridges, here... essentially,
00:13:56.21 this is best seen here... and these are called trabeculae.
00:14:00.16 And let's just come back to the previous slide and, again, these are these ridges that we're
00:14:06.07 seeing here in cross-sections and these are the trabeculae, and we'll be looking... in
00:14:11.04 the second part of this series, we'll be looking at how these trabeculae, these cardiac trabeculae,
00:14:16.22 are forming.
00:14:18.11 So, essentially, we also of course... we want to look at these organs in three dimensions
00:14:24.06 and with time, but also at the single-cell level.
00:14:27.23 And so, for example, we and others have generated transgenic lines where one can look at the
00:14:32.23 actin filaments, here on the right... or sorry, on your left... or the microtubules, here
00:14:38.23 on your right.
00:14:40.04 And we'll be coming back to this actin filament transgenic line in the second part of this
00:14:45.10 series.
00:14:46.12 So, essentially, as I said, imaging has been a key part, a key tool, for zebrafish heart
00:14:53.18 research, not only in terms of visualizing the events that are occurring, but also in
00:14:59.17 terms of really leading to a hypothesis that then can be tested either genetically or embryologically.
00:15:07.24 As I mentioned, forward genetics was the main means by which people were identifying mutants
00:15:12.18 and interesting genes that essentially regulated the process of heart development and function.
00:15:18.14 The morpholinos, which are modified and stable oligos that are used as antisense regions
00:15:26.11 were identi... were essentially... came about in 2000.
00:15:30.20 We had very efficient transgenesis using the Tol2 transposon around the same time.
00:15:36.23 Chemical biology, and we won't say very much about that, around 2003.
00:15:42.00 And then reverse genetics, using zinc finger nucleases, TALENs, CRISPRs, essentially just
00:15:47.08 a few years old now.
00:15:49.02 Now, in the third part of this series, we'll be looking to see and... how, in fact, when
00:15:55.07 people were using morpholinos, or these antisense reagents, to look at gene function, they identified
00:16:01.05 a number of phenotypes and some of these genes that were analyzed using morpholinos were
00:16:06.07 revisited using these reverse genetic techniques, mainly the TALENs and CRISPRs, and, very often,
00:16:12.21 the phenotypes resulting from these two different approaches appear to differ with the mutant
00:16:18.23 phenotype being much milder, and so we'll be trying to understand how some of these
00:16:23.17 differences may occur.
00:16:25.15 So, as I mentioned, besides imaging, the large... idea... the other major advantage of the zebrafish
00:16:31.19 model is the ability to do large-scale screens, be they genetic or chemical screens, and I'll
00:16:37.08 just focus here on the genetic screens, and, again, just focus on the heart as the model
00:16:43.12 system.
00:16:44.12 So, I've shown you this movie of the wild-type heart, and on the right you can see a mutant
00:16:50.11 heart.
00:16:51.11 This mutant is called cardiofunk, and what you see is that the blood is essentially moving
00:16:55.14 back and forth between the two chambers, the atrium and the ventricle.
00:17:00.01 And it turns out that this cardiofunk mutation affects an actin gene that's expressed in
00:17:04.13 the myocardium, and this leads to aberrant contractivity of the myocardium, which in
00:17:10.13 turn, then, leads to defects in the maturation of the heart, specifically, in this case,
00:17:16.01 in the formation of the valves that prevent retrograde blood flow, and I'll have more
00:17:19.15 to say about that in a few minutes.
00:17:21.08 But, the point here is that the concept of this tight interaction and crosstalk between
00:17:27.12 formation and function is, of course, present in many... many organ systems, but is specifically
00:17:33.12 promin... prominent in the development and formation and function of the heart.
00:17:39.22 So, one can of course a... get more sophisticated in terms of analyzing the various phenotypes
00:17:46.09 and also identifying mutants with the interesting cardiac function defects, and here, for example,
00:17:52.10 I'm showing you GCaMP, a GCaMP line which is used to look at the calcium flux.
00:17:59.05 And GCaMP is a modified GFP protein.
00:18:01.17 And, again, you can see, in this case, right here is the ventricle and you can see these
00:18:07.04 muscular protrusions, which are the trabeculae.
00:18:11.11 So, for example, I... we identified in one of our screens a mutant that we call dococ,
00:18:17.19 and you can see, again, the ventricle beating away, and in the mutant you can see that this
00:18:21.18 ventricle is fibrillating.
00:18:24.18 And one can look at this more closely using this calcium line that I introduced earlier.
00:18:29.24 And, essentially, you can see that there is a partial block along the... essentially,
00:18:34.11 the left side of the heart; the right side of the heart appears to propagate the calcium
00:18:40.15 wave normally.
00:18:41.15 You can see this much better uhh.. if you look at, or much more easily, if you look
00:18:44.22 at an isochronal map.
00:18:45.23 And so, again, a partial block only on one side of the heart, and this mutation affects
00:18:51.02 the connexin protein, and when the mutation in the... the mouse orthologue was made, a
00:18:57.04 similar phenotype was seen.
00:18:58.12 So, again, suggesting that, in fact, mutations in the human gene may also lead to the same
00:19:03.17 kind of phenotype that is seen in this zebrafish mutant.
00:19:08.03 So, with that I want to come to the last point of this lecture, which has to do with the
00:19:16.10 endothelial layer.
00:19:18.09 And the endothelial layer is a vertebral invention, and here we're looking at the endothelium
00:19:22.19 in green, and we're looking throughout the embryo.
00:19:25.09 We're looking at the blood cells moving through the vasculature and these blood cells are
00:19:29.09 labeled in red.
00:19:30.18 And you can see... right under the eyes, you can see essentially the heart pumping away.
00:19:37.00 And, essentially, as I mentioned, the endothelial layer is a vertebrate invention and was key
00:19:44.07 in terms of the evolution of the cardiac pump.
00:19:47.12 And so, if you start with... on the left, what is a... simple chordate... heart of a
00:19:52.14 simple chordate, like a tunicate, and this uhh... chordate lacks endothelium, and essentially
00:19:59.00 this heart can pump blood in both directions.
00:20:01.02 It's more like a peristaltic pump.
00:20:03.03 And, as you move to look at, for example, the fish heart right here, you can recognize
00:20:08.10 a number of new features including the subdivision into chambers, the atrium -- the collecting
00:20:13.12 chamber -- and the ventricle -- the pumping chamber, as I mentioned before.
00:20:17.00 There's also specializations of cells in the boundary, at the AV boundary, the atrioventricular
00:20:21.24 boundary, both in terms of the myocardium -- the muscle cells -- as well as the endocardium
00:20:26.24 -- the endothelial cells.
00:20:29.02 And then, you may have seen from the movies that there is a coordination of contraction
00:20:32.06 in the sense that the atrium contracts first and then there's a slight pause and then the
00:20:36.19 ventricle contracts.
00:20:37.20 And, of course, moving then from fish to amniotes, of course, there was a subdivision of both
00:20:43.13 the atrium and ventricle, leading to the four-chambered heart.
00:20:47.03 So, as I mentioned, the endothelium, which is known as the endocardium in the heart,
00:20:53.07 is a vertebrate invention, and it allowed a number of features...
00:20:57.06 again, listed here, and we'll go through three of them in turn.
00:21:01.02 So, it allowed a more efficient pump by allowing the complete occlusion of lumen, it directly
00:21:07.13 participated in the formation of valves between the chambers to prevent the retrograde blood
00:21:11.09 flow, and it created... it allowed increased muscle mass by providing the signals for the
00:21:17.12 formation of trabeculae.
00:21:19.08 And this is prior to the formation of the coronary vessels, which feed the muscle wall.
00:21:24.18 And these trabeculae are also, of course, key in terms of the formation of a specialized
00:21:29.21 conduction system, but we won't say more about this today.
00:21:33.22 So, let's start with the fact of the this more efficient pump.
00:21:39.12 There was probably the initial selective advantage for this appearance and selection of the endocardium
00:21:48.07 in the heart.
00:21:50.00 So, imagine that... again, starting with a simple heart like the tunicate heart, which
00:21:55.01 consists of a single cell type, which are the muscle cells, as these muscle cells contract,
00:22:00.21 here, they lead to the formation of a finite lumen.
00:22:03.22 Now, if you bring in a second cell type, which are the endocardial cells, and assume that
00:22:09.19 these cells can be distorted or deformed, as the outer layer contracts, then this inner
00:22:15.10 layer will essentially completely occlude the lumen, and, presumably, then allow the
00:22:21.16 complete emptying of this chamber.
00:22:23.20 And, in fact, if you look at the zebrafish heart and, again, this is SPIM imaging by
00:22:28.17 Jan Huisken, you can see that, indeed, in the green you can see the endocardium and
00:22:34.03 these... the red cells contract, and here we're looking at the atrium right here...
00:22:39.10 as the red cells contract, the green cells come together, completely occluding the lumen.
00:22:45.16 And you can see this a bit better by just focusing on the endocardium, that's here,
00:22:49.17 these cells, and again, of course, we're looking in a single plane but these cells are coming
00:22:53.05 together and... from the 3-dimensional organ, and you can then lead to the complete occlusion
00:23:00.14 chamber and complete emptying of this chamber.
00:23:05.00 This is not a new concept in the sense that the cardiac jelly, which essentially is a
00:23:10.14 specialized extracellular matrix that lies between the endocardium and the myocardium,
00:23:17.12 had been proposed to play this role in terms of... if you assume there's... there's...
00:23:22.12 there's a finite and stable volume of cardiac jelly, then of course as the outer layer,
00:23:28.09 the myocardial layer, contracts, then this will also help lead to the complete occlusion
00:23:33.19 of the lumen.
00:23:34.23 So, moving on then to the second point about the evolution of the cardiac pump, and the
00:23:41.23 fact that the endothelium was key in terms of the evolution of this organ, I will show
00:23:47.23 you how, in fact, it is directly involved in the formation of the valves between the
00:23:52.13 chambers.
00:23:53.13 So, let's first look at the retrograde blood flow as it appears in early stages.
00:23:58.24 As I mentioned, the zebrafish heart starts beating at 24 hours post fertilization, and
00:24:03.18 you can see... at 55 hours, you can see, again, this... the atrium, right here, and the ventricle,
00:24:10.08 right there, and the atrioventricular canal, right here.
00:24:13.06 And you can see that as the ventricle contracts a few blood cells are moving back towards
00:24:18.06 the atrium.
00:24:19.06 Now, if you look at 70 hours, you can see, in fact, that there is no retrograde blood
00:24:24.03 flow, and you can also notice that there's now a thickening of the atrioventricular canal,
00:24:30.18 essentially, that helps prevent this retrograde blood flow.
00:24:35.03 And the... these are... the shape of these cells is diagrammed right here.
00:24:40.00 So, essentially, again, looking at this in... more closely, one can see that, in fact, this...
00:24:47.11 at the AV canal, as you can see right here, there's these cells, the endocardial cells,
00:24:54.06 have assumed a different shape, now, and are forming what looks to be a S-shaped structure,
00:25:01.10 possibly through invagination of these cells or partial epithelial-to-mesenchymal transition,
00:25:07.13 as we've gained evidence for, and, essentially, then these structures can then get remodeled
00:25:15.07 into these valve leaflets that you can see, now, very clearly at 7 days, again, between
00:25:22.20 the atrium and the ventricle.
00:25:24.16 And you can see that these valve leaflets are very productive and very efficient at
00:25:29.16 preventing retrograde blood flow.
00:25:31.19 So, essentially, then... the... one of the key questions that we and others are trying
00:25:38.04 to address is essentially trying to get a single-cell resolution understanding... and,
00:25:43.24 again, we're looking here... this is the atrium and the ventricle, and the atrioventricular
00:25:48.16 canal.
00:25:49.16 And you can see, in green, again, we're labeling the endocardial cells, and you can see these
00:25:53.21 are single endocardial cells right here, and you can see how, essentially, these cells,
00:25:59.14 as they acquire different shapes and as various genes and mechanisms are essentially affecting
00:26:06.21 their behav... behavior, these cells are now forming these specialized structures called
00:26:11.07 valves, that prevent retrograde blood flow.
00:26:13.17 And so, this is really the goal of this... this investigation now to understand... not
00:26:21.06 only identifying the genes possible for the formation of this structure, but really understand
00:26:26.06 how these genes drive the behavior of each cell as it forms this evolving structure.
00:26:33.03 And, very similarly, in Part 2, what we'll be looking at is this process of cardiac trabeculation,
00:26:40.14 as I mentioned.
00:26:42.00 And, again, as you can see in this cartoon, trabeculation is an interesting process.
00:26:48.10 As you can see, trabeculae are mainly formed, here, in the outer curvature of the ventricle,
00:26:54.13 but not in the inner curvature, right here.
00:26:58.00 And, also, not in the atrium.
00:27:01.00 And so we'll look at, not only in terms of the cell behavior of these cardiomyocytes
00:27:05.19 in this.. as they form these cardiac trabeculae, but we'll be looking at some of the genes
00:27:11.08 and also, again, coming back to a process called epithelial-to-mesenchymal transition.
00:27:18.02 And then, in the third part... and, as I mentioned, the original way that genes were analyzed
00:27:23.18 in zebrafish was through forward genetics and then people started using... about ten
00:27:27.24 years later, they started using these antisense reagents, and analyzed the function of a large
00:27:33.13 number of genes, and when the reverse genetics techniques became available, just a few years
00:27:39.04 ago, some of these genes that had been analyzed, whose function had been analyzed using morpholinos,
00:27:44.22 we'll revisit it using TALENs and CRISPRs, and, essentially, the phenotypes resulting
00:27:51.16 from these mutations were often milder than the phenotypes that had been observed using
00:27:56.03 this antisense reagents, and so we'll be trying to understand what may be the basis for this
00:28:02.10 discrepancy between antisense-induced phenotypes and mutation-induced phenotypes.
00:28:09.11 And this will bring us to essentially look at the question of mutant versus antisense
00:28:15.12 phenotypes.
00:28:16.12 And, with that, I'd like to thank the many past and present members of the lab, as well
00:28:22.09 as the funding agencies, both when the lab was at UCSF and, more recently, at the
00:28:29.14 Max Planck Institute in Bad Nauheim.
00:28:31.10 Thank you.

Part 2: Cardiac Trabeculation

00:00:13.05 Hello.
00:00:14.05 My name is Didier Stainier.
00:00:15.05 I'm a Director at the Max Planck Institute for Heart and Lung Research in Bad Nauheim,
00:00:17.22 Germany.
00:00:18.22 And, today, we're going to be talking about cardiac trabeculation, which is a means by
00:00:22.04 which the heart increases in size, mainly incre... increases its muscle mass, thereby
00:00:28.10 allowing the growth of the organism.
00:00:30.21 And we're going to be talking about this process in the zebrafish model system, which is the
00:00:34.22 model system that allows live imaging and it's also a fairly amenable genetically.
00:00:39.15 And, just to give you some introduction about the zebrafish heart, it starts beating at
00:00:44.17 24 hours post fertilization.
00:00:46.20 Here, we're looking at 62 hours post fertilization, and one can recognize a number of features.
00:00:53.08 The chamber right here is called the atrium and it's the collecting chamber, then the
00:00:58.24 blood moves into the ventricle, which is the pumping chamber.
00:01:02.19 And, here, we're labeling just the myocardium, or the cardiomyocytes, so, the muscle cells
00:01:08.06 of the heart.
00:01:09.13 And we're using a microscopy technique called selective plane illumination microscopy, or
00:01:14.16 SPIM, which was revived right here at EMBL, and Jan Huisken, a former postdoc in lab,
00:01:21.00 was instrumental in this technology.
00:01:24.23 We can label multiple cells, of course, and here we're looking, in dark red, at the cardiomyocytes,
00:01:30.17 the muscle cells.
00:01:32.22 In green, we're looking at the endocardial cells, or the endothelial lining cells in
00:01:37.02 the heart, and then, in bright red, the blood cells.
00:01:39.09 And, again, we're looking at SPIM imaging, so looking at single-plane images, and this
00:01:44.11 is now at 5 dfp -- 5 days post fertilization.
00:01:47.14 And, again, here we're looking at the atrium and the ventricle, and you can see these right
00:01:55.05 here... these are the so-called trabeculae, these are multicellular protrusions that form
00:02:01.04 in the lumen of the ventricle.
00:02:04.12 And you can see, and we'll come back to this point later on, that this is now the inner
00:02:08.20 curvature, and here... and this is... out here is the other curvature, and these trabeculae
00:02:14.11 are formed in the outer curvature, but not in the inner curvature, and not in the atrium.
00:02:20.09 Now, you can see, also, and we'll come back to this later, the different blood flow patterns
00:02:26.03 in the ventricle in terms of the inner curvature versus the outer curvature.
00:02:32.03 So, one can also, of course, look at this organ in three-dimensions, with time, so,
00:02:37.24 in four dimensions.
00:02:38.24 So, imagine you're sitting inside the ventricle looking out and this is what you'd be looking
00:02:43.15 at.
00:02:44.15 So, these are... these muscular structures right here, and these are what we were looking
00:02:50.09 at, essentially, in cross-section earlier on, so these are these muscular ridges, the
00:02:55.09 trabeculae.
00:02:57.17 And so, essentially, taking a cartoon from the mouse heart, and this is just looking,
00:03:03.12 again, at the ventricles, right here, and you can see in these ventricles these ridges,
00:03:09.03 which are the trabeculae, both in the left and the right ventricle.
00:03:13.16 Looking at this at high magnification, one can recognize the compact layer cardiomyocytes
00:03:19.00 from the trabecular cardiomyocytes, and then, right adjacent to that, right here in blue,
00:03:24.08 are the endocardial cells.
00:03:26.04 And, very similarly, in the zebrafish, one can distinguish between the compact layer
00:03:31.03 cardiomyocytes -- it's a very thin monolayer or cells -- and these ridges, which were...
00:03:37.12 we were looking at earlier, which are called the trabeculae.
00:03:41.02 So, essentially, as I mentioned, a number interesting patterning questions or developmental
00:03:49.12 questions about the formation of trabeculae and, just looking at this cartoon, as I mentioned,
00:03:55.23 in the atrium, the collecting chamber, there are no trabeculae, but in the ventricle these
00:04:00.21 trabeculae are formed in the other curvature but not in the inner curvature.
00:04:05.17 And, when we look back in the movie in a few minutes, you'll see, again, these different
00:04:09.20 flow patterns, and it's more of a perturbed flow in the other curvature, more of a laminar
00:04:15.00 flow in the inner curvature.
00:04:17.09 And this already suggests that flow may in fact be important for the formation of these
00:04:22.22 trabecular structures.
00:04:24.11 So, this is... so, I'll just show you a few movies of trabeculation over time -- we'll
00:04:29.17 just look at three different time points -- and trabeculae start forming at around 60 hours
00:04:35.00 post fertilization, usually directly adjacent from the atrioventricular canal.
00:04:39.24 And, again, you recognize not only the trabeculae, right here, but I also want to point out your
00:04:45.18 attention in the... essentially, in the atrium, the separation of the space between the red
00:04:52.10 and green cells, between the myocardial and endocardial cells, is filled by a specialized
00:04:57.17 extracellular matrix called the cardiac jelly.
00:05:01.00 And you can see this more clearly, again, at 5 days -- this movie I showed you earlier
00:05:06.01 -- looking, essentially, again, here, at the atrium, you can see that the space between
00:05:11.20 the green and red cells is fairly substantial in this stage, especially in the atrium but
00:05:17.19 not in the ventricle itself.
00:05:21.10 So, the later stage that we can look at easily, using this microscopy technique, is 14 days.
00:05:28.10 And, again, you can see the compact layer of cardiomyocytes right here, a very thin
00:05:32.18 layer, and, mostly, myocardial mass is in fact in the trabecular layer.
00:05:37.24 So, not only are most of the cardiomyocytes, most of muscle cells, in the trabecular layer, but,
00:05:44.16 now, if you look at them at the extra... ultrastructure, specifically looking at sarcomeres using,
00:05:49.14 now, a line which labels actin filaments, you can see that these trabecular layer cardiomyocytes
00:05:55.04 look very different from the compact layer cardiomyocytes.
00:05:58.21 You can see, in the compact layer cardiomyocytes, these actin filaments are loose, they form
00:06:04.00 a loose mesh of filaments, whereas you can see these pronounced sarcomeres in the trabecular
00:06:09.18 layer.
00:06:10.18 And, in the next movie I'll show you, in fact, again, this is a SPIM movie focusing on a
00:06:16.13 single plane and, of course, as the heart contracts you'll see the trabecular layer
00:06:20.21 and compact layer cardiomyocytes coming into focus, so you can appreciate the difference
00:06:25.18 in the actin filament distribution in these two different cell types.
00:06:30.05 So, not only are most of the cardiomyocytes in the trabecular layer, but the ones that
00:06:36.02 are doing most of the work, because they have more pronounced sarcomeres, are also the trabecular
00:06:41.18 layer cardiomyocytes, so they of course very important for the function of the heart as
00:06:47.17 it pumps blood throughout the organism.
00:06:49.16 And so, again, imagine you're sitting inside the ventricle looking out, and here in yellow
00:06:55.20 we've outlined all the... essentially, painted these trabecular layer cardiomyocytes.
00:07:01.01 Here, in red, we have essentially, then, labeled the nuclei of these cardiomyocytes, and I'll
00:07:08.00 just rotate this so you can see how these cardiomyocytes, these trabecular cardiomyocytes,
00:07:13.21 are in fact filling the entire chamber.
00:07:16.21 Looking at the shape of these cardiomyocytes by doing cell transplantations at early stages,
00:07:21.16 early embryonic stages, you can see, for example, that the trabecular layer cardiomyocytes,
00:07:26.05 shown right here, present a forked appearance, whereas the compact layer cardiomyocytes present
00:07:32.13 a more smooth appearance.
00:07:34.01 And, again, this uhh... allows these trabecular cardiomyocytes to form these multicellular
00:07:39.22 muscular ridges that essentially bring the wall of the ventricle together as they contract.
00:07:46.01 So, let's sort of think about the cell biology underlying this trabeculation process, specifically
00:07:53.13 how, essentially, a compact layer cardiomyocytes, as you can see here, essentially would move
00:08:00.04 out towards the lumen and start seeding this trabecular layer.
00:08:04.24 And so one can imagine a number of scenarios, including the two shown here.
00:08:09.14 So, one would be a simple delamination way by which this red cardiomyocyte, would essentially
00:08:16.10 migrate away from the compact layer and start seeding the trabecular layer.
00:08:22.14 Another process might be oriented cell division, whereby, again, this red cardiomyocyte would
00:08:27.12 divide, one of its daughters would essentially seed the trabecular layer, and the other cardiomyocytes
00:08:33.03 would stay in the compact layer.
00:08:35.14 So, we carried out a number of experiments, including making small clones of labeled cells
00:08:41.00 -- or labeling small clones of cells.
00:08:43.18 We also did some sophisticated lineage tracing using the Cre technology.
00:08:49.17 We looked at the proliferation of these cardiomyocytes at this stage in the fish embryo -- there
00:08:54.06 is very little proliferation of cardiomyocytes, in fact.
00:08:57.21 And all the data we generated in fact pointed to the first model, the delamination model,
00:09:03.12 being mostly at play.
00:09:05.22 And so, thinking about this delamination process and thinking about, essentially, how these
00:09:11.08 cardiomyocytes may be oriented in the apical-basal orientation, we labeled one cardiomyocytes
00:09:19.04 using a basal marker that came from a C. elegans forward genetic screen called Par1, and so
00:09:25.15 you can recognize that this red marker now lies closest to the lumen, but of course it's
00:09:31.08 separated from the lumen by a layer of endocardial cells and also a layer of extracellular matrix,
00:09:38.02 the so-called cardiac jelly.
00:09:39.15 So, essentially, on the abluminal side, so, away from the lumen, we had the apical side,
00:09:46.17 and then on the luminal side of the cardiomyocytes we had the basal side.
00:09:51.07 And so, if you think about this process of delamination, it's possible and it's driven
00:09:56.02 by a process called apical constriction, whereby the apical surface, or abluminal surface,
00:10:01.21 of the cardiomyocyte would constrict and would... this would allow a drive to delamination of
00:10:08.00 the cardiomyocytes.
00:10:09.08 And so, David Staudt, a student in the lab, a MSTP student in the lab, decided to look
00:10:13.13 for apical constriction, again using live imaging.
00:10:17.21 And after spending a number of years developing a processing software for live... and using
00:10:26.18 this software for imaging and image reconstruction, he was able to generate movies such as this,
00:10:33.14 where he was able to in fact find evidence for apical constriction, but in addition he
00:10:38.05 found evidence for the formation of protrusions, and these turned out to be basal protrusions.
00:10:43.05 So, again, looking and focusing on the red arrow, here, this is a single cardiomyocyte
00:10:48.06 right here, and you'll see that in this overnight movie this cardiomyocyte is sending out basal
00:10:54.15 protrusions.
00:10:55.22 And about 50% of the cardiomyocytes are in fact sending these protrusions.
00:10:59.23 So, looking at this process in cross-section, and labeling here a few cardiomyocytes, you
00:11:06.15 can see in purple, right here... again, we're looking now at a surface view, so this is
00:11:12.22 the abluminal surface of these four cardiomyocytes.
00:11:16.05 In purple is the one that's shrinking its abluminal surface, and so again... you can
00:11:21.10 see over time that this surface, this abluminal or apical surface of the cardiomyocyte, is
00:11:27.11 becoming smaller.
00:11:28.13 And if you look in cross-section you can see, in fact, that this red... or, this purple
00:11:34.02 cardiomyocyte starts in the compact layer, shrinks its apical surface, sends out basal
00:11:39.19 protrusions, and then eventually it delaminates completely, loses contact with the abluminal
00:11:46.08 surface of the heart, and then is then ready to start seeding the trabecular layer.
00:11:53.04 So, essentially, what we have, then, is a process of delamination driven by basal protrusions
00:12:00.24 and apical constriction, and so we were ready, then, to start looking for factors regulating
00:12:05.18 this process.
00:12:07.12 And we initially took a forward genetic screen, so we conducted a pilot screen, and essentially
00:12:13.16 we identified a number of mutations where trabeculation was defective, but it turned
00:12:18.07 out that all of these mutations also affected the function of the heart, the contractility
00:12:24.04 of the heart, and we... as we saw in the first part, when you affect the contractility of
00:12:29.00 the heart, you also affect its maturation, not only the formation of the valves but also
00:12:32.23 the formation of trabeculae.
00:12:34.14 And so this was not a productive way to identify genes that regulate the formation of trabeculae.
00:12:40.03 So, we went back and looked more carefully at the mouse literature, and when mutations
00:12:46.20 were initially first made in the mouse, one of the... two of the genes that were initially
00:12:53.00 mutated are shown here.
00:12:54.18 This is uhh... neuregulin, the ligand for the receptor ErbB2.
00:13:00.10 And you can see, again, the wild-type hearts here, these ventricles that I showed you before
00:13:05.02 with the pronounced trabeculae.
00:13:07.10 And in the mutant birth, in terms of the ErbB2 receptor as well as the ligand, neuregulin,
00:13:12.18 you can see, in fact, that these trabeculae are missing.
00:13:16.05 And so this led to a model, backed up by additional data, that essentially suggested that neuregulin,
00:13:23.19 which is expressed, as you can see here, by the endocardial cells, will bind its receptor,
00:13:29.19 ErbB2, which is expressed by the cardiomyocytes, and this will lead to a series of events that
00:13:36.10 leads to the formation of these trabeculae.
00:13:40.00 And so we wanted, then, to use the zebrafish to see if we could get additional insights
00:13:43.17 into this process, and fortunately we were able to identify... to find an ErbB2...
00:13:49.19 ErbB2 mutant from Will Talbot at Stanford and this mutant had come from a screen for
00:13:55.14 myelination mutants.
00:13:56.23 And you can see, again, on the left they are the wild-type and on the right the mutants,
00:14:03.13 and you can see that, both at early and late stages, these mutant hearts are completely
00:14:07.20 lacking trabeculae.
00:14:08.23 So, very much like the mouse phenotype... as shown by the mouse phenotype, ErbB2 zebrafish
00:14:17.00 mutants completely lack trabeculae.
00:14:20.15 One of the questions we wanted to ask is, what would happen if you took a mutant cell
00:14:25.18 and put it into a wild-type heart?
00:14:28.22 Would that mutant cell now essentially go along for the ride and be able to populate
00:14:34.12 trabeculae, or would it always stay in the compact layer?
00:14:39.08 And the way you address this question in zebrafish, the way you make genetic mosaics, is by cell
00:14:43.12 transplantation at an early embryonic stage, as shown here.
00:14:48.04 And so, essentially, you take fish that are transgenic and expressing different colored
00:14:53.15 transgenes, fluorescent transgenes, in their cardiomyocytes, you take a few donor cells,
00:15:01.02 right here, and you put them into the host, and essentially, then, you let the embryo
00:15:06.00 develop and look at its heart.
00:15:08.23 And in a number of cases, then, you have contribution of the donor cells into the host heart.
00:15:16.10 And so, for example, this is a control experiment where we've taken red cells and put them into
00:15:23.13 a so-called green heart, and you can see that these red cardiomyocytes now are populating
00:15:29.20 both the trabecular layer as well as the compact layer.
00:15:34.18 If you do the experiment, now, and start by taking mutant cells, right here, in red, and
00:15:41.23 put them into a wild-type heart, in green, you can see that all the red cells are in
00:15:47.01 fact in the compact layer.
00:15:48.20 There are no red cells in the trabecular layer.
00:15:52.21 Conversely, if you now take wild-type cells, in green, and put them into a mutant heart,
00:15:58.17 in red, you can see that the only cells in the trabecular layer are in fact green.
00:16:03.06 There are no red cells in this trabecular layer.
00:16:06.01 So, there's a very strict requirement, a cell-autonomous requirement, for ErbB2 in the formation of
00:16:12.04 trabeculae.
00:16:14.14 And so, essentially, this brings me to the first summary, and I've shown you that, at
00:16:19.03 the cell biological level, these cardiomyocytes appear to form trabeculae by delamination,
00:16:26.16 and this delamination process appears to involve basal protrusions and apical constriction.
00:16:32.04 And it requires ErbB2 signaling.
00:16:34.13 I mentioned the role of flow before, because, in fact, these trabeculae only appear in the
00:16:40.10 outer curvature, not in the inner curvature, and flow patterns are very different in these
00:16:44.18 two regions of the heart.
00:16:46.17 It turns out that if you block flow or contractility, in fact, prevent the formation of these trabeculae.
00:16:53.23 And we'll come back to this point later on, trying to understand how, in fact, flow and
00:16:58.02 contractility regulates cardiac trabeculation.
00:17:02.09 So, these uhh... experiments... further experiments have taught us a number of things.
00:17:11.06 One is that ErbB2 signaling regulates actin dynamics and, essentially, it involves...
00:17:18.04 actin dynamics is involved in a number of processes, including the formation and spen...
00:17:22.12 stabilization of these basal protrusions, and ErbB2 signaling is required for these
00:17:28.05 actin dynamics.
00:17:30.09 From work in a number of model systems, including looking at breast cancer, ErbB2 has also been
00:17:35.21 implicated in regulating epithelial-to-mesenchymal transition.
00:17:40.11 And so one of the questions we'll address in the next few minutes is whether, in fact,
00:17:44.13 there is also an epithelial-to-mesenchymal transition in this process of trabeculation.
00:17:48.24 And, clearly, ErbB2 signaling is going to be involved in regulating many other processes,
00:17:54.13 and we're now busy trying to identify, using a phosphoproteomic approach... identify some
00:17:59.15 of the downstream pathways of ErbB2 signaling during the process of cardiac trabeculation.
00:18:06.19 Data in a number of other model systems, including chick and mouse, have implicated a number
00:18:11.15 of other signaling pathways in cardiac trabeculation, and these signaling pathways include Notch,
00:18:16.08 Bmp, and ephrin, and we're now also looking at some of these pathways using some of the
00:18:21.14 advantages that I mentioned, including the live imaging and the ability to get single-cell
00:18:26.08 resolution of this process in this beating heart.
00:18:29.18 So, I want to finish off with asking three questions.
00:18:34.11 As I mentioned, one major issue is this apical... basal... basal polarity, and this question
00:18:39.20 of whether epithelial-to-mesenchymal transition is in fact taking place during this delamination
00:18:44.17 process.
00:18:45.20 We'll look at the neuregulin ligand and, by looking at this question, we'll be able to
00:18:52.15 address whether, or, in fact, why, trabeculae occur only in the ventricle and not in the
00:18:59.09 atrium, and also what the role is for cardiac function, that is, flow and contractility
00:19:05.24 in the formation of these trabeculae.
00:19:07.14 And then, finally, I'll just show you a few slides about one of the challenges that these
00:19:12.09 cardiomyocytes face as they delaminate and seed this inner layer, this trabecular layer.
00:19:18.06 So, let's start with the first one, this question of apical-basal polarity.
00:19:22.22 So, the working hypothesis is very simple, and essentially states that, before trabeculation
00:19:30.04 starts, these cardiomyocytes are in fact polarized in the apical-basal axis.
00:19:35.09 So, the proteins present on the apical surface, or abluminal surface, would be different from
00:19:41.17 the proteins present on the basal surface.
00:19:45.06 And the second hypothesis would state that, then, ErbB2 signaling is in fact regulating
00:19:51.17 the loss of this apical-basal polarity as the cells start to delaminate.
00:19:56.06 And so, to address this hypothesis, what Vanesa did is generated a transgenic line that labeled
00:20:04.08 specifically the apical membrane of these cardiomyocytes, here shown in green, and another
00:20:10.09 transgenic line that specifically labels the red membrane, shown here, the basal side.
00:20:16.05 And I'll just focus on the apical marker and... which is shown right here.
00:20:22.02 And so let's just focus... so, this is now prior to trabeculation right here and this
00:20:27.02 is after trabeculation, and here, in red, we're labeling essentially the entire membrane
00:20:33.11 of cardiomyocytes, so all around the cortex, and here, in green, is this apical mark.
00:20:38.16 And we're using a podocalyxin protein, which is the apical protein.
00:20:44.15 And you can see very clearly that this green marker, now, is labeling only one surface
00:20:50.09 of the cardiomyocytes, and that's the a abluminal surface.
00:20:53.07 And this would be... for example, just right here would be a single cardiomyocyte.
00:20:57.24 This is prior to trabeculation, and then after trabeculation you can see, now, that this
00:21:03.04 apical marker has become relocalized all around the cortex of cardiomyocytes that have left
00:21:09.06 the compact layer and have now entered the trabecular layer.
00:21:13.19 So, we can use this line to also do some live imaging, and here I'm just showing you a few
00:21:20.13 snapshots before I show you the movie, but, again, just focus on the arrow right here
00:21:25.15 and, of course, as this abluminal surface is shrinking, you have of course consolidation
00:21:31.05 of the GFP -- there's the GFP-podocalyxin protein -- on this shrinking surface and that's
00:21:37.22 why, of course, it becomes much more prominent, more visible in the confocal microscope.
00:21:44.05 So, just focus on the arrow again and... in this movie, and you can see the accumulation
00:21:52.01 of this apical marker on the shrinking part of this cardiomyocytes that eventually, then,
00:21:59.00 will leave this compact layer and enter the trabecular layer.
00:22:03.02 Now, we can use this apical marker, as shown here, to quantify the number of cardiomyocytes
00:22:10.08 in the compact layer that are undergoing this apical constriction.
00:22:15.12 And we can look at this... using this method, we can see, in fact, what the role of ErbB2
00:22:22.08 signaling, as well as the role of flow, is in the formation or the appearance of these
00:22:28.05 apical constrictions.
00:22:29.23 And what we see when we block ErbB2 signaling -- in this case, we're using a small molecule
00:22:34.08 inhibitor to block this receptor tyrosine kinase -- and you can see that blocking ErbB2
00:22:39.23 signaling leads to the loss of the appearance of this cons... apically constricting cardiomyocytes.
00:22:45.14 And, similarly, if you block flow -- in this case we're blocking contractility using a
00:22:51.02 myosin ATPase inhibitor, using a small molecule called BDM -- you can again lose the appearance
00:22:56.18 of these apical constrictions.
00:22:59.08 So, essentially, what I've told you is that, in fact, yes, these cardiomyocytes are polarized
00:23:05.18 prior to trabeculation, that the abluminal surface is where the apical surface of the
00:23:13.06 cardiomyocytes is, and the other surface, the basal surface, is in fact the luminal
00:23:18.05 surface, and this is the same in the mouse as well we've shown also.
00:23:23.06 Now, we can visualize live and quantify this apical constriction, as the cells depolarize
00:23:30.17 and delaminate, so essentially the cells lose this apical-basal polarity as they start delaminating,
00:23:39.07 and we've also seen that, in fact, in our...
00:23:42.03 I won't have time to show you the data, but that Notch signaling is activated in the cardiomyocytes
00:23:46.17 that are directly adjacent to those undergoing apical constriction, and so one interesting
00:23:51.03 question about trabeculation is what selects the cardiomyocytes that in fact undergo this
00:23:57.14 process, this behavior.
00:23:58.17 As I mentioned, about 50 percent of the cardiomyocytes undergo or exhibit these basal protrusions,
00:24:04.07 but only a subset of them will in fact delaminate and start seeding the trabecular layer, and
00:24:08.19 so it's been a question in the field as to what selects the ones, the cardiomyocytes,
00:24:13.01 that go into the trabecular layer versus those that remain in the compact layer.
00:24:19.06 And then, finally, and because Notch is activated in these cardiomyocytes that stay in the compact
00:24:24.11 layer, we wanted to test whether, in fact, there was a role for Notch in this pathway.
00:24:29.02 But, in our hands, Notch signaling -- blocking of Notch signaling -- does not appear to increase
00:24:34.06 the number of cardiomyocytes undergoing apical constriction or those ending up in the trabecular
00:24:39.06 layer.
00:24:40.08 So, this brings me to the second point I want to talk about, which is the ligand issue.
00:24:46.09 And, in the mouse, neuregulin-1 has been shown to be the ligand that binds ErbB2 and drives
00:24:53.11 trabec... cardiac trabeculation.
00:24:55.13 So, in the fish, we started by looking at neuregulin-1, and we made mutations in this
00:25:01.15 gene, but, in fact, as you can see here... we're looking at the wild type on the left
00:25:05.14 and the mutant on the right, and you can see that there are no differences between the
00:25:09.11 mutant and wild type.
00:25:11.00 So, neuregulin-1 is, in fact, not required for cardiac trabeculation in zebrafish.
00:25:16.24 We turned our attention to the other neuregulin genes, and specifically to neuregulin-2a.
00:25:23.07 The reason we looked more specifically at this gene is because, much like the ErbB2
00:25:28.20 mutants, these mutants die between 8 and 12 days post fertilization.
00:25:33.03 We're using an allele from Steve Ekker and Matthias Hammerschmidt that's, in fact, a
00:25:39.19 protein trap, and so this allows us to look at the expression of this gene by looking
00:25:45.21 at... or get a sense of the expression of the gene by looking at the expression of protein,
00:25:51.09 here shown in red.
00:25:53.12 And you can see that, again... compared to the wild type on the left, you can see that
00:25:58.00 the mutant, the neuregulin-2a mutants, are in fact lacking trabeculae.
00:26:02.05 Here, in red, is the expression of this protein trap with neuregulin-2a, and so one can look
00:26:09.10 at... essentially, at higher magnification you can see, again, just focusing on these
00:26:14.03 arrows right here, the yellow arrows are cells that are expressing neuregulin-2a, these are
00:26:20.14 endocardial cells, and they're directly adjacent to these forming trabeculae.
00:26:26.07 So, these uhh... endocardial cells that express neuregulin-2a, we are of course expecting
00:26:35.05 them...
00:26:36.05 expecting to see them in the ventricle, which is where trabeculae occur, but, surprisingly,
00:26:41.08 we also saw evidence for neuregulin-2a expression in the atrium.
00:26:46.07 And so the question remained, then, why are they no trabeculae in the atrium?
00:26:52.05 And so to address this question, we took a slightly different approach, and that is that
00:26:58.01 we expressed a neuregulin-2a specifically in cardiomyocytes.
00:27:03.07 So, this is now ectopic expression -- as I mentioned, neuregulin-2a is normally expressed
00:27:08.01 in the endocardial cells, not the cardiomyocytes, not the myocardial cells.
00:27:12.12 So, we took this transgenic approach and so, essentially, this is just the way... that
00:27:17.04 this... this construct was designed, and so we're using tdTomato, here, to label the cells
00:27:23.01 that are expressing neuregulin-2a.
00:27:24.21 And, essentially, what you see is the... as you ectopically express neuregulin-2a in cardiomyocytes,
00:27:33.05 that you can not only rescue the mutant -- the neuregulin-2a mutant -- but you're driving,
00:27:40.08 essentially, a trabeculation-like behavior in the chambers, both in the ventricle as
00:27:46.24 well as in the atrium, as you'll see much better in this cartoon.
00:27:50.18 So, again, on the left is the... essentially, the control heart -- no transgene expressed
00:27:55.20 -- but when you express, now, ectopically, neuregulin-2a in the cardiomyocytes, you can
00:28:01.21 see it birth this... this trabeculation-like behavior in the ventricle, but also right
00:28:07.15 here in the atrium.
00:28:09.04 Essentially, these data tell us that atrial cardiomyocytes are competent to respond to
00:28:14.23 neuregulin-2a.
00:28:15.23 So, the other question I wanted to address using this tool is, why cardiac contractility
00:28:23.19 and flow are, in fact, required for cardiac trabeculation.
00:28:27.02 And so these are some of the data initially generated.
00:28:30.10 So, if you now block cardiac contraction using, for example, a mutation in cardiac troponin
00:28:36.06 T, or use this ATPase inhibitor I mentioned earlier, BDM, you block contraction and, as
00:28:43.00 you see, you also block the formation of trabeculae.
00:28:47.06 And it turns out that the reason this occurs is because, now, if you look at the expression
00:28:52.13 pattern of neuregulin-2a, again, using this protein trap allele I mentioned earlier, you
00:28:57.22 can see, again, the wild type on the top and the mutant at the bottom... so, in this case,
00:29:03.11 we're using a morpholino against cardiac troponin T to block contraction, and you can see that
00:29:08.09 you're losing most of the expression of neuregulin-2a, especially in the ventricle.
00:29:14.08 And, as you lose neuregulin-2a, much like you see in the neuregulin-2a mutant, you essentially
00:29:20.08 fail to form trabeculae.
00:29:21.17 So, here, we're expressing neuregulin-2a, specifically in non-contractile cardiomyocytes.
00:29:26.21 As you can see, we're using cardiac troponin T morpholinos to silence the heart, and then
00:29:33.05 we're overexpressing... and you can see by this red fluorescence, overexpressing neuregulin-2a
00:29:38.05 in the cardiomyocytes, and this leads to a trabeculation-like behavior, again indicating
00:29:43.23 that these cardiomyocytes, despite being non-contractile, are competent to respond to neuregulin-2a.
00:29:49.13 So, in summary, what I've shown you is that, in terms of the ligand driving trabeculation
00:29:56.07 behavior in the zebrafish, unlike in mammals it's neuregulin-2a, not neuregulin-1, which
00:30:03.24 brings in interesting evolutionary issues.
00:30:06.14 That neuregulin-2a is expressed in all endocardial cells, including those in the atrium.
00:30:12.05 Of course, there may be differences in level of expression, but this protein trap is not
00:30:16.20 the ideal reagent to look at this.
00:30:21.01 And also it... that neuregulin-2a expression is related positively, by flow, to cardiac
00:30:27.10 contractility.
00:30:28.10 That is, if you lack flow or cardiac contractility, you lose neuregulin-2a expression.
00:30:33.15 And then, finally, that the cardiomyocytes in the atrium, as well as the non-contractile
00:30:38.24 cardiomyocytes, are in fact competent to respond to neuregulin-2a signaling.
00:30:44.22 So now, I'll finish off with the last point I want to make here, which is the challenge
00:30:50.02 that these cardiomyocytes are facing as they delaminate from the compact layer.
00:30:55.13 And, of course, they need to remain attached to their neighbors as they do this in order
00:31:00.14 to form, essentially, a fully functioning organ.
00:31:06.22 And so we started looking at a cadherin, N-cadherin specifically, which is a major adhesion protein
00:31:13.07 in cardiomyocytes.
00:31:14.15 And, as you can see here... so, this would be, now, the single cardiomyocyte right here,
00:31:21.07 and on the lateral sides of these cardiomyocytes -- and just to remind you, these are now...
00:31:26.08 now epithelial or epithelial-like cells with a clear apical-basal polarity -- and you have
00:31:31.21 N-cadherin expression specifically on the lateral sides.
00:31:35.07 And then, as you move a few hours later, you can see that there's redistribution of this
00:31:42.08 N-cadherin protein, and this is better seen if you just look at the surface view.
00:31:46.20 Again, this would be, now, a single cardiomyocyte in the inset, here, on the left, and you can
00:31:52.02 see, of course, then, the lateral distribution of this N-cadherin-GFP protein.
00:31:56.21 And now, as you move a few hours later, you can see redistribution so that it's not just
00:32:02.13 on the lateral sides, but also has come, now, either to the apical or the basal side.
00:32:07.09 And you can distinguish this by looking at the cross-section and so, very clearly, what
00:32:11.18 you see -- and, again, this will be a single cardiomyocyte in the compact wall, a single
00:32:17.02 cardiomyocyte in the trabecular layer -- and that you have redistribution of N-cadherin,
00:32:21.22 now, to the basal side of compact layer cardiomyocytes.
00:32:25.18 And, of course... so, just to show you in a cartoon fashion, we start with N-cadherin
00:32:31.18 on the lateral side of cardiomyocytes, and then redistribution to the basal side, the
00:32:39.09 trabeculating cardiomyocytes now have N-cadherin all around their surface, and this allows,
00:32:44.02 then, the adhesion between the trabecular cardiomyocytes and the compact layer cardiomyocytes.
00:32:51.16 And so, again, just to summarize what I've shown you is that this process of cardiac
00:32:58.06 trabeculation, which is required to increase the muscle mass of the heart before the formation
00:33:05.15 of the coronary vessels, requires ErbB2 signaling, it also requires flow and contractility, and
00:33:13.01 the way flow and contractility are required is by regulating the expression of the ligand,
00:33:18.09 neuregulin-2a.
00:33:19.09 In the case of the zebrafish, that's the... the ligand for ErbB2.
00:33:24.03 That we have, in terms of the cell biology of the process, we have a clear apical-basal
00:33:30.12 polarity of these cardiomyocytes prior to delamination, and there's an appearance of
00:33:36.17 basal protrusions as well as apical constriction to drive the delamination process.
00:33:41.07 And, I won't have time to go into it, but essentially the endocardial cells are not
00:33:45.18 only secreting neuregulin-2a and activating the ErbB2 receptor, but they may also play
00:33:52.07 more active... a more physical role in the trabeculation process, possibly by making
00:33:58.15 stable junctions, stable contacts, with the cardiomyocytes that are in fact delaminating
00:34:03.18 from the compact layer.
00:34:04.24 And... well, I want to really acknowledge many past and present members of the lab,
00:34:12.09 who have been contributing very significantly to these projects, as well as the funding agencies,
00:34:18.01 both initially at UCSF and now in Germany.
00:34:21.21 Thank you.

Part 3: Genetic Compensation

00:00:13.06 Hello.
00:00:14.06 My name is Didier Stainier.
00:00:15.06 I'm a director at the Max Planck Institute for Heart and Lung Research
00:00:17.10 in Bad Nauheim, Germany.
00:00:19.05 And today, in this third part, we're going to be talking about the phenomenon of genetic
00:00:22.19 compensation in the context of vascular development in the zebrafish embryo.
00:00:29.24 Historically, gene function in zebrafish has been studied initially through a forward genetic
00:00:36.01 approach, namely... mutagenizing the genome -- introducing mutations randomly -- and then
00:00:43.16 doing phenotypic screening.
00:00:46.20 More recently, morpholinos -- antisense oligos -- have been used to knockdown gene function.
00:00:54.01 And then, more recently, just a few years ago, using the zinc finger nucleases as well
00:00:58.15 as TALENs and the CRISPR/Cas9 technology, mutations have been introduced in specific
00:01:03.15 genes to study not only the genes that were studied previously using morpholinos, but
00:01:09.13 also other genes.
00:01:11.18 And the studies that I'll be telling about today were essentially inspired by the fact
00:01:17.08 that looking at specific genes that have been studied both using morpholinos, as well as
00:01:23.08 reverse genetic techniques, it became apparent that in many cases the phenotypes induced
00:01:30.05 by mutations were much milder than those induced by the morpholinos technology.
00:01:36.16 And so, essentially... just to give you a little background, morpholinos are, as you
00:01:40.23 can see here, modified oligos, they are highly stable, they bind RNA, and they're used to
00:01:46.16 blocked either translation or splicing.
00:01:50.18 And so the question, then, is why mutant phenotypes are in fact often milder than antisense phenotypes...
00:01:59.15 So, mutant phenotypes versus antisense phenotypes...
00:02:01.23 I will also be using the word morphant for morpholino-induced, so we'll be talking about
00:02:07.23 mutant versus morphant phenotype, and mutants are often referred to as knockouts and antisense
00:02:14.15 as knockdowns.
00:02:16.03 So, let's start by some history of the antisense approach, antisense technology.
00:02:22.03 And... more than thirty years ago, people working in developmental biology were using
00:02:27.20 or started using antisense RNA to essentially block in function.
00:02:32.06 This was birthed in the context of frog embryo as well as the fly embryo.
00:02:37.11 But this period was fairly short-lived, probably because people were concerned about off-target
00:02:42.17 effects, and so people moved to, now, overexpressing either wild-type or dominant-negative versions
00:02:49.19 of genes or proteins.
00:02:51.24 And, essentially... for example, in this case, dominant-negative activin receptor certainly
00:02:59.13 was known and... to interfere with other proteins besides the activin receptor, and yet approaches
00:03:06.17 such as this gave us important insights into developmental processes, and so it's important
00:03:12.03 to realize that, while no reagent is perfect, certainly these reagents can be used to make
00:03:19.18 important insights into biological processes.
00:03:24.20 And so, in a zebrafish, this antisense, using the morpholinos, was introduced in 2000, at
00:03:32.23 the same time as it was introduced in the frog.
00:03:35.23 And a few years later, a number of guidelines were written up essentially to try and...
00:03:40.22 as best as people could, essentially control for these morpholino studies and specifically
00:03:48.08 trying to avoid and recognize off-target effects.
00:03:53.17 And so, essentially, with this in mind and, as I said, as the various reverse genetics
00:04:01.05 techniques became available, people started seeing, essentially, important differences
00:04:06.09 between the mutation-induced phenotypes and the morpholino- or antisense-induced phenotypes.
00:04:13.03 And this was further emphasized by a larger study from Nathan Lawon's lab, where they
00:04:19.08 essentially looked at a large number of genes and, again, found a poor correlation between
00:04:24.17 morpholino-induced and mutant phenotypes in zebrafish.
00:04:28.06 And so, essentially, this is not specific, I should mention, to the zebrafish field.
00:04:33.17 In fact, if you will now look at antisense work in the mouse... is now using transgenesis
00:04:39.17 to drive antisense trans... transcripts, essentially, again, you see more severe phenotypes from
00:04:47.14 the antisense approach than from the mutation approach.
00:04:51.14 And so it's clearly a question that spans beyond, or goes beyond, just using morpholinos
00:04:58.01 and probably applies to all antisense work.
00:05:01.22 So, we decided to revisit this issue in detail, and we picked this gene called Egfl7 for a
00:05:10.06 number of reasons, but mostly because, in three different settings, using the antisense
00:05:15.06 approach, this gene had been implicated in the... playing an important role in
00:05:21.03 vascular development.
00:05:22.19 This was in zebrafish, in frog, as well as in human endothelial cells.
00:05:27.00 And yet, in the mouse mutant, there was no phenotype, not discernible phenotype.
00:05:33.03 Just to give you a little background about this gene, in encodes an ECM protein, it's
00:05:36.20 expressed mostly by endothelial cells, and it's apparently expressed by tumor cells in
00:05:42.11 human cancers, and for this reason was a good drug target, and Genentech, in fact, had a
00:05:48.22 clinical trial for one of the humanized monoclonal antibody against this protein.
00:05:53.02 So, as I mentioned, in zebrafish, where the first work was done on this gene, when you
00:05:59.05 knock down Egfl7 using morpholinos, you see severe defects in a number of processes including
00:06:06.02 vascular tube formation, and you also get pericardial edema, indicative of a failing heart.
00:06:13.07 A similar phenotype was seen in the frog embryo, again using an antisense approach, and also
00:06:22.18 in human ES cells, human endothelial cells.
00:06:25.20 And, as I mentioned, in the contrast to these studies, to these findings, the mouse mutant
00:06:34.00 was in fact phenotypically normal.
00:06:36.09 Now, this was a little complicated, initially, by the fact that there is an... in the Egfl7,
00:06:44.00 as you can see here, there is a microRNA, a microRNA-126, embedded in this locus.
00:06:50.06 And this microRNA is also expressed in endothelial cells, and so in fact the original mutant
00:06:55.07 that was made deleted this microRNA as well, and this led to the appearance of vascular
00:07:01.13 phenotypes, when, in fact, when specific knockouts, specific mutations were made either in the
00:07:08.07 microRNA or the Egfl7 gene itself, it was realized that, in fact, the microRNA was the
00:07:15.20 one responsible for the phenotype seen in the original mutant.
00:07:20.05 So, essentially, the bottom line is that we have severe phenotypes using antisense technology
00:07:27.24 in the fish, frog, and in human cells, but no phenotype in the mouse.
00:07:33.02 And so we went on to make, using, now, TALEN technology, an Egfl7 mutant, and we identified
00:07:42.01 a number of mutant alleles and focused on two alleles: one is the delta-3, which removes
00:07:47.08 a proline, another one is the delta-4 that leads to a premature stop codon.
00:07:52.15 And this is the main allele, mutant allele that we'll be using for the rest of this study.
00:07:58.07 And we used high-resolution melt analysis to develop the very rigorous and reproducible
00:08:05.18 genotyping protocol, which, as you'll see, is essential for the studies I'll be telling
00:08:10.13 you about.
00:08:12.05 So, the... much like the mouse mutant, in fact, the zebrafish mutant shows a very mild
00:08:18.23 if any phenotypes; only about 5% of the mutants show this transient hemorrhage that you can
00:08:25.11 see, here, in the head of the mutant, but... essentially, looking at both the trunk as
00:08:31.03 well as the head vasculature, essentially one sees no discernible or no major phenotype.
00:08:38.19 We also made and analyzed maternal zygotic mutants and, again, these mutants, these maternal
00:08:46.10 zygotic mutants did not show a more severe phenotype than just the zygotic mutants.
00:08:51.02 So, essentially, as I said, we now have a situation where the mutants in zebrafish,
00:08:57.21 much like the mutants in the mouse, show very mild vascular defects, if any; only about
00:09:03.00 5% of mutants showed this defect.
00:09:05.01 But, as you can see here, again, from the original data using the antisense technology,
00:09:09.22 we have both severe vascular defects, vascular tube defects, that is, so, defects in vasculogenesis,
00:09:17.13 as well as defects in angiogenesis that leads to the sprouting of new vessels.
00:09:23.02 So, essentially, we have... as had been seen/observed for several other genes, now... we have profound
00:09:30.16 phenotypic differences between the Egfl7 mutants and the Egfl7 morphants, and one can think
00:09:36.21 about a number of simple or trivial explanations why that would be.
00:09:41.03 It's possible that the mutant allele we had made was a hypomorph, and it's also possible
00:09:45.14 that the morpholino that had been used for these studies were inducing off-target effects,
00:09:51.00 and this is the phenotypes that, essentially, people had been looking at.
00:09:54.20 It's also possible that there was a more interesting observation, and so we decided to look further
00:10:01.10 and try and understand the discrepancy between the mutant and the morphant phenotype.
00:10:05.23 So, essentially, the first question then is, is the mutation a null allele?
00:10:13.11 And you might think this is a simple question to answer, but in fact the genome has come
00:10:19.17 up with many different ways to bypass mutations, especially mutations that lead to stop codons,
00:10:26.16 premature stop codons in the 5-prime end of the gene.
00:10:30.22 So, for example, it's been observed now, several times, that downstream ATGs, or even non-ATG
00:10:37.07 codons can be used for the initiation of translation.
00:10:41.08 We can... we've also observed, and I'll show you in a minute, exon skipping.
00:10:46.01 And then, in terms of secreted proteins like Egfl7, certainly one can also imagine scenarios
00:10:52.07 where unconventional secretion pathways are used for truncated proteins, for example.
00:10:58.22 Let me show you an example of exon skipping, again, that's been observed several times
00:11:04.06 in the field.
00:11:05.06 And, in this case, we made mutations in exon 2 and, as you can see, again, here, these
00:11:11.17 are now two different mutations -- there's a mutation 1 and mutation 2 -- and in the
00:11:16.20 mutant you can see, essentially... they're right here... that there's a smaller band
00:11:21.22 that's also present in the heterozygote, and this band comes from the skipping of exon 2,
00:11:29.02 as you can see there.
00:11:30.09 So, is it... essentially, as I said, many ways by there... by which the genome can circumvent
00:11:37.07 what looks to be severe lesions.
00:11:40.12 So, in terms of our lesion, our delta-4 mutation, specifically, what we did is first look at
00:11:49.08 the RNA levels, as shown here, and you can see there's a reduction in the delta-4 allele
00:11:55.03 compared to wild-type or the delta-3 allele, so there's an about 50 percent reduction in
00:12:00.10 the transcript level, possibly through nonsense-mediated decay.
00:12:04.08 If you look at the protein... we expressed both the wild-type and the mutant protein
00:12:09.00 in cells, and, as it is a secreted protein, you can see that most of the wild-type protein
00:12:14.13 is present in the medium; if you look at the mutant protein, you can see there's a reduction
00:12:19.09 in the level of expression, but you can see very little protein, in fact, secreted.
00:12:23.21 So, these two data together suggest that this allele, this delta-4 allele that we had generated,
00:12:29.19 could in fact be a severe allele.
00:12:32.13 How about the second question -- what about off-target effects caused by morpholinos?
00:12:38.03 Now, in order to do this, we're going to be injecting the morpholino into this mutant
00:12:43.16 allele that we made.
00:12:45.09 And essentially the reasoning here is that, if this allele, this mutant allele, is null,
00:12:50.19 the morpholinos... any additional phenotypes that's seen from morpholino injection, should...
00:12:55.23 should by definition be an off-target effect.
00:12:59.02 So, essentially, before we do that, before we inject this morpholino into the... the
00:13:05.03 Egfl7 mutants, we want first to introduce a myc tag in the Egfl7 locus, again, by gene
00:13:13.10 editing following cleavage by TALENs, and this is to allow us to look at the efficiency
00:13:19.16 of the morpholino at different doses.
00:13:22.22 And so by Western blot analysis, then, after injecting one nanogram of this morpholino
00:13:29.08 into these transgenic embryos, one can see, in fact, there's about an 80% reduction of
00:13:35.06 protein levels, Egfl7 protein levels, using one nanogram of morpholino.
00:13:40.12 We chose this small dose of 1 nanogram because if you inject higher doses, as you can see
00:13:46.14 here, you'd essentially induce the expression of p53, which has been reported to be indicative
00:13:56.05 of an off-target effect from morpholino injections, and so, essentially, one nanogram does not
00:14:01.04 cause p53 induction, but two will, and so we stuck with one nanogram.
00:14:05.23 So essentially, now, we are ready for the experiment, so we're going to be injecting
00:14:10.16 this Egfl7 morpholino into Egfl7 mutants in the following manner.
00:14:15.02 We're going to be crossing hets, injecting one nanogram of the morpholino and then taking
00:14:21.01 32 affected embryos and genotype them.
00:14:24.22 And so let's sli... first look at the various scenarios that... and the outcomes of what
00:14:31.04 one would predict.
00:14:33.10 So, essentially, if the mutant allele is not null, then the mutant embryo should be more
00:14:39.17 sensitive than the wild type to the morpholino injections.
00:14:41.16 Let's say, for example, there's 20% gene function left, you, you know, inject the morpholino
00:14:47.20 at one nanogram, the mutant embryo should be more sensitive.
00:14:51.08 If the mutant is null and the morpholino phenotype is due to off-target effects, then essentially
00:14:57.03 the genotype of the embryo should not matter; the mutant and wild-type embryos should be
00:15:02.03 equally sensitive to the morpholino injections.
00:15:06.08 However, if the mutant is null and the morpholino phenotype is not due to off-target effects,
00:15:09.23 then the mutant embryo should be less sensitive than the wild type to the morpholino injections.
00:15:15.08 And this is exactly, in fact, what we saw.
00:15:17.23 So, we... as I said, genotyped 32 affected embryos, and we would expect, through Mendelian
00:15:25.01 segregation, 8 of them to be mutants, but in fact we only found 3 of them, here, shown
00:15:29.18 in these red curves, and so this indicates that, in fact, the Egfl7 mutants are less
00:15:35.22 sensitive, they're somewhat protective... protected to the Egfl7 morpholino.
00:15:41.23 And so, in fact, these are the data on our different experiments.
00:15:45.17 In the control experiment, you can see Mendelian segregation of the various genotypes, but
00:15:50.04 when you inject the morpholino you can see that, out of 32 injected embryos, fewer than
00:15:55.06 8 of them are in fact showing a phenotype.
00:15:58.24 And you can of course, then, look at this retrospectively -- after you've genotyped
00:16:02.10 the embryos, go back to the pictures that you took -- and this is, for example, here,
00:16:06.23 a wild-type embryo that was injected with one nanogram of the morpholino.
00:16:10.21 In this case, a mutant embryo… you can see that the mutant embryo
00:16:14.23 does not show any angiogenesis phenotype.
00:16:17.11 We're looking, here, at the formation of these vessels, here, in the trunk, that form angiogenesis,
00:16:22.12 the so-called intersomitic vessels, and you can see a clear phenotype in the wild type
00:16:27.02 but not in the mutant.
00:16:30.06 So essentially, we have this situation where the mutants do not show a severe phenotype,
00:16:35.05 but if you inhibit translation you see a severe phenotype.
00:16:38.09 This is by, now, using these morpholino antisense.
00:16:42.13 What about, now, if you inhibit transcription?
00:16:45.10 What will you see?
00:16:46.10 And the way we did this -- inhibiting transcription -- was to take advantage, again, of a recently
00:16:51.21 developed technique called CRISPR interference, and we're using, now, a dead version of CAS9,
00:16:57.20 one that doesn't have nuclease activity, to block transcription.
00:17:02.06 And so, here are the experiments, first to show that in fact we can block transcription
00:17:07.05 to about a 50% level, and so here are the guides used against both the template and
00:17:14.12 non-template strands.
00:17:15.14 And, in fact, when you use this approach, you can phenocopy the morpholino-induced phenotypes,
00:17:22.07 or the morphant-like phenotypes.
00:17:24.01 So, again, these are, again, control and two experimental, and you can see defects in the
00:17:31.00 intersomitic vessels, as shown here.
00:17:34.10 So, what we have is a situation where the mutants don't show a phenotype, but if you
00:17:39.02 use a morpholino to block translation, or you use this CRISPR interference to block
00:17:44.06 transcription, you see a severe vascular defects.
00:17:47.08 So, essentially, the hypothesis then became one of gene compensation, and to use a classical
00:17:54.08 example, one from the muscular dystrophy field, when mouse mutants were made for the dystrophin
00:18:01.05 gene, the utrophin gene was up regulated, and so one needs to make the double mutant
00:18:06.19 to see the kind of phenotype that Duchenne patients exhibit.
00:18:11.16 And so the hypothesis, then, in our case was that there was in fact the activation of a
00:18:16.09 network, a compensatory network, that would buffer against deleterious mutations, and
00:18:21.09 this compensation was present in mutant embryos but not in morphants or in
00:18:26.02 CRISPRi-injected embryos.
00:18:28.08 And so we used, then, proteomics and transcriptomics to test this hypothesis, and let's go... look
00:18:34.09 first at the proteomics, so we're comparing wild type, mutant, and morphants.
00:18:39.09 And what we found is, essentially, a single protein... this is now comparing mutant to
00:18:45.00 wild-type... we found a single protein, Emilin3a, that's upregulated in the mutant compared
00:18:51.23 to the wild type, but interestingly this Emilin3a... 3a was not significantly upregulated in the
00:19:00.05 morphant, compared to the wild type.
00:19:04.18 Looking at the RNA levels, we found, in fact, not only Emilin3a, but other family members,
00:19:11.03 including Emilin3b and Emilin2a, and you can see, again, that they are upregulated in the
00:19:18.03 mutant compared to the wild type, but not in the morphant.
00:19:22.12 Similarly, when we use CRISPRi, we did not see... we did not see upregulation of these
00:19:27.09 genes, Emilin3a... 3a, 3b, and 2a.
00:19:32.07 What are these Emilin genes?
00:19:34.22 We know, in fact, that, like Egfl7, Emilins are negative regulators of elastogenesis and
00:19:41.15 one of the main and functional domain of Egfl7, shown here in yellow, is in fact an EMI domain,
00:19:48.19 and this name, EMI, comes in fact from the Emilin genes.
00:19:53.05 Now, on these genes, the upregulation of these Emilin genes is in fact important functionally,
00:19:59.13 and as it... can it explain, in fact, the lack of phenotypes in Egfl7 mutants?
00:20:02.12 And the way we addressed this question is by essentially making Egfl7 morphants and
00:20:11.12 then rescuing them with wild-type as well as mutant Egfl7, as well as Emilin2 and Emilin3.
00:20:18.10 And, as you can see here, again, this is now the number of effect... in green, we're looking
00:20:23.20 at the phenotypically normal embryos, which are a few, when you inject the Egfl7 morpholino.
00:20:31.09 When you come in with Egfl7 RNA for rescue, you can see that this frequent number...
00:20:37.08 frequency increases.
00:20:38.08 If you use a mutant version of Egfl7, you fail to rescue and, again, much like wild-type
00:20:45.14 Egfl7, you can partially rescue the Egfl7 morphant phenotype by using these Emilin2
00:20:52.22 and Emilin3 genes.
00:20:55.13 This uhh... compensation phenomenon that we observed in zebrafish, this difference between
00:21:00.21 knockout and knockdown, or morphant and mutant... mutant and morphant embryos, is also observed
00:21:08.19 in yeast.
00:21:09.19 There was a recent study from Orion Weiner's lab, where they looked at the bem1 gene.
00:21:14.16 If you look at the mutation in bem1, as you can see here, essentially... that causes a
00:21:20.24 very minor phenotype, but now if you use optogenetics to drive this protein away from its site of
00:21:26.09 action, then you see, now, more severe phenotypes, including cell cycle arrest and cell lysis.
00:21:32.24 So, to summarize and essentially provide some outlook on these studies, clearly there are
00:21:39.04 some morpholinos that phenocopy mutations, at least at the morphological level, for example,
00:21:44.14 cardiac troponin T, that we used extensively in Part 2 to block contraction of the heart.
00:21:51.19 There are other morpholinos that do not phenocopy mutations, so there are possible... a number
00:21:57.11 of possible explanations, including the fact that mutant alleles could be hypomorphic and
00:22:03.09 that certainly could be the case for many mutations that were induced in the 5-prime
00:22:07.22 end of genes.
00:22:08.22 There are going to be morpholinos that, in fact, do cause a number of off-target effects,
00:22:15.24 even if used at a relatively low dose.
00:22:19.15 And then, in some cases, for example, as we just observed with Egfl7's, we can have compensation
00:22:27.16 in the mutants but not in the morphants.
00:22:31.15 What about morpholinos?
00:22:33.08 How, now, with this in mind, and with the ability to essentially, then, mutate any gene
00:22:40.12 using TALEN or CRISPR/Cas9... how should we think about using morpholinos?
00:22:47.06 And the argument would be that, in fact, to find a morpholino that causes no off-target
00:22:52.18 effects, and probably the best way to do this is to find a dose and a sequence of morpholino
00:22:58.02 that has no effect in the corresponding null mutant embryos, or maybe, even better yet,
00:23:02.09 in embryos that are lacking the morpholino binding site.
00:23:07.13 Since we do see differences between the morphant and mutant phenotypes, a question of course
00:23:12.22 arises as to which of these phenotypes is the real phenotype, and which tool to use.
00:23:17.10 And we would argue that, of course, both of these tools, the mutation approach as well
00:23:22.05 as the morpholino or antisense approach, should be used; even if they give you different answers,
00:23:27.00 both of these answers could in fact be correct, especially if the antisense reagent has been
00:23:32.05 validated previously.
00:23:34.00 Now, the zebrafish is particularly well suited to do this kind of work, that is, to compare
00:23:40.06 mutant and morphant phenotypes, and one way we are thinking of using it is essentially
00:23:45.22 to identify members of the network.
00:23:48.01 So, for example, in this context, the context of the work I just described, you might think
00:23:53.00 of Emilins as being part of a network with Egfl7.
00:23:56.18 And, Emilin could in fact... because it can at least partially compensate for the lack
00:24:02.07 of Egfl7, it can be by definition seen as a modifier gene.
00:24:06.07 And so, essentially, the idea now is to take the genes implicated in vascular development
00:24:11.03 or vascular biology and see if the morphants and mutants for these genes show different
00:24:16.14 phenotypes, and, if they do, can we identify the compensating genes and thereby identify
00:24:21.13 the modifier genes?
00:24:22.19 Now, mechanistically, there are a number, of course, of interesting questions including,
00:24:28.01 what are the mechanisms of this transcriptional upregulation?
00:24:31.01 So, in the Egfl7 mutants, how does Emilin transcription get upregulated?
00:24:38.14 What is the trigger for this upregulation?
00:24:40.11 What are the mechanisms between the trigger itself and this transcriptional upregulation?
00:24:46.10 And so this is sort of some of the ongoing work now in the lab, and let me just give
00:24:51.19 you a few slides... show you a few slides about what we're doing, not only in zebrafish
00:24:56.21 but also in mammalian cells.
00:24:59.07 We now have a number... identified a number of genes where essentially we see this RNA
00:25:06.10 level upregulation in the paralog, the non-mutated paralog.
00:25:10.06 And, in four of these cases, we've also seen that the heterozygous embryos, as well as
00:25:15.07 the mutant embryos, show in fact this increase in mRNA levels.
00:25:19.08 So, for example, I've told you about, in this case... earlier, about Egfl7 and the upregulation by...
00:25:26.08 of Emilin.
00:25:27.22 We've also looked at vegfa mutants and we see upregulation of vegfa-b.
00:25:35.13 If you looked at the heterozygous embryo, again, for example, looking at Egfl7, right
00:25:41.19 here, you can see that the heterozygous embryos show an intermediate level of upregulation
00:25:46.18 compared to the mutants, and of course here is the wild type, and similarly vegfa also...
00:25:53.12 heterozygous embryos also show this intermediate level of upregulation.
00:25:57.13 Now, of course, we've also identified and seen... observed genes that, when mutated,
00:26:03.10 do not cause the upregulation of the paralog, and some of these cases are shown here.
00:26:09.11 And so, with that, I'll thank the.. and acknowledge the people who have been driving these projects,
00:26:15.12 including the original paper, Rossi and Kontarakis et al, and also Mohamed's work, who is now
00:26:22.16 working to essentially look further into this phenomenon of compensation using zebrafish
00:26:29.05 mutants, but also using mammalian cell lines, where we observe similar compensation of phenomenon,
00:26:35.00 trying to see if we can get it to mechanisms, again, not only to identify the trigger, but
00:26:40.03 also the mechanism between the trigger and the transcriptional upregulation.
00:26:44.06 And I also want to thank the funding bodies who are supporting this work.
00:26:48.15 Thank you.

Videos in this Talk

Part 1: Vertebrate Organ Development: The Zebrafish Heart
Audience:
- Student
- Researcher
- Educators of Adv. Undergrad / Grad
Part 2: Cardiac Trabeculation
Audience:
- Student
- Researcher
- Educators of Adv. Undergrad / Grad
Part 3: Genetic Compensation
Audience:
- Student
- Researcher
- Educators of Adv. Undergrad / Grad

Speaker: Didier Stainier

Total Duration: 1:27:39

Recorded: February 2017

All Talks in Development and Stem Cells

Talk Overview

How does a fertilized egg develop into a complex multicellular organism such as a fly, mouse or human? During zebrafish heart development, for example, cells must proliferate, differentiate, move and come together to form a complex organ. Didier Stainier explains that zebrafish are an excellent model organism in which to address this question because their eggs are externally fertilized, they produce many offspring and the embryos and larvae are translucent. In addition, specific cells can be fluorescently labelled making it easier to image organ development in live fish. Taking advantage of these characteristics, Stainier and his colleagues performed large forward genetic screens to look for mutants in zebrafish heart development. Their findings provide insight into the evolution and development of the vertebrate heart.

In his second lecture, Stainier describes work from his lab investigating the formation of trabeculae in zebrafish hearts. Trabeculae are multicellular protrusions into the lumen of the ventricle that allow the heart to increase in muscle mass and thus pump more forcefully. Interestingly, trabeculae form only in the ventricle, not in the atrium, and only on the outer curvature of the ventricular lumen. For trabeculae to form, cardiomyocytes must delaminate from the outer layer of muscle cells and proliferate in the lumen. Stainier discusses how his lab identified factors regulating this process including the important roles of blood flow and contractility.

Gene function in zebrafish has been investigated by 1) randomly mutagenizing the genome, 2) knocking down genes with antisense oligos or 3) more recently, by specifically mutating a gene of interest with gene editing tools. Interestingly, phenotypes obtained by antisense knockdown are often more severe or different than those obtained by gene knockout. In his last lecture, Stainier presents work from his lab that compares knockdown vs knockout of the egfl7 gene in zebrafish (causing severe vs mild vascular defects) and asks why this difference in phenotypes occurs. He walks us through the experiments which show that in the case of egfl7, and numerous other genes, gene knockout effects are compensated by upregulated transcription of paralogous or related genes. This finding raises many questions about how this phenomenon occurs and Stainier’s group continues to investigate this and related questions.

Speaker Bio

Didier Stainier

Dr. Didier Stainier is a Director (Principal Investigator) in the Department of Developmental Genetics at the Max Planck Institute for Heart and Lung Research (MPI-HLR), in Bad Nauheim, Germany. His lab uses the zebrafish as a model to study development of the cardiovascular system and pancreas, and the mouse as a model for lung development…. Continue Reading