(beeping) (uptempo music) – So, I like to start by
sharing a personal story. All scientists have I think a memory, a story that they can go back to when they think about how they
became interested in science and I wanted to share mine. This is mine here. These are baby leatherback turtles that are hatching from a nest not far away from where I was born and raised in Puerto Rico. And when I was a child there I always found fascinating
that these animals are coming out of the nest and they haven’t been to turtle school and they know exactly where to go. They’re beelining it towards the ocean. There’s something really
profound happening here also, very important for the turtle
but invisible to our eyes, which is that these turtles
are forming a memory. They’re forming a memory of this beach. These are leatherback turtles, they grow up to be about
the size of a Volkswagen, about nine feet long, and they will travel the world’s oceans and decades later when they have to make what is perhaps the most important decision in a turtle’s life which is where it’s gonna lay its eggs, it will remember this beach
because that’s where it was born and he’ll come back to and
lays its eggs at that place. So with that I’d like to
coarse-grain behaviors and essentially group them into innate or hardwired behaviors like
these animals’ capacity to know to go towards the
ocean when they’re born. An experiential memory is either capacity to remember this specific
vision come back to it. Now, innate behaviors are facilitated by the developmental program that leads to the formation of the
neural circuit architectures, and experiential memories
are facilitated by a pretty complex interplay between that architecture that develops and the environment that
the animal is experiencing, what the animal is seeing. But I wanna emphasize that both of them are ultimately facilitated by the architecture of the nervous system. So what do I mean by the
architecture of the nervous system and this is essentially what I mean. If you look at a mouse
hippocampus for example you’ll see this incredibly
and exquisite organization of the neurons. This is actually image
here using a technique called Brainbow by Jeff
Lichtman at Harvard. And this organization raises all sorts of interesting questions one of them which my lab is interested on is how is it that this organization
is actually established? And in thinking of how this
organization is established, if you take for example the
case of the human brain, you have about 80 billion neurons, that’s more neurons in
a healthy human brain than you have stars in the Milky Way, and about 100 trillion synapses. And during development you have hundreds of millions of
neurons almost simultaneously during development specifying fate, growing out axons,
connecting to each other, finding each other in a
very specific way to lay out this amazing architecture. So, how does that happen? What are the organizing
principles that rules this organization that ultimately
underpin human behaviors? Now we look at this
question and we don’t do so in turtles or humans or mice actually, we do so in a worm. It’s a tiny nematode called
C. elegans, it’s over here. One of the characteristics
that that C. elegans has that we really like is
that it’s transparent. So it allows us to look at the organization of the nervous system which is very finely
organized as you can see here. Now I wanna acknowledge that C. elegans is far simpler than a healthy human brain. So instead of a hundred billion neurons, C. elegans has exactly 302 neurons and instead of 100 trillion synapses C. elegans has approximately
7,000 synapses. So some of you might be wondering, how is it that we can extract
any useful information out of such a simple system that will tell us anything useful about a complex organ like the human brain. And to address this I like
to use a quote by Diderot that said, “A worm is only a worm. “But that only means that
the marvelous complexity “of its organization is hidden from us “by its extreme smallness.” Now, I’m not as eloquent as Diderot so I’ll say it in a wordier way, when nature finds a solution to a problem it recycles that solution
over and over again. So the fundamental principles that allows for the proper functioning
of the nervous system in this tiny worm are not unlike
the fundamental principles that facilitate the
functioning of our own brains. And to exemplify that I’ll
just give a few examples. Studies done in this worn by other groups ranging from the development of neurons like cell fate specification
or axon guidance to fundamental physiology, having to do with biogenesis of neurotransmitters or neurotransmitter release
or neuron stuck to each other to even system’s levels examination of sensory perception and behavior, all done in this nematode
have actually shed light on important and fundamental
aspects about how our own brains and brains
in higher metazoans work. It’s really true that evolution kind of conserves these principles. So, using that as a foundation we wanted to essentially
examine this question of how this nervous system comes together and both our studies and the
studies that I just mentioned benefited from the fact that this is also the only animal for which
we have a wiring diagram. Now we’re hoping this is gonna change soon and you’re gonna hear other talks today that are working very hard to make this. You know, I think of the
past now we’re very happy that that’s the case but today this is the only animal for which we have that wiring diagram. So, what is the wiring diagram? It’s the ability to know
where each neuron is, essentially the morphology of the neuronal and what a neuron is connecting to. And we use this wiring diagram to examine two fundamental questions. One of them is how is it that the brain of this animal is organized and the second one is how
does this organization comes out during development? Now we’ve had the wiring
diagram for about 30 years. The first wiring diagram was done in 1986 and we still have these two questions that we need to address and
I’ll explain what we know and what we don’t know in the next slide. The way that that wiring
diagram was originally done was that somebody took, a group of scientists led by
John White took this animal and essentially fixed it and sliced it like a salami. So imagine that you’re slicing
all over the animal here and then you get these cross sections and then they did electron microscopy on those cross sections. And by segmenting each of the neurons which we are putting here
in different colors by hand, this is 1986 they segmented them by hand, they were able to recreate
the connectivity map for every single cell. Now this was this was incredibly
powerful for the field and it has benefited our studies but I just wanna emphasize
that all of this was, it wasn’t digitalized. It was all done by hand. So you could go narrow by narrow and you can kind of tell the
shape and who it connected to but we didn’t have a
systems level understanding about how this happened. Until recently where Scott Emmons at Albert Einstein University
and his students Steve Cook actually went in and did
the same segmentation that John White did 30 years
ago but with computers. What that allows us to
do is to be able to know who’s neighboring who
in a quantitative way, in a way that we can actually analyze. And we can analyze using
tools not unlike the tools that are being used by companies
like Twitter or Facebook to understand interactions
between human beings. So our collaboration included
computational scientists Smita and Alex and what they
did is that they took that data and we worked together with them in using clustering algorithms. Again similar to a cluster
in network algorithms that are used to understand
personal interactions. And cluster neurons with
similar contact profiles, so essentially neurons that
are contacting each other as they’re traveling together in the nervous system in fascicles. And indeed iterative
clustering to understand how that structure of
the brain is organized. And you end up with essentially
a flow diagram like this where in one corner you
have all the neurons and this cluster in algorithms iteratively bring them together. So, this different
clustering of the different subgroups of neurons
that eventually cluster into these larger families, what it represents, what underlies that is real biology of how these neurons are interacting with each other. Real biology that was inaccessible before we were able to digitize these programs. And we’re capable of
overlaying that on top of the original EM micrographs. Each of these pseudo colors
here that we have placed represent a community of neurons that are interacting together
more than the other neurons. And I’m gonna just walk you
through the brain of the animal and this is gonna be playing a movie so that you can see how
the different communities are actually snaking and how
they relate to each other and how they come together. As this movie plays I’ll explain that what I’m showing you here is actually reproducible across the animals for which we have connectomes. So this has been a really fun project because we were able to work with the computational biologist and inform the algorithms with the biology and use the algorithms to
inform our own biology. We were able to find four main bundles that account for 83% of
the whole connectome. Now we have a pretty good idea about how these different brain
regions are organized and these neighborhoods, I’ll
just mention that they reveal important biological insights. So there’s a reason why
they’re organized this way that have to do with
development and function. I could give a one hour talk about this so I’m gonna use one
example to present this and I’ll talk about
how these neighborhoods led to insights about
how this organization is actually established
during development. And this gets to a fundamental question that is a blind spot in the field which is how do you go from
a group of cells in an embryo to this organized connectome in the worm? So this is essentially a question that we wanted to understand, and the reason that this
has been a blind spot is that for many organisms this part of the
development happens inside, like for example in mammals
it happens inside the uterus. So the organisms are
essentially inaccessible. In C. elegans they lay eggs
so the organism is accessible but there were a number of challenges that have prevented in the
past like four or five decades to image embryonic neurodevelopment. So this is a pretty large blind spot, we have this information
but we didn’t know how this was happening. The reason that imaging neurodevelopment in C. elegans was
challenging were multifold. One of them is that embryos, when they’re inside the
egg they’re very quick, they’re moving very quickly. So if you’re looking for example at the development of a single neuron using even the fastest microscopy that we had available at the time which is spinning disk microscopy for the aficionados in the audience. This is a single neuron
and you get a motion blur where it looks like four neurons. Because the animal is moving and you’re taking different pictures so it looked at you
taking four neurons there. The other problem that is
hard to capture in an image is the animals just die. So when you’re imaging for
prolonged periods of time which is what you want to
do to be able to capture these different neural
developmental events, many of the embryos just die
so you end up taking snapshots at different points in development that you have to stitch together later. And neurons are actually pretty thin. They’re near the
diffraction limit of light. They’re traveling in ways that are not convenient for imaging. They’re not traveling in planes, they don’t care about our imaging needs. They’re actually snaking in
through the whole nervous system so we needed methods that allow us to have good resolution not just in two dimensions but in all three dimensions. So to address this we
establish a collaboration with a fabulous microscopist Hari Shroff and his team of scientists. And Hari essentially, we had
these discussions together and he came up with the
design of this microscope. I won’t go into the nuts and bolts about how these microscopes work. But what I will tell you is the aspects of this
microscope that makes it usable for our studies. This microscope is called
a light sheet microscope which essentially means that instead of imaging a single point of light like many other microscopes do, it creates a whole sheet of light so it’s much faster
than regular microscopes because it’s like in Star
Trek when they scan you versus scanning you point by point. So it’s faster. Because it’s faster it exposes
the animal to less light, so less photo toxicity, also the way that light is generated to be able to image these animals exposes the animals to less light. And the other aspect of this is that for the microscopists in the audience, so people who have done
microscopy they will recognize that when you take a
three-dimensional image of an object in a microscope, the resolution in X, Y
is always really good but when you turn it the
resolution you see it’s always bad. And that has to do with
just the physics of light which I’m not gonna go into but I will say that X, Y and Z is relative to how you’re standing. So you guys right now are in my Z but if I was standing this
way that will be my Z. So if I had two images where the Z varied then I could combine them to essentially have an isotropic image. So if I look at this object like this, this is my X, Y and then
I look at it this way and this is my X, Y and I
combine those two images, it’s the same resolution
in all three axis. And this is essentially what we’re doing with this microscope
because we have two cameras, this one illuminates, this one image and then they take turns. You end up with two images
that you can then fuse and the Z of this axis is
different from this one. Again, I’m discussing this superficially because I know Phil Keller will
discuss this in more detail. We have this microscope working in my lab and this is a slide that I put here just to acknowledge two institutions, the Marine Biological Laboratories and the University of Puerto Rico which were meeting places
where technologists like, like microscopists like
people from Hari’s group and biologists from my group
were able to come together and exchange ideas, and I think these places are gonna be increasingly more important for interdisciplinary collaborations necessary to address these problems. What can we do with this? So this is actually the brain of the animal here in an embryo, the tail, the head and that’s the brain, that ring that you see in there. And I’m gonna play a movie
where we can actually see the development of this
animal in real time. I mean this is, as a
biologist this is spectacular that we can see this. You’re gonna see the
animals, it starts to move, it starts to twitch as the
nervous system comes online. But even when it’s twitching
you can get very crisp images. Now, if you look at subsets of neurons which we’re going to do here
using subcellular markers or markers that are more restricted you can still see the,
you can see neurites, specific neurites growing
into the brain of the animal, this is all taken again in live animals. And essentially you
can stop it at any time and you can rotate it so and with two-dimensional projections because I’m using a screen but these are three-dimensional data since you can see resolution is very good in all angles that you look at it. And you can continue in movie. So this gives us unprecedented access to the events that are leading to the formation of this nerve ring. And then we had another collaboration that was really enabling
which is that in C. Elegans we know the lineage of every single cell, we’ve known that since
the animal was developed as a model organism, that’s how the original studies of programmed cell death were done. And this person, this is Zhirong Bao, he’s a scientist of Sloan Kettering who’s a computational biologist and a developmental biologist. And what I’m gonna be playing here is an embryo that is in a four cell stage. Now this is like playing a movie that you know what’s gonna happen because we know the lineage. We know for each one of these cells we know who their progeny is gonna be because those studies
have been done before. But what he did is that
he trained a computer to be able to recognize this in real time. So as the animal is developing the computer can keep track
of all of these nuclei and it knows what each
cell is gonna become. So why is this important to us? It’s important for two reasons. One of them is because
when we’re marking neurons as we’re gonna be doing here, we can tell their identity
because if we have nuclei that we’re labeling in
the background in red then we can keep the
identity of every single one of these neurons that are emerging. But the other aspect is that we have an internal coordinate system. As the neurons are growing out and crossing the different nuclei we know where each neuron is. So if we take different embryos and we’re imaging different embryos we can overlay them because we essentially have a multi-point
internal coordinate system. Does that make sense? Okay, so what can we do with that and here I’m gonna summarize the work of a postdoc
in my lab, Mark Moyle, this is five years of work
summarized in one slide. So, essentially what we can, what Mark did is he was
interested in identifying the pioneering neurons
that lead to the formation, that trail-blazed the formation
of the brain of the animal. So he looked in embryos,
he labeled all membranes and he found the first
membranes that are formed that are part of the brain. He traced them back to these cells here, we’re pseudo coloring
here so you can see them. Then if he looked at the
nuclei he can actually identify by name and last name
those specific cells. Then he can image as I mentioned before the development of the brain so he can take these cells and kill them and the hypothesis is if you kill them because they’re the first cells, if the first cells are important then you shouldn’t be
able to develop a brain. And we get these brainless animals in which we cannot see, essentially the brain is greatly abrogated so these neurons are
actually very important. And it turns out when we look back in the maps that we have created with the computational biologists that this pioneer cells
are actually the cells that are here in purple. And they’re like seamed cells, they’re cells that are holding together the whole nerve ring. And with that I’d like to essentially bring to your attention
how we can both look at how the brain is organized and how the organization is
established during development. And I’ll finish by saying that what we have been able to do
with this and our aspiration is to create what would
be the first map of metazoan neurodevelopment for any animal. So we’re systematically tracking all of the incisions of these neurons and creating this virtual
embryo that allows examination of all of these decisions in the context on the developing embryo which we think will be enabling for neuroscience. And with that I’ll finish by thanking the people that did the work. This motley crew and you
for listening to me today. Thank you. (audience applauding) – So, I’m gonna be talking about the complexity of the single cells that make up the brain. We’ve heard a lot about
neuronal populations and how they’re organized
which is quite amazing. But if you go, if you zoom in and look at any one of those cells whether they’re astrocytes or neurons or microglia within the brain, inside those cells is a plethora
of subcellular organelles that are playing a huge role
in how the brain is operating. I just want to draw your attention to some of these organelles who if they don’t function properly can lead to many of the
neurodegenerative diseases that we’re aware of. So for instance mitochondria
that you can see here, defects in mitochondria
lead to Parkinson’s, many of the Parkinson’s disorders. Lysosomes, proteins that
comprise some of these lysosomes are responsible for
frontal temporal disorders. And the endoplasmic
reticulum which you will hear a lot about today is very much at play in where mutations in proteins
that shape that organelle underlies a variety of
spastic paraplegia disorders. In order for us to really get I think at this neurodegenerative
aspect of the brain and how it ages and deteriorates, we need to understand how these organelles that comprise the cellular components which is the unit of the
whole system of the brain are organized. So, this is a classic
transmission electron micrograph, a 90 nanometer slice through a cell to really reinforce the
complexity of these organelles. What we wanna know is
more information about the three-dimensional
organization of these organelles and how proteins are
dynamically distributed. That’s one thing we wanna know and I’m gonna be talking
about other layers of information that we’re
now beginning to get in terms of the
organization of this system thanks to high end microscopy technology. So let’s start with the
3D organelle shapes. This is as I mentioned is a transmission electron micrograph
slice of 90 nanometers. What that means is within
that 90 nanometers slab you have no Z information, everything’s flattened. But thanks to technologies like the Focused Ion Beam
Scanning milling system that in combination with
scanning electron microscopy we can now slice through the
cell at very thin sections, four nanometer in this case. And we’ve used that on
collaboration with Harald’s beautiful FIB-SEM system at Janelia to begin slicing through parts of the cell to reconstruct particular organelles, in order to understand how these
organelles are shaped in 3D and how they communicate
with other organelles. What you’re looking at
here is just the fine complex architecture of
the endoplasmic reticulum in a small, the edge of the cell. In this panel right here
what you’re looking at is the confocal volume that we’ve sliced up in what you would see if you were using a confocal microscope. So you can see there’s a huge difference between the reality of this organelle as revealed by imaging at this very fine three-dimensional architecture versus the typical
image that you would get if you are using conventional
confocal microscopes. Now we’ve been working with Harald to look in more detail
at these organelles. In this case, we’re looking at
plasma membrane, mitochondria ER, endosomes that we can segment out as you mill through these slices. This is a two micron volume of the cell, four nanometer voxels and you can color code each
of these different organelles to see how they’re arranged
relative to each other. Now the mitochondria that
you can see here in green are intimately communicating with the ER which is shown in red and
there’s a lot of crosstalk between these two organelles that’s absolutely critical for
calcium handling in the cell for reactive oxygen exchange, as well as lipid and other
types of communication. Well, how do we localize
proteins on these organelles? And the approach that we’ve been taking really builds from really
the 20 years or more of work that people have done using fluorescent protein technology, which allows you to tag
proteins of interest and then look at how they’re distributed. Now in order for us to get high resolution protein as well as lipid distribution to understand the fine architecture in a three-dimensional section of a cell, we’ve applied Lattice
Light Sheet Microscopy developed by Eric Betzig and colleagues, where essentially you take
ultra thin vessel beams to create a thin 2D optical lattice which is then used as
a sheet to pass through your cell of interest. What this allows is ultra
thin slicing through a cell which is at a much smaller scale than big neuronal slices, sections. And because we have
such a thin light sheet that’s basically almost the same dimension as the X, Y lateral resolution that you get with the microscope, you have isotropic resolution. And again it’s low, it’s
relatively low to photo toxicity because it’s a lattice sheet of light rather than a full sheet. So we combined the lattice
light sheet microscope with a point localization
or PALM-like imaging where individual molecules, in this case we’re
looking at lipid molecules that bind and then
dissociate from membranes. And whenever they’re
bound they create a spot on the surface of that membrane that we can fit with very high accuracy by point PSF centroid fitting. And when we do that we can reconstruct the entire sort of organelle distribution, map it out in 3D through
lattice light sheet imaging with a combination of plotting out all the individual
distributions of these lipids that have been using super
resolution imaging docked in and put in place. Now, now that we have an
image of all these organelles in this you can see mitochondria here and this web-like structure represents the endoplasmic reticulum, we can now come in and dock in particular proteins of interest. I’m just gonna zoom in on this area here because what we’re
particularly interested in or we’re interested in in this study was how proteins that are part
of the endoplasmic reticulum which is this large structure that expands throughout the cytoplasm,
it’s this tubular meshwork, how its organized. And so what we can do
with this technology is superimpose our distributions
of fluorescent proteins that we’ve specifically
genetically engineered and tagged onto our fluorescently acquired image using this lattice light sheet system to dock in where these
proteins are localized. And this is an example where we have again, all of the membranes of the cell that have been painted out if you will using the lipid, our lipid probe and single molecule
super resolution imaging. And we’re now correlating it with the diffraction
limited image of Sec61 beta tagged with an mEmerald
of fluorescent probe. And you can see how they align. This is exciting to us because it really sets the stage
for beginning to investigate a whole slew of different
proteins and how they localize within this cellular system. So we think that fits in and
Lattice Light Sheet-PAINT or PALM-like correlative approaches will allow us to really
gain deeper information about how all of these
organelles are shaped in three dimensions, and how different proteins
might distribute on them. But one of the challenges
that we still face is how many organelles are arranged relative to each other
in a living cell context. In particular that’s
important for understanding how different organelles
are contacting each other and communicating with each other. We know for instance the
endoplasmic reticulum which is this sort of snake-like structure that you can see in this EM image is contacting virtually
every other organelle within the cell and communicating with those organelles through lipid trafficking, ROS, essentially exchange of
reactive oxygen species, calcium signaling. Many other types of
communication is going on between these organelles. And our problem with
trying to understand that is that we haven’t been
able to look simultaneously at all of these organelles
in a living cell context. We can see them clearly
with electron microscopy but if we do fluorescent
time lapse imaging we’re limited to imaging two or three of these organelles at one time because of this problem of
overlap in emission spectrum in among the different
fluorescent proteins that are available. So here we have classic
fluorescent proteins that have different emission spectrum, CFP, GFP, YFP. These are the emission
spectrum for each of these different fluorescent probes. And the problem is their
emission spectrum is overlapping. And what that means is that if you tagged a particular organelle with
different fluorescent probes like the ones that I just mentioned which are the most widely
used fluorescent probes, and then you image essentially, you go across the emission spectrum to look at any particular
wavelength of light which population of
organelle you’ve looked at, you can see that you have overlap at any particular wavelength. So that means if you were
imaging for instance at 438 you’d see three different, you would not be able to
distinguish mitochondria, ER and lysosomes from each other. It would just be one big blur. And hence you could not distinguish how these different
organelles are behaving relative to each other. So to overcome that challenge, two postdocs in the lab
decided to employ a technology called Multispectral
Imaging to try to unravel this overlapping emission spectrum. And a strategy that they
used was essentially take, if you know the the emission spectra of each of these different fluorophores you can then query an
observed pixel spectrum that’s a combination of one
or more of these fluorophores, and then use linear unmixing to decipher what combination of and in what abundance any one or two or more of these fluorophores would give rise to this
particular spectrum. And using that we can at
each pixel of our image unmix to determine which
fluorophore is giving rise to the signal that we’re
observing at that pixel. And so we combine that with
Lattice Light Sheet Microscopy to be able to image simultaneously six different organelles
within the cell over time. In order to do that we’ve
essentially introduced six different laser lines that
cover the visible spectrum that allows us to excite the fluorophore that we’ve tagged on each one
of these different organelles. And then we do our linear
unmixing algorithms to determine each of the specific spectra
associated with each organelle. And this is what you can see in the case of these six different
organelles that we’ve introduced fluorescent tags for. Now this is in a single cell so we can superimpose all of these signals on top of each other to simultaneously see in a three-dimensional space because we’re using
that lattice light sheet to move through the
whole volume of the cell, how all of these
organelles are distributed. Now with this technique, we can begin to zone in on really specific measurements in terms of organelle distribution, localization, essentially connectivity. This is just a set of values for the number of organelles
that each of these different populations
of organelles represent. So for instance on average in
the cell that we’re looking at there’s about 89
lysosomes, 186 peroxisomes, 157 lipid droplets. The ER occupies by far the
largest volume in this cell among all of these different organelles. It’s about 30 times the
size of the Golgi apparatus which is involved in
the secretory pathway, eight to nine times the
size of the mitochondria which is involved in energy
production within the cell. Now we can also come in and segment these individual organelles to look at how they are connected to each other, how they’re contacting each other in order to begin to
understand the communication, the cross communication
or crosstalk in activity that we know is so important for how cells are operating and communicating with other cells in their environment. For instance the endoplasmic reticulum controls the secretory pathway together with the Golgi apparatus. It’s what’s secreting
the paraneural network that we heard earlier about. It’s critical for us to really understand how these organelles behave
relative to each other and from these types of segmented images, we can create what we’ve
been able to describe what we call the organelle interactome. Where we just measure
the pairwise contacts between these different organelles in our images of these cells. And from that we can see the
frequency of communication that different organelles
have with each other. You can see here that the ER is by far the most communicative
of all of the organelles, it’s contacting everyone. Importantly if you look
at a single cell over time you can see that this
organelle interactome is conserved over fairly
significant periods of time, and that is despite the fact that any particular contact that we see is relatively transient. I should emphasize that this interactome changes dramatically if we perturb the cell in different ways, we can depolymerize microtubules or starve cells in different ways, and we dramatically change
the way these organelles are interacting with each other. Now this is a movie
where we’ve segmented out the mitochondria as an
example of an organelle that is intimately communicating with the endoplasmic reticulum. On the right hand side
represents the surface, all of the surface sites of mitochondria where we see ER signal. So the ER is wrapping around
the surface of the mitochondria and intimately
communicating with virtually all of those mitochondrial
elements that we see in the cell. This contact we think is what’s up, we think there’s calcium flux between the ER and mitochondria. That calcium is playing an important role for mitochondrial output, how much energy is being produced by the mitochondria. We also know there’s lipid and
cholesterol being transferred across these contact sites, and importantly reactive oxygens, ROS, reactive oxygen species
is being trafficked across those contacts which could play a big role in the disulfide bond formation
and protein remodeling occurring in the endoplasmic reticulum. Now let’s focus in on the ER for a second. It occupies 25% of the cell cytoplasm. We can measure that using
our lattice light sheet three-dimensional reconstruction. What is interesting is that if we look at, we plot out the position of the ER over a 15-minute time period which you can see in this movie, what we find is that the
ER has pretty much explored the entire cytoplasm over just 15 minutes. So it is a very dynamic organelle that has lots of capability
of communicating. Now in my final two minutes or one minute I wanna take you through how
fast these organelles can move. We know that the ER has the capability of exploring that cytoplasm. Let’s look at its dynamics
at higher resolution. We can do this using a TIRF-SIM system. We can see these tubule
matrices move incredibly fast. Interestingly the tubes themselves undergo an oscillatory activity, that’s ATP and GTP-dependent so it’s not just thermally driven. And finally we can actually come in and start mapping out individual proteins that move or diffuse along the surface of the endoplasmic reticulum. That’s what you’re seeing here. Each of these yellow spots
represent a halo tagged protein that is associated with
the membrane of the ER and we can begin to map out the trajectories of these proteins, and that’s shown here for another ER resident protein Sec61 beta. If you sum up all of these trajectories you see that these proteins and this is a transmembrane
protein you’re looking at, will explore the whole
surface of that ER freely. Now in my final movie here, I just wanna show here’s an example where we’re mapping out a protein
that actually interacts, it’s on the surface of the ER but it is part of a tethering complex that brings the ER close
to the mitochondria. And what we can see is when we track these individual molecules
as they diffuse across the surface of the ER, we can see that as they move
across the area of the ER that’s in close proximity of mitochondria, they slow down significantly. Consistent with a transient interaction of this tethering protein with the target protein on mitochondria. So with that I wanna end and say that this field is really being
significantly impacted by the high resolution
technology that’s now available that scans from electron microscopy up to the fast imaging technology. People in my lab have
been greatly impacted by the physicists at Janelia, Eric Betzig and Harald in particular who’ve really provided the technology that allow us to do this type of work. Thank you. (audience applauding) – So as you know, life sciences are always trying to make best use of light for many different purposes. For instance we use x-ray
beam for crystallography and the infrared light for
vibrational spectroscopy. But most biologists use visible light. And chromophore is a structure unit so that can absorb certain visible light, and it is responsible for color and in many cases it has
the pi-conjugation system. And single and double
bonds appear alternately so that the Pi electrons
can be delocalized. So electrons oscillate or
sing on the chromophore and that’s quite important
for the interaction with the visible light. And my laboratory is engaged in technological innovation in bio-imaging and principally using
fluorescent proteins. And I’d like to introduce to you, so the most classical
the fluorescent protein Aequorea GFP. And long time ago, so in 1962 Osamu Shimomura discovered protein and from the light-emitting
organ of the jelly fish. And just 30 years later in ’92 it’s seen then it was cloned by Doug Prasher. And in ’94, so the heterologous
expression with Aequorea GFP was achieved by Martin
Chalfie at Columbia, okay. And then this is the primary
structure of Aequorea GFP, just 238 amino acid. So there’s no chromophore
that I defined a moment ago in the peptide. But from this three amino acid there’s serine, tyrosine, glycine so three reactions occur
spontaneously, okay. And cyclization, dehydration
and oxidation reactions to make the Pi-conjugation system. So this is the GFP’s chromophore and which observes blue light, okay? But this light doesn’t say
anything about its fluorescence. This is a crystal
structure they call a GFP and the beta-barrel structure. And the 11 stranded beta-barrel with one alpha helix inside, and the chromophore is
formed on the helix, and that the barrel is very robust. And the chromophore is packed inside and it takes very rigid structure so that explains why GFP’s fluorescence quantum yield is very high. And the mutagen studies get
done by Dr. Roger Tsien, so produced quite a few color variance and the the more useful and brighter. And I stayed in Roger’s
lab from ’95 to ’98 and I used CFP and the YFP and as the donor and acceptor
for FRET so to create. So the genetically encoded
indicator for calcium ion, Cameleons as Mark mentioned kindly. So now I’d like to discuss
why calcium imaging is so interesting and appealing. So there should be two major reasons. First, calcium signaling
is very, so dynamic. The concentration of free calcium ion it changes so greatly so it should be fun. To observe for instance
here calcium oscillations. The second, so due to the endogenous and the very abundant
calcium buffering systems we can express a huge
amount of calcium probes so without affecting intracellular calcium dynamics very much, okay? We can increase the signals, signal-to-noise ratio. Here we expressed a large
quantity of Cameleon in excitatory neurons
with the forebrain mouse. And we signed the skinless
head with excitation light and to get these calcium readout. And through this intact
skull and the video rate for more than 30 minutes. So this readout reflect
spontaneous neuronal activities and which are composed
of multiple oscillations of different frequencies
and very symmetrical. And the mouse was half awake so when we gave visual stimulation so we saw an evoked response in the path, in the visual cortex. Now back to, so this slide. GFP is chromophore and it has a phenol ring. So it comes from tyrosine residue, okay? Well, almost all of the fluorescent proteins now available so do have phenol rings in
there chromophores, okay? And the exceptions, the CFP and the BFP, so they have indole ring or imidazole ring coming from tryptophan and the histidine. But phenol ring containing
chromophore is most common. And that there is an equilibrium between protonated and ionized state of the phenol hydroxyl group. And that they then absorb
400 and 480 nanometer light. And in many cases, so the
ionized form fluorescent, okay. Therefore, GFP basically has bimodal absorption spectrum. And now a circular permutation. And a long time ago just serendipitously Geoff Baird and in Roger’s lab found, so this site 144, 145, so the midpoint of the beta
strand number seven (mumbles), so circular permutation. So circular permuted GFP can synthesize chromophore very well, and now so the chromophore. The equilibrium is quite
sensitive to a neighboring event such as a calcium-dependent
protein-protein interaction. So this is the operational
principle of GCaMP or pericam. The very important point is that so that here calcium-binding
changes absorption spectrum not fluorescence quantum yield, okay. As a kind of, a future perspective, so the GCaMP should be well combined with photo-acoustic imaging. Now I have to say and
not only the jellyfish but some of them are done
in animals or marine animals so can produce quite similar
fluorescent proteins, okay? And while we have cloned
many new fluorescent proteins from those animals. By using this, so these
green red proteins, we some time ago we
developed cell cycle probe. And we used the Ubiquitin Oscillator. Cell cycle-dependent proteolysis, so to create the cell cycle probe which labels individual
nuclei in genome phase red and those in S, G2, M phases green. That’s FUCCI and the
green and red indicate, so the go and stop. That S phase entry, so
like a traffic signal and when the FUCCI was
introduced into HeLa cells, so derived from malignant tumor we observed a cell cycle operation in the other time at a single cell level. But when we did the same thing using benign tumor derived cell line we saw very clear contact inhibition. And upon reaching convertible urea all of the cells became red and then stopped so their proliferation. And we introduced scratch
and so some of the cells or that one in green
after some rate and time, so they reenter the cell
cycle and to build a gap. And we prepared FUCCI transgenic mice, so this is a coronal section of an embryo and in the brain, so neural stem cells at the green nuclei line
in the ventricular zone but post-mortem neurons
had only red nuclei in the cortical (mumbles). And the green and the red
signify cell proliferation and the differentiation. And during the embryogenesis with time, so green to red ratio decreased, so the earlier the more proliferation, the later the more differentiation. And after the burst so the
red signal became predominant but even in adult tissue
so we can identify very few green nuclei and the cells with the green nuclei corresponded to tissue stem
cells like a neural stem cells found in the dentate gyrus, okay. And when we used zebrafish
embryo very transparently so we can obtain cell cycle
profile in four dimensions. So this is a segmentation period of embryo and a side view on head and tail, and again with time green
to red ratio decreased but some other organs
like retina and the brain, so very green. And in top view in addition to two eyes, so we saw a notochord. So differentiating notochord and in notochord we discovered, so the cell cycle transitions
and moved, traveled from head to tail like this. So this G1, this transition
moved towards here. And we think, so the cell cycle regulation in the notochord is
linked to somite genesis which is also characterized
by head to tail progression of differentiation. But anyhow, it’s quite very amazing that we can visualize that
the middle of the body from outside so that’s just because, so the fish embryo’s a bit transparent. And by contrast, mouse embryos, well, mammary animals where
including us are turbid. Yes, visible light
scatters very much inside so our body so like now mentioned, okay? Now, tissue clear identification and long time ago we discover that the membrane for Western Blotting analysis it became transparent when
soaked in 4 molar urea. And we knew that, so the GFP beta-barrel, it’s a brave and tough, so rigid and tolerant of high
concentration of urea. We started, so the clearing
and devolution 2011, so with our paper on ScaleA2 which is the first and urea-based aqueous solution for tissue
clear identification. And then it was followed by many explosion of many new techniques, and also two years ago we
reported some modification ScaleS, ScaleS that contains
sorbitol in addition to urea and we verify that ScaleS can preserve the fluorescent signal
and also outer structures, so subset of outer
structures like a synapse. I think researchers should be aware of the critical trade-off between effective clearing and the tissue signal preservation. And also we invented AbScale method so which enables three-dimensional
immunohistochemistry. And some time ago the Saito group at BSI generated a single APP
knock-in mouse models of Alzheimer disease. And we applied AbScale method, so to the brain of aged APP knock-in mice. And to visualize A beta-plaques we used Alexa488 labeled and antibody. And there was a brain got
prepared from a nine-month old and APP knock-in mouse and very sparse. But in 18-month old, so we saw, we observed so many
immuno-labeled A beta-plaques and yes, even the experiment, we transcardially perfused
the entire vasculature with Texas Red-lectin to label blood vessels
with red fluorescence and then we immune-stained
the A beta-plaques with green and to obtain this, so the 3D perspective. Very comprehensive perspective of how A beta-plaques interferes
with blood vessels. Then we took interest
in a special association of A beta-plaques with microglial cells. And we applied so the dual color AbScale so to a brain slice prepared
from APP knock-in mouse. And we analyzed the interaction between these two object by light sheet microscopy observation. And we measured the distance
from each microglia center to the nearest plaque age in the 3D space. Those measurements should
provide the information about stability of plaque
neuritis, inflammatory state. For instance, so two neighboring
plaques with similar size but different association
with microglial cells, very direct to contact
versus very isolated. We think they suggest acute
neuritic state versus obsolete. And in a similar way we 3D imaged plaque microglia association in
post-mortem brain samples of AD patient of more than 50 years old. And in these three brain
samples we observed, detected very clear cored
plaques, green ones. And the cored plaques without and with microglia association. But in the remaining
six of the brain samples we didn’t see any cored plaques. So then we became interested
in diffused plaques. So diffused plaques are
basically undetectable, not detectable in the
two-dimensional image. But our new software can re-slice, so the 3D reconstruction systemically and with different sizes, different thickness, okay? And that helped us to identify diffused plaques of different sizes. And the diffused plaques are usually assumed to be very
primitive and very isolated from any inflammatory cells, but we found, so most of
them can show the sign of neuritis with considerable association with microglial cells. Clinically, diffused plaques are ambiguous but they’re potentially
very, very important, okay? And Scale plus FUCCI we
studied vascular niche for neural stem cells
in in the dentate gyrus. And that we were able… So neural stem cells or
proliferative neural stem cells we used FUCCI S, G2, M phase marker and the blood vessels
were labeled with red. And the green nuclei,
so the neural stem cells were indeed localized
inside the dentate gyrus not outside of it. And we again performed
distance measurement in a three-dimensional space. We measured distance
from each green nucleus, so to the nearest blood vessel, okay? And then we came to a conclusion, so the neural stem cells are more closely associated with blood
vessels than matured neurons. And Rusty has been saying about hippocampal neurogenesis is operated by exposure to enriched environment and we performed the
comparative experiment, comparative experiment and
large-scale 3D reconstruction. And our data showed the
enriched environment increased so the number or density
of neural stem cells but did not change so their
association with blood vessels. This a comprehensive data I think, I hope. All right, so this slide is an interplay between light and life. Well I remember that Roger Tsien, Roger used to remind us
nature is very kind to us, to researchers and that nature still
remains the best source of bio-imaging tools, okay? And this is my last slide and only a scientist
with a respect for nature would have been permitted to surpass it in English, Japanese and Chinese. And thank you very much for attention. (audience applauding) (uptempo music)


1 Comment

Lowell-James Hicks · February 3, 2018 at 3:55 am

Fascinating, but Miyawaki really needs some coaching in articulating his obviously high-level command of English if he is to present in that language.

Leave a Reply

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