Molecular and Cellular Mechanisms Underlying Human-Specific Evolution of Cortical Connectivity


(lively piano music) – [Narrator] We are the paradoxical ape. Bipedal. Naked. Large-brained. Long the master of fire,
tools, and language. But still trying to understand ourselves. Aware that death is inevitable. Yet filled with optimism. We grow up slowly. We hand down knowledge. We empathize and deceive. We shape the future from our shared understanding of the past. CARTA brings together experts
from diverse disciplines to exchange insights on who we are, and how we got here. An exploration made possible by the generosity of humans like you. (uptempo music) (upbeat music) – My lab is primarily interested in understanding some of the
molecular and cellular basis for the development of the cortex. But unlike some of the previous speakers, we’re focusing on the development
of neuronal connectivity and how circuits assemble
basically in the developing brain. And so, I’m gonna move
fast on the introduction because I think you’ve heard a lot about the general interest of
understanding what makes us human and in particular, what makes
our brain human-specific. Another way to phrase that question as you heard before is to ask, what happened to our
genome since we diverged from our common ancestor more
than seven, 10 millions years, there’s actually debate about that exact time for divergence from our common ancestor. So, to the first question, what makes our brain human-specific? You have heard a lot today about the possibility that what makes our brain different from our most
common living relatives like chimps and bonobos, or to other higher mammals,
would be brain size or at least the size of the cortex relative to the rest of the brain. I would argue that despite the
fact that that’s the question that most people are interested in, it cannot be the entire
answer for a simple reason. Which is that if you look
at brain size expansion including increases in neocortical size, many mammals have succeeded at this without any real gain
in cognitive abilities. In fact, brain size is best
correlated to body size not to cognitive abilities, right? So, brain size and neocortex expansion were probably an important
step during evolution and it happened many times
during mammalian evolution but it can’t be the only explanation for what makes our brain human-specific, at least in emergence
of cognitive abilities. Many are open remaining possibilities, neuronal composition, right? That the type of neurons that are produced that Rick and others have
talked about earlier. Circuit connectivity, the
total number of synapses made between neurons are things that will influence circuit function and probably cognitive abilities. Something that very few
people have talked about, non-neuronal cells, astrocytes, microglia. There’s actually some very recent evidence suggesting that differences in composition or total number of non-neuronal cells might be different between human and non-human primate. So today I’m gonna try to convince you that we have evidence,
probably the first evidence suggesting that a specific
gene duplication event has pretty significantly changed both the number of synapses
made between neurons, and we think actually affects circuit connectivity and circuit function. So, how did we get
interested in this question? We got interested in this question based on the work of
several groups including a beautiful work from Evan
Eichler that you’ll hear about just after me so I’m not
gonna bore you with details about what Evan did. But back in early 2000s
work from Jim Sikela and then refined and expanded by Evan identified gene duplication events that are specific to the human lineage. Gene duplications, new gene
copies that are not present in any other non-human primates
or in any other mammals. So essentially what Evan has shown, shown here on this graph. So what Evan has shown is that in humans you have a whole list of gene, about 30 genes that have
extra copies essentially. Those copies are specific to humans, they’re not found in non-human primates. And unfortunately or fortunately for us, most of those genes have
completely unknown function. That’s not entirely true. There is one gene in that list, SMN1 which is the gene that’s mutated in spinal muscular atrophy. Turns out that humans and humans only have an extra copy of this and it’s actually
absolutely fascinating work that’s been done on this. There’s a drug treatment that
just emerged very recently actually taking advantage
of the second copy, the human specific copy of SMN1. If you’re interested I
can tell you more about. But for the vast majority,
all the other genes have largely unknown function. And so, except that around the time when this first work was published, there was one gene in that list, SRGAP2 listed here, that we were probably the only lab working on this gene at that time. And we had some interesting data suggesting that this gene was actually playing important functions during brain development, at least the ancestral copy of this gene. But when Jim Sikela and Evan
Eichler published these results were struck by the fact that there are human-specific copies of this gene. And so, we embarked on
trying to characterize what those human-specific
duplications of SRGAP2 do. So as a paradigm basically many questions arise from
from this observation that there are human specific
gene duplicates for those. The first one is which of those genes are expressed in the human brain? Second question is what is the function of the ancestral copy of these genes? I mentioned most of those genes have completely unknown function
during brain development so we had to start with this. And the third question
that’s truly interesting is, is the function of the
human specific paralogs related to the function
of the ancestral copy? Or are those new copies basically acquiring some completely
independent function, some completely new functions independent of the function
of the ancestral copy, right? Both scenarios could be true. So I’m gonna summarize here essentially about 11 years of work from my lab, from multiple people in my lab. And then during the rest of the talk I’m gonna unpack and
tell you a bit more about how we got to those conclusions. So essentially SRGAP2A
is an adapter protein containing three functional domains. This bottom end which is essentially a homodimerization motif. A central domain called the RapGAP domain. It’s a domain that inactivates
small GTPas called Rac1. And an SH3 domain that’s essentially a protein-protein interaction motif. All mammals from rodents
to non-human primates and humans have this gene. It’s expressed, it’s highly
expressed in the brain, it’s expressed largely in only in neurons and as I’ll show you later on, it’s actually very unreached at synapses. But it turns out that about two to four million years ago, two gene duplication events at least made this new copy that’s
human-specific emerge, that we had to call SRGAP2C. And Evan will probably tell
you a bit more about that. And so, this coding sequence
is truncated actually, it doesn’t express the
full-length protein. It’s truncated, it expresses
90% of this bottom end which remains able to
bind to the ancestral copy and largely inhibit its function, okay? So its main function, so it turns out that we get essentially the same phenotype when we inactivate the SRGAP2A or when we induce the
expression of this gene, okay? And the phenotypes, we
get our delayed excitatory and inhibitory synaptic maturation, this phenomenon called neoteny, which is essentially defined by retention of immature features for longer periods of
time during development. We get increased density of both excitatory and inhibitory synapses. And finally, we get increased
cortical connectivity. This is largely unpublished, probably won’t have time to
tell you much about this. So, how did we reach these conclusions? So the first thing that we
did a long time ago in 2009, we published expression pattern for SRGAP2 and this very simple
observation that the protein in mouse cortex is expressed
actually peaks at P1 but is expressed largely at postnatal time when between, during the first
two weeks of postnatal life when the animals are basically
forming a lot of synapses. And so, when this very talented postdoc, Cecile Charrier, joined the lab, she discovered that in fact SRGAP2 protein is very unreached at synapses. This is high-resolution images of individual synapses in mouse neurons and I hope you can
appreciate the fact that essentially there is very
little SRGAP2 in small synapses and lots of SRGAP2
protein in large synapses. Those synapses here are visualized, those excitatory synapses are visualized by this protein Homer1. And it turns out that at that time we didn’t know but in
fact those two proteins in function interact. So, what is SRGAP2 function
in synaptic development? So, neurons are absolutely amazing cells if you think about it, right? Those cells are gigantic. A fibroblast that people
have talked about before would be the size of this
cell body here, right? So they’re gigantic cells and a graduate student
in the lab, Dan Iascone, has developed some new
methodology to actually at single cell resolution in vivo visualize the presence and the topography of all synapses, excitatory
and inhibitory synapses, made into a single neuron, okay? And so, this allows us to understand both to have very quantitative ideas of where synapses are localized in neurons, how many they form, and you know, for a
typical neuron like this, this layer two, three pyramidal
neurons in the mouse forms somewhere between 4,000 and
6,000 excitatory synapses. And somewhere between 1,000 and 1,500 inhibitory synapses, right? So, staggering numbers
for just a single cell. And so, if you zoom up on those green dots that you see here, those are individual
spines, dendritic spines. I’m gonna talk a lot about
those in the rest of the talk. Those protrusions, those micron, the scale is a one micron here. They’re tiny protrusions. They essentially constitutes
individual synapses, if you look at an EM picture
of a single spine like this, these protrusions, this
is the site basically. A single spine constitute
a single synapse, it’s the postsynaptic receiving
end basically of a synapse so this is an axon with those neurotransmitter-filled vesicles here that release neurotransmitter
in the synaptic cleft, that then activates
postsynaptic receptors here. So, essentially those
synapses are the basis for neuronal communications, right? And decades of work by hundreds of labs have identified essentially
a zoo of proteins, hundreds of proteins that are localized pre and postsynaptically here. The dendritic spines. Those proteins, the main
job of those proteins including this Homer/Shank
interacting proteins, I’m gonna tell you in a second, is to basically create
a very complex scaffold to anchor and stabilize the
neurotransmitter receptors here that sense neurotransmitter release here. And so, we discovered that SRGAP2 look is actually directly
interacting with Homer here, and in fact I’ll show you later on SRGAP2 seems to control
the scaffolding properties of those proteins and controls
the rate of accumulation of those neurotransmitter receptors such as AMPA receptors during development. So how did we discover this? Takayuki Sassa, the first postdoc who was involved in this
project started by making a knockout for SRGAP2A for
the ancestral copy in mice. So, we made this knockout in order to ask what happens when we delete this gene? And the main thing that happens
when you delete this gene is essentially that if
you look at those spines, the size of the spines I forgot to mention is proportional to how mature it is, how many postsynaptic receptors have accumulated there. If you look at an animal
three weeks after birth, a wild-type animal, those spines are already mature. So they have mature synapses. There’s a lot of data that has shown this. But then if you look at the heterozygous, so if you delete one copy
or two copies of the gene, you have significantly smaller spines at this juvenile time point. They have longer necks. So the neck that connects the spine head to the dendrites is longer, and most interestingly,
there’s a pretty significant increase in density of those spines. But interestingly enough this
size effect is transient. So if you look at P65 at 40 days later, those spines reach maturation. So it’s just a delay in maturation. Those spines when you inactivate SRGAP2 seems to be much slower
at reaching maturation. Instead of being mature at P21 they’re mature at P45. But in the adult basically
the spines remain having a longer neck and very significantly
higher numbers basically, very higher, significantly higher that. So what happens if we
humanize mouse neurons for SRGAP2C, right? I briefly mentioned before that our model based on
studies we published before was that the truncated version, the human-specific version of this gene would bind to and inhibit the function of the ancestral copy, right? So the prediction would be that if we express
SRGAP2C in a mouse neuron that expresses SRGAP2A, we would get phenotype that
look exactly like this, right? And that’s exactly what we found. Essentially if you look at
juvenile animals control or SRAGP2C humanized mouse neurons, you get this delayed maturation, right, but they actually reach maturation at P65 instead of P21. And they retain longer necks
and higher spine density. So this was a very interesting
result for us, right? This is because essentially we know that those three features
delayed maturation, a longer neck which actually has very
important function, and a higher density are three phenotypes that are known to
characterize human neurons compared to either mouse
neurons or non-human primates. So beautiful work from Javier
DeFelipe, Rafael Yuste, my colleague at Columbia, have shown that if you look
at spine density for example between human or non-human primates, or if you look at spine density between human and mouse neurons, you essentially get 30% to 50% more dense, higher density of spines and those spines have longer necks. And we also know that
synaptic development in humans is profoundly neotenic. So basically very prolonged maturation, very delayed maturation. So this is interesting, right? Essentially with just
introducing a single gene that’s human specific in mouse neurons, we’ve seen a copy three major aspect of synaptic development. Delayed maturation, higher density and change in the morphology
of those spines, right? But we have a problem which is that we know that changing and increasing the density of excitatory synapses cannot be the only explanation since beautiful work from
Javier DeFelipe and many others have shown that there is an
amazing degree of conservation in the ratio between excitatory
and inhibitory synapses in the cortex, right? So, we know that if SRGAP2, if that’s the only thing SRGAP2 is doing, increasing the number
of excitatory synapses, we would have a problem, right? We would have a major imbalance between excitation, inhibition which is the landmark of many neurodevelopmental disorders for example such as autism or epilepsy. So either SRGAP2 was doing the job and was coordinating the maturation of both excitatory and inhibitory synapses or some other genes, maybe some of the other gene duplications that I mentioned before would do the job. It turns out that nature
has selected SRGAP2 probably for a reason. It’s because SRGAP2 does
exactly the same thing it does at excitatory synapses and inhibitory synapses. So we discovered this
because more recently because now we can actually visualize inhibitory synapses with amazing accuracy using for example this gephyrin GFP. Gephyrin is a protein
that’s exclusively localized at inhibitory synapses. And using single cell technology, single cell genetic labeling technology we can not only visualize with
submicron precision spines, we can actually localize
those inhibitory blue tones either the ones that are made
on the dendrite shaft here or even more rare blue
tones that are made directly on spine heads here. I can tell you more about this later on. And remarkably, we find
that the same thing happens then for spines if you look
at inhibitory blue tone, if you down regulate
SRGAP2A, the ancestral copy, or you introduce and humanize mass neurons for SRGAP2C expression, you get higher density
of inhibitory clusters. Those inhibitory synapses
are much smaller early on but ultimately reach maturation at P65. And there’s a change in
their actual localization, the striking increase in the one, the inhibitory synapses made on through the spine heads here. So increased density
and delayed maturation. It’s pretty remarkable. In fact, if you plot the normalized increasing density for spines and for inhibitory
synapses between control and humanized mouse neurons, we get remarkably similar effects, right? So the conclusion here is that nature has selected the
gene duplication event that of a gene that essentially controls the co-evolution of inhibitory and excitatory synapses, right? This is really I think a very important take-home message here. So, I will finish here by taking a minute to recap everything I told you today. We’ve identified a gene and SRGAP2A which in its ancestral form
expressed in all mammals controls three very important aspect of synaptic development. It controls the density of
excitatory inhibitory synapses through work, I didn’t
have time to show you. We were actually able to do
structure function analysis in vivo using gene replacement strategies. And so, we can actually dissociate how SRGAP2 regulate spine density and inhibitory synapse density from how it controls the maturation of excitatory or inhibitory synapses. And we discovered that the
human specific copy of this gene which is this truncated form which remains able to
bind to the ancestral copy and inhibit its function
inhibits all three major functions of SRGAP2, right? So what are the future
directions on this project? So, there are three future directions. The first one is what are
the upstream regulator of this protein? I didn’t tell you anything
about how this gene is regulated and how the protein itself is regulated. We have some evidence of
what’s upstream of this. The second very important question is, where are these increased
number of connections coming from, right? So this gene basically, the emergence of this humans-specific copy increases the total number of synapses by 30% or 40% which is very significant. Where are those connections
coming from, right? We have some ideas about this already. And finally probably the most important is what aspect of circuit function has been affected by the emergence of this human-specific copy? That’s something that’s much
more difficult to approach but believe it or not, we actually have some very
interesting results about this. So finally, the people who did the work. The work was largely carried out by a very talented postdoc
in my lab, Cecile Charrier, who now has her own lab at Ecole Normale Superieure
in Paris in France. But she was assisted by many people and many people in my lab, at least those four people listed here are involved in current
aspect of the project. And finally I’m excited
since I moved to Columbia about four years ago, Columbia is creating a
third campus actually called Manhattanville, and very recently we went
from an architectural sketch to an actual building that will host 55 labs,
about a thousand people. So, the Zuckerman Mind
Brain Behavior Institute. We’re very excited to
move into this building hopefully this fall. Thank you for your attention. (audience applauding) (upbeat music)

, , , , , , ,

Post navigation

3 thoughts on “Molecular and Cellular Mechanisms Underlying Human-Specific Evolution of Cortical Connectivity

  1. I am so grateful that I have the opportunity to follow lectures like this, watching from a different continent. Thank you guys.

Leave a Reply

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