In the last lesson we saw how postsynaptic conductances evoked by glutamate release
from presynaptic terminals were converted into postsynaptic potentials and
integrated in complex ways across the dendritic arborization.
We found that most unitary excitatory postsynaptic potentials (EPSPs)
that originate from a single action potential in a single presynaptic neuron
that most unitary EPSPs are small in amplitude.
They have an amplitude typically of 1 mV and the last 10-20 msec in duration.
An excitatory neuron is typically at rest at around -70 mV, so something like
30 mV away from action potential threshold.
In order for a postsynaptic neuron to contribute to further computation
in the brain, it must receive those excitatory synaptic signals
depolarized beyond threshold and fire action potentials of its own
at the appropriate time.
This means that many synaptic inputs must be integrated in that individual neuron.
In order to understand how that neuron integrates sensory information,
we need to know where its presynaptic neurons are.
We need to map the wiring diagram and the connectivity of how synaptic input
arrives onto postsynaptic neurons.
In this lesson we'll begin to think a bit about how synaptic wiring diagrams
can be constructed at least in a probabilistic way and how that might
relate to sensory processing in the mammalian brain.
We'll begin by considering sensory processing in the mammalian brain
and its dependence upon glutametergic synaptic signals.
We'll think about vision and touch: two important senses for us humans.
If we think about the sense of touch, there are mechano-gated ion channels
sitting in our skin inside axons of sensory neurons.
Touch on the skin causes action potential firing in the sensory neuron and these
sensory neurons release glutamate from their synaptic terminals
somewhere here in the brain stem.
The brain stem neurons in turn receive those excitatory glutametergic signals
and send them on to the thalamus, the somatosensory part of the thalamus
deals with the sense of touch. Neurons there receive the
glutametergic signals from the brain stem neurons and in turn they generate PSPs,
action potentials and send the sensory signals up the neocortex
where it's thought that our conscious precept of touch is driven by the
activity of the neurons in the neocortex.
Again, most of that activity is driven by glutamatergic synapses.
And so we have here a pathway of brain stem synapse and thalamic synapse
and neocortical synapses, all of which are primarily
glutamatergic in nature and essential in order to convey
a sensory signal from the periphery to the neo cortex, where it might generate
sensory percepts.
Similarly, if we think about our sense of vision, this also depends heavily
upon glutamatergic synapses.
Within the eye, the photoreceptors that detect photons arriving
from the visual field, they release glutamate
on to second (inaudible) so called bipolar cells
in the retina.
These bipolar cells sense the glutamate and in turn release glutamate at their own
synaptic terminals, on to retinal ganglion cells.
It's the retinal ganglion cells that then receives its glutamatergic
inputs as far as action potentials and transfers the visual information
first to the thalamus where again a glutamatergic synapse
conveys that information further to the visual cortex.
Again, glutamate is the second molecule there.
And so glutamatergic synaptic transmission is a fundamental importance
of getting sensory information from the periphery to the neocortex
where it drives sensory perception.
And that's true for all of our senses.
In this video I'd like to focus particularly on one sensory pathway
that has some specific features that are useful for detailed analysis
in terms of microcircuits and synaptic connectivity.
We'll think about the <i>Mouse Whisker System</i>
and the mouse is an attractive animal in terms of neurophysiology
because of its extensive genetics and its extensive use
in biological research.
Mice are nocturnal animals, they live underground in tunnels
and so vision is not one of their most developed senses.
Instead, one of the ways in which they gather
a lot of sensory information about their immediate surroundings
comes from their array of whiskers that are present
on the snout of the animal.
These are arranged in a highly organized manner
and using their whisker system, the animals are able to navigate
in tunnels, they can see whether a hole is a sufficiently large aperture
for them to enter in, they can detect the shape
of objects and their sense of texture discrimination
with their whiskers rivals that of the human finger tip.
So this whisker system is an immensely sensitive organ
through which the mouse can learn a lot about its immediate
spacial environment.
Deflection of one of these whiskers activates mechano gated ion channels
sitting here in the whisker follicle and that drives actual potential
firing here down the trigeminal nerve.
There's a first glutamatergic synapse in the brain stem and neurons
in the brainstem receive that glutamatergic signal
postsynaptic potentials are generated, some of the neurons in the brainstem
fire action potentials and those signals in turn,
are relayed to the thalamus.
Again, thalamic neurons here receive the glutamatergic input
post synaptic potentials are generated and integrated
in the thalamic neurons, action potentials are fired
and subsets of those thalamic neurons and information is then further relayed
to the primary somatosensory cortex where presumably,
whisker sensory perception takes place for the mouse.
So there are three glutamatergic synapses that relay
sensory whisker information from the periphery
through to the neo cortex of the mouse, and in particular
it's the primary somatosensory neo cortex of the mouse
that deals with whisker sensation.
The organization of the primary somatosensory neo cortex of the mouse
is remarkable in the sense that it has obvious anatomical units
that appear to be in a direct one to one correspondence
with the layout of the whiskers on the periphery.
So the whiskers here, the hairs that eminate from individual
whisker follicles are laid out in a highly stereotypical patent
where every mouse has a same organization of the whisker lay out
and indeed mice and rats have also the identical layout.
So because it's highly stereotypical we can then label these whiskers
in specific ways and so this is called the <i>A</i> row, the <i>B</i> row, the <i>C</i> row,
<i>D</i> row, <i>E</i> row and equally they're considered to be
archs <i>1,2,3</i> and <i>4</i> in terms of the whiskers and so this one here
that's highlighted in yellow is a so called <i>C2</i> whisker.
If we now look at the new cortex across the horizontal extent
of the neo cortex when looking down on the face of the neo cortex
across a so called somatotopic map, we'll see that there are corresponding
units here drawn in blue that correspond to the whiskers.
Again, highlighted here in yellow is one particular anatomical unit.
The so called <i>C2</i> whisker representation and these round things here
are called barrels.
So this is called the <i>C2 barrel</i>.
Present in primary somatosensory cortex and because of its obvious analog
in terms of the whiskers on the periphery one might think that sensory information
relating to the <i>C2</i> whisker is primarily processed in the <i>C2</i> barrel column
in the primary somatosensory cortex, and indeed, at the various stations
in the brain stem and the thalamus there's a nice representation
also of the individual whiskers in terms anatomically identifiable units.
Now if we look at a section of the brain where we cut through
the thickness of the brain and so before we were looking
at the top of the brain and we saw that there was
a nice C2 representation of the whisker.
Now we look inside the brain and we see that it has some depth.
It's about 1mm thick, the neocortex in this particular location
in the mouse.
In here maybe you can make out the walls of the C2 whisker representation
that's here, high cell density represented here
by this fluorescent stain of DAPI that stains the cell nuclei
and you can also see it here in this stain of neurons
present in this air, you'll see that there are these walls here
that represent these limits here of the C2 whisker representation
and now this would be the surface of the brain up here,
and here we go deeper into the brain and this is the thickness of the neocortex.
Its about 1.2mm thick.
Now there are some obvious differences at different depths in the neocortex
and people have divided the neocortex into different layers,
and so there's a layer one, the most superficial,
the outside area of the brain that has very few cells in it.
It's mainly composed of synapses, axons, dendrites and synapses.
The cells themselves begin here in layers 2 down to layer 6,
and you'll see that there are different densities of cells in different layers.
And here we've labeled some individual neurons
that are present of these different layers and you can see
that there are some different morphologies of the cells.
And so there are very small cells that are present here
in this so called <i>Layer 4</i> area that's actually where the main sensory
input arrives in the neocortex and in the superficial layer,
there are so called <i>small pyramidal</i> cells small excitatory neurons,
and in the deeper layers, there are much larger cells
with much more extensive dendrites.
And it turns out that the microbiology, the electrophysiology
and the synaptic properties of the cells in different layers are different,
and so we can think of these as different cell types.
All of these neurons here are glutamatergic neurons
that have their axonal outputs release glutamate.
So they are of the one cell type in terms of their neuro transmitter
but we can subdivide them into many different subtypes
of excitatory neurons and one of their ways
of subdividing them is their laminate position
within the neocortex that tells a lot about the synaptic connectivity.
Which is what we're going to look at further.
Now in order to functionally map the wiring diagram
of the C2 barrel column, we first need to identify it
in the living brain.
It turns out that the anatomical barrel map that we've looked at so far
is nice in anatomical sections but when you look at the living brain,
you simply see the surface blood vessels without an obvious indication
of where different whisker representations are.
However we can perform a simple optical imaging technique where we can shine
a red light on the surface of the brain, we can then stimulate the C2 whisker
that then causes activity in a localized region of the brain,
the C2 whisker representation, and that changes in turn,
the optical properties of this area of the brain.
There are changes in blood flow, damaged from neuron activity,
and there are also swelling of different processes
that affects how light enters and leaves this area of the brain.
So, simply by shining a red light on the surface of the brain,
stimulating the whisker, we can see changes in deflection of this red light
and it turns out that when an area of the brain
is activated, less light is reflected, and at the level of hundreds of microns,
we can now map the location of the C2 whisker, it causes a response
in the small area here of the brain.
We can then furthermore put a bit of red dye at this location
inside the brain and at later times we can cut horizontal sections
of the brain and see that the red dye that we placed inside the brain
is located inside the map of the whiskers and in particular,
it corresponds to the C2 barrel column inside the neocortex.
As one would expect, there's a nice mapping
of the functional location of where C2 whisker evokes responses
and the anatomical map that matches them nicely.
You can also see that the spacial extent of this whisker representation
is really at this level of hundreds of microns.
The C2 whisker representation is maybe some 200 by 300
micrometers in extent, and as mentioned, the neocortex
is about 1.2 mm thick at this location.
So we can then do this type of experiment,
inject some red dye, cut brain slices
through the thickness of the cortex,
and we can then begin to analyze how are neurons that are located
in different layers communicate with each other.
So here's a picture of a living brain slice,
we injected red dye to label the C2 barrel column,
you might be able to make out the so called barrel walls here,
present inside the layer 4 of the neocortex,
and in order to analyze synaptic connectivity,
we can then simultaneously record from a number of different nerve cells.
In this particular experiment, we're recording from six neurons
that are located here in layer 5 of the neocortex,
so some depth from the surface of the brain would go
to the infragranular layers and here we've got
six wholesale recordings that are located on six different neurons,
so in each case there's a wholesale glass electrode that's making contact
with this neuron as a membrane potential recording and so there's a continuity
between the patch electrode and the cell membrane here,
so we can record the membrane potential.
In order to study synaptic connectivity, we can then initiate action potential
in one of our cells.
We can simply inject current into <i>Cell 1</i> cause it to depolarize,
fire an action potential, and then see what happens in these other neurons,
whether any of them respond with post synaptic potentials.
In this particular experiment, <i>Cell 1</i> released glutamate
and caused post synaptic potentials in <i>Cell 3</i>.
That's what you see here, an action potential in <i>Cell 1</i>,
and a post synaptic potential in <i>Cell 3</i>.
This 80mv refers to the action potential trace and the 2mv
refers to the post synaptic potential so the post synaptic potential,
as expected is somewhere in this 1mv range lasting some tens of milliseconds,
whereas the action potential is brief, it is one millisecond depolarization
of about 50 to 80mv.
So, <i>Cell 1</i> is synaptically connected to <i>Cell 3</i>, there's a synaptic connection.
But what's not shown here is that, that same action potential didn't cause
any effect of the membrane potential in cells 2, 4, 5 or 6
so there's a single connection here from <i>Cell 1</i> to <i>Cell 3</i>
and four of the others have no connection and so <i>Cell 1</i> has a 20% probability
of having post synaptic partners, only one out of five
of these possible cells were synaptically connected to <i>Cell 1</i>.
So we can think of that as a 20% probability
of finding a synaptic connection.
We can then of course, go through the different cells and stimulate <i>Cell 2</i>
and see whether any of these other cells responded and in fact it turns out
that <i>Cell 2</i> drives a small postsynaptic potential in <i>Cell 1</i>.
Note though, the scale is rather different in the time course of the PSP
is also different from <i>Cell 3</i>.
There's some diversity in the properties of the postsynaptic potentials
as we've already seen.
We then go through all the different cells stimulating cell 1, 2, 3, 4 and 5 in turn,
and seeing who responds postsynaptically to that action potential.
That then allows us to draw a small wiring diagram
for these particular six neurons that we recorded from,
here you see the dendritic structure of those cells, here we've separated them,
adding different colors so you can see the structure of the different neurons,
they're all excitatory, pyramidal neurons and they're connected
in this particular wiring diagram pattern where <i>Cell 2</i> connects to 1, one to three
three connects to six, but it receives an input from <i>Cell 5</i>,
four connects to both five and six and so there's a divergent output here
from <i>Cell 1</i>, causing postsynaptic potentials
in both of these two partner cells that are postsynaptic territory.
So we can then repeat this experiment over and over again in different layers
recording cells in <i>Cell 2</i>, in <i>Cell 5</i>, we have four
and simply see what the probability of finding synaptic connections
is across a population of neurons.
We can't record from all the cells at the same time so it becomes
a statistical argument in the end where through analyzing
many different brains, focusing on the one particular area
of the brain and nicely delimited column of the neocortex,
we can get the statistical wiring diagram
of how neurons in different layers, different cell types if you like,
communicate with each other at a statistical level and that then,
helps us understand how sensory information
is processed in this part of the brain.
So here's the end result of the analysis after recording from hundreds of pairs
of neurons we find that there's a connection probability matrix
that we can construct where we take the location
of the presynaptic neuron in terms of which layer
of the neocortex it's in, and when we look and see
how often that connects with postsynaptic neurons
again, ordered according to which layer the SOMA is located in.
And if we now start with neurons that are located
in presynaptic layer 4, we'll see that they make connections
onto postsynaptic cells in layer 2,3,4,5A,5B
and a little bit in layer 6.
So layer 4 neurons send the axons and make synaptic connections
with neurons in all other layers of the neocortex.
On the other hand, if we think postsynaptically about neurons
in layer 4, we'll see that they don't receive
any input from layer 2,3,5 or 6 and they almost exclusively
receive their input from other layer 4 neurons
So layer 4 is a very tight local processing area
when neurons within layer 4 like to talk to each other
and they also like to send signals to other parts of the neocortical column
but they don't want to receive information from these other layers.
We can also see that there's a strong diagonal component where neurons
within layer 2, within layer 3, within layer 4, 5A and within layer 5B
like to talk to each other. They like to talk
to their neighboring cells that are sitting right next to each other.
That's one of the ways in which they make connections,
simply by being close to each other.
Beyond that, there's further specificity.
The superficial postsynaptic neurons, they get information
from presynaptic layers 4,3 and 2.
And so here, you'll see that the layer 2 and 3 neurons
are receiving information from these circuits here.
Probabilistically, that's where they're most likely to be getting
their synaptic input from.
On the other hand, if we look here in the deeper layers, in layer 5,
you'll see that they get synaptic inputs from almost all other layers,
they get information from layer 2,3,4,5 and perhaps a little bit from layer 6.
So these neurons here are integrators, they receive information
from all other parts of the cortex and presumably, they use that information
to distribute some further processed signals to other parts of the brain.
There's a higher degree of specificity in the directions of which
sensory information gets distributed to different cells in the neocortex
and what remains to be learned from further experiments
is what the purpose of that computation is within the neocortex.
If we now look at the amplitudes of the postsynaptic potentials
that are found inside the C2 microcircuit, we find that there's a rather interesting
distribution of unitary EPSPs.
Most of the synaptic connections are very small in amplitude.
They're sitting there somewhere at around 100 or 200 mV.
So, an action potential in a presynaptic neuron on the whole gives a rise
to a relatively small postsynaptic potential
of the postsynaptic neuron. It's just a few hundred mV.
Almost nothing compared to the amount of depolarization that's necessary
in order to get this postsynaptic cell to fire an action potential,
it needs some 30 mV of depolarization, and that 100 mV is going
to make a small impact.
However there are also some rare synapses within the neocortical microcircuit
that are much, much larger and so we can get synapses
that are at 5 and even 10 mV in amplitude in response to a single action potential.
These synapses then could give rise to a considerable impact,
postsynaptically, and contribute strongly to driving postsynaptic
action potential firing.
And so we are left in an interesting situation
where there's a few very strong synapses
and there are very many, very weak synapses
and at this time, it's still not completely clear
which synapses are most important in terms of driving neocortical
microcircuit activity.
Computational studies suggest that one can remove
all of these small synapses and it makes very little impact
upon the neocortical microcircuit activity.
It may therefore be that larger amplitude synapses are the main drivers for reliable
sensory processing in the neocortex, and it may be that a very sparse set
of connections, which are large in amplitude, might dominate
over a seed of very small connections that make almost no impact
upon the larger neocortical microcircuit.
Large unitary EPSPs might be a fundamental importance
for reliable signaling in the brain.
Further indications that some of the liable processing
might occur through these big amplitude connections,
comes if you're looking at the variance of the EPSP amplitudes.
in here we compute the coefficient of variation which is equal
to the standard deviation over the mean.
In terms of the trial by trial EPSP amplitudes that we measure
for the same connection in response to a given action potential,
and you'll remember that there's some variance.
We've already discussed that large synapses might have
low variance whereas small synapses have much more variable responses.
That then, is another indication that if you want to have
reliable sensory processing, you might want to use
these big connections that not only have a big impact
on the postsynaptic neuron but they're also reliable
with low variance from one trial to the next.
So here in this video we've seen some important aspects
about glutamate synapses.
First of all we've seen how glutamatergic synapses
are essential for relaying all sorts of sensory information
from the periphery to the neocortex where it's thought that sensory percepts
are generated, so glutamatergic signaling is essential
for sensory perception.
We've also seen that within the micorcircuits
of the neocortex, there are highly specific circuits that are made
where some types of neurons like to talk to other types of neurons
but might not want to receive information from other types of neurons,
and so there's a great deal of cell type specificity in terms of
where glutamatergic signals are made and how the sensory processing
is likely to occur within the neocortex.
In addition we've seen that there are some extremely large amplitude
synapses that are one or two orders of magnitude, larger in amplitude
compared to the large sea of small synapses
that are present in the brain and these larger amplitude synapses
might be particularly relevant for reliable sensory processing
within cortical microcircuits.
These synapses may in part be determined by genetic mechanisms.
The different cell types express different genes
and so genetically, one can distinguish different cell types
in the neocortex and there may well be genetic factors that skew
the wiring diagram of the neocortex in one way or another.
It's also very likely that experience makes a big difference to the wiring
of the neocortex.
We learn how to perceive the world around us
and it's likely that some of that learning as to how to interpret
the sensory information occurs in the wiring diagram
of the primary sensory areas.
So the distribution of large amplitude and small amplitude
synaptic connections in the brain may in part relate to experience
depend on learning.
Over the next video, we're going to see how
synaptic plasticity through different patterns
of activity can alter the strength of synaptic connections.