THURSDAY, April 1, 2004
Session 2: Neuroscience, Brain, and Behavior I: Brain
Development in Children
Thomas M. Jessell, Ph.D., Professor of Biochemistry and Molecular Biophysics,
Columbia University, and
Investigator, Howard Hughes Medical Institute
CHAIRMAN KASS: Session 2 of our meeting, Neuroscience,
Brain, and Behavior I: Brain Development in Children. From
reproduction and the beginning of bodily life to the brain
and the functions and flourishing of mental and emotional
life. Everyone senses the great excitement of neuroscience
and modern psychology, methodical investigations by the mind
of the mind to increase the human understanding of the human,
prospects for addressing mental illness and aberrant behavior,
and improving our abilities to make the most of our native
mental capacities. Everyone also vaguely senses that those
exciting new discoveries might be accompanied by new ethical,
social, and philosophical challenges precisely because these
studies touch so directly on many of the powers and activities
that make us who we are: awareness, memory, imagination, desire,
motivation, thought, feeling, and moral judgment itself.
The Council is interested in learning about the developments
in neuroscience and psychology present and projected that
might raise significant ethical and social issues either because
of the technological interventions to which they might lead,
or because of changes in human self-understanding that they
might invite. In our Beyond Therapy report, we explored some
of the psychotropic drugs that alter behavior, memory, and
mood, in the service of goals beyond therapy. But we sense
that this was just one of a number of challenging issues that
might emerge from this field. At the last meeting we got
a beginning in this area with introductory presentations from
Robert Michaels on neuropsychiatry, by Jonathan Cohen on brain
imaging and reward and decision, and you'll recall that
we decided at that meeting before proceeding further to find
out which aspects of this large domain were most worthy of
our attention, that we should try to learn something about
just the foundations, about normal brain and mental development,
especially in children. So today and the rest of today is
about those foundations, the scientific foundations of normal
brain and mental development in infants and in children.
The questions behind these discussions are really these.
What can neuroscience and developmental psychology tell us,
either now or soon, that might be relevant for understanding
how best to raise flourishing children and adults? What can
neuroscience and developmental psychology tell us that might
be relevant for understanding and preventing disordered or
dysfunctional children or adults? But the explicit discussion,
as opposed to these larger questions, the explicit discussion
will be the science itself, neuroscience in the morning, psychology
in the afternoon, and at the end of the day we'll have
a chance to take stock of what we have done and how far we
It's my great pleasure to welcome Dr. Thomas Jessell
who is Professor of Biochemistry and Molecular Biophysics,
the Center for Neurobiology and Behavior, and Howard Hughes
Investigator, at Columbia University. He is a distinguished
researcher in the field of brain development. It's a
great pleasure to welcome you to the Council. Apologies for
detaining you, and we very much look forward to your presentation.
Dr. Jessell has asked me to announce that he would very much
welcome our interruptions if we have questions or things on
clarity. He wants to make sure that we're following.
Dr. Jessell, the floor is yours, and thank you again.
DOCTOR JESSELL: Well, first I would like to thank Dr. Kass
for this opportunity to participate in this discussion of
brain science and its relation to behavior. And just to echo
these comments that I really would encourage interruptions,
comment, dissent, clarification. If I lapse into jargon,
then please bring me back to clarity.
Perhaps in the light of the first session this morning,
an interesting place to start this discussion is in fact early
in embryonic development, because it's at a relatively
early stage in development that nervous systems form, that
nerve cells are generated, and that those nerve cells begin
to acquire identities that drive the subsequent formation
of connections within the developing brain. And one of the
things that I'd like to suggest or argue in my comments
this morning is that to a large extent the precision with
which those connections form governs the intrinsic behavior
or repertoire of any organism, whether it be a child, any
type of mammalian organism, even lowly invertebrates. Behavior
is in large part dependent on the nature of the circuits that
exist within those nervous systems.
Now that is not to say that the wiring diagram, if you like,
of the nervous system is the only determinant of behavior.
I think we've understood for many years that this provides
an architectural substrate upon which environment and experience
then mold those connections and refine those connections to
produce the full, rich repertoire of behaviors that any given
organism is characterized by.
So one of the things I'd like to discuss this morning
initially at a relatively reductionist level is really to
review what the field in general has learned about the nature
of brain development during embryogenesis and postnatal development.
What is the relationship between genes, neurons, circuits,
and behavior? What are the molecular determinants that begin
to establish the structure of the nervous system? Then discuss
the interplay between these genetic programs and environmental
influences. When does the nervous system cease to develop,
if you like. When is the pattern of connections established
in a way that cannot be changed? And our views on that have
also changed rather dramatically over the last 10 to 20 years.
So we're dealing with questions of plasticity of the nervous
system in relation to circuits and the way that those circuits
influence behavior. So these are all normal developmental
processes, if you like, and perhaps if we have time at the
end I will turn to the way in which those normal developmental
processes become subverted in brain disease, in various types
of neurological and psychological, psychiatric disorders.
Now those are very large problems, so I think at best I will
be really scratching the surface on some of those later issues.
So just to put things into a general context. Now, I hope
that everybody can see some of these images, because some
of my comments will really be dependent on them. And those
on my left are in danger of being attacked by this laser pointer,
but I think we're working at the moment. So what you
are looking here is in very superficial terms the process
of human brain development as a function of the nine months
of gestation. And so somewhere within the third to fourth
week, the nervous system begins to form. So cells in one
of the major germ layers of the embryo, the ectoderm, make
a decision that they're going to acquire neural character
as opposed, for example, to acquiring the properties that
eventually give rise to skin. So the first steps in the formation
of the nervous system occur at a surprising early stage.
And then as you can see because this is drawn relatively to
scale, what happens over the subsequent eight months of embryonic
development is a process by which the nervous system at least
in large part accumulates large numbers of cells. So the
nervous system is exceedingly small at these early stages,
but as you can see, by the time an embryo is born at nine
months of gestation, there is a very highly organized nervous
system which is much, much larger than the anlage that was
seen at three to four days. And so what we're really
looking at at this period, and this process continues, not
just in embryonic life, but in postnatal development. Neurons
continue to be added.
So in a sense what the field of developmental neuroscience
at these early stages has been trying to address are two basic
questions. So one of these relates to numerology. So these
are estimates, but the human brain is thought to contain some
1011 neurons. This is a staggeringly large number
of neurons. And it also contains, and again this is simply
an estimate, at least 1,000 different classes of nerve cells.
So by comparison, with most other tissues and organs in the
body, the extent of cell diversity inherent in the nervous
system is observed at a much more extreme level than in the
liver, or in the heart, or in other tissues. And so one of
the major questions is to understand how during development
the nervous system controls the number of nerve cells that
And the other thing that this image shows is that not all
nerve cells, once they're produced, are the same. Individual
nerve cells have discrete functional properties, and some
of those functional properties are inherent or derived from
the fact that the morphology, the shape, the structure of
those nerve cells is markedly different. So on the right-hand
side of the screen here we're looking at one nerve cell
that is found in a part of the brain called the cerebellum
that is involved in motor control and many other functions.
And you can see that that cell, that neuron, looks very different
from a neuron found in the retina that is involved in visual
processing. So one of the things that development does is
ensure that not only do you generate very large numbers of
neurons within the developing human brain, but you make these
neurons different from a very early stage in fundamental structural
ways that presumably influence their later functional properties.
And I'll talk a little bit about what we currently understand
about how these two problems of generating neurons and making
neurons different are achieved at a developmental level.
Now these neurons don't exist in isolation. The functional
property of the neuron is dependent on that neuron becoming
incorporated into a functional neural circuit, a network of
neurons interconnected in different regions of the brain such
that the combined activities of many neurons in different
parts is necessary to produce a given behavior.
So the problem becomes even more complicated at this point.
So if we look at a typical region, for example the cerebral
cortex, what we're seeing here is a complex, dense mesh
work of neurons. And this has been appreciated from classical
anatomical techniques for the last century or more. So each
of these individual neurons as you can see from their morphology
is subtly different from their neighbors. One of the questions
is how do these differences in morphology, these differences
in structure, relate to the types of connections that they
So the two general questions that I want to begin to address,
at least in the first part of my comments, are really going
to be the issue of the impact of genes and genomes on neurons,
on neuronal identity, and the way that those neurons begin
to assemble interfunctional circuits. So if you like, this
is a reductionist view in which many aspects of brain organization
are achieved through a genetic program to result in a wiring
diagram of the brain.
So how are neurons generated? This is perhaps the fundamental
problem, the event that has to proceed if all other aspects
of circuitry and behavior are going to be achieved. And this
is a problem that has been appreciated for, again, almost
a century. But it's only very recently that we have any
molecular understanding of the way in which a cell in a primitive
ectoderm acquires a neuronal property. And this is a process
that is reiterated many, many times to produce this vast diversity
of neurons that exists in the human brain.
And so if we look from classical studies by Pasko Rakic
and others, what we're looking at here is a small region
of the developing brain, in this case a region of cortex.
And so in this region what you can see is a layered, a striated
structure here. One particular region called the ventricular
zone, which you can't really see properly here, is the
site of the precursors that will give rise to nerve cells.
These are neuro-precursors if you like. They're stem
cells in a particular context of neural development. So these
cells lie at one side of this epithelium, this sheet of neural
cells. And those cells divide under tightly controlled programs.
And the process of division here will determine whether that
precursor cell continues to remain a precursor cell with a
capacity to divide further, or whether that cell will leave
the cell cycle and acquire so-called post-mitotic neuronal
property. One of the characteristic features of neurons is
that once they're generated, they never undergo further
divisions. So this is a crucial event in determining the
number of cells that populate the nervous system to determine
how many of the precursor cell population remain, and how
many post-mitotic non-dividing neurons are produced.
And the classical view until a few years ago had been that
these cells divide in this restricted germinal or ventricular
zone, and then they migrate through a complex process series
of cellular events away from this germinal zone into their
eventual settling positions in other regions of the nervous
system. And one of the ways that they're thought to undergo
this migration process is along a series, if you like, of
tram lines, or structural elements called radial glial cells
which act as conduits for this migration process. So really
there are two processes going on here at these very early
stages. The process of cell proliferation coupled with the
decision to give rise to a neuron, and then the migration
of the neuron along these processes.
Until very recently, and part of the reason I'm mentioning
this, is to indicate that some of our thinking on these processes,
even though the problem has been identified in classical terms.
Some of our thinking and some of the information is really
very, very recent. So much of what I'm saying is really
a current 2004 view of these processes. But this is still
a field that is in a state of considerable flux.
So the classical view, which is shown on the left-hand image
here, is that neurons achieve their final position within
the nervous system through this migration. They ensheath,
they intertwine around these radial glial cells, and that's
the way they move. But the radial glial cell and the neuron
have been thought of distinct cell types with no relationship.
What we now know from work in the last five years or so from
many people is that this is certainly true, but this is an
oversimplification, because it turns out that the radial glial
cells, these structural elements, not only serve as a scaffold
for neuronal migration, but they also represent in many ways
the precursors of neurons themselves. So the radial glial
cells have a dual role. They not only generate neurons, but
they then act as a scaffold. And I'll just give you one
example of this new view. So here we're looking almost
in real time at one of these radial glial cells. You can
see its processes spanning the length of the future cortex.
And if we follow that cell in real time, although I'm
showing it in a series of static images, what you can see
is this cell divides at around Time Zero. And then there
are two cells. And as we follow with time, one of the progeny
remains in the ventricular zone, presumably to remain a precursor
cell, but the other cell begins to migrate along this structural
element. And by using various genetic markers, some of which
are not showing actually on this slide here, what we can see
- there's a slight technical problem here, because this
cell would be labeled Red with a neuronal molecular marker.
So this type of real-time imaging here is showing that radial
glial cells are in fact the precursors of neurons. So this
is a piece of cell biology that has emerged over the last
few years that has radically changed the way that we think
about neuronal production in the developing brain.
So can we move then from this cellular description of where
neurons throughout the central nervous system, throughout
the brain, come from, into a molecular understanding of this
decision. What are the genes that determine whether a cell
remains proliferative, or whether it exits the cell cycle
and acquires a neuronal fate?S And over the last decade there
have really been very substantial advances in understanding
the nature of genes that drive this neuronal differentiation
process. And one can show that some of these genes are sufficient
when expressed in a proliferative cell to produce the neuronal
phenotype, to produce the neuronal fate. And I'm showing
you one example of this. The nature of the genes don't
really matter here. But what we're looking at is a top-down
view of an early vertebrate embryo. And so this has a left
side and a right side. And on the left side, if we concentrate
on, for example, the image in Panel B, you can see a few blue
dots here. Those are the neurons that have been generated
under a normal developmental condition. And you can see that
there's a sea of non-neuronal tissue in which are interspersed
individual nerve cells. So in this experimental situation,
a single gene that is one of these genes that promotes neuronal
formation, has been introduced into these precursor cells,
and the consequences for neuronal production have been examined.
And what I hope you can see is that to the right of this dotted
line there is a vastly greater number of nerve cells that
have been generated as a consequence of introducing that one
gene. So neuroscientists now are beginning to be able to
control this early decision as to whether to give rise to
a proliferative cell or whether to generate a nerve cell.
And these are genes that are operating in the embryo normally,
and this is part of a genetic program that in a sense is determining
the fact that the human brain has 1011, not 1012,
not 1010 neurons. So there's a tightly orchestrated
genetic program that ensures the production of neurons.
This image indicates, at least with this sort of color-coding,
that all neurons are the same. So there are really two problems
here. One, as we saw, to generate large numbers of neurons.
But the second is to make those neurons different. And so
if we consider that the human brain at the time of birth contains
some 1,000 different cell types, how are those different cell
identities? They're all neurons, but each of them acquires
a characteristic identity that suits its particular later
function. How is that diversity of neuronal identity controlled?
So if there are 1,000 different classes of neurons, does that
imply that there must be 1,000 different signaling mechanisms
that are operating with one pathway of signal, one signal
per neuron, which would provide extreme constraints on the
number of genes necessary to generate this diversity. Or
has the nervous system evolved more efficient ways of generating
diversity from a relatively small number of signaling systems.
So as a generality in the nervous system, as in the embryo
as a whole, the way in which cells acquire their different
identities is through the intersection of two different pathways.
The fate of no cell in the nervous system is really preordained
from an early stage. So there is no such thing as a central
nervous system homunculus. Cell identities are acquired because
of their position in an early developing neural epithelium.
And what their position really is doing is defining the environment
to which that individual neural precursor cell is exposed.
So if you're in one position you're exposed to a different
set of signals then being in a different position. So really,
neuronal generation and neuronal identity involves the intersection
of environmental signals with programs of gene expression
that are going on within the nerve cell itself. And it turns
out that this vast diversity of neuronal identities, 1,000,
2,000, whatever that number turns out to be, is generated
through a surprisingly small number of environmental signals,
where those signals operate in combinations. So the nervous
system has evolved efficient ways of using a small number
of signals to generate this vast degree of diversity.
And I just wanted to give you one principle by which diversity
comes, how you can generate many different cell types, neuronal
cell types, from just one signal. And it turns out that some
of these signals, which are secreted proteins produced by
one cell which influence neighboring cells, have been termed
morphogens. So what we're looking at here is an early
primitive region of the nervous system. This particular region
is going to give rise to the spinal cord, but if you looked
in the brain, you'd see a very similar morphology.
So this is the nervous system. These are non-neural tissues.
There are localized sources of these secreted factors in non-neural
tissues. You can see by the shading of the color blue here
which represents a site of gene expression. And one of these
proteins, which has been known as Sonic Hedgehog for various
reasons, is a secreted protein. That protein functions as
a morphogen. So what a morphogen is is a protein that can
induce or change or compose different cell fates as a function
of the different concentrations of that one substance that
a cell is exposed to. So the exposure of a naïve precursor
neural progenitor cell to a low concentration of this factor
will induce one neuronal fate. But as you double the concentration
of that factor, that same recipient cell acquires a different
fate. And so in this way, one single signaling factor, by
acting at different concentrations can induce different classes
of neurons. And so for example, the local source of this
factor establishes a concentration gradient throughout the
early nervous system, the neural epithelium. And in this
particular example at least five different classes of neurons
- their identity doesn't really matter - are generated
in response to this one factor through twofold differences
in concentration. So this type of mechanism is used again
and again in the nervous system. In some cases it's concentration,
in some cases it's the convergent exposure to two different
signals that results in differences in identity. But numerically,
a relatively small number of these signals imposes the vast
degree of neuronal diversity that we see within the nervous
So I want to switch now. And you'll see that this is
going to be of necessity a relatively superficial overview
of all of these processes away from generating neurons. Let's
assume that you have generated vast numbers of neurons in
the early nervous system. How do those neurons begin to form
connections? Because one of the things that these early programs
achieve are neurons with distinct identities. And in reality,
what the identity of a neuron does is enable it to extend
a long process, kind of an axon, from the site at which it's
generated into the vicinity of its target cell. So neurons
need to connect. And so in some cases - and I'm going
to show you an example from the visual system throughout some
of these comments - you can see that the distance over which
a nerve cell has to project in order to connect to its eventual
target is very considerable. Some several orders of magnitude
longer than the diameter of the cell body at that cell itself.
So here we're looking at a sort of schematic view of some
aspects of the visual system. These are the eyes. And in
the retina there are neurons which process visual information,
process light, convert light into electrical signals. And
the job of these retinal neurons is to take that information
that is arriving from the environment, from the periphery,
and to transfer that information into higher centers in the
central nervous system. And in many ways it does this by
extending this long process called an axon from the site at
which the cell is generated into its eventual target regions.
And then once it reaches that target region, it then has to
choose particular sets of target neurons with which to form
So there are many challenges here for a developing neuron
in order to be able to establish an appropriate wiring pattern.
For example, if you focus on this retinal neuron, it has to
make a decision to leave the retina, to extend this process
out of the retina, along the optic nerve. It then has to
make a decision as to whether to cross, as most axons do,
or stay on the same side of the brain. So there are important
guide post or path-finding decisions. It then has to reach
an appropriate position within its target region. And then
it alters trajectory, and then reaches an appropriate depth
within the target region. So all of these are different environmental
problems that the developing neuron has to negotiate in order
to reach its target. And again, over the last several years
we've begun to understand the complexity of this organization,
again, has been appreciated for over a century. But what
we haven't known are any of the mechanisms, the molecular
mechanisms that ensure these precise patterns of connectivity.
And the last 10 to 15 years has revealed almost a greater
degree of molecular information than is currently processable.
But I want in very brief terms just to go through some of
the thinking about how neurons, once they have acquired an
identity, begin to be able to connect with target cells.
So the business end of the neuron is a sensory motor apparatus
which is called the growth cone. So the neuronal cell body
gives rise to a long axon. And at the end of that axon is
a highly motile cellular structure, the growth cone. And
so what we're looking at here are various views of growth
cones. So the axon would then move out of below the screen
and some many yards away would be the cell body. But what
you can see is that the growth cone which really senses information
in the environment and converts that sensory information into
a direction, into a vector or trajectory, is a highly organized
structure. And the general view is that the ability of the
growth cone of the neuron to navigate this complex terrain
en route to its target depends on the expression of a variety
of receptor molecules which are sensing information in the
environment, mediating that information through these receptors
and converting that into directed behavior.
And so what this involves, both the reception of information
in the environment and the conversion of that into a motor
response. The growth cone has to crawl physically through
a variety of substrates. So in this process of extension
of the axon, there is a highly complicated cellular machinery
that is occurring where, for example, the actin cytoskeleton
is tightly linked to these receptors on the cell surface.
So that information from the environment is converted through
the actin cytoskeleton into directed movement. And we're
beginning to understand a great deal about the way in which
neurons, growth cones, use their cytoskeletal, use their structural
properties to influence their direction of growth.
The other thing that we've begun to understand is the
nature of the environmental symbols that influence the behavior
of axons, behavior of growth cones. Now I'm going to
show you an example from an isolated in vitro experiment that
the assumption is that same mechanisms are going on in the
developing brain in situ.
So what we're looking at here in this panel is the axon
of a growing neuron and here is the growth cone. You can
see that this is a highly elaborate process, the small protrusions
called Filopodia that if this was a real time image are constantly
changing, constantly moving. These are the sensory apparatus.
This is the sensory apparatus that is looking for cues in
We know a lot about the molecules in the environment that
influence the structure and the direction of axon growth cone
movement. So the details of this don't matter, but we
now understand many, many classes of molecules that are distributed
at key points along the trajectory of a growing axon and that
influence direction directly by impinging, by imposing, changes
in shape and motility in the growth cone.
So what we're looking at in the bottom panel is the
exposure of the same growth cone to a protein that we know
guides axons and the way that it guides those axons is by
inducing a very dramatic collapse of this complicated cellular
structure. So essentially it prevents growth cones and axons
from moving in that direction because the growth cone is necessary
for motility in a directionally sensitive manner. This will
then prevent growth continuing in this direction and then
the axon will move off in another direction.
Some of these molecules influence guidance if you like in
a negative, repellant way. They form no-go zones in which
axons are simply incapable of moving. Others act in a more
positive way. That is they act as lures or attractants to
direct in a positive way axons towards a particular target.
Through the combination of these negatives factors and positive
factors, it's thought to guide axons from the site in
which the neuron is generated in a set of complicated steps
to the vicinity of its targets. So this have been a very
dramatic advance in understanding as we will see later on
and may have implications in therapy in the context of regeneration
in the adult state.
Once an axon through these guidance mechanisms has reached
the vicinity of its target cell the next challenge it faces
is actually to select which of the small subset of neurons
it's going to form stable connections with. There may
be 1,000 different potential target neurons for that one neuron,
one axon, to connect with.
How does it choose which of those 1,000 neurons it's
going to form a stable association with? Now this is an area
that even though has been the subject of intensive study is
still relatively poorly understood. It's only in the
last five years or so that we have any understanding in the
central nervous system in the brain of the mechanisms for
forming functional synaptic connections because communication
between nerve cells depends on the formation of these functional
connections. There are many problems inherent in this process
of recognition of how you recognize the target cell to form
So in various schematic ways, there are sort of where type
questions. That is this is one target cell. This in-growing
axon has chosen to form a connection on that one target cell,
but not on a neighboring cell which would be off the screen
here. So that is the first type of question. Which cell
do you choose to form a connection with?
The second type of question which is very important for
the function of the neuron is where do you form the connection.
Do you form the connection close to the cell body of that
target neuron or do you form because neurons are such highly
polarized structures way out from the distal processes of
the so-called dendrites of the neuron? So there are questions
of sub-organization as well as which cell you choose.
Frankly we don't have a good understanding in molecular
terms of how these processes operate even though we do know
that there's a high degree of stereotypity of selectivity
in those processes. So what the field currently trying to
understand in a way that I think will eventually be important
for understanding how circuits control behavior and how those
behaviors are eroded in disease is understanding the biochemical
mechanism that is essentially a recognition process between
the arriving neutron, the pre-synaptic neuron and its pos-synaptic
target. We have little information yet on the nature of the
molecules that drive this process.
But interestingly and I'll come onto this later, the
few molecules that have been implicated in this process are
known to be mutated in certain forms of human neuro-developmental
disorders. So it begins to suggest that actually understanding
the machinery of synaptic connectivity may say a lot about
the normal and pathological function of the human brain.
I'll elaborate on that point.
So the final step in this brief overview of the wiring diagram
in the brain occurs after neurons have made connections with
their target cells. You might imagine that this is the end
of the process. Once you have generated the neuron, you've
managed to reach the target region. You form the connection
with the target cell. The job of development is done. But
it's very far from done. As we'll see, there are
many processes that have to operate.
One of those processes relates again to this basic problem
I started with of numerology. That is in order for a circuit
to function appropriately, there has to be a precise matching
of the number of neurons that are generated in one region
and the number of target cells or target neurons that are
found in a completely different region.
And one of the very surprising things that emerged some
50 years ago is that the nervous system is extremely wasteful
in the generation of neurons. In fact in most regions of
the nervous system, two to three-fold more neurons are generated
than are eventually incorporated into functional circuits
which would seem a wasteful process.
But one of the rationalizations of that process has been
that perhaps one of the things that permits one to do is to
achieve numerical matching between two sets of interconnected
neurons. If you generate neurons in larger numbers, those
that are generated in excess become dispensable and can be
eliminated. There is a lot of evidence, cellular and molecular
evidence, that this type of process occurs and this is generally
known under the term of neurotrophic factor hypothesis.
The basic idea that seems to operate is that you have a
restricted target cell and many neurons that have the potential
to innovate that target cell. But what happens again and
again during brain development is that only a small set of
those neurons that are originally generated eventually form
stable connections with that target cell. The reason at least
that this occurs in part is that one of the things the target
cell is doing is communicating back to the neuron.
There is a two-way process of communication and the target
cell is producing factors that support the survival of the
neuron and it's producing those factors in limited amounts
and limited availability. So in a sense, this target cell
produces only half as much of this survival factor as is needed
to support the entire compliment of neurons. As a consequence
through a competitive process, only perhaps 50 percent of
those neurons that are alive in the vicinity of the target
cell form stable connections. What we'll see in the next
section is that this process of competition, excess production
and competition, is really paramount in linking the genetic
programs of neuro-wiring to experience and activity in the
way that experience and activity sculpt these basic projections.
Without going into details, we know the basic elements of
this competition, target-celled derived support factor hypothesis.
In many cases, we know the nature of the molecules that promote
cell survival. One of the dramatic advances in all biology
over the last 10 to 20 years has been the appreciation that
in fact the default state of most cells in the body is not
to survive, but to die.
There's an intrinsic death program that is inserted
into every cell. Cell survival really only is permitted under
conditions in which that intrinsic death program is suppressed
and is subverted and the first inkling of that information
in fact derived from these sorts of experiments in the nervous
system. So we know the nature of these neuronal survival
factors and we know a lot about the biochemical mechanisms
by which they promote cell survival. This is the first example
really in a normal developmental epoch of ways in which competition
and interactions between cells shapes this genetic programming
of the wiring diagram.
I'm going to move now unless there are questions or
comments that I could elaborate on from this simplistic but
nevertheless I think valid idea that genes in large part program
the basic wiring diagram of all nervous systems from invertebrates
to vertebrates, from mouse to man. Many basic aspects of
the connectivity are programmed in a way that involves serial
developmental decisions to give this basic neural blue print.
But at that point, the precision of connections, which connections
are maintained, which connections are reinforced, which are
eliminated then begins to be impacted in an extremely profound
way by environmental signals.
So I want to begin to talk now about how activity of nervous
systems, how the experience and the environment shape these
connections. Again I'm going to use the classical example
which emerged from studies by Hubel and Wiesel in the 1960s
and the 1970s focusing on the organization and structure of
the visual system and the influence of environmental deprivation
on the structure because I think that there are many, many
examples of this general paradigm but that the first realization
came from these classical cellular and physiological studies.
We're now back in the visual system. These are the
eyes. This is the optic nerve projecting to a set of relay
nuclei to a second order neurons in the thalamus and those
thalamic neurons then relay visual information from the eye
via this one relay station to the cortex. So most vertebrates
can extract information from the visual world and use two
eyes to do that.
How does the brain extract information from the left visual
field, the right visual field, keep that information separate
as it begins to form these projections, form these connections,
and where is that information integrated? One of the important
things that occurs is that you keep the information from the
left eye and the right eye separate at many of the early processing
stages in the developing brain. What you can see is as one
traces through and then looks in higher magnification, that
an input from say the right eye here will eventually project
to a region of cortex that is spatially distinct from that
region of cortex that receives information from the left eye.
What the great discovery of Hubel and Wiesel together with
Vernon Mountcastle in the somatosensory system was that the
cortex is basically organized in columnar fashion. Information
doesn't arrive in an indiscriminate, anatomically mixed
fashion. Information from the two eyes is physically separated.
One can see that physiologically as Hubel and Wiesel first
saw that but you can also see that anatomically.
For example, if we provide a tracer into one eye such that
that tracer is incorporated through all of the neurons that
form that right eye circuit, you can then see the consequence
of the projection of that right eye in relation to the unlabeled
left eye. So that if we look here, this is a cross section
through a mammalian cortical structure. This is true of humans.
This is true of primates. This is true of cats, many mammalian
organisms. What we are looking at are the terminal fields
of axons that receive information let's say from the right
eye. They would be shown as these light patches here.
What you can see is the cortex consists of the mosaic of
light and dark patches which is a reflection of information
coming in from the right eye as opposed to the left eye.
This is a relatively mature state. This is what 13 week post-natally
in this cat experiment here.
But what is interesting is if you look at a much earlier
but still post natal stage, what you can see on the top here
is that these stripes or banded patterns don't exist.
What the implication of that is is at this early developmental
stage information from the left and right eyes is arriving
in the same spatial region within the developing cortex.
What you can see is that over the intervening 11 weeks of
post natal development in a so-called critical period, the
information, the axons from the left and the right eyes are
segregating to form these distinct domains.
This is a process of development of connectivity that is
occurring in the post-natal period which then follows most
of the earlier embryonic patterning mechanisms that are described
in the first section. What is particularly dramatic about
this sort of structural organization of visual input is that
it's critically dependent on visual experience. An early
set of classical experiments looked at the consequences of
eliminating visual input into one of these eyes and examining
anatomically and physiologically the consequences for the
structure of the brain. What happens if you deprive one eye
of the normal sensory information that it receives? And you
see very dramatic changes.
So we are now going to switch orientation and we're
going to be looking at top down images of the brain, of the
cortex with sections or slices cut through in this plane.
What you see here is in the top panel - I hope you can see
that from the back - that you have again these alternating
stripes, so-called ocular dominance columns of left eye, right
eye where the width of each of these stripes which is an anatomical
representation of information coming from the two eyes is
roughly equally proportioned, that equal black stripes and
white stripes here. This is a diagram showing this mosaic
What you can see in these intervening panels is that structural
reorganization of information in the brain in the cortex has
a consequence of monocular sensory deprivation, depriving
one eye of visual input. For example here, and it really
doesn't matter, what you can see in this case is if the
eye that was represented by these dark stripes was deprived,
you can see here that there's a massive over representation
of the structure, the inputs, from the eye that is still receiving
visual information. The white patches greatly exceed the
Conversely if you design the experiment in a different way
and you deprive the light patch eye, you can see that the
dark patches prevail. This is very dramatic evidence that
visual experience produces a structural change in the brain.
One of the things that one can now begin to understand is
the nature of the process by which environment and experience
produces structural changes in the central nervous system
in the cortex. This is just a diagram of what we were looking
at in real images before.
Early on the information from the two eyes is essentially
overlapping over this post natal, critical period. The information
gets segregated into distinct ocular dominance columns in
the developing cortex. And that under conditions where sensory
information is eliminated from one input, then the other input
predominates. So these are gross structural changes in the
organization of the brain that are the consequences of presumably
the inability of experience and activity of those neurons
to maintain a spatial territory.
You can actually see this influence of experience and the
environment at the level not of just domains of the cortex
but individual neuronal branching patterns and morphologies.
These are just four diagrams showing essentially the normal
change in development of an axon, a nerve that is going to
innovate the cortex and the consequence of deprivation.
On the left-hand side here, we've traced the terminal
projections of one of these axons arriving in the cortex conveying
visual information in a young post natal animal. You can
see that the region of space or cortex occupied by this axon
that it's relatively broad that the density of terminals
of branches in any given region is relatively small, relatively
Now as the animal matures under the consequences of normal
visual experience you can see that there's a dramatic
change in the shape of that projection. A much smaller region
of the cortex is now innervated, but that region that is innervated
now receives a much higher density of axonal branches and
synapses. So this is the process of post natal structural
reorganization that is occurring.
We know that this structural reorganization is again dependent
on experience upon sensing the visual world because if we
deprive one eye then you can see that at the level of an individual
neuron those structural changes are very dramatic. The neuron
has few axons, few terminals. We are seeing this level of
structural reorganization in the nervous system as a consequence
of visual experience at many, many levels.
Here we're looking - and I'm going to emphasize
this point because I think it's one of the major features
in the way in which environment, the experiential world influences
the structure of the mature nervous system - at the level
of the axons that are arriving in the cortex, but we can also
see this at a finer level of resolution if we look at the
target neurons that are going to receive that information.
I'll do this first. Here is one of these cortical neurons.
The cell body, it has a very extensive set of processes called
dendrites and the synaptic input, the information that is
coming into this neuron is occurring on a series of very small
micro structures that come out from one of these dendrites
which are so-called dendritic spines. The spines here, these
protrusions, are really the units of synaptic information.
Each of these spines is going to be occupied by an incoming
We know that some of the same activity experience dependent
changes that I was showing you in the previous slide at the
level of the pre-synaptic terminal also influenced the dynamics
of these spines. Whether these target neurons have many spines,
have few spines, these are very dynamic cellular structures
which come and go as a consequence of activity. Both on the
pre-synaptic input side and the post-synaptic target side,
experience and activity in the circuit is having an dramatic
impact on the structure.
This information all of which is derived initially from
the visual system raises the question of whether this is something
special about the way the visual system works, or whether
in fact all sensory systems are subject to the same structural
change in organization of the brain as a consequence of sensory
information coming in from the environment. There are many,
many examples as one moves out of the visual system that the
same thing is true. I'll show you just one example from
the auditory system.
This is again a slightly varied organization here, but we're
now looking at a map of a region of auditory cortex. This
is the region of the brain that processes normal sound, auditory
information. The wonderful thing about the auditory system
as opposed to the visual system is that sound is tuned according
frequency. Different regions of the auditory cortex are tuned
to optimal frequencies.
There's a so-called tonotopic map. Different sound
frequencies get projected to different regions of the auditory
cortex. By applying a single sound you can make that region
of the cortex that receives that particular sound frequency.
That's what we're looking at here in an early post
natal rat in fact this is.
Now the relatively large region of cortex associated or
devoted to that particular tonotopic frequency. If we look
some 34 days later, the region of cortex that is devoted to
that same sound frequency has reduced in a very dramatic way.
This looks a lot like what we saw in the visual system. We
know that auditory experience, the sound that we experience,
over that critical post natal period has a dramatic influence
on how this auditory, tonotopic map forms. Because if we
now apply to an experimental situation, apply in a sense a
white noise preventing the auditory system from extracting
normal variation in sound frequency, then applying this distracter
noise produces a very dramatic slowing of the normal maturation
of this cortical map.
We're now looking at the same map here but under conditions
in which a distracter noise had been applied. You can see
that not only does deprivation of visual information but deprivation
of important auditory information erodes the normal formation
of the structure of the brain, the structure of the cortex.
You see this same information, this same principle, occurring
in all sensory systems.
What this means I think is that during this critical post
natal period information coming from the environment and from
the visual senses, the auditory senses, the tactile senses
and the somatosensory system, taste, smell, all of this information
is converging on the central nervous system and providing
these structural refinements in the basic wiring diagram that
are essential for accurate perception of the environment,
accurate performance of motor commands in response to that
information. This I think highlights what we've known
for many years that this critical post natal period is essential
for the appropriate structural information within the nervous
CHAIRMAN KASS: This sounds as if this is mostly
a certain kind of openness is then restrained and pruned and
that there's a greater sort of amorphousness which then
becomes more specialized as a result of this experience.
Have I gotten that particular point right first?
PROFESSOR JESSELL: Yes, perhaps the easiest way
of showing it is just to go back to this image which is essentially
a pruning image. What is happening at these post natal stages
in development is that the map is roughly correct. So visual
information is coming into the visual cortex and not to the
auditory cortex. But the fine details of the working of those
cortical structures is not sufficient to achieve this rough
map. You have to proceed. You have to achieve a precise
map. That precision is achieved by this pruning process.
So in a sense what we're looking at here as you go from
young to mature is the pruning of one neuron. That has functional
consequences because then you achieve a much greater degree
of point to point precision in connections. This pruning process
is activity, is experience driven. Without that experience,
without the range of information coming in from the outside
world, this pruning process fails to occur. We can see that
at the single cell level and we can see that at the level
of the auditory example.
CHAIRMAN KASS: May I just follow up? Would there
be - I don't even know what I'm asking here, but is
there something that you could call pre-natal experience that
is at work even before? Sound I suppose is still transmitted
intrauterinely. Visual would be a different matter. But
do we have environmental and "intrauterine" experiential
things that are already doing this or does the major effect
take place after birth?
PROFESSOR SANDEL: And what about the Mozart effect?
Even if that's not true, would that be an instance of
PROFESSOR JESSELL: Yes. I haven't mentioned
it and stressed the post natal period, but there is strong
evidence that this activity-driven wiring of the nervous system
occurs during fetal development as well as post natal development.
There is very strong evidence in one of the articles I distributed.
Carla Shatz is one of the people who has really provided
some of the strongest evidence that in fetal development the
retinal ganglion neurons that are first arbiters of processing
of visual information are firing in patterns during embryonic
development that is thought to be important for setting up
perhaps the first aspects of organization superimposed on
this basic genetic wiring program.
That is certainly true for the visual system. I think there
is good evidence that is true in the auditory system. Whether
it's true in some of the other sensory systems, the evidence
is less. So this process even though it's been most heavily
studied at a post natal period for reasons of accessibility
probably is beginning during the embryonic stage and when
it begins you can put limits on to in the nature of the basic
connections that have to be formed before the information
from the environment can even reach relevant areas of the
cortex. But it's a relatively early process.
DR. ROWLEY: But if I can follow on with that. It
was my impression from reading Carla's article that what's
formed are some of these intermediate connections or the way
stations or the relay stations rather than the specific connection
with the fine processing neuron in the cortex.
PROFESSOR JESSELL: Yes.
DR. ROWLEY: In her example in the visual cortex.
So you set up the systems that are necessary already before
birth so that these intermediate processes that connections
have been made and then the pruning of the final one is what's
then done after birth.
PROFESSOR JESSELL: Yes, that's absolutely correct.
The system that Carla is looking at which is shown on this
slide back here, I've been concentrating on the cortex
which is the final target. Carla has looked at this intermediate
thalamic relay station because those are the targets of retinal
ganglion neurons that are spontaneously active during fetal
I suspect that wherever you look you will find evidence
that the activity experience dependent structural changes
traditionally this has been looked at first in cortical structures.
We now know it's true in subcortical structures in the
thalamus. I suspect even at the level of the first sensory
relay station, perhaps even in the retina itself, activity
is having an influence in terms for example of the morphology
of neurons or fine details. I suspect that this is happening
in a very pervasive manner and where you look depends on what
type of experimental system accessibility of that particular
region of the nervous system.
PROFESSOR SANDEL: You've been speaking about
the plasticity in the early post natal period. Do we know
when it stops?
PROFESSOR JESSELL: Yes.
PROFESSOR SANDEL: When experience dependent changes
no longer affect the structure of the brain?
PROFESSOR JESSELL: That's the next section.
So maybe I can come onto that. Yes. The answer is that if
you'd asked me this or asked the field this ten years
ago, the idea would have been that this critical period closes
at some point two to three months after birth in an experimental
animal. It perhaps closes in the first several post natal
years in the human infant. What we now know is that to some
extent that process of plasticity persists throughout adult
life. I'll give you some examples of that. Probably
I think as a generality the nervous system is most plastic
at early stages and plasticity is essential for connectivity
in the way that we've seen in an immediate post natal
period. But there are striking functional elements of plasticity
albeit at a reduced level even in the mature organism.
CHAIRMAN KASS: Dr. Jessell, would you like to complete
the presentation? We can take questions then.
PROFESSOR JESSELL: If they are burning questions,
now is a good point. We've reached the end of this stage.
CHAIRMAN KASS: Bill Hurlbut and then Gil.
DR. HURLBUT: In saying that there is a longer phase
of plasticity at some point whether now or in your further
presentation, can you talk about the possibility of actual
revisions like the slide you showed Mike Merznich's information?
I know Mike is very interested in revisions of neural process.
PROFESSOR JESSELL: Yes, I will comment on that perhaps
in this context of the adult stage and what some of the Merznich
observations, some of those implications are. Yes.
CHAIRMAN KASS: Gil, very briefly. Then I think
we should let Dr. Jessell complete.
PROFESSOR MEILAENDER: Yes. Whether I
can make sense of this question or not I don't know.
I don't belong to the party of rationality. But could
you or would you use in describing this process of development
a word like optimal? Is there an optimal point anywhere along
the way? Or is that not the kind of word that you would use?
Is it just a process that you describe in which one couldn't
find a place to use that sort of word?
PROFESSOR JESSELL: I think what we've learned
is that development as I'm describing it is a process
of gradualism. That there is constant change as a consequence
of influence of the environment. To use the word "optimal"
presupposes to some extent that we know how the circuit should
function in all its aspects. What we see are erosions of
performance in behavior that presumably can be linked to changes
in circuits. So that is suboptimal.
What the optimal wiring of these circuits are I think we
don't understand. So dramatic perturbations I think we
can interpret. As we'll see, maybe it's a matter
of give and take that in considering for example the plasticity
of the adult nervous system. One is balancing two forces.
One is that you want some constancy in connections because
that constancy allows you to perform at a high cognitive level.
Learning and memory are thought to be encoded at least in
part in the constancy and the strengths of individual connections.
So constancy there is a desired attribute perhaps something
On the other hand if you think about regeneration of the
nervous system under conditions of lesional injury, then the
nervous system, that constancy becomes a liability in terms
of the ability to reform connections. Where optimal is along
that sort of line, I don't know and I don't think
anybody perhaps knows at this point.
I'd now like to come onto the question that we were
just discussing which is over what period does plasticity
really persist. Is there as classically viewed a critical
period and once you reach the end of that critical period
in post natal development, the wiring diagram of the brain
is fixed and activity simply operates on the structure that
exists without having the ability to change it further? Work
from Michael Merznich and many other people I think has very
dramatically changed our view of the extent of plasticity
in the adult brain.
What I was going to do is to talk about three aspects of
this balance between constancy or stability and change in
an adult context in three different sections. One comes back
to what we were talking about earlier in terms of just the
generation of neurons. The second is the reorganization of
circuits. The third is the regenerative capacity of the nervous
system as a function of age.
One of the dogma that has been overturned in the last 10
to 15 years relates not only to the reorganization of circuits
but to this idea that neurons are born early in development
and then once they're born the nervous system has acquired
its mature cohort of neurons and there's not much you
can do to change that situation. I think that now we know
that that is not the case and that there are certain privileged
sites within the brain that are capable of new neuronal production
even in the adult stage. This obviously has many implications
in a clinical, in a therapeutic way.
If Rusty Gage who I think was due to be here, he would expounded
on this view because he's been one of the major contributors.
But in his absence, I'm going to really review very briefly
some of the evidence that I think provides persuasive evidence
that at least in the mammalian brain although I think whether
this is really true and the extent to which it's true
in human brain is still a matter of debate that in the mammalian
brain, in mice and many experimental animals, neurons are
produced in the adult nervous system, in the adult brain.
But they are not produced everywhere. There seem to be
two privileged sites where neurons can be produced. One of
those is in a region in the forebrain in the subventricular
zone. The details really don't matter, but many of these
neurons will then migrate into the olfactory bulb, one of
the regions that is involved in processing sensory information.
It's been known for many years that neurons that convey
odor perception from the external world turn over even in
the adult state. These are the primary sensory neurons.
What has been appreciated more recently is that they are
target neurons, the neurons that have to receive that incoming
odor information also turn over at some rate. So the olfactory
system for whatever reason and one might speculate on that
is a privileged site within the human brain in which neurons
are constantly being replenished. What impact that has on
one's ability to respond to odorant information to pheromone
information, the environment, is something that could be discussed.
What this shows very clearly is one site in which neurons
are produced in the adult brain contravening existing dogma
that had persisted for many decades.
There is also a second site and so far only a second site
of new neurogenesis and that is found in a cortical structure
called the hippocampus. This is an area which has been implicated
in many interesting cognitive functions, learning and memory
amongst them, but also mood and emotional disorders. It's
very clear here that there is a small group of precursor cells
that in the adult animal in just the same way that we've
seen in the embryonic period of neurogenesis, there are progenitor
cells which make the decision whether to leave the cell cycle
and become post mitotic neurons in the adult.
You can visualize them with various molecular tricks. What
people now think that at some slow rate these neurons are
produced. They begin to acquire all the hallmark structural
features of neurons and there is some, albeit weak, evidence
that these neurons can actually contribute, reconstitute themselves,
into functional neural circuits although I think it remains
unclear as yet what the functional consequence or contribution
of these adult generated neurons is.
But this has been a very surprising set of observations
that initially met with some resistance and then some over
enthusiasm and interpretation. But I think that it's
now clear that these two regions of the adult mammalian brain
can produce new neurons.
There are further implications of this observation in terms
of plasticity of the mature brain because the production
of neurons for example in the hippocampus in the adult brain
doesn't proceed at a constant rate. It itself, this process
of neurogenesis, can be influenced by environmental stimuli.
Rusty Gage is one of the people who've show that rearing
experimental animals in deprived or enriched environments
can have a relatively significant impact on the rate of new
neuronal production in this region of the hippocampus. One
of the experiments that Gage performed is to rear mice in
a relatively deprived experimental laboratory environment
or to give them enriched environments or environments in which
exercise and motor systems are activated at a much higher
You won't be able to see this because of this technical
problem, but maybe if we concentrate of the middle panels
here, each of these black dots represents that region of the
hippocampus where new neurons are produced and each black
dot is a newly generated neuron under these control conditions
and under conditions of exercise or an enriched environment.
Here you can see that stimulating the environment of the animal
produces a significant increase in the rate of new neuronal
production. Again this is a much later example of the way
in which an animal's environment influences plasticity
in the mature state.
DR. ROWLEY: I was wanting to ask you to define more
precisely the age both of the animals and of the equivalent
age in humans.
PROFESSOR JESSELL: Yes. In these animals, these
are mice in this case. You can see the same thing in experimental
rats. These are so-called mature which would be two to three
to four months of age, something like that. The issue with
humans is, I think, at a more fundamental level and there
are people more expert than I who could address this as to
whether the same process that we see in lower mammalian species
occurs in a significant way in the adult human brain. I think
that is a matter of debate.
It's clear that there are progenitor cells that are
proliferating in the adult human brain. The question as I
understand it is whether those progenitor cells produce, overt
fully authenticated post-mitotic neurons in the human brain
at the rate that they do in the lower mammalian brain. This
phenomenon can be seen at six months of age in mice. This
is a very robust phenomenon. It's not the trailing edge
of a developmental process. I think new neurogenesis is really
occurring at a relatively constant level probably until middle
age for the mouse. But again there are fundamental issues
as to what extent - and maybe other people will have views
on this, to what this really occurs in the human brain.
So one form of plasticity change in the human brain as a
consequence of the environment is seen through simply modifying
the environment. But there are also experiments which suggest
that pharmacological intervention can also influence the rate
of neurogenesis in the hippocampus at least in experimental
Eric Nessler and Ronald Dumond, some years ago, not too
long ago, came up with a very striking observation that anti-depressant
treatment using a serotonin uptake inhibitor, Fluoxetene,
has a dramatic effort on the rate of neuronal production in
the hippocampus in this same system. I've just taken
one image which is exactly the same type of experimental approach.
Under the control conditions, there are relatively few of
these black dots. You probably can't even see them there.
Under Fluoxetene treatment, then the rate of neurogenesis
increases in the hippocampus. This suggests that clearly
these drugs have behavioral consequences and raises the question
of whether some of the behavioral consequences of mood-altering
drugs are in fact a consequence of changes in the rate of
new neuronal production. Clearly these drugs may have many
effects. So this could be an incidental rather than a causal
My colleague at Columbia, Rene Hen, has performed an interesting
series of experiments recently that were reported in Science
last year which actually while not proven at least keep open
the idea that adult neuronal production is in fact perhaps
a causal contributing factor to changes in the behavioral
consequences following serotonin uptake inhibitors. This
is a very active field at the moment, but I think it have
many people galvanized about thinking not only about natural
experiences but also the way in which pharmacology and drug
treatment influences this aspect of brain plasticity and behavior.
This is a recent set of observations so the consequences of
this I think remain to be discussed and thought about. This
is one example which really takes the information.
DR. HURLBUT: Those studies are on mice.
PROFESSOR JESSELL: These are on mice.
DR. HURLBUT: And are these mice that are in any
sense depressed so to speak or are they in an enriched environment?
PROFESSOR JESSELL: Yes. There are a behavioral
assays that the Hen group have done in particular where they've
used certain behavioral, open field trial responses. It's
a complicated issue. One of the things that the Hen study
tried to do is actually prevent adult neurogenesis and then
ask about the behavioral consequences of Fluoxetene administration.
One of the ways they tried to do that is to irradiate the
hippocampus, killing the progenitor cells, preventing neurogenesis
and they have some behavioral evidence that many of the behavioral
consequences of Fluoxetene treatment are belated as a consequence
That raises all sorts of additional questions about whether
irradiation is specific and whether you can contribute the
effects of irradiation to the loss of neurons and whether
the behavioral parameters being measured in mouse really are
a reflection of the state of elation or depression as viewed
in a human. I think all of these are open questions and hopefully
will emerge with further studies.
DR. HURLBUT: So essentially you are saying that
there is to date no evidence that an otherwise normally functioning
enriched mouse would have increased rates of neurogenesis
under an SSRI.
PROFESSOR JESSELL: I think many of these things
are suggestive correlations at the moment that drugs and environmental
experience that change the behavior of the mouse change neurogenesis.
The Hen experiment to my knowledge is the experiment that
most closely links causality or implies causality in the process,
but I think Hen himself would not argue that that case has
been proven at the moment. I think there are experimental
designs that will emerge over the next two to three to four
years that will establish causality or not. Then I think
things become interesting.
This has been plasticity in the adult context viewed from
neuronal production. There is extensive evidence that aspects
of circuitry are also plastic beyond the traditionally viewed
critical period. We've discussed the Merznich type of
experiments and Michael Merznich has been the person who has
promoted this idea most persuasively. It again comes back
to mapping in regions of the cortex, particular regions of
a peripheral sensory receptive field.
Merznich has done this to a great extent in the somatosensory
system where individual regions of the body surface can be
mapped to particular locations in the somatosensory cortex.
Then the basic gist of these Merznich type of experiments
is to change the level of information that is applied to that
peripheral receptor field by training in way or another and
ask in a mature central nervous system in the mature brain
whether the cortical representation of that peripheral receptor
field changes using physiology and to a lesser extent using
Then there are striking examples that the cortical representation
of a given receptor field for example on the digits in a primate,
in a monkey, changed dramatically as a function of training.
So here is the receptor field on the digits of a monkey.
This is the normal region of cortex that receives that information.
This darkly shaded area represents the normal area, the normal
region of cortex, that receives information from those three
receptor fields on the digits. After training you can see
a very significant expansion in the area of cortex that processes
This is a functional reorganization of the nervous system
as a consequence of experience or training. It raises the
issue of what is the structural basis of this. Is this a
structural reorganization in the way that we've seen at
earlier stages of development in the visual system or is this
really a functional reorganization? Are some synapses normally
just not operative and sitting there in a quiescent state?
But upon training, these synapses suddenly start to work?
So at a physiological level, you see a change in the map.
These are questions that I think are currently being examined.
The extent to which this clear and established functional
reorganization which is the important thing reflects a structural
change or reflects changes in the efficacy of synaptic communication
between neurons. Merznich's group has some evidence for
structure, but whether structure accounts for all of the dramatic
changes in mapping of sensory inputs I think remains to be
seen. Nevertheless this very clearly, I think, shows that
there can be reorganization of circuits as well as simply
the generation of neurons at a much later stage in an adult
Then finally I want to deal with, if you like, the flip
side of plasticity in the adult nervous system. The ability
of the nervous system to adapt to a changing environment probably
is a good thing. It allows you to optimize perhaps circuitry
in relation to particular environmental conditions.
But one of the things the nervous system is not good at
doing is reorganizing in the wholesale way so all of the processes
I described at early stages in development, the extension
of axons, the ability to form connections, the ability to
refine those connections in a very dramatic way, decrease
dramatically with age. One of the major problems from a clinical
perspective is the relative constancy here. I've focused
on examples were things can change. Neurons can be produced.
But by and large, the mature nervous system is a relatively
static structure. This has dramatic consequences following
damage or injury to particular regions of the nervous system.
In one well-talked about example in spinal cord injury, one
of the reasons that people with high spinal cord injuries
recover so poorly is because of a poor regenerative capacity
of the nervous system. This is if you liked limited plasticity.
Damage to a particular region, all of the descending fibers
from the brain that are necessary to activate the motor system
to restore motor control, the extension of those processes
across a lesion or scar region is extremely poor at the moment.
This again has been appreciated for 100 years.
Over the last ten years I think partly through work on developmental
mechanisms, there is now some hope that plasticity in this
context can be enhanced dramatically by changing the environment
in which regenerating axons are trying to grow. One could
view this in two ways. One is that perhaps the failure of
adult neurons to extend axons effectively to recover function
is an intrinsic property of those neurons. They just somehow
with age lost the capacity to grow axons. That probably is
not the case.
A major influence on the inability of axon regeneration
is nothing to do with the neuron itself, but is the fact that
the environment of that neuron is now extremely non-permissive
for axon growth whereas at an earlier stage in development,
it was highly conducive to growth. The last five years or
so has seen very dramatic advances in understanding the molecular
basis of that non-permissive adult environment. It turns
out that many of the supporting glial cells in the nervous
system - we focused here on neurons - in particular a class
of cells called oligodendrocytes express proteins on their
surface that will inhibit axon regeneration.
Since most axons are trying to grow in the mature central
nervous system in the highly myelinated oligodendrocyte rich
environment, the reason they don't grow well at least
in part is because oligodendrocytes with maturity acquire
proteins that inhibit axon regeneration.
Again the details don't matter, but this slide is really
just to show that there are now some molecular reality to
proteins that are expressed on the myelinating oligodendrocyte
and receptors expressed on the sensitive neuron. Many studies
now are underway to try and eliminate thes proteins and see
whether that has the ability to enhance some regeneration.
This is viewed from a clinical perspective to recover motor
function for example in cases of spinal cord injury, but I
think it has interesting implications because it comes back
to this issue about what is optimal in terms of stability
and plasticity in the mature state. If you could find ways
of allowing perhaps in order to maintain faithful function
of neural circuits, you have to find a way to cement them
The downside of that fixation process is the inability to
recover function from more dramatic traumas to the nervous
system. What happens if we can now unleash plasticity to
a much greater extent than had previously been possible?
What will be the consequence in terms of other aspects of
nervous system function, cognitive processes? Will one generate
complete disorder? Will animals become more intelligent as
judged by behavioral methods under conditions where they can
reorganize circuits in the mature state?
I think these are issues that one is only just beginning
to scratch the surface with. With the technical abilities
that come in part through molecular biology over the last
five years, one will likely be able to manipulate circuitry
in the adult central nervous system in a much more profound
way than has been possible for the last century. Thinking
about how one then designs, at least initially experimental
tests to examine the consequences of reorganization in the
mature state, I think is something that needs considerable
thought. The last thing I want to do is if I still have five
minutes. I can stop at this point.
CHAIRMAN KASS: Why don't we stop and take a
few questions if that's all right?
PROFESSOR JESSELL: Okay.
CHAIRMAN KASS: Janet.
DR. ROWLEY: I'd like to hear the rest of what's
been prepared please.
CHAIRMAN KASS: Okay. All right.
DR. ROWLEY: We all can have a shorter lunch.
PROFESSOR JESSELL: This will be two minutes so hopefully
it won't impose too much. What I've talked about
so far has really been normal developmental processes, the
influence of environment and genetic programs and predetermination
on the structure and function of the nervous system. But
this has as we've begun to talk about in regeneration
a clinical consequence. One of the other things that has
emerged I think from understanding normal molecular genetic
programs that have developed is that those processes are likely
to be precisely the processes that go awry in many neurological
and psychiatric disorders. This is a very new field, but
I think there are enough small examples to indicate how this
will really change or produce a convergence of clinical neuroscience
and these basic developmental mechanisms.
One example just comes from looking at many classically
defined neurodevelopmental or cognitive disorders. In the
few cases where genes associated with those disorders have
been identified, many of those genes affect precisely the
processes that we've been talking about this morning,
the process of neuronal identity through genes that control
cell identity, transcription factors, the nature of signaling
factors, the nature of synaptic proteins, so two examples
that I won't stress.
Some of the proteins that I've mentioned briefly that
are involved in synapse formation, how one neuron communicates
to another, it's been shown that mutations in those genes
are associated with forms of autism and with the Asperger
Syndrome. Signaling factors that are involved in neuronal
communication have been associated, albeit not completely
persuasively, with schizophrenia.
In the case of transcription factors, one has the
surprising example that quite highly complex, cognitive disorders
relate to genetic defects in genes that are DNA-binding proteins
that control patents of gene expression. So two good examples
of that exist.
One is a syndrome known as Rhett Syndrome which again if
you like is a variant form of autism. There is a gene that
probably accounts for the vast majority of Rhett Syndrome
cases. These are children who acquire gradually over the
first two to three years of post natal life progressive cognitive
impairments, stereotypic motor behaviors. They lack function
of a gene involved in methylating many target genes. Some
of those target genes are now known and one of those are the
trophic factors that we talked about in the context of keeping
An even more striking example is in the case of genes that
influence human ability for language and integration of spatial
information. So there is a syndrome described by Tony Monaco
and by Vaga Cadan associated with facial abnormalities and
with dyspraxias. There is a clear pedigree. So these people
have abnormal impaired language acquisition. The gene associated
with that pedigree turns out to be a transcription factor
which is expressed and involved in the determination of neuronal
This raises very profound questions of taking a disease
which appears highly specialized in a cognitive manner and
finding out that the gene that causes that is a gene expressed
at a relatively early stage of development, probably involved
in establishing some aspects of circuitry. One of the challenges
that I think is going to face the field is now understanding
causally how the mutation in that one gene influences a relatively
specialized set of cognitive behaviors.
Another example is Williams Syndrome which is associated
in contrast to for example Down's Syndrome and this KE
language defect with in fact an over elaborate use of language
by children. They have a very extensive vocabulary. They
are extremely articulate, but they have very poor ability
to extract and integrate information from a visual world.
This is a very highly localized cognitive disorder that has
been studied by Ursula Belugi in particular. All of the children
who have this cognitive disorder have a small chromosomal
deletion that probably eliminates in large part a transcription
factor that controls cell differentiation.
So here is another example where very specialized cognitive
dysfunctions are mapping to genes that we can now begin to
understand in terms of their role in developmental processes.
Yet there is a very large gap between understanding the nature
of the gene, the neuron, the circuit and the behavior. But
I think over the next several years as human genetics begins
to give us more information on the nature of these behavioral
disorders many of which affect this early childhood period,
we're going to be faced with trying to link in very relevant
clinical context some of the early developmental mechanisms
that we've talked about with some of these behavioral
events. Perhaps that's all I want to say. I'll stop
CHAIRMAN KASS: Thank you very much. I'd like
to run over and as Ms. Janet says we'll have a slightly
shorter lunch because we shouldn't waste the opportunity
of having some conversation with Dr. Jessell after this very
comprehensive and very stimulating presentation. Thank you.
Dan Foster please would you start and could we get the lights?
DR. FOSTER: I have just - I don't know whether
you can answer this in 30 seconds, but it's always intrigued
me. From the very beginning, you've talked of a necessity
of scaffolding and we move here in the spinal cord and you
have some sort of signaling, a way that neurons can grown
and so forth. I understand a little bit about oligodendrocytes
and astrocytes and jagged and notched and so forth.
My question is something different in terms of therapy.
At least in rodents, it looks like that if you inject a neurogenic
stem cell even in a peripheral vein that it's going to
circulate and get to the brain and then somehow will target
to a damaged cell - it might be a stroke or a glial blastoma
- you put a human tumor in there. So the more I listen to
you it's almost as though they are chemotactically or
someway attracted rather than going on scaffolding. I'm
just dying to know what you think about how that process goes
in ten words or less.
PROFESSOR JESSELL: The first slide I showed I think
illustrates the nature of the problem a little bit. At early
embryonic stages the brain is small. The adult brain is extremely
large. So just distance imposes great challenges. The environment
of the adult brain is completely different from the embryonic
brain. Many of those scaffolds that we talked about have
disappeared in the adult brain.
So in an example that you describe where cells seem to be
smart enough to get to the right place, I think we have very
little understanding of how that homing behavior is actually
achieved given that the normal developmental cues are now
missing. I would say that the evidence that that happens
in a highly efficient way is not so great.
I think it can happen whether it happens as efficiently
as it would in a normal developmental context so maybe only
one in a million perhaps by some random probabilistic nature
ends up in the right place and then those cells proliferate
in an appropriate environment. The shorter answer is no one
knows how this process occurs, but whether it occurs in an
efficient way is, I think, also a matter of debate.
DR. FOSTER: But just to follow, I mean I saw a picture
in PNAS where you put a green fluorescent protein on and you
were tracking infiltrating cells from a glial blastoma and
there were a lot of these neurogenic stem cells that got into
PROFESSOR JESSELL: Yes.
DR. FOSTER: Not only in the mass but moving in there
and honing in on it. So maybe you're saying -
PROFESSOR JESSELL: I'm completely receptive
to the idea that perhaps under conditions where in a tumor
the blood brain barrier has broken down, one of the things
the tumor does is produce attractants that then guide cells.
I think there is likely to be that process.
CHAIRMAN KASS: Dr. Carson.
DR. CARSON: Yes, early on, you were discussing the
periventricular glial cells which divide and then give rise
to neurons. There's always left behind a glial cell.
Now is it possible later on to provoke that residual glial
cell to differentiate?
PROFESSOR JESSELL: Yes, this relates to why most
of the adult nervous system doesn't actually produce neurons,
only these two little epicenters. So does that mean that
the precursor cells simply don't exist in those other
regions or whether they exist and they're quiescent and
something needs to happen in order as you say to provoke them?
There's a lot of evidence, for example, in the spinal
cord, one of the areas that is not noted for neuronal production.
If you damage the spinal cord, ependymal cells around the
central canal will start to proliferate and start to produce
glial cells. So there clearly are environmental stimuli or
trauma that can kick these cells into action.
What they can't do still even under those conditions
is produce neurons. They seem to produce glial cells. So
is there a single gene? As we saw there are single genes
that produce neurons. Is there just one gene missing from
those cells that impairs their ability to produce neurons?
These are things that I think will emerge over the next five
years as many people are looking at this as now some of the
molecular players in the normal developmental process can
be applied to the conditions of cell regeneration in an adult
CHAIRMAN KASS: Mike Gazzaniga.
DR. GAZZANIGA: Dr. Jessell, a wonderful talk. I
wish my mentor, Roger Sperry, could have been alive to hear
your first section of it and that beautiful molecular work.
When one hears the sweep that you gave today, sometimes people
can misunderstand the extent to which the people come to think
of the brain as a set piece to the extent to which it can
be infinitely plastic and change. Plasticity is used in this
funny way. We all still are capable of learning Leon tells
me. So plasticity in that sense is going on all the time.
But plasticity in the infrastructure of the nervous system
aspect is probably not as much as we would like to think it
is. So when we hear these beautiful examples of Merznich
and of Rusty Gage and so forth, I would be interested in your
opinion on a scale of the set piece - I think those were your
words - versus the changing at the edges.
PROFESSOR JESSELL: Yes.
DR. GAZZANIGA: And as you go through maybe by decade,
just give us a sense of your feel for that.
PROFESSOR JESSELL: I think there's no region
of the nervous system in which the two extremes of a genetic
predeterminism and environment don't contribute. I think
the question is what are the relative contributions of those
types of programs according to a particular region.
I tend to agree with you that there is a remarkable degree
of molecular genetic programing of connectivity. The more
one understands the more remarkable that observation is.
In regions like the spinal cord where motor commands are essential
for processes like locomotion, I think there is very compelling
evidence that many basic features of those sensory-motor integration
processes can get wired up in the absence of experience, in
the absence of activity.
One of the early sets of experiments as Sperry did before
he worked on the visual system really demonstrated that that
essentially if you misconnect so that the sensory information
coming from an extensor muscle is now rerouted to a reflexor
motor neuron, an adult animal has a hard job in adapting to
that surgically induced misconnectivity. So what that says
which I think was the basic thesis of Sperry is that essentially
there are strong molecular cues that drive the specificity
of circuitry. That will occur in the spinal cord. It will
occur in the cortex.
What I think we're seeing in the cortex for example
but probably in other regions is that the extreme view that
in the adult state all of those connections are fixed at a
functional and structural level is beginning to be eroded
a bit. But the real challenge I think is to work out to what
extent the environmental influence what are the constraints
achieved by these early molecular programs on which environment
Now all of these experiential influences have to have a
molecular basis. This is not some mystery. So presumably
environment and activity is changing connectivity through
a molecular program. If we knew more effectively the nature
of that molecular program - what does the firing of an action
potential in the neuron really do biochemically to that cell,
what molecules change - then we will be able to integrate
more effectively the plasticity environment together with
these molecular programs. To me that's one of the great
things to emerge from understanding the molecular biology
of development. It allows you to perhaps demystify some of
the environmental influences that clearly do exist.
CHAIRMAN KASS: Janet, please.
DR. ROWLEY: Well we spoke about this earlier in
our general discussion and the release of this report and
the one issue that was not touched on directly is the whole
issue of cloning for therapeutic uses. I would be interested
in your opinion on how you think embryonic stem cells for
example could be used therapeutically either to deal with
those problems of the central nervous system or of course
for spinal regeneration. The second part of that is what
kind of research is needed, what kind of support for that
research is needed to see whether embryonic stem cells do
have a potential in dealing with these very serious problems.
PROFESSOR JESSELL: So this is something that in
a slightly different context we in the lab experimentally
got very interested in because what we do on a day-to-day
basis is work on spinal cord organization and development
and an important set of neurons there are motor-neurons which
really mediate all the central nervous system control of action
and movement. There is a reasonably good understanding of
the normal developmental processes that take a naive progenitor
cell and convert that to a motor neuron.
So one of the things we asked is if we really understand
that process as it occurs in the normal embryo, can we then
apply that information in the context of other precursor cells
like embryonic stem cells and apply the right embryonic signals
in the right order and the right time and the right concentration
and simply ask the question of whether you can now convert
an embryonic totipotent stem cell and can drive that to a
motor neuron using normal developmental signals? Rather surprisingly
that turns out to be quite easy to do. So the same signals
that operate in the normal embryo will operate in the context
of an ES cell such that you can produce fully differentiated
motor neurons from embryonic stem cells.
One of the things that I think development has taught is
how to manipulate other classes of progenitor cells along
defined pathways. That raises all of the questions that you
mentioned about what use is that going to be in regenerative
medicine. It may be that the motor neuron is not the ideal
neuron in which to test this because of the details of circuitry
that are necessary.
So we come back to this question of if you put these cells
back in, how are they going to find their right targets in
a very different adult environment. But in some diseases,
in demyelinating disorders where one is trying to reintroduce
oligodendrocytes or in Parkinson's disease where there
is almost proof of principle that if you can put back dopaminogic
neurons, you can ameliorate their motor deficit.
If you could make infinite numbers of purified dopamine,
mid-brain dopamine neurons, with that be an interesting route
to self-therapy in that particular neuro-degenerative disorder.
I think all of those issues are soluble now. What I think
we've learned that molecular biology has taught us how
to turn embryonic stem cells let alone the issue of adult
But I think embryonic stem cells I think can be converted
to any class of central nervous system cell that one now is
interested in on the basis of the normal developmental program.
To what extent that information is useful in a clinical context
I think will very much depend on which disease one is thinking
about. Somewhere it won't be useful and there is somewhere
where I think it's promising.
But the technology just from the point of view of cell differentiation
I would say exists today. So if one wanted to generate one
class of CNS cell, I think there are good ways now of thinking
about how to do that.
DR. ROWLEY: Then I ask a second part. What's
needed to move this field along?
PROFESSOR JESSELL: Yes. Much of the work that is
at an experimental stage has been done on mouse embryonic
stem cells. Mouse cells, ES cells, are remarkably constant
from cell line to cell line. The first thing that is needed
is to ask whether the developmental potential of human embryonic
stem cells mimics or reflects that in the mouse. I think
what we know at the moment is that many of the existing human
ES cells, until Melton's recent studies, have shown a
remarkable variability in their properties in differentiation
along neuro-pathways, a much greater diversity of properties
than would be predicted simply from the mouse experiments.
One of the virtues of the report from Melton and his collaborators
in generating new cell lines is that many of these cell lines
now are going to increase the probability that some of them
will behave like their mouse counterparts. We have actually
been working together with the Melton group to extend some
of the work on mouse embryonic stem cells to ask which of
those human lines that Melton has generated behave most closely
to their mouse counterparts. That's a first step.
DR. ROWLEY: But then this brings the point which
some of us have again dealt with tangentially that these are
not cell lines that you can study with Federal funding.
PROFESSOR JESSELL: No, that's right. There's
DR. ROWLEY: So that one of the critical things that's
needed is Federal funding for these new cell lines as they
become available to see whether or not they are useful.
PROFESSOR JESSELL: I would absolutely endorse that.
They are technical limitations.
DR. ROWLEY: And have you tried any adult stem cell
lines to see whether they too can produce functional neurons?
PROFESSOR JESSELL: Yes. So using the
mouse embryonic stem cell as a positive control, that is we
can now turn those cells into motor neurons at will with the
addition of two chemical factors. You are now in a position
to ask whether adult progenitor cells from the nervous system
can recapitulate the properties of the mouse embryonic stem
cell as we currently understand how to do that. If you do
then adult neuro-progenitor cells will not generate motor
neurons under the conditions that the mouse embryonic stem
What that tells us is there are some constraints on adult
progenitor cells. It doesn't mean it can't happen
in the future. It just means as of today I think the developmental
repertoire of an adult neuro-progenitor cell is very much
more limited than its embryonic counterpart. Just in a practical
sense what you can do with embryonic stem cells today in the
context of neutral pathways of differentiation far exceeds
what can be achieved with adult neuro-progenitor cells.
CHAIRMAN KASS: Could I ask before? We're going
to have to draw to a close fairly soon, but stepping back
from some of the particular phenomena that you've described
and looking at their possible human-social-education significance
because we'll be talking about this later on, and the
question is exciting and wonderful though this is scientific
matter. If one wanted to think about the possible human implications
of this over the next decade or so - I know you're not
a prophet, you are a scientist - but the question is what
are the one or two areas or questions that you would like
us to keep our eye on when we think of the significance of
One thing I gather has to do with the knowledge of some
of these horrible behavioral and mental disorders and the
capacity to be able to learn some of their sources and perhaps
ultimately to intervene. The other vaguer thing when Michael
Sandel talks about Mozart and the like and there's all
kinds of faddishness out there with respect to taking advantage
of this alleged early environmental stimulation to enhance
the pre natal capacities, could you say again along those
lines and have we missed or did I miss something, some other
areas of significance?
PROFESSOR JESSELL: To me the greatest challenge is that
the very few, relevant interesting human behaviors, do we
have any idea about the principles of circuitry that underlie
those behaviors? We don't yet know. We have through
imaging methodologies, through other methods of intervention,
we have a sketchy idea of some regions of the brain that are
more involved in learning and memory, mood disorders. But
the nature of the circuitry and the ability to intervene in
that circuitry and relate that to important forms of behavior,
I think is still at a very, very primitive state.
So one of the things that I haven't emphasized but I
think at a practical level is going to have important consequences,
both experimentally and in terms of studies of human behavior,
is what molecular biology of circuit assembly has given one
so far are a large number of genes that show that neurons
that behave differently are molecularly distinct. The advantage
of that is that certainly in experimental animals that is
going to give you an unprecedented way of manipulating the
circuit on the basis of a neuron by neuron change which hasn't
been possible with conventional anatomy and conventional physiology.
So for example if we knew in some region of the cortex a
gene that defined one very small subset of functionally coherent
neurons, how do you get at those neurons and how do you tell
what their contribution to circuit and behavior is at the
moment? Through the gene what we can now do as a field is
introduce genetically encoded proteins which allow you to
manipulate that set of neurons in a coherent way. For example
you can introduce proteins that allow you selectively and
conditionally to take those neurons out of the circuit and
look the consequences for behavior. Or you can introduce
sensors that allow you to selectively activate those neurons
and leave all of the other neurons intact and look at the
consequence of that particular micro circuit for a given behavior.
So what I think will emerge from this is a much clearer
understanding of what behavior in any context really means
in terms of the details of the circuit. What are the core
elements of the circuit that contribute to that behavior?
If one knew that, then I think the diversity of range of human
behaviors that one sees in the normal population together
with pathological behaviors one can start now to analyze that
from the level of single neuron or dysfunctions. What is
the relevant aspect of circuit that we're trying to look
I think, and Michael will have thoughts on this, my feeling
is at the moment is that we simply don't know enough about
the basic circuitry that underlies many of these interesting
behaviors even to know how to approach the problem. What
I would hope is that link between circuit and behavior becomes
consolidated over the next ten years.
CHAIRMAN KASS: Thank you. Look we are 12:50 p.m. We
have guests coming at 2:00 p.m. I don't know how quickly people
will serve lunch where we're going. If Dr. Jessell might be
willing to simply standby for people who didn't get a chance
to ask their questions, I fear that if I don't discharge you
now I won't see you back here before 2:30 p.m. and that's
not permitted. So, Dr. Jessell, thank you for an enormously interesting
and stimulating presentation. It's a massive amount of stuff
to present in a kind of coherent way. I learned a great deal and
I trust my colleagues did too. Thank you and we'll reconvene
at 2:00 p.m.
(Whereupon, at 12:48 p.m., the
above-entitled matter recessed to reconvene at 2:03 p.m. the