Neuroanatomy lesson

Document Sample
Neuroanatomy lesson Powered By Docstoc
					                                                                   Caput 2009

                      The neurobiology of fear

Speaker’s notes to accompany PowerPoint presentations for the Master’s course
                    “Molecular and Cellular Neurobiology”

                                 Bruce Jenks
            Donders Institute for Brain, Cognition and Behaviour
                      Radboud University Nijmegen
To introduce the topic “Neurobiology of Fear” I will first introduce the brain region
intimately involved in orchestrating fear responses in mammals.

This brain area, part of the limbic system, is called the amygdala.

Early neuroanatomists gave it this name because its overall shape resembled that of an

The amygdala has a central role in emotional processing (pain, fear, anxiety, desire) plus
other functions. such as social memory, at least in rodents.

A number of psychic disorders, such as stress disorders and disorders that clinicians have
designated as emotional dysregulation and borderline personality disorder, are associated
with dysfunction of the amygdala.

For these reasons the functioning of the amygdala has been the object of many
biomedical research programs.

In introducing fear I will quickly review how the brain processes sensory input to
produce appropriate physiological and behavioral responses.

Processing of sensory input to generate behavioral outputs

Most sensory information, after it has found its way to the primary sensory cortex, is then
processed in deeper areas of the cortex known as the intermediate cortical fields.

In this processing appropriate output programs are selected.

These programs are programs that have been learned by the animal.

Following this processing the information is then sent to a very ancient part of the brain,
the limbic system (also called the visceral brain; visceral = inward feelings as opposed to
conscious reasoning).

Here the incoming information from the cortex is integrated with programs of the limbic

These limbic programs are inborn i.e. they have been developed during the evolution of
the species.

The limbic system then sends out programs to other parts of the brain where the final
programs are selected.

These are programs of the autonomic nervous system (controlling heart rate, breathing
etc.) and the neuroendocrine system.

Particularly important in assembling some of the final programs is the hypothalamus.


So, there is a general hierarchy in generating physiological and behavioral responses,
with the cortex regulating the limbic system, which in turn regulates the hypothalamus
and other brain areas.

The amygdala, being part of the limbic system falls under control of the cortex.

The part of the cortex intimately involved in the regulation of the amygdala is the
prefrontal cortex.

As outlined below, brain imaging is being applied to demonstrate prefrontal-amygdala
relationships in brain function.

Brain imaging studies of the amygdala

Brain imaging methods (fMRI and PET scans) are finding wide application in studying
human brain function, including the functioning of the amygdala.

A recent fMRI study by Beauregaurd et al has beautifully demonstrated PFC control on
the amygdala.

The first BOLD image shows the activation of the amygdala in a male volunteer (a link is
provided that gives the principles of fMRI and the BOLD image, if these are unfamiliar
to you).

The BOLD image shows that the right amygdala is activated while the subject is
watching erotic film clips.

Such activation is in keeping with the idea that one of the functions of the amygdala is
involvement in construction of emotional responses.

The fact that only one of the anmgdala is activated is not surprising; In brain regions that
have bilateral symmetry there is often a division of labor, the left being responsible for
some functions the right for others.


The second image is the BOLD image generated when, as instructed, the volunteer tries
to suppress emotional response while watching the film clips.

Clearly he has been successful in suppressing activation of the amygdala.

As already mentioned, the hypothalamus is down stream from the amygdala in generating
physiological and behavioral responses, in this case sexual arousal.

The hypothalamus is clearly activated while watching the film clips and this activation is
successfully suppressed, when the volunteer receives instructions to do so.

Images of the prefrontal cortex during the attempted suppression shows that specific
areas of this brain area are activated.

The idea generated is that the prefrontal cortex controls the amygdala. In this case regions
activated in the PFC are suppressing activation the programs of the amygdala.

The idea that the prefrontal cortex (PFC) controls the amygdala brings me to
consideration of the criminal mind.


Another example of prefrontal control (accessed via link)

fMRI of the threatened brain has also delivered data consistent with the idea that the PFC
exerts control over the amygdala.

In this case human subjects moved themselves, via a keyboard, through a virtual maze in
which there was also an “intelligent” predator hunting them.

The predator was simply a red dot but the subjects were highly motivated to avoid getting
caught by it as, if they were caught, they were given a painful electric shock.

If the predator was a long way off it was noted that there was a BOLD signal in both the
PFC and the amygdala.

The idea here is that the PFC and the amygdala are cooperating in orchestrating
avoidance behavior. (a response involving some thought).

If however the predator was suddenly very close then only the amygdala but not the PFC
displayed a strong BOLD signal.

The idea here is that this is a visceral, hard-wired, defense response.

The above interpretations for this data are supported by the location of the BOLD signals
in the amygdala.

In the case where the predator was a long way off the lateral amygdala displayed the
strongest signal.

This part of the amygdala is programmable and (as you will be seeing later in the lecture)
is in part the location of fear memories.

When the predator was close the central amygdala became active.

This part of the amygdala contains the hard-wired output programs of the amygdala.

So, when the predator was far away the PFC controlled the amygdala to select the best
program to avoid the predator.

When the predator was close the amygdala fired off a defense program with no
participation of the PFC.


Note: The prefrontal cortex is a large brain area (particularly in humans) which has a
number of functions besides controlling the amygdala. A link is provided to a podcast
discussing a fMRI study looking at the role of the PFC in creativity (musicians creating
new melodies versus playing a set piece).


Nice kids?

A recent structural MRI study (to measure volumes of different brain areas) has
highlighted again a possible relationship between the PFC and the amygdala.

The interaction of 137 young teenagers with their parents were assessed (duration of
aggressive behavior) and this was the size of the PFC and amygdala was determined in
each participant.

A significant positive association between the volume of the amygdala and the duration
of aggressive behavior was found, for both males and females.

A male-specific association was also found between the volume of the prefrontal cortex
structures, in which two areas were found to be smaller.

(It has been estimated that the PFC is not fully mature until around the age of 25…an
interesting estimate as this corresponds to the age at which car insurance companies will
allow rental agencies to rent cars to young people).

The criminal mind

The prefrontal cortex is very susceptible to brain damage during physical and
psychological trauma.

In the criminal mind it is thought that the connections of the cortex with the limbic
system, in particular the amygdala, are disrupted.

It has been found that many criminals have a history of physical and mental abuse as
children, with evidence that this has lead to prefrontal brain damage.

The lack of proper control of limbic (amygdala) programs is thought to contribute to
criminal behavior.

This subject has recently been extensively reviewed (Brower & Price, J. Neurol.
Neurosurg. Psychiatry).

Given are some of the statements made in this review article concerning links between
PFC damage and criminal behavior.

Presumably the inappropriate or reflexive aggressive behavior mentioned are the
aggressive defense/protective programs of the amygdala.


If the link between prefrontal damage/malfunction and asocial or criminal behavior is so
strong, one might wonder if, perhaps, society should not intervene before a crime is

These legalistic/moral issues will not be considered here, but a reference to a review
article on this topic is included in the presentation for anyone wanting to pursue this
matter further.


Overall, brain-imaging experiments support the concept of cortex-limbic system
interaction in generating behaviors.

To go after the molecular and cellular mechanisms of fear one must leave the human and
go over to animal models.

Before considering the amygdala’s role in generating fear responses I will first consider,
in a very general way, how fear is studied in the laboratory.

How is fear studied in animal models?

One of the very important experimental paradigms in studying the mechanisms of fear
(ie.where in the brain is it perceived, where in the brain is it remembered and where it the
brain are fear responses generated) is something called “Classical Fear Conditioning”.

To explain what CFC is, I want to first consider what occurs when pain signals are sent to
the brain.

The painful sensory input mobilizes physiological and behavioral responses to the input.

These are usually protective or defensive responses against the pain e.g. fight or flight
responses, collectively called “Fear Responses”.

In classical CFC the painful sensory input is given at the same time as another input,
which can be a sound, a visual input or an odor.

This second input is innocuous whereas the painful input is noxious.

This association of the innocuous with the noxious can be done several times so that
brain will associate the innocuous input with pain.

Following conditioning, you only have to give the innocuous input and the entire
protective or defensive response will be generated by the brain.

The brain has learned to see the innocuous input as a danger signal and makes the
appropriate responses.

The innocuous input, in CFC jargon, is called the conditioned stimulus or CS.

“Conditioned” because the animal has had to be “conditioned” (manipulated) to respond
to this input with the fear response.

The painful input, to which the CS had been matched during training to generate the
conditioned fear response, is called the unconditioned stimulus or US.

The US is often an electrical shock.

The simple nature of the learning task, and the readily measured physiological changes
that accompany it, have made the study of fear conditioning a very attractive model for
the study of learning and memory consolidation.


You are probably already familiar with the basic principles of CFC (especially if you
have followed my neurobiology course).

CFC is used with the fruit fly to study mechanisms of learning and memory.

Fruit flies love rotting fruit; the odor of rotting fruit will normally attract flies.

If however the odor of rotting fruit has become associated with a painful experience (i.e...
during fear conditioning) then the fly will actually fly away from the odor, even if it is no
longer associated with the electric shock (i.e the fly remembers!).

It associates the odor with danger.

In the conditioning the odor is the CS and the US is an electric shock delivered via a
metal grid.

After several CS-US associations the CS alone will cause the animal to go away from the
source of rotting fruit.

It has learned that the odor indicates danger i.e. changes have occurred in its brain as to
how it processes and responds to the odor.

This simple system has become a very powerful model for studying the basic
mechanisms of learning and memory.

In particular it has been used to identify the genes responsible for memory.

Chemical treatments are used to cause mutations in flies, flies displaying deficiencies in
memory (memory mutants) are selected and the genetic make-up of the memory mutants
is then analyzed.

The memory mutants are flies which can not remember the US-CS association (they
quickly forget to associate the odor of rotting fruit with danger and thus will quickly
return to flying towards the odor once the US is no longer given).


Another well-known example of CFC is with the snail Aplysia (also discussed in the
Neurobiology course).

An analysis of this model system has made major contributions to our understanding of
the mechanisms of learning and memory.

Here a light touch to the siphon (the CS) is associated with a painful prick to the tail (the

Normally, with the light touch to the siphon, the gills retract only very slightly.

The animal does not see the gentle touch as dangerous.

With the US the gills retract quickly and completely into the animal (defensive response).

Following CFC a light touch to the siphon is sufficient stimulus for the gills to become
fully retracted.

It has been shown that sensory neurons from the siphon and the tail converge on motor
neurons responsible for gill withdrawal.

The US-CS conditioning strengthens the synapse between the siphon sensory neuron and
the motor neuron.

Thus the animal has learned (via this strengthened synapse) to orchestrate a defensive
response (gill withdrawal) following the fear conditioning.

Analysis has shown that changes take place in the synaptic strength between the siphon
sensory neuron and the motor neuron running to the gill during fear conditioning.

The synapse becomes stronger and thus, following conditioning, a gentle touch to the
siphon will cause complete gill withdrawal.

The memory of the association between touch and danger lies in the strength of the

The simple associative learning demonstrated here with the fruit fly and Aplesia, is often
referred to as Pavlovian fear conditioning, for obvious reasons.

[Podcast Link: Discussion of a study in which caterpillars are fear conditioned and later
the moth tested for memory of the fear conditioning. The result: the moth remembers, so
apparently the neural networks storing the fear memory have been left in tact when the
animal undergoes metamorphosis!]


In turning attention to vertebrates, we must turn our attention to the amygdala, the brain
area that orchestrates the fear response.

This has proved to be the brain area where the CS and US converge to create a memory
of fear

Activation of the human amygdala can be visualized with fMRI as a consequence of
classical fear conditioning.


In this experiment we begin with healthy college students.

A stimulating bar electrode was attached to the subject's wrist and electric shocks
delivered transcutaneously

Each subject's tolerance level was determined before the fMRI experiment.

During the experiment non-emotional visual images were presented in a random fashion
via a mirror (a yellow square or a blue square).

During the acquisition phase of the fear conditioning one of the colored squares (e.g.
yellow) was associated with an electrical shock.

The blue square was never associated with a shock.

The first BOLD image shows the activation pattern of an individual to a yellow versus a
blue square during the early extinction period (extinction is the period following the
conditioning period; here the CS and US are no longer paired and the CS is gradually
lost; in the early stage of extinction the CS is still strong).

Clearly there was a bilateral activation of the amygdala in this particular individual when
the yellow square was shown.

In another individual only the left amybdala was activated and in yet another only the

All subjects (11) showed activation of the amygdala although apparently there are
individual differences in which amygdala (right, left or both) that is activated.

[note: Such differences fit with clinical findings that damage to either the right or left
amygdala is sufficient to impair conditioned fear acquisition in some human subjects].


With the experiment outlined above we have just about reached the current limit of what
we can do to study fear in human subjects.

To get at the cellular and molecular mechanism of fear we have to go over to animal
studies, most of which concern the rat.

In animal studies we also do classical fear conditioning, often associating a sound (tone)
with a foot shock.

One of the big advantages in animal studies is that electrodes can be brought into the

These can be used to record neuronal activity during acquisition (when the US-CS
pairing is conducted) and extinction (period following the US-CS pairing, when
responses to the CS alone are determined).

Or they can be used to activate neurons in the amygdala to see if we can elicit fear
responses in the animal.

Likewise, they can be used to induce lesions, to see if we can eliminate acquisition or its
conversion in fear response outputs.

Via iontophoretic electrodes or mini-injectors, neurotransmitters or neurotoxins can be
introduced to the amygdala and the effects of such treatments on fear conditioning

Also, in vitro electrophysiological approaches can be applied.

For example, brain slices containing the amygdala, can be brought in vitro and neurons
within the slices patched so that the generation of EPSP in neurons of the amygdala can
be studied.

In this way the integrative function of these neurons can be analyzed.

Such studies can also be conducted on cultured neurons derived from the amygdala.

These types of studies are revealing the cellular and molecular mechanisms of anxiety
and fear.

They may ultimately be instrumental in developing new pharmacons for the clinic to treat
malfunctioning of the amygdala.


One example of animal studies, which is more or less the animal experiment equivalent
of the human study of LaBar et al just discussed, is that of Quirk et al.

They implanted an electrode bundle into the amygdala of the rat, each bundle consisting
of 20 wire electrodes.

With these they could record the activity (action potential firing rate) of neurons within
the amygdala.

[note: having 20 electrodes increases the chances of recording some neurons; the
electrodes pick up action potentials of neurons near by; some electrodes may not be close
enough to pick up any action potential, others may pick up the activity of more than one
neuron (differentiated by the height of the potentials); Quirk et al occasionally had an
electrode that was picking up the firing activity of three different neurons]

During pre-conditioning, called sensitization in Classical Fear Conditioning jargon, they
recorded the response of a large number of neurons to what would ultimately be the CS (a
5 kHz tone of 80 dB for 2s).

During sensitization the US was also given (a foot shock of 5 mA for 0.5s), but these
were unpaired with the Cs.

Some neurons responded to the tone, other neurons did not (those that did respond,
responded time after time to the tone).

Illustrated in the presentation is the response of 5 different neurons to the tone in an
individual rat during the preconditioning (sensitization).

The spike rate (action potential frequency) is given in the Z direction, with the peak of
the most active neuron (neuron 5) being 18 spikes per the 10 ms (for representational
purposes the data was processed in lots of 10ms, the so-called bin width).

The animal then undergoes conditioning where during the last 500 ms of each CS a US
was given.

A total of 20 CS-US pairings were given, followed by a 1h rest period.

During extinction, a number of CSs were again given, but these were not paired with the

Comparisons of the responses during the early extinction trials with the sensitization
trials gives a measure of the effects of paired as opposed to random presentation of tone
and shocks.

16 of 55 neurons recoded (from a number of rats) were found to have conditioned (i.e
displayed significantly stronger activity during extinction compared to sensitization).

For the neurons shown (from a single rat), neurons 3, 4 and 5 clearly show bigger
responses and neurons 1 and 2 appear to have been recruited to be tone responders.

Overall, the response of responding neurons increased during the conditioning.

During early extinction the responses remained strong, despite the fact that there was no
more pairing of the CS and US.

Thus, some sort of memory process has occurred whereby the previous association of the
CS and US is remembered.

It is the molecular and cellular events that are occurring during the acquisition of this
memory that forms the molecular and cellular basis of fear.

The ultimate aim of this presentation is to describe what is currently known concerning
the mechanisms involved in constructing this memory, and its extinction.

At first glance one might assume that loss of fear response during extinction simply
reflects the loss of the memory (i.e. the memory of the association of the Cs and Us).

In fact, recent studies have shown that the animal does not forget, it actually learns not to
respond (i.e it forms a new memory that says the Cs and Us are no longer paired).

This new memory is stored in the PFC. How the PFC is controlling the amygdala to
block the fear response will be considered in detail latter.

Neuroanatomy lesson

In going after the cellular and molecular mechanisms of fear I will first briefly consider
the neuroanatomy of the amygdala.

The information presented here has been assembled largely from the recent review by


Different structural and functional subsystems can be discerned within the amygdala.

The basolateral complex (BLA) receives sensory information coming from other parts of
the brain.

In fact it is the lateral amygdala that is most important in this regard.

Numerous tracing studies have shown that most sensory afferents enter this part of the

This input includes auditory, visual, somatic, spatial and olfactory.

The somatic input includes normal somatosensory information (ultimately originating
from sensory cells in the skin), as well as gustatory (taste) input and input from a higher
brain center called the insular cortex.

Spatial information comes from the hippocampal formation (besides its role in learning
and memory, the hippocampus is involved in contextural orientation of animals).


The information is in some cases partially processed by other areas of BLA and it is then
sent to the second major subsystem of the amygdala, the central nucleus of the amygdala

The CEA projects to many brain areas. It is the amygdala’s interface with the fear
response system (i.e. it drives the fear responses).

Many of these are defense responses such as acoustic startle and freezing.

Startle and freezing responses are found in most mammals including rats and humans.

Startle is a sudden involuntary movement or twitch made in response to e.g. a loud noise.
It involves somatic muscle flexes, presumably preparing the muscles for the possibility of
fast contractions (e.g. if the decision is made to fight or flight)

Freezing is a total immobility except for movements associated with breathing
(presumably a defense response so a predator does not see you).

Via connections to the brain stem the CEA can bring about changes in the cardiovascular
system and breathing.

Via connections to the hypothalamus the CEA can, at the same time it is inducing startle
or freezing, activate the autonomic nervous system (increased heart rate and breathing)
and the hypothalamo-hypophyseal adrenal complex leading to typical stress responses in
the animal (elevation in glucocorticoids leading to a more active and alert animal).

The amygdala is built for fast responses

The amygdala is very special with respect to how it receives sensory input.

To illustrate this I will first quickly review how the brain in general handles sensory
information to produce motor and behavioral output.


Most sensory input goes through the thalamus on its way to the primary sensory cortex.

Here neurons carrying sensory information synapse on neurons that then carry the
information to higher brain regions.

Olfactory input goes directly to the olfactory cortex.

On the output side the information goes first to the intermediate cortical fields (location
of learned programs) and then to the limbic system (innate programs).

Again, olfactory information is the exception, it goes directly to the limbic system.; infact
it goes directly to the amygdala of the limbic system.

The information from the limbic system is then passed to the hypothalamus where final
autonomic and neuroendocrine programs are selected.

The limbic system is also responsible for activating appropriate motor programs.

In this way well integrated motor, autonomic and neuroendocrine responses to sensory
input can be made.


The above is the general way in which sensory information is handled within the CNS to
ultimately produce output.

The amygdala is special in that it receives all the sensory information, not only via
cortex, but also directly via the thalamus.

Neurons carrying sensory information have been found to run from the thalamus and
innervate (primarily) the lateral amygdala..

Direct thalamic-amygdala connections means that the amygdala can orchestrate
extremely fast responses to incoming information (all the processing time in the cortex is

Admittedly the information it is getting is quite raw, but the amygdala’s function is to
respond to dangerous and painful situations where the speed of the response might
literally be the difference between life and death.


The above can be illustrated by considering a hiker who suddenly sees what might be a
snake on the trail immediately in front of him.

Not only does the amygdala get the information fast, this brain area has direct
connections to the brain stem so it can bypass the hypothalamus in activating the
autonomic nervous system.

Likewise, the amygdala can directly activate protective or defensive motor responses
without consultation with the rest of the limbic system.

In this case the response would be to freeze (with the hope that the snake won’t see him).

Finally, more detailed (processed) information arrives from the cortex (via the prefrontal
cortex) and one is able to judge if the threat is real or not.

This example was taken form the presentation of Joseph LeDoux in the web symposium
of the IBRO.

In the words of LeDoux: “Your’re better off treating a stick as a snake than a snake as a

So, the amygdala, in being wired to be the fast responses, has protected the individual
from a potentially dangerous situation.

It is the fast reaction system of the brain, getting its input directly from the thalamus and
directly innervating appropriate brain areas to orchestrate fast motor and autonomic


Movie of Freezing: Transgenic mice which have had the dorsal aspect of their olfactory
bulb destroyed lack the hard wiring for an innate fear of the smell of the fox. Note the
freezing in the wild type animal.


Lesion experiment demonstrating involvement of amygdala in fear conditioning.

To demonstrate that the amygdala is really involved in fear conditioning I want to show
an example of a lesion experiment whereby the amygdala, or at least part of it, is knocked
out and, consequently, the animal fails to fear condition.

This type of experiment has been done many times over many years, dating back to the
early 1980s.

The example I want to show is fairly recent (Tazumi and Okaichi, 2002) not only because
it demonstrates a role of the amygdala but also it says something as well about the model
presented here, namely that auditory and visual input to the amygdala may not be going
to the same place as previously illustrated (and reported in most text books).

In this experiment electrodes were brought stereotaxically into the LA of rats and a 60s
current given to kill the neurons in this brain region (Sham LA lesioned animals were

prepared at the same time whereby the electrode was placed in the LA but no current
given; all brains were examined morphologically at the end of the experiment to confirm
the specificity of the lesions).

The animals were then given 4 or 7 days to recover from the operation.

For the experiments each animal was placed in a so-called “shock box” which had an
electrified metal grid floor to administer a US.

This box was placed inside another box that was suspended on 4 soft sponge balls; this
latter box was very sensitive to any movement of the rat inside (i.e. it would shake very
easily); this box was fitted with a displacement sensor that converts movement of the box
into a current, which is measured.

Thus, startle (sudden movement), one of the fear responses, can be measured.

This entire setup was placed within yet another box, this last box being sound-attenuated
and background lit with a 5W incandescent lamp.

This box also has a speaker with built-in amplifier and a 25W incandescent lamp installed
directly above the shock box.

These were for administering an auditory or visual CS respectively.

Tazumi and Okaichi conducted two different experiments, one using the auditory CS and
the other the visual CS.

I’ll consider the auditory experiment first.


The animals were first habituated to the experimental setup for a couple of days.

Then, to obtain the base-line startle response, the animals were given the CS 10 times on
day 1 of the experiment.

There was only a weak startle response, given in mV, in both the lateral amygdala
lesioned animals (LAx) and the sham operated animals (Sh).

On day two and three the animals were given the CS paired with the US (pairing protocol
given in presentation).

They received this CS-US pairing 10 times each day.

On the fourth day they again received the CS alone and the startle response once again

The sham operated animals showed a clear increase in startle response; the LAx animals
failed to show this increase.


The same experiment was conducted with a visual CS (light) substituting for the auditory

Here both the control and LAx animals showed an increased startle response.



Clearly the LA is involved in auditory fear conditioning.

This was not a surprise.

An auditory CS has been the preferred CS over many years of fear conditioning research.

Consequently the mechanisms of auditory fear conditioning are much better understood
than that of some of the other sensory systems used to produce fear conditioning.

For example, the neural connections between the thalamus and LA that carry the auditory
sensory information into the LA have been described in great detail.

Such detail is still lacking for visual input.

The results of the experiments here indicate that the LA is not site of visual input to the

That is not to say that the amygdala has no role in visual fear conditioning.

It has been demonstrated countless times that total amygdala lesions block visual fear

The present experiment indicates that the site of input of visual information is apparently
not the LA itself.

The basolateral amygdala (the area directly adjacent to the lateral amydgala) has been
suggested to be the site of visual input; neuron tracing experiments should ultimately
show exactly where such information enters the amygdala.

a scenario of what might be happening

Auditory input comes into the LA and, apparently, visual input goes into another area of
the amygdala, possibly the BL.

Pain (brining in the US) apparently goes to both areas.

The coupling of the auditory and the pain in the LA causes changes in this area (to be
discussed shortly) such that, after fear conditioning, only the auditory input is required to
generate a defensive output from the central amygdala.

A lesion in the LA eliminates the possibility for auditory fear conditioning.

The BL is intact and thus the coupling of the visual CS with the US can still occur in the
LA lesioned animals. Consequently the animal can display visual fear conditioning.


There are regional differences within the LA with respect to auditory input

So, from the above we must conclude that the pathways of auditory and visual input into
the amygdala are different and that visual does not enter via the LA, as depicted, but
perhaps via other areas of the basolateral complex.

Even within the LA there are regional differences in auditory input.

This can be nicely seen in the study of Quirk et al that I discussed earlier.

Quirk et al placed their electrodes in different places in the lateral amygdala.

In their recording they could determine the latency in the various neurons recorded
(latency = time between the initiation of the tone and the firing of the neuron).

Neurons in the dorsal region of the lateral amygdala had extremely short latencies, which
could only mean that they were receiving the auditory information in an almost direct
way and not via the auditory cortex and intermediate cortical fields.

This observation fit with the fact that the dorsal region of the lateral amygdala has been
shown to be directly innervated from auditory centers of the thalamus.

Ventral regions of the amygdala have much longer latencies, indicating that they are
receiving auditory information via the auditory cortex or, alternatively, that they are
responding to neurons of the dorsal lateral amygdala.

The picture that is developing…..

The picture that is developing is that the amygdala receives sensory input, directly from
the thalamus and more processed input from the cortex.

On the basis of this input the amygdala orchestrates output (autonomic, motor and

On the input side, research has shown that sensory input from either the cortex or the
thalamus can lead to fear conditioning.

Lesion input from the cortex has no effect on fear conditioning.

Likewise, lesion input from the thalamus has no effect.

Lesion inputs from both the cortex and the thalamus, and fear conditioning fails to occur.

Presumably thalamic input leads to fast responses which the more processed input from
the cortex can modify, if deemed appropriate.

In the absence of thalamic input the input from the cortex is capable of driving the
complete set of responses in the amygdala.


On the response side of amygdala function, electrodes have been placed in the central
amygdala to activate neuronal networks.

It was noticed in these types of experiments that the responses evoked were complete fear
conditioned responses (i.e. well integrated automomic, motor and neuroendocrine
programs were activated).

This indicates that the down-stream responses programs are “hard wired”, just waiting for
activation by the amygdala.


The current thinking, based on the results of more than 20 years of experimentation, is
that in the amygdala the CS and the US act on the same set of neurons.

Before fear conditioning (pairing with the US) the CS generates a EPSP which is smaller
than that generated by the same input after conditioning.

Thus the output of the neuron, frequency of action potentials, is lower than that
displayed after conditioning.

The output after conditioning is sufficient to generate the fear response.

In this scenario the synapse carrying the CS has been strengthened during the
conditioning and remains strengthened after the conditioning.

Thus, following conditioning the same CS that lead to a small EPSP before conditioning
now leads to a much stronger EPSP (i.e the synapse can remember the conditioning).

The evidence for this scenario comes from a multitude of electrophysiological

More recently, the question as to how CS synapses become strengthened (i.e. the
molecular and cellular mechanisms of synaptic strengthening) during fear conditioning
are being addressed.

It is this question that I will now consider.

How does synaptic strength become strengthened?

The fact that the combination CS and US strengthens the synapses makes one
immediately consider involvement of a coincidence detector.

A coincidence detector is a component of a signaling system that can detect coincidence
between two events and change signaling properties of the system as a consequence of
the detection of the coincidence.

The best characterized coincidence detector of the vertebrate brain is the NMDA

The NMDA receptor is the critical signaling component in learning and memory
mechanism in the hippocampus and elsewhere in the CNS.

This receptor is a ligand-operated ion channel.

The ligand for this receptor is the neurotransmitter glutamate, the most abundant and
most important stimulatory neurotransmitter of the CNS.

The presence of glutamate, however, is not sufficient for opening of the ion channel of
this receptor.

This is because there is, within the channel, a Mg2+ ion that blocks the channel.

This Mg2+ block is lifted only if the membrane undergoes a depolarization.

So, in acting as a coincidence detector, two events must occur for activation of the
NMDA receptor.

There must be a release of glutamate and, at the same time, the membrane must be

Only then does the NMDA receptor channel open.

The receptor is a Ca2+ channel and thus, as a consequence of coincidence detection, Ca2+
flows into the cell.

With respect to the NMDA receptor working in the amygdala, the CS delivers glutamate
in the CS synapse and the paired US causes a general depolarization of the neurons via a
so-called back propagating action potential (BPA).

The combination of the BPA and the presence of glutamate in the CS synapse leads to a
so-called supralinear Ca2+ signal in the CS synapse.

As a consequence of the strong Ca2+ signal in the postsynaptic CS synapse the Ca2+
initiates an intracellular signaling cascade that ultimately leads to synaptic strengthening.

In fact voltage-operated dCa2+ channels are oalso involved in creating the Ca2+ signal
which, in the case of coincidence, contributes to the generation of a so-called supra linear
Ca2+ signal. This topic was covered extensively in the earlier Neurobiology course. For
those would did not follow this course, link to the topic Supralinear Ca2+ signaling is

Explanation of link
For activation of the NMDA receptor two criteria must be met, namely (2) membrane depolarization, to
relieve the Mg2+block of the channel, and (2) the presence of the ligand, glutamate.
When these two criteria have been met, a process known as coincidence detection, then the channel opens
and Ca2+ flows into the neuron.
The idea is that this Ca2+ signal goes on to induce LTP, an important mechanism in leaning and memory.
To induce LTP the Ca2+ signal generated during coincidence has a very special property, it is said to be
The easiest way to explain supralinear Ca2+ signaling is to show an experiment that demonstrates it.

Demonstration of coincidence detection leading to supralineral Ca2+ signaling

The tissue preparation
In this demo the tissue preparation is the same as that described earlier, the hippocampal brain slice
containing CA1 pyramidal neurons, innervated by the Shaffer collaterals.
The equipment
Again, a stimulating electrode is placed in the Schaffer collaterals and a cell attached patch electrode
(recording electrode) is placed on one of the pyramidal neurons (the stimulating electrode is moved around
until EPSPs are picked up by the recording electrode).
A multiphoton laserscaning microscope (MPLSM) is also included in this experiment to measure the Ca 2+
signal in the postsynaptic spine (The Ca2+ probe is introduced via the patch electrode).
To find an appropriate spine innervated by the Schaffer collateral (for the Ca2+ recordings in this
experiment) the preparation is first viewed with the MPLSM at low power.
When the stimulating electrode fires the Schaffer collateral weak Ca 2+ signals will be seen in spines
innervated by the collateral activated.

One of these spines is then selected for high power line scanning which gives high temporal resolution on
Ca2+ signaling events in the spine (1-2 ms resolution).
Thus, with activation of the Schaffer collateral we have a Ca 2+ recording in one of the spines and an EPSP
recording from the cell body of the neuron.

Events in the synapse during an EPSP
The synapse possesses NMDA receptors, AMPA receptors and voltage-operated Ca2+ channels (VOCCs).
Release of glutamate in the synapse activates some of the AMPA receptors.
The influx of sodium causes a depolarization, see by the recording electrode as the EPSP.
Glutamate also binds to the NMDA receptors but these, for the most part, remain closed due to the Mg 2+
The depolarization caused by the AMPA receptor is not strong enough to cause global NMDA receptor
activation but some NMDA receptors do open, probably those near an AMPA receptor (the strength of
EPSP dissipate with distance).
This gives rise to a weak Ca2+ (Note that only the synapses that generate an EPSP generate a Ca2+ signal;
this was how the appropriate synapse for Ca2+ imaging was located).
The idea that AMPA receptors lead to opening of NMDA receptors which subsequently gives rise to the
Ca2+ signal is supported by data showing that treatment with either AMPA or NMDA receptor blocker
block the Ca2+ signal (note that the EPSP, as recorded by the patch electrode, is blocked by the AMPA but
not the NMDA receptor blocker).

Generating a Ca2+ signal with an action potential

A second way of generating a Ca2+ signal in the synapse is as a consequence of the neuron firing off an
action potential.
This can be demonstrated in the preparation by using the recording electrode to generate a strong
depolarization (i.e. use the recording electrode briefly as a stimulating electrode).
The depolarization causes an action potential to propagate along the membrane of the axon towards the
This same depolarization causes and action potential to go in the opposite direction, through the dendrites
and into the spines.
This is called a back-propagating action potential (bAP).
Not all neurons possess the ability to generate bAPs; this ability is strongly associated with neurons that
display synaptic plasticity.

In all synapses the strong depolarization of the bAP lifts the Mg2+ block on the NMDA receptors.
No NMDA receptors become activated, however, because there is no glutamate.
The bAP does activate VOCCs in all the synapses, thus leading to a weak Ca 2+ signal in each synapse.
This can be measured by the MPSLM in the imaged synapse.
The Ca2+ signal can be blocked by VOCC blockers, thus showing their origin though opening of VOCCs.

Generating a Ca2+ signal with an EPSP plus a bAP
In this case an EPSP is fired (through activation of the schaffer collateral) and shortly thereafter (within a
few ms) a bAP is fired off (through use of the recording electrode as a stimulating electrode).
Initially everything looks the same as previously outlined (EPSP leads to opening of AMPA receptors and a
few NMDA receptors, thus leading to weak Ca2+ signal in those synapse innervated by the Schaffer
collateral; bAP activate VOCCs leading to a weak Ca2+ signal in all synapses).
The bAP lifts the Mg2+ block in the NMDA receptors; this has no consequences in most synapses but, in
the synapses activated by the EPSP, this leads to a massive activation of NMDA receptors.
In these latter synapses there is both a strong depolarization (via the bAP) and there is glutamate present
(from activation through the Schaffer collateral).
Thus these synapses show a strong Ca2+ response.

Time for some mathematics
In this demo induction of an EPSP via the Schaffer collaterals lead to 2 Ca 2+ channels being open.
Induction of a bAP lead to 2 Ca2+ channels being open.

One might therefore have expected that the consequences of a combination of EPSP plus bAP would be 4
channels open.
In the demo there is in fact 7 Ca2+ channels open.
Thus the combination is said to be supralinear.
This can be seen the Ca2+ signal recorded by the MPLSM line scan.

Consequences of supralinearity
Synapses that display plasticity possess Ca2+ calmodulin dependent protein kinase II (CaMKII).
This kinase exists as a complex of 12 kinase molecules, (i.e. it is a dodecameric holoenzyme) arranged in
two back-to-back rings (one ring illustrated).
Kinases within the complex can transmolecularly phosphorylate other kinases within the complex, leaving
them in a Ca2+ calmodulin-independent active conformation.
This property means that the response of this complex to different concentrations of Ca2+ is very different.
In the weaker Ca2+ signals generated by the EPSP or the bAP alone the enzyme remains absolutely
dependent on the presence of Ca2+ calmodulin in order to display kinase activity.
In the supralinear Ca2+ signal the enzyme becomes “supra” activated (ultimately independent of Ca2+
It is this supralinear Ca2+ signal, acting via CaMKII, which ultimately leads to synaptic strengthening in the
process of synaptic plasticity.

Some real examples
The preceding story was constructed from many reports in the literature.
To conclude this section I want to look at a couple of studies and examine the data that supports the above
The first study is that of Koster and Sakmann, 1998.
Confocal laser scanning microscopy (CLSM), used to image Ca2+, nicely illuminates the dentrites and
Note that when bAPs (denoted as AP) are induced all spines and dendrites give and enhanced Ca 2+ signal.
When an EPSP is used to activate then only one spine gives a signal (thus only this spine is in synaptic
contact with the Schaffer collateral that was activated).

Next, the Ca2+ signal generated by (1) an EPSP, (2) a bAP, (3) a bAP followed by a EPSP and finally, (4) a
EPSP followed by a bAP, are illustrated.
Note the supralinear signaling when the EPSP-bAP protocol is used. (results of 8 independent experiments
When an AP is given first, followed by an EPSP then a sublinear Ca 2+ signal is generated.
Sublinear signals are thought to be important for LTD.
The exact mechanism generating sublinear signals is not well understood.
It is thought that the bAP, besides lifting the Mg2+ block of the NMDA receptor, sets in motion intracellular
events that inhibit the NMDA receptor.
The idea is that the Ca2+ signal generated by the VOCCs activates processes that ultimately render the
NMDA receptor inoperative.
When the Schaffer collateral is activated some ms later (thus glutamate in the synapse) the NMDA receptor
is not in a state to respond and thus a weaker EPSP is generated.


The integrative mechanisms involving NMDA receptors, AMPA receptors and VOCCs, just described for
CA1 pyramidal neurons, are found in various brain regions (in general neurons displaying these
mechanisms are neurons which display strong plasticity).
As a second example I have selected a study concerning so-called spiny neurons in the somatosensory
In the first image the Ca2+ filled recording electrode can be seen at the left; spines are clearly visible at
higher magnifications.
In the second image a stimulating electrode to induce an AP in the spiny neuron can be seen.

Because a second stimulating electrode, rather than the recording electrode, was used in this experiment to
induce the AP the recording electrode could be used to record the AP (something that could not be done in
the earlier experiment).

The first results shown give the EPSP generated by activating fibers that innervate the spiny neuron,
together with the Ca2+ signal generated in the innervated synapse.
This enhanced Ca2+ signal could be blocked by both AMPA receptor and NMDA receptor blockers.
Next, the AP generated by the stimulating electrode near the spiny neuron is shown, together with the Ca 2+
signal generated as a consequence of this AP.
This signal could be blocked by VOCC blockers.
Finally, the results are shown when an EPSP is induced followed shortly thereafter by a AP.
Clearly the Ca2+ signal generated is supralinear.
Results from a number of experiments show that if the AP follows the EPSP by anywhere up to 50ms then
a supralinear Ca2+ signal is generated.
If the AP precedes the EPSP then a sublinear Ca2+ signal occurs.

What’s the significance?

In considering the significance of coincidence detection and supra- and sublinear signaling I want to
consider once again the CA1 pyramidal neuron and Schaffer collaterals.

Consider first the situation where action potentials come in on two independent Schaffer collaterals, greater
than 50 ms apart.
Each collateral causes a slight Ca2+ increase in their respective postsynaptic target spine.
The first EPSP has dissipated before the second EPSP enters the cell body and hillock (where addition of
EPSP takes place)
Under these circumstances nothing spectacular happens.

Now consider the situation where the same action potentials come in less than 50 ms apart.
The EPSP are additive and reach threshold.
An AP is fired and a bAP also propagates out of the cell body into the dendritic tree.
In most of the spines the bAP will generate a weak Ca2+ signal (through opening of VOCCs).
In the synapses responsible for generating the EPSPs however a supralinear Ca 2+ signal is generated via the
NMDA receptor (there is both the presence of glutamate and a depolarization, the two criteria required for
activation of this receptor).
This coincidence detection by the NMDA receptor ultimately leads to LTP (and synaptic strengthening).

Now consider the case of a third collateral in which an AP comes along about 50 ms after the first two in
the above example.
This generates an EPSP in the synapse that comes after an AP.
Consequently there will be LTD in this synapse.


So, we have strengthened and weakened synapses.
Remember the idea of how the brain might work, as illustrated by artificial neural networks.
Through the BPA the ANNs can store information i.e they can learn and recall)
The idea is that LTP (and LTD) is used in neural networks of the brain (such as in the hippocampus) for the
construction of memory.
So the significance of supra- and sublinear Ca2+ signaling is that they may very well reflect the first
step in a signaling cascade that leads to learning and memory

What is the evidence that the NMDA receptor is involved?

Over the years considerable evidence has accumulated for involvement of the NMDA
receptor in fear conditioning in the amygdala.

This evidence has come from both in vivo experiments where the behavior of whole
animals has been observed, as well as in vitro where electrophysiological studies have
been conducted on amygdala neurons in brain slices.

The work I want to consider here is fairly recent studies by Brauer et al where both in
vivo and in vitro evidence was presented for NMDA receptor involvement in fear

The in vivo evidence

The in vivo evidence comes from experiments whereby the lateral nucleus of the
amygdala was bilaterally cannulated.

These cannulas could be coupled to a perfusion system and LA infused with either
vehicle or receptor agonists/ antagonists such as the NMDA receptor antagonist AVP.

The cannulas were implanted at least 5 days before the experiment so the animals had
plenty of time to recover from the operation.

The actual experiment began with an habituation period whereby the animals were
habituated to the chamber in which the CS-US pairing and CS behavioral responses were
to be conducted.

As well, the animals were habituated to having the cannulas attached and unattached to
the infusion system.

The following day both the left and right amygdala were infused with AVP; the animals
were put into the test chamber and 10 to 15 minutes later received, in quick succession,
three CS-US pairs.

One hour later they were tested for their response to the CS (tone testing); tone testing
was also conducted 3h 8h and 24 h after the CS-US pairing.

The parameter measured was freezing (commercial equipment, containing electrified grid
and sound equipment and in which digital video can be analyzed by a computer to
determine freezing, is illustrated; alternatively, freezing can be quantified by trained

The total seconds of freezing during each CS was determined.

The results clearly shown that AVP treatment lead to significantly lower levels of
freezing compared to vehicle only treatment.

The conclusion: behavioral responses are dependent on the NMDA receptor.

The in vitro evidence:

For the in vitro evidence freshly prepared brain slices (400 M thick), containing the
amygdala, were brought into a holding chamber under an upright microscope.

Whole cell patch-recordings were made on a total of 71 LA neurons.

In some cases AVP was introduced to the bath solution surrounding the neuron.

A stimuli to the neuron was delivered each minute through bipolar stainless steel
electrode that activate fibers coming from the auditory thalamus.

This electrode was also used to deliver a high-frequency tetanus to the neuron (the
tetanus is known to induce a general depolarization of the neuron).

Following the tetanus the bipolar electrode was again used to deliver a stimulus each
minute to the LA neuron.

The strength of the EPSP (expressed as EPSP slope) was calculated for each stimulus.

The results… in the control group the EPSP were clearly increased after the tetanus.

In the AVP treated group there is little or no increase in the strength of the EPSP.

The increase strength is evident for many hours (or in fact, days), a clear example of the
phenomenon called long-term potentiation or LTP.


So, from the in vivo experiment we conclude that the NMDA receptor is required for fear

From the in vitro experiment we conclude that the NMDA receptor is required for LTP.

Putting the two observations together and it would seem that NMDA receptor-dependent
LTP in the lateral amygdala reflects the construction of fear conditioning.

Or, to put it another way, the fear is stored in the form of the strength of synapses in the

This is a hypothesis.

It is the same hypothesis, based on the same circumstantial evidence, that has been made
for LTP representing learning and memory in the hippocampus and LTP being
responsible for drug addiction in the reward center (lateral tegmental area) of the brain.

In the hippocampus this hypothesis has been further backed up experiments where by the
NMDA receptor has been knocked out in a tissue-specific and time specific way
(consequently, no LTP and no memory) and in experiments where NMDA receptors have
been over-expressed in the hippocampus (stronger LTP and better memory).
Thus the role of the NMDA receptor and LTP in memory processes in the hippocampus
are on a firm experimental basis.

While, this type of evidence is still to be collected for the role of NMDA-dependent LTP
in fear conditioning, the circumstantial evidence is already very strong.


The picture that is developing….

The coincidence of CS and US leads to activation of the NMDA receptor and
consequently there is an influx of Ca2+ into the cell.

Now what?

For the “now what” lets first take a look, via the link provided, at what is known to
happen in the hippocampus during LTP (review from the Neurobiology course).

Summary of link

Ca2+ influx via the NMDA receptor leads to the activation of CaMKII which phosphorylates various
proteins; included among these proteins is adenylyl cyclase; phosphorylation of AC leads to its activation
and the cAMP generated leads to activation of PKA, which in turn phosphorylates proteins, often the same
proteins that are being phosphorylated by CaMKII; among the known targets for the kinases is the AMPA
receptor, phosphorylation of which increases its sensitivity to glutamate, thus strengthening the synapse;
another target is CREB which is activated via the phosphorylations; Activated CREB leads to the
production of new proteins; phosphorylations are responsible of early LTP and gene expression (new
proteins) is required for late LTP; the phosphorylations are also known to initiate a cascade of intracellular
events which leads to a recruitment of AMPA receptors to the membrane .

How is it in the amygdala?

While events down stream from the Ca2+ influx have not been as well worked out for the
amygdala as the hippocampus, it would seem likely that similar, if not identical events
are occurring.

The neurons of the amygdala display both early and late LTP similar to what has been
found in the construction of memory in the hippocampus.

 Protein synthesis is required for late LTP, as revealed in electrophysiological
experiments in the presence of protein synthesis inhibitors.

Thus one talks about short-term memory and long-term memory.

Likewise, a generation of cyclic AMP seems to be one of the down stream events, as
PKA inhibitors block LTP.

Notice that in this case the short-term memory is largely eliminated by this treatment.
This is in keeping with the idea that the short-term memory is dependent on
phosphorylations (which will still occur in the protein synthesis block experiment shown
above, but will be largely eliminated in the PKA inhibitor experiment)

In the hippocampus there is a sensitization of AMPA receptors and a recruitment of
AMPA receptors during LTP.

How is it in the amygdala?

As outlined in the experiment below, there is evidence that similar events may be
occurring here too.


Animals were fear conditioned (CS, sound; US, footshock) 10 times per day for 2 days;
unpaired control animals were also prepared.

Just before sacrifice on the third day the animals were tested for startle reflex; the fear
conditioned animals showed a clear startle response.

Brain slices were prepared and stimulatory electrodes were placed in the tissue slices in
the fiber tract known to carrying auditory input from thalamus to the LA (remember that
these fiber tracts have been well described); neuron cell bodies in the LA were patched to
made recordings.

The electrophysiological data showed stronger EPSPs in the neurons of fear conditioned
animals compared to those of unpaired and naive animals.

Likewise contribution of AMPA receptors to the EPSP was greater in such animals
(NMDA receptors were blocked in order to determine AMPA receptor contribution).

So clearly AMPA receptor signaling is increased with fear conditioning.

What is causing this strengthening in AMPA receptor signaling?

In the rat hippocampus it has been found that AMPA receptors become phosphorylated
during a tetanus and that phosphorylation increases the opening probability of the
receptor (thus leading to stronger AMPA generated depolarizations).

This is likely the case in the amygdala as well because both systems (amygdala and
hippocampus) possess the same signaling hardware and display the same intracellular
signaling cascades.

In recent years it has been shown that an important component of synapse strengthening
in the hippocampus is recruitment of AMPA receptors to the synapse.

Recent studies have demonstrated that this is also the case the amygdala.

Receptor recruitment in the amygdala during fear conditioning

First a review of the situation in the hippocampus

The AMPA receptor is constructed from subunits. There are four subunits, the products
of four different genes, which have been designated Glu1, Glu2, Glu3 and Glu4 (Glu =
glutamate, the endogenous ligand for the receptor).

In the adult mammalian brain the most common AMPA receptors are composed of
mixtures of Glu1 with Glu2 (designated Glu1/Glu2) or Glu2 with Glu3 (designated
Glu2/Glu3). The Glu4 subunit is particularly highly expressed during development.

It has been demonstrated in the rat and mouse hippocampus that the expression of
Glu1/Glu2 and Glu2/Glu3 AMPA receptors on the membrane of CA1 pyramidal neurons
occurs under very different circumstances.

In vitro it has been demonstrated that Glu1/Glu2 type AMPA receptors are recruited to
the membrane during experimental protocols that induce long-term potentiation (LTP) in
the pyramidal neuron; Glu2/Glu3 type receptors are plugged into the membrane in a
constitutive fashion to replace old, damaged receptors.

The critical structural difference between the receptors that determines induced versus
constitutive delivery lies in the C-terminal of the Glu1 and Glu2 subunits. Glu1 has a
long intracellular C-terminal tail, the very tip of which can bind PDZ group 1 proteins;
Glu2 has a short C-terminal tail capable of binding with PDZ group 2 proteins.
Recruitment studies have shown that Glu1 is apparently dominant over Glu2 such that
Glu1/Glu2 AMPA receptors are induced to the membrane (during LTP) whereas
Glu2/Glu3 AMPA receptors are constitutive. Altering the C-terminal of Glu1 such that it
can no long bind to PDZ group 1 proteins blocks its induced recruitment to the membrane
(a Flash move is provides showing the two modes of AMPA receptor recruitment).


Experimental evidence for the above scenario has come from experiments in which genes
coding for Glu1 and Glu2 subunits have been altered and, following either in vitro or in
vivo infection of CA1 pyramidal neurons with the gene constructs, the behavior of the
receptors have been followed.

Two alterations have been important in these experiments (illustrated on the Glu2

First, green fluorescent protein has been attached to the N-terminal of the receptor
subunit so that cells that have been successfully infected with the gene construct can be
easily identified (simply view the tissue preparation under UV light to identify
fluorescent neurons).

Secondly, within the P-region (pore region) of the subunit a site directed mutation has
been introduced such that a receptor containing this subunit will display different
electrophysiological properties than a normal receptor. This is thus termed an
electrophysiological tag. In the experiments with the hippocampus the mutations caused
the receptor to display a lower inward rectification. Inward rectification is an
electrophysiological property of neurons which can be easily determined by constructing
a so-called I-V plot form electrophysiological data collected from the recorded neuron.

The electrophysiological tag in the receptor recruitment experiments of the CA1
pyramidal neurons was an important component of the methodology because it showed
when receptors derived form the gene constructs have been recruited to the membrane
(i.e. if there was a decrease in inward rectification then receptors derived from the gene
construct have been plugged into the membrane).

Now, a look at AMPA receptor recruitment in the amygdala

The cellular and molecular mechanisms of learning in the amygdala are looking
remarkably similar to the model developed earlier in the hippocampus (i.e NMDA
receptor initiating events leading to stronger AMPA-dependent synapse).

Thus, a logical question is does fear conditioning involve recruitment of AMPA receptors
to the membrane.

This question has recently been examined by Rumpel et al using methods similar to those
used in the earlier hippocampal studies described above.

The hypothesis was that Glu1 containing AMPA receptors would be recruited to the
membrane of pyramidal neurons of the lateral amygdala during fear conditioning, and
that this recruitment would be a crucial component memory construction of that fear.

Gene constructs for the analysis of fear conditioning

Rumpel et al made two gene constructs to test the above hypotheses.

The first construct was the Glu1 receptor subunit in which GFP had been attached to the
N-terminal (so cells successfully infected and expressing the gene construct could be
easily identified) and an electrophysiological tag had been introduced in the P-region (in
this case the mutation was such that receptors containing expressed subunit would display
increased inward rectification).

They termed this construct their “plasticity tag” because expression of this subunit on the
membrane of a neuron would indicate that the neuron had undergone a plastic event (if
their hypothesis was correct).

The second gene construct was very simple, namely GFP coupled to the C-terminal of the
Glu1 receptor.

The purpose of the GFP was to identify neurons that had been successfully infected. The
purpose of the C-terminal was to bind to PDZ group 1 proteins, thus blocking the
possibility of Glu1 containing receptors to bind to these proteins. They thus termed this
their “plasticity blocker” because, if their hypothesis was correct, this construct would
block LTP and thus would block fear conditioning.

I will first consider experiments conducted with the plasticity blocker.

Experiments with the plasticity blocker

The gene construct was introduced into the living animal using a so-called amplicon
vector of the herpes simplex virus.

The infectious viral particles were introduced into the lateral amygdala using standard
stereotactic methods.

The amplicon possessed an appropriate promoter such that the gene construct containing
the plasticity blocker would be immediately expressed in infected cells.

One day following the injection the animal was sacrificed and thick brain tissue slices
prepared (500 M). One day is sufficient time for infection of the cells and expression of
the protein carried by the viral vector.

Shown is the tissue under a microscope using normal transmission and epifluorescence
(UV light source). The lateral amygdala is indicated.

Under the epifluorescence the lateral amygdala displays fluorescence, indicating that
cells have become infected and are expressing the gene construct for the plasticity

At higher magnification cell bodies of pyramidal neurons can be seen; Some of these are
infected and other are not.

Under the microscope it is very easy to patch infected and non-infected neurons and
examine their electrophysiological properties.

Depolarizations of both infected and non-infected neurons generate action potentials; the
electropysiological properties of which are essentially identical; this establishes that
infection of the neurons has had no adverse effects on their function.


Having established that the infection has no adverse effects on the infected neurons
Rumpel et al were ready to establish if the plasticity blocker would block LTP (it should
block LTP if their hypothesis is correct)

To do so they patch-clamped infected and non-infected pyramidal neurons in the lateral
amygdala of their tissue slice and placed a stimulating electrode in the ventral stiratum
where the thalamo-amygdala fibers are found (i.e. the fibers bring thalamic input to the
lateral amygdala pyramidal neurons).
[note: the patch electrodes are measuring the strength of the EPSPs generated as a
consequence of action potentials generated in the innervating fibers].

The LTP profiles of the infected and non-infected neurons were different.

The infected neurons:
The infected cells displayed LTP but synaptic strength returned to basal level within 40
min. This is equivalent to early LTP as seen in CA1 hippocampal pyramidal neurons. It is
a consequence of phosphorylations occurring in the postsynaptic neuron as a
consequence of activation of kinases (which in turn is a consequence of the activation of
the NMDA receptor). One of the likely targets of the kinases is AMPA receptors;
phosphorylation increases the opening probability of the receptors and thus the
magnitude of the EPSP is increased. Phosphorylations of proteins are transitory with
endogenous phosphatases removing the phosphorylations within 40 to 90 minutes. With
removal of the phosporylations the size of the EPSPs of the infected neurons falls back to
basal levels. Any participation of new Glu1 containing AMPA receptors in the synaptic
strengthening would be blocked by the presence of plasticity blocker tying up all the PDZ
group 1 domains required for receptor recruitment.

The non-infected neurons:
The non-infected neurons also displayed an early and rapid LTP, most likely due to
phosphorylation of existing AMPA receptors as outlined above. The non-infected
neurons also displayed late LTP; The conclusion from this experiment would be that at
least part of this late LTP is due to new Glu1 containing AMPA receptors being plugged
into the membrane to participate in creating EPSPs, thus the non-infected neurons display
late LTP.


Having established that the plasticity blocker can attenuate LTP, Rumpel et al were now
ready to go after the question “Will infection of the plasticity blocker attenuate fear

Fourteen to twenty hours after bilateral infection of the lateral amygdala with either the
plasticity blocker gene construct or a construct containing GFP alone (to act as a control),
animals were submitted to a single footshock-tone paring. They were then tested for
freezing 3h and 24h later.

The results were clear, significantly less freezing at 3h and at 24h of the plasticity blocker
injected group versus the GFP injected group..

Clearly treatment with an agent which blocks AMPA receptor recruitment to the synapse
attenuates fear conditioning (i.e. the construction of the memory of the pairing).

These results strongly support the hypothesis that memory processes in the amygdala,
like in the hippocampus, involve recruitment of AMPA receptors to the synapse.

Experiments with plasticity tag

With the plasticity tag Rumpel et al were able to determine if fear conditioning was
associated with increased rectification in infected cells (as predicted if the hypothesis that
AMPA receptors are recruited to the synapse is correct).

Infected animals were submitted to shock-tone pairing; control infected animals were
submitted to the same number of foot shocks and tones but unpaired. Pairing clearly lead
to a robust freezing response (response of animals to the first minute in the testing cage,
designated silence, is also shown).

The crucial question is: do the plasticity tag infected neurons, on average, display
increased rectification compared to the average rectification of non-infected neurons.

To answer this question brain slices were prepared after behavioral testing. I/V plots were
prepared for a large number of tagged and non-tagged cells and the degree of rectification

The results: Cells expressing the plasticity tag had a significantly higher degree of inward

The conclusion: Glu1 containing receptors are recruited to the membrane during fear

Apparently memories are made in the amygdala the same way they are made in the

In summary….

The issue of molecular and cellular mechanisms underlying emotional learning and fear
has recently been reviewed.

Presented here is the summary figure from that review. The model presented looks
remarkably similar to the models developed in the hippocampus for construction of

The dynamic duo of the NMDA receptor and the AMPA receptor take center stage.

So to does the voltage-operated Ca2+ channel (VOCC); Back propagating action
potentials created by the US are responsible for activating these channels, together with
the NMDA receptor in a CS-US pairing. Simultaneous activation of both types of Ca2+
channels lead to “supralinear” Ca2+ signaling, and thus to LTP.

There is evidence that metabotropic glutamate receptors also participate in generating
LTP; which is again is similar to events in the hippocampus.

Down stream of the ion channels and receptors we have kinases, such as CaMKII and

These are responsible for phosphorylations which generate the short-term memory

They also initiate events which lead to gene expression and long-term memory (LTM).

Fear extinction

We’ve seen how a fear memory is constructed.

I now want to turn attention to how you forget a fear memory, in other words fear

For the subject of fear extinction I want to introduce Gregory Quirk. He is one of the
world authorities on fear extinction and is one of the participants of the web symposium
on the Neurobiology of stress, fear and anxiety. The title of his presentation is “learning
not to fear”.

Much of what I have to tell on fear extinction is taken from his presentation in this


First of all, let’s look at exactly what fear extinction is…

Shown here is a slide from Quirk’s presentation showing extinction.

The association of the tone and the footshock is gradually lost when they become

One might rather naively think that the animal is forgetting the association during
extinction…in other words he is loosing his fear memory.

Quirk makes the case that in fact extinction is not forgetting, but is in fact a new learning
process whereby you learn that the US and the CS are no longer associated, and thus the
CS is no longer to be feared.

So the idea is that the fear memory stays but is suppressed by a new memory (a memory
that says: the CS is no longer to be feared).


The idea that extinction is a form of suppression of a memory was already seen by

This is again from the presentation of Quirk.

Here we see volume of salvia secretion during extinction (when the bell is no longer
associated with the sight or smell of meat).

Pavlov makes the case that the association of CS and US is not lost during extinction
because it spontaneously recoveries after a break of two hours (so it was there, being
suppressed, during the CS immediately before the 2h break).

In other words, extinction is new learning that inhibits the conditioned response.

[Note: As far as I can see, from the data presented an alternative explanation could be
that the dogs run out of saliva during extinction; following a 2h break there is sufficient
build up of saliva in the salivary gland to once again respond; Quirk does not mention
this possibility but, as you will see in a moment, he does present convincing data that the
basic concept, first outlined by Pavlov, is correct].

[What could have been added to Pavlov’s the protocol to make the data more


Where is extinction learning being stored?

We have already seen earlier in our discussions of the amygdala that the prefrontal cortex
can exert control over programs executed by the amygdala.

It had already been shown in animal experiments in the early 1990s that while lesions of
the prefrontal cortex has no effect on acquisition of fear conditioning, it blocked

More recent data, from Quirk’s group, show that it blocks extinction in a very special

In these experiments animals were electro-lesioned 1 week prior to fear conditioning and
subsequent extinction training.

Lesions in the ventral medial PFC has no effect on the acquisition of extinction, but, as
seen on the day after the extinction trials, it block the consolidation of extinction.

In other words, the vmPFC is not the site of learning the extinction (that is a function of
the amygdala), but it is the site of the storage of that memory.

When comparing the percent freezing recovery on day 2 there is essentially no difference
between the vmPFC lesioned animals and control animal that have not undergone
extinction training.

Is the NMDA receptor involved?

We have seen how the NMDA receptor is involved in the construction of a fear memory.

The question arises, is this same receptor type involved in the construction of an
extinction memory?

This question was also addressed by Quirk’s group using a systemically injected NMDA
receptor antagonist (CPP can cross the BBB and thus act on central receptors).

The results: little or no effect on extinction learning (if anything, treatment with the
antagonist seems to improve it) but a dramatic effect on extinction memory!

Again, when one compares the percent rebound of freezing on day 2 between an animal
that has not undergone extinction training with the CPP injected animals, there is
essentially no difference.

So, blocking the NMDA receptor has completely blocked the ability of the animal to
construct a memory of the extinction.

To summarize thus far:

Extinction involves both NMDA-independent and NMDA-dependent processes.

One of the NMDA independent processing is clearly learning extinction (indeed, CPP
treated animals seem to learn faster!).

It is clear from the data that CPP-treated animals can form short-term memory of the
extinction training.

For the NMDA-dependent processes we have long-term memory of the extinction

The learning and the STM of extinction training is believed to be a function of the
amygdala alone, although the mechanisms involved are largely unknown (there is
evidence that the endocanibinoids might be involved in STM process).

As you have just seen, the site of LTM of extinction would appear to be the vmPFC.


Next question : Can we see neuronal activity in the vmPFC associated with the extinction

To study this Quirk and coworkers implanted 8 channel electrodes into the vmPFC, or
more specifically, the infralimbic area.

At the time of implantation they would adjust the position of the electrodes such that at
least 3 of the 8 wire electrodes would record action potentials of neurons.

They recorded a total of 31 infralimbic neurons in this manner.

Following implantation of the electrodes the animals were given one week rest and then
submitted to fear conditioning while recording firing activity of neurons in the
infralimbic area.


Shown is the firing activity of a “typical” neuron during conditioning, extinction and day
2 of extinction.

Clearly the extinction training of day 1 was remembered on day 2 and this was associated
with firing of the infralimbic neuron. This same neuron showed no activation during
conditioning or extinction training of day 1.

This already gives the idea that perhaps the activity shown by this neuron has something
to do with the existence of extinction memory on day 2.

The firing of this (and other) neurons was within 100 ms of the tone.


You will notice in the data that some animals remembered the extinction training better
than other animals.

The next question Quirk and coworkers tackled was “Is there any relationship between
the level of activation of infralimbic neurons and the level of memory displayed?”

If the infralimbic neurons are indeed involved in this memory then one would expect that
higher levels of activity of these neurons in animals displaying the lowest degree of
freezing on day 2.

Quirk and coworkers processed their data to look at this possibility.


There was indeed a correlation.

Animals displaying, on average, high responses in their infralimbic neurons to tone on
day 2 displayed lower recovery of freezing (i.e. they could remember they had nothing to
fear from the tone).

Animals displaying, on average, low tone responses in their infralimbic neurons
displayed high recorvery of freezing (i.e. they couldn’t remember that the tone no longer
signified that they were about to receive a foot shock)

For further analysis of this data they divided their animals into two groups, those
recovering less than 50% of their day 1 freezing response on day 2 (low recovery group)
and those recovering more than 50% of this response (high recovery group).

Analysis of the two groups separately, in bar graphs, clearly revealed that the animals that
displayed low recovery of freezing displayed high tone responses in their infralimbic
neurons on day 2.

Thus, the infralimbic neurons seem to signal extinction memory.

The same data set also shows what the infralimbic neurons do not signal.

They do not signal acquisition (i.e. they do not fire during early habituation; trial blocks


they do not signal in extinction training (i.e. they do not fire in late extinction; trial blocks

So, they are therefore not taking part in the extinction training per se, but apparently are
taking part in storing the memory of the extinction training.


When the experiments were completed high voltage was sent through the electrodes to
create lesions which could be located in postmortem brain sections to locate the exact
position of the electrodes.

So far we have been examining results from animals where the electrodes were in the
infralimbic region of the vmPFC.

In some animals the electrodes were in the prelimbic area and in others in the medial
orbital area.

Only neurons of the infralimbic region showed tone responses, thus showing that there is
a high degree of spatial specificity in which neurons are involved in extinction memory.


Next question:

If infralimbic activity signals extinction memory, can infralimbic stimulation “simulate”
extinction memory?

In other words, can we go into the brain of the intact animal with electrodes and activate
infralimbic neurons to induce extinction?

To test this, animals were equipped with bipolar stimulating electrodes unilaterally into
the right infralimbic area (neurons from this area were known to project to the LA).

The stimulation protocols used mimicked the activation recorded in the previous
experiments, with 100 to 400 ms latency between the tone and the start of the activation.

The protocol was Day 1 conditioning, Day 2, extinction training with or without IL
stimulation and Day 3, testing for extinction memory.

The critical questions are, first, does IL stimulation on day 2 induce lower freezing (thus
indicating that these neurons are indeed involved in suppression of the fear response) and
do IL stimulated animals remember this on day 3.

The results show at the very first stimulus-tone paring during extinction training that the
stimulation is very effective in reducing freezing behavior.

Thus stimulation of the infralimbic neurons can substitute for extinction training

Thus the first question is answered with a very clear YES!

Animals that were stimulated unpaired with the tone were identical to unstimulated
control animals.

On day 3 the animals can clearly “remember” the stimulus-tone paring.

The most probable explanations here is that the sensory information concerning the tone
is entering the infralimbic area and casing release of glutamate in synapses on infralimbic
neurons at the same time that the stimulating electrode induce high frequency discharge
is generating back propagating action potentials.

With the above two events occurring simultaneously the criteria for activation of NMDA
receptors is met and synapses undergo LTP.

Thus on day 3 the tone is sufficient to fire the infralimbic neurons and suppress the
activity of the amygdala.

Altogether these experiments provide strong evidence that the infralimbic neurons are
involved in extinction memory.


To summarize thus far…….

There is strong converging evidence that the infralimbic area is involved in the
consolidation and storage of fear extinction.

This evidence comes form is from three experimental approaches.

First, there were the experiments showing that lesioning of the infralimbic nuculi
eliminate fear extinction memory.

Secondly, single-unit recording show a strong correlation between the degree of
activation of infralimbic nerouns and the magnitude of the fear extinction memory.

Thirdly, as just demonstrated, stimulation of infralimbic neurons, with protocols that
mimic the tone-activity relationship of single-unit recordings, induces fear extinction


So, the picture that is emerging from the work of Quirk is that not only does the basal
lateral nucleus of the amygdala project to the central output nucleus, but the infralimbic
area does as well.

Normally a tone coming into the BAL will not activate the output neurons of the CeA.
After fear conditioning however, such connections have been strengthened and the tone
now activates the output of the CeA and we have the fear response (e.g. freezing).

The idea from Quirk’s research is that with extinction connections from the IL to the CeA
are strengthened and these connections lead to an inhibition of CeA output neurons.

Thus, following extinction training the tone, while still activating the BAL (i.e. the fear
memory is not forgotten), also activates the IL neurons and subsequently the output of the
CeA is inhibited (i.e the animal has learned not to fear).

This scheme begs the question: Do IL neurons indeed inhibit activation of CeA neurons?

Quirk and coworkers looked at this question as well in anesthetized rats.

These rats had stimulating electrodes placed in the BAL and recording electrodes placed
in the CeA to monitor discharge activity (action potentials) of the CeA output neurons.

As well they had stimulating electrodes placed in the infralimbic area of the vmPFC.

The results: activation of BAL neurons lead to high frequency discharges of CeA neurons
(top trace).

 Note: activation of BAL apparently causes some electronic noise (recording artifact) in
the recording electrode which is convenient because one can see on the recording exactly
when the stimulus was given.).

This BAL induced activation is blocked is the IL neurons are activated within 600 to 800
ms of the BAL activation

[Note: activation of the IL also causes noise in the recording electrode; this is seen only
on the second trace because the stimulations of the IL in subsequent traces are off the
scale of the data presented.]

Clearly, the answer to the question is, yes, activation of IL neurons can inhibit activation
of CeA output neurons is the signal from the IL comes within +/- 600 ms of the signal
from the BAL.


Next, I want to look at just what kinds of connections the IL is making with the CeA and
the output neurons in this region.

What is the nature of the IL-CeA connection, direct as illustrated here, or indirect,
perhaps even involving some other brain center.

A probable answer can be found in a study by McDonald et al who did anterograde
tracing (cell body to terminal) of neurons of the IL.

A micro-electropipette was brought into the IL of an anethesized rat and the anterograde
tracer injected by ionotophoresis (the tracer carries a charge and with a small current
carries the tracer into the tissue).

The tracer was Phaselous vulgaris leucoagglutinin, abbreviated PHA-L (L because there
also is an E form). . This is a complex glycoprotein found in seeds of Phaselous vulgaris
(red kidney beans). It is a neurotoxin, presumably found in the seed to discourage birds
and other animals from eating the seeds.

Evolution has engineered it to be taken up in cell bodies of neurons and then hitch a ride
via the anterograde transport system of the neuron to gain access to the nerve terminals.

Immunocytochemistry, using antibodies against the toxin, is applied to tissue sections to
find the terminal fields of neurons (colored image is from website of commericial
company marketing antibodies against PHA-L).

McDonald et al. found PHA-L widely but weakly distributed throughout various regions
of the amygdala.

However, they found an extremely strong signal in a region between the basal and central
nucleus of the amygdala, a region known to contain so-called “intercalated projection
neurons” which have been established to project to output neurons of the central

These interneurons are GABAergic, and thus inhibitory.

Thus the idea has developed that the IL would project to the ITC inhibitory interneurons,
thus activating them and consequently leading to an inhibition of the Central amygdala
output neurons.


Next question:

Do the ITC interneurons indeed inhibit CeA output neurons?

An answer can be found in Royer et al., doing electrophysiology on amygdala slice

They were actually stimulating in an area of the Basal Lateral Amygdala (BL) known to
project to the ITC area via glutamate containing neurons. They were recording output
neurons of the CeA.

They also had a micropipette so they could introduce an AMPA receptor antagonist in the
ITC region (the idea being, if the ITC neurons are projecting an inhibitory signal to the
CeA they can cut it off with the AMPA receptor antagonist because stimulatory input to
the ITC will fail to activate the ITC neurons).

The results:

In control experiments (no AMPA agonist in ITC region) the CeA output neurons clearly
displayed a IPSP.

When experiments were conducted in the presence of the AMPA receptor antagonist
there was no IPSP component to the response of the CeA neurons.

Evidence that GABA is causing the IPSP was collected in experiments where bicuculline
was introduced to the Ringers solution superfusing the tissue preparation.

Bicuculline is a GABAa receptor antagonist and, in its presence, the CeA output neurons
failed to generate a IPSP (note: recordings at different membrane holding potentials
(indicated on the right in the figure) are shown; the IPSP generated at holding potentials -
38mV and -43 mV are clearly attenuated in the presence of bicuculline).

Altogether it is concluded that that the concept that IL input would go via ITC
interneurons is probably correct (although the experiments where IL neurons are used to
activate the ITC neurons in tissue preparation have apparently not yet been done).


To summarize

Fear Conditioning strengthens BL connections to the CeA which leads to activation of
fear response programs in the brain stem and hypothalamus.

In fear extinction the IL to ITC is strengthened.

So, the fear conditioned pathway is still intact and strengthened….

But is blocked by the IL/ITC pathway before it can give an output.

In other words, through fear extinction the animal has learned not to fear.


And how is it in humans?

Several fMRI or PET studies have been conducted with post-traumatic stress disorder
(PTSD) patients.

In the study illustrated war veterans, who has undergone similar degrees of trauma, were
put in two groups, namely those displaying PTSD and those that did not.

During imaging the subjects were shown words that reminded them of their trauma.

The non-PTSD subjects showed a BOLD signal in the vmPFC during such experiments,
indicating that this brain region has been activated during recall of the trauma.

The PTSD patients failed to show such activation in this brain region.

Apparently they have not learned not to fear!

These finding could ultimately have clinical implications. As you have just seen, animal
studies have shown that IL activation can lead to fear extinction in the absence of
extinction training.

Possibly, pharmacons might someday be developed that can specifically target the PFC to
help lessen the fear response in PTSD patients.


Before stopping with my lectures on fear, I want to make a few concluding remarks on
humans, in particular the relationship between the amygdala and the hippocampus.

For this I want to return to the fMRI studies of La Bar et al. demonstrating activation of
the amygdala during fear conditioning in the human.

Elizabeth Phelps has extended these kinds of studies to include measurement one of the
fear responses, namely sweat production, measured as the skin conductance response

[These experiments are outlined in her presentation in the Web Symposium on Fear.]

It is well known that during a fear response there is an increase in skin electrical
conductance, reflecting increased autonomic activity in the sweat glands.

In the experiment the subjects showed a highly significant increase in SCR when viewing
the blue (threat) square. This was increased even further when the electrical impulse
finally hit.

The first (upper) graph is showing the response from a normal subject.

The second graph is showing a typical response from a person who has a damaged
amygdala. Also included here is a movie getting the reaction of this person when shown
that she has an abnormal fear response.

Except for the first time blue was flashed (where she was caught by surprise) she knew
“Blue means pain…I’m about to get shocked!” She knows this because this knowledge is
a function of her declarative memory in the hippocampus (I declare…blue means
pain…). The hippocampus has made an association between blue and pain. The
amygdala, which is damaged in this patient, does not make this association.

Notice that pain induces a fear response, so the ability to generate such a response is
intact. What is missing is the learned component of the fear response.

She knows she should be sweating and wonders why she isn’t.


Compare this response with the response of someone with hippocampal damage (like
patient HM, discussed in the Neurobiology course).

Here the response is normal, but the person can not make a conscious connection
between blue and pain. Thus, at the time the blue color appears they are wondering why
they are sweating.

In other words, they can’t remember blue means pain but the amygdala does its job.

For the patient with amygdala damage, they know that blue means pain, but their
amygdala fails to give the proper respond leaving them wondering why they don’t sweat
like everyone else.


Another point Phelps touches upon in her Web Symposium presentation is the question
“Why are events with emotional content remembered better?

We have already discussed the output of the amygdala to the brain stem and
hypothalamus to drive the fear responses.

I have already mentioned that the hippocampus sends input to the amygdala to
incorporate contextual information into the fear response.

The new element is that the amygdala affects the hippocampus, in a positive way, to form
new memories.

Animal studies have shown that amygdala has a stimulatory effect on the process of
memory consolidation, whereby the hippocampal short-term memory is sent to the cortex
for long-term storage.

Thus, through the action of the amygdala, memories with an emotional content become
more readily consolidated (they become burnt into the memory circuits of the cortex).


Contextual fear conditioning

The best way to illustrate the interaction between the amygdala and the hippocampus is
to consider an experimental protocol called “contextual fear conditioning”.

We have already discussed extensively fear conditioning where the Cs is often auditory
(i.e. auditory fear conditioning).

Consider the case where there is no auditory Cs. Rather, the fear conditioning takes place
in a certain (well defined) environment (shown in this presentation as a rat in a cage with
certain objects marking the environment).

The environment is the Cs. After fear conditioning, if the animal is returned to this
environment it will show all the signs of fear memory (freezing, startle etc).


A recent study by Huff et al beautifully illustrates the activation of the hippocampus
during contextual fear memory and they show, moreover, that the amygdala is
responsible for that activation.

Their protocol for contextual fear memory was quite standard….rats were brought from
their home cage to the test cage in black buckets. They were immediately placed in the
test cage.

In some cases the animals were given an immediate foot shock and then immediately
returned to the black bucket and immediately returned to the home cage.

Other animals were given 5 min to explore the test cage (i.e. they were given 5 min to
build-up a “context”).

Following the 5 min the animals were foot shocked and then returned (via the black
bucket) to their home cage.

After 60 min in their home cage all animals were killed and their brains dissected and
processed for quantitative RT-PCR for the immediate early gene cFos.

Shown is the relative cFos mRNA levels in the hippocampus in animals from the home
cage, animals subjected to a immediate foot shock and animals that were submitted to the
contextual fear conditioning protocol.

Clearly, the contextual fear conditioning has led to an induction of cFos expression
(indicating activation of neurons).


The next question for Huff et al was, is the amygdala involved in this activation (it could
be that simple exploration is responsible)?

To test this (in separate experiments) this they “knocked-out” the amygdala just before
submitting the animals to the contextual fear conditioning protocol.

They did this by injecting the parmacon muscimol (a GABAa receptor agonist) into the
basal lateral amygdala (the BLA is known to possess GABAergic interneurons which
inhibit BLA activity).

The result show no increase in cFos activity in contextual fear conditioned animals
injected with muscimol, whereas those animals injected with vehicle (solvent) alone
displayed full cFos activation.

Their conclusion…the amygdala is necessary for the activation of the hippocampus seen
during contextual fear conditioning.


Next question: is the amygdala doing this directly or are indirect routes involved?

Thus question has been tackled by Paz et al.

Their study was based on two previous observations.

First, in going from the sensory cortices to the hippocampus, information passes through
the rhinalcortex, with two subdivisions, the perirhinal and the entorhinal cortex. Here
electrical recording have revealed some sort of bottle neck in information flow.

They also noted that there are robust projections of neurons from the basal lateral
amygdala to the peri- and entrorhinal cortex.

They placed recording electrodes in the peri- and entrorhinal cortex, as well as in the
BLA itself, and then subjected animals to emotional experimental protocols that activated
the BLA.

They found that BLA activation increased impulse transmission form the peri to the
entrorhinal cortex and suggest that this may be important to amygdala enhancement of
hippocampal function (i.e. the BLA facilitated information flow into the hippocampus for
construction into memories).

In a News and Views item on this study, Quirk and Gonzalez appropriately titled their
article “Keeping the memories flowing”.

Also included in the PowerPoint is a link to a Nature Podcast giving an interview with
Denis Pare, one of the senior authors of the study.


For my final remarks on the neurobiology of fear, I want to show a remarkable human
fMRI study.

This is again work from Phelps, and discussed in her presentation in the web symposium.

In humans, we have already seen how the amygdala lights up in imagining studies when a
BOLD image is made of the CS versus –CS.

In a study by Phelps et al they considered the possibility that the US might not be
required to fear condition a person, just the threat or anticipation of the US might be

In other words, let the subject think he is going to get a painful US and see if this leads to
fear conditioning.

The experiment was conducted almost the same as that described earlier.

A stimulating bar electrode was attached to the subject's wrist for delivering electric

They were told that during the experiment they would receive between 1 and 3 electric

During the experiment yellow squares and blue squares were presented.

They were told that the electric shock, if it came, would come when e.g. the blue square
was displayed (threat condition) but not when the yellow was displayed (safe condition).

Shown is one of the threat versus safe BOLD images from this experiment.

The left amygdala has been activated, as has the left insula, a region of the cortex that is
known to communicate directly with the amygdala. The insula is a higher brain center
center that receives cortical and thalamic (sensory) input.

From this and other images they propose that the imagined and anticipated pain, when
viewing the blue square, results in a cortical (insula) representation of fear.

This is relayed from the insula to the amygdala.

The amygdala in turn would orchestrate a fear response.

In fact we may be seeing part of the response in the BOLD image.

On the right hand side of the brain the premotor cortex involved in motor responses has
been activated. Perhaps the person is getting ready to run or to take some other defensive


In their experiment Phelps et al did record one activity associated with fear, namely skin
conductance response (SCR)

In the experiment the subjects showed a highly significant increase in SCR when viewing
the blue (threat) square compared to the yellow (safe) square.

Clearly, fears that are imagined and anticipated but never experienced can have a
profound influence on our behavior.

It is to be hoped, in conditions where these fears become pathological, that the human
and animal-model research described in this presentation will ultimately help in
developing new strategies and new pharmaceuticals to treat these pathologies.