Neural mechanisms of habituation and sensitization in Aplysia

Brain mechanisms of behavior (PSY 445)

Lecture 9

Required reading:
Reading: Neuronal Mechanisms of Habituation and Dishabituation of the
Gill-Withdrawal Reflex in Aplysia
Vincent Castellucci, Harold Pinsker, Irving Kupfermann and Eric Kandel, 1970.
http://www.neuro.uoregon.edu/wehr/coursepapers/Castellucci-Kandel-1970.pdf

Further readings:
Chapter 10, Learning and memory systems in Aplysia, in Behavioral Neurobiology by
Thomas Carew.

Overview:
This paper is an elegant example of the scientific method, in which they start with a
simple form of learning, and step by step, reveal the underlying neural mechanisms with
a series of experiments.

The animal:
Aplysia californica is one of the workhorse model sys-
tems of neuroethology. It is studied by countless labs
(there are at least 3800 scientific articles about Aplysia,
and Google returns at least 400,000 hits for a search
on ‘Aplysia’). The Castellucci et al. paper that we read
is one of the series of papers on Aplysia that led to the
Nobel Prize being awarded to Eric Kandel in 2000. The
Aplysia californica inking.
creature used for these elegant experiments is a pur-
plish brown sea slug that lives along the Pacific coast.
It is commonly called the sea hare, and it lives in the
intertidal zone (the region between high tide and low tide). This marine gastropod mol-
lusk is usually between 6 inches and a foot long, and is covered with a gummy slime
which is virtually impossible to wash off your hands after you have handled one. Aplysia
californica prefers the warmer waters of the southern California coast, so you won’t see
one out along the Oregon coast. They eat seaweed, especially red or purple seaweed.
Their main defense against predators is to release a dark purple ink, and they derive the
pigment for this ink from the purple seaweed they eat. Their predators include crabs and
sea anemones.

The behavior:
Aplysia breathes underwater with a delicate gill, similar to those used by fish. The ani-
mal draws water over the gill by means of a siphon. In case of danger, Aplysia with-
draws its gill and siphon out of harm’s way, and covers them with a protective mantle.
This reflex can be easily reproduced in the laboratory. If you pin a slug to a dish, and
then touch the siphon gently with a paintbrush, the animal will retract the gill. This gill-
withdrawal reflex is very easy to quantify, using an “electric eye” similar to those that
Neural mechanisms of habituation and sensitization in Aplysia 2

cause a chime when you walk

through the doors into a store like
Radio Shack. You can place a pho-
todiode underneath the gill, and
measure the amount of light falling
onto the photodiode. When the gill
withdraws out of the way, more light
hits the photodiode, which is easily
measured as a voltage change.

What causes this reflex in the ani-

mal’s natural environment? If it’s a
crab trying to take a bite out of the
siphon, then withdrawing the gill is a Touch the siphon, and the gill withdraws.
life-or-death escape response for
the slug. However, Aplysia lives in
the kelp forest. Imagine a forest of kelp strands swaying back and forth in the waves. It’s
likely that an Aplysia could be constantly touched on its siphon by harmless stimuli in its
natural environment. Exhibiting a full-blown gill-withdrawal response for each of these
repetitive touches would be counterproductive and would interfere with the slug’s ability
to breathe normally. Instead, Aplysia displays a simple form of learning, called habitua-
tion, which allows it to learn to safely ignore repetitive touch stimuli and only withdraw its
gill for surprising stimuli. Habituation refers to a progressive decrease in responding to a
stimulus when that stimulus is repeatedly presented. After a long rest without any stim-
uli, the response will recover to its original level. Habituation is a form of non-associative
learning, that is, it does not involve the association of one thing with another (for exam-
ple, learning to associate the ringing of a bell with food is a form of associative learning).
Aplysia gill-withdrawal displays three forms of non-associative learning: habituation,
dishabituation, and sensitization. Dishabituation occurs when you take an animal that is
completely habituated to repeated touching of its siphon. If you then pinch its tail, the
animal will respond to a siphon touch with a full-blown gill-withdrawal response. In other
words, the habituated response is restored to its original level. Similarly, sensitization
refers to an enhancement of a response following the presentation of a novel stimulus.
For example, after you pinch the
tail, touching the siphon will pro-
duce a larger gill withdrawal than
normal. The difference between
sensitization and dishabituation is
that sensitization refers to a rested
Motion of gill (photodiode signal).
animal, whereas dishabituation re-
fers to a habituated animal.

The circuitry underlying gill withdrawal

The touch stimulus activates mechanoreceptors in the siphon skin. These sensory neu-
rons make direct connections to the motor neurons that cause gill withdrawal. The sen-
sory neurons also make connections to excitatory and inhibitory interneurons that con-
Neural mechanisms of habituation and sensitization in Aplysia 3

nect with the motor neu-

rons, forming an indirect
pathway from sensory to
motor neuron.
The nervous system of
Aplysia, like other inverte-
brates, consists of a series
of ganglia connected to
each other by nerve cords Both direct and indirect sensory-motor pathways.
called connectives. The
neurons involved in
the gill withdrawal re-
flex are located in the
abdominal ganglion.
One of the huge ad-
vantages of working
with simple inverte-
brate model systems
is that many of the
neurons are individu-
ally identified. For ex-
ample, neuron R2 is
in the same location,
and has all the same Gill withdrawal circuitry is in the abdominal ganglion.
synaptic connections,
from animal to animal.
The same is true of many, many other identified neurons. R2 is especially large (over 1
mm in diameter) and therefore easy to record from intracellularly (one afternoon, all the
students in my undergraduate neurobiology course recorded intracellularly from R2). L7
is one of the motor neurons that cause gill withdrawal. If you record from L7, and re-
peatedly touch the si-
phon, you can see
compound postsynap-
tic potentials (PSPs)
that decrease in am-
plitude with repeated
touches. This is the
neural correlate of the
habituation of the gill
withdrawal response. Identified neurons in the abdominal ganglion.
Another huge advan-
tage of working with
an invertebrate model system like Aplysia is that you can simplify the preparation, in-
stead of using the entire intact animal. Castellucci et al. used a reduced preparation in
which they removed the abdominal ganglion along with a small piece of siphon skin that
was still attached to the ganglion via the connective. This reduced preparation can be
Neural mechanisms of habituation and sensitization in Aplysia 4

kept alive in a dish

for hours with a
little oxygenated Decreasing PSP in motor neuron L7.
seawater. In this
way they could
chase down the mechanisms underlying the habituation of the PSP in L7, by using a
sequence of alternate hypotheses and decisive experiments that are truly elegant ex-
amples of the scientific method in action.

Scientific inference
Everyone knows what the scientific method is. You start with a question or a problem.
You formulate a hypothesis about what is really going on. Then you devise an experi-
ment to test your hypothesis, do the experiment, and then interpret your data as either
supporting or refuting the hypothesis. This usually leads to the next question, and you
iterate through the method again. For example, you come out of the house in the morn-
ing and your car doesn’t start. What’s wrong with it? That’s the question you start with.
Hypothesis: maybe it’s a dead battery. Experiment: test your hypothesis by jump start-
ing the car. Prediction: if your hypothesis is correct, a jump should start the car and it
should run fine after that. If your hypothesis is incorrect, a jump won’t start the car. Re-
sults: jump starting works, and your car starts. Conclusion: it was indeed a dead battery.
Despite how simple and obvious this scientific method sounds, you would be surprised
and perhaps a bit dismayed to learn how rarely the scientific method is rigorously prac-
ticed in the laboratory. Even well-seasoned scientists make logical errors all the time,
such as formulating the hypothesis that a jump start will get the car running. That’s a
prediction of the outcome of an experiment, not a hypothesis about the mechanism un-
derlying the problem. The difference is that the prediction leaves out a clear explanation
of the true hypothesis about the underlying mechanism (a dead battery). It’s also critical
to come up with all possible alternate hypotheses that could explain the problem, and
these must be falsifiable hypotheses. A hypothesis that cannot be disproven by an ex-
periment is useless. The Castellucci et al. paper is a beautiful example of the process of
scientific inference, in which they start with the simple learning phenomenon of gill-
withdrawal habituation, and formulate a series of pairs of alternative hypotheses, per-
form the experiments that disprove one of them, thereby leading to the next pair of hy-
potheses. The hypotheses form a logical tree, and the experiments lead us through the
branches of this tree, step by step. Let’s follow the steps.

Habituation
The first step is the statement of the problem or question. In this case the habituation of
the behavior gill withdrawal is correlated with the decrement in the L7 PSP. What
causes the PSP to decrease with repeated stimuli? This is our starting question. Castel-
lucci et al. consider two possible explanations. Either some excitatory input is decreas-
ing, or some inhibitory input is increasing. These aren’t the only possible explanations,
but they are a useful starting point. What experiment can distinguish between these
possibilities? Synaptic excitation and inhibition are caused by ionic currents that have
different reversal potentials. Excitatory reversal potentials are higher than the resting
membrane potential, whereas inhibitory reversal potentials are lower than the resting
Neural mechanisms of habituation and sensitization in Aplysia 5

Logical inference "PSP in L7

$ driving force

! "EPSP #IPSP

"RIN n.c. RIN

! measure RIN
n.c. EPSC "EPSC

minimal
"Exc. IN
! "Exc. SN stimulation

heterosynaptic homosynaptic dual

!
presynaptic inhibition depression intracellular

Theytree
The logical conclude that
for the homosynaptic
experiments indepression
Castellucciat et
theal.SN!L7
synapse is one of the
of
mechanisms of gill-withdrawal
functional modi-
habituation.
tential in the motor neuron
capable undergoing could be could in turn be produc
membrane potential. That’s ficationwhy However, the
excitation
(1-3). relevance
is normally either byand
depolarizing,
produced a decrease
inhibition in ex- is crease in the input resis
27 /24
of synaptic plasticity to a specific in- citatory synaptic input or by an in- motor neuron. We tested
normally hyperpolarizing.stanceBy hyperpolarizing
of behavioral modification the cell
has with
creaseanin intracellular
an underlying inhibitory current post-injec-
ity by measuring the resi
tion, we can change the neverdriving been force. If weWe
demonstrated. hyperpolarize
have de- below
synaptic the (IPSP)
potential reversal masked potential
by motor neuron with intrace
scribed behavioral parameters of ha- the EPSP. The role of incrementing polarizing pulses and fou
for inhibition, then inhibitory synaptic
bituation potentials
and dishabituation (IPSPs)
of the gill- will become
postsynaptic depolarizing.
inhibition in the motor This decrement of the complex
is what Castellucci did. They
withdrawalfigured
reflex out
in thewhat
intact the reversal
Aplysia neuronpotential for inhibition
could be excluded because sim- wasassociated
by with a change
(4), and we have examined their cellu- ilar EPSP decrement occurred when resistance of the motor
looking at spontaneous IPSPs, and hyperpolarizing them
lar correlates in a semi-intact prepara-
until they reversed and be-1B). These findings can
the membrane potential was hyperpo-
came depolarizing. When tionCastellucci
(5). We now hyperpolarized
describe experiments L7 well
larized below
well beyond this membrane
the equilibrium po- po-resistance changes at a
in the isolated in which we tential for the spontaneous IPSP's in from the microelectrode
tential, they saw that the EPSP still decreased with repeated
ganglion touches. This means that
the motor neuron.
have simplified the neural circuit of the body, but they do rul
the excitatory potential (EPSP) must
reflex in order to be decreas-
investigate the cellu- The decrement in EPSP amplitude changes in input resistance
ing. It doesn’t rule out a change in inhibition,indicate
lar mechanisms. Our data but that that the PSP alterations
both habituation and dishabituation of changes in the synaptic
they ignore that to instead the follow up with their next
gill-withdrawal reflex in Aplysia in-
. Al A2 motor neuron.
question. What’s causingvolve thechanges
decrease in the in the exci-
effectiveness of a ,^--*^L^t~~~ 1^^ MotorN
Stim Rec
A decrease in the excita
istim
/ /^y'-},
specific set of central excitatory syn- ' input in the motor neur
tatory synaptic potential?apses between the sensory neuron and Stim Stim
caused either by an incre
the motor neuron. These plastic tion of excitatory inter
contribute to the complex
The decrease in the EPSP could
changes befrom
result caused by ande-
homosynaptic
a decrease in synaptic eff
pression and heterosynaptic facilitation,
actual reduction in excitatory synaptic input, or it
respectively. vidual afferent excitatory
could be caused by a decreased
The abdominalinputganglion
resistance.
was isolated demonstration of decreme
except for a strand of the siphon nerve B mentary monosynaptic ex
Input resistance is the property
which remainedof a neuron
attached that
to a small Control to the motor neuron would
determines how large ofpiece
a postsynaptic
of skin from potential
the tactile isre- dence for the latter po
ceptive field of the gill-withdrawal re- therefore simplified the af
caused by a given synaptic current. In other
flex (Fig. 1A). A localized tactile or the reflex pathway by e
Decremented
words, as described by Ohm’s
electrical law, thewas
stimulus voltage
applied to the
\ f
L7, unitary and presumab
and a double
produced is just the resistance times the current.
skin, barrel microelec- aptic EPSP's produced
trode was inserted into one of the stimulation of mechanorec
Thus the smaller voltageidentified
response motorcould
neuronsbe(usually
due to L7, Facilitated
skin. We found a respons
the neuron lowering its resistance, or(6)it could be and
Fig. 1A, part 2) for recording the skin by using a tac
for measuring the membrane resistance. and we then applied a w
due to less current coming in. How
In some could we dis-
experiments we also impaled 500 msec
10 mv
stimulus to this portion of
tinguish between these two possibilities?
the cell Castel-
bodies of the mechanoreceptor established the elementar
neurons of the gill-withdrawal reflex. No change
Fig. 1. in input
(A) The isolated resistance.
ganglion prepara-
tion. Part 1, the whole animal and a the EPSP's in these ex
lucci et al. measured input resistance by passing
In experiments in the intact and in of showing that the threshold
typical region (indicated black)
semi-intact preparation (4, 5), we used the tactile receptive field which was used was all-or-none. The EPS
a jet of seawater that lasted from 500 in the isolated ganglion experiment. Part to be monosynaptic since
to 800 msec as the tactile stimulus to 2, composite set-up for the isolated gan- fixed shape and constant
glion preparation. The ganglion was re-
elicit the gill reflex. To facilitate the moved from the animal with a piece of were not abolished in solu
analysis in the isolated ganglion, we skin from the tactile receptive field still calcium content which t
have used brief (5 msec) mechanical attached to a strand of the siphon nerve. polysynaptic inputs by
or electrical stimuli to the skin. These The region innervated by such a strand, threshold of interneurons.
when tested with tactile stimuli, is indi-
brief stimuli produced complex excita- field The elementary EPSP
Neural mechanisms of habituation and sensitization in Aplysia 6

When they intracellularly stimulate only a single sensory neuron, they still see
the EPSP decrease in the motor neuron. This rules out presynaptic inhibition.

current pulses, and showing that they produced the same voltage deflection before and
after habituation. This means that there was no change in input resistance, and the de-
creased EPSP must therefore be due to decreased excitatory synaptic current (that is, a
decreased EPSC). In other words, the decrease must be presynaptic to L7.

Take another look at the

circuit diagram. A quick
look reveals that there
are two sources of excita-
tory input to L7: (1) exci-
tatory interneurons, and
(2) the sensory neurons
themselves. Which one of
these is decreasing? How Both direct and indirect sensory-motor pathways.
could we tell? One way is
to stimulate only the sen-
sory neurons, and see if
the strength of those inputs decreases with repeated stimulation. The first approach that
Castellucci used to selectively stimulate only the sensory neurons is called the minimal
stimulation technique. The idea is to electrically stimulate the siphon skin, which will ac-
tivate sensory neurons. For a strong stimulus, the sensory neurons will be strongly acti-
vated, and will in turn activate the excitatory interneurons. Thus both the direct and indi-
rect pathways are activated for strong stimulation. But as we turn down the stimulus in-
tensity, the sensory neurons will fire fewer and fewer spikes, and activate the interneu-
rons less and less. At some minimal stimulation intensity, the sensory neurons will pro-
duce only a subthreshold response in the interneurons, and the interneurons will no
longer fire action potentials. At this minimal stimulation intensity, the PSP recorded in L7
will be purely from the sensory neurons. In order to convince us that this is the case,
Castellucci report that the threshold for these PSPs was all-or-none, and that they had a
fixed shape and latency. In addition, they saw no effect when they reduced synaptic re-
lease probability by increasing the calcium concentration in the bath, which would re-
duce the probability of transmission via the interneurons. Together, these observations
make it very likely that they were recording only monosynaptic inputs from the sensory
Neural mechanisms of habituation and sensitization in Aplysia 7

neurons. And since this EPSP still showed a decrease with repeated stimuli, they con-
clude that the synapse from the sensory neurons onto L7 is actually decreasing in
strength.

They then propose a last pair of hypotheses to explain the decreased EPSP from the
sensory neurons. It could be that the decreased strength of the synapse is occurring
just at that synapse, and that no other neurons are involved. This is called homosynap-
tic depression (i.e. a decrease in strength at the same synapse). In contrast, it’s possi-
ble that presynaptic inhibition is involved. Even though the EPSP in L7 is monosynaptic
and doesn’t involve interneurons, they can’t rule out the possibility that some other in-
terneurons are activated which don’t synapse directly onto L7. If these interneurons
presynaptically inhibited the synapse from sensory neuron to L7, it would be a hetero-
synaptic circuit rather than homosynaptic depression. How can we guarantee that only
the sensory neurons are activated, without electrically stimulating some other interneu-
rons? As luck would have it, they discovered the place where the cell bodies of the sen-
sory neurons were located in the ganglion. They then recorded intracellularly from both
a sensory neuron and L7 at the same time. With this dual recording, they could inject
some current into the sensory neuron to cause it to fire an action potential. They could
then observe the postsynaptic response in L7 caused by a single spike in a single sen-
sory neuron. When they did this repeatedly, the EPSP decreased. This means that the
synapse itself is decreasing in strength, homosynaptically, and that no other neurons
are required.

The next logical step would be to ask whether it’s the presynaptic terminal that is releas-
ing less transmitter, or the postsynaptic side that is less sensitive to transmitter. Al-
though Castellucci et al. mention these two possibilities, they don’t come up with an ex-
periment to test between them. How might you do so? This is another example of the
“next step” experiment that might be an interesting point of departure for a term paper,
as you read any article of interest. It turns out that homosynaptic depression at this syn-
apse is presynaptic, and that less transmitter is released. The critical experiment was to
puff on tiny amounts of the transmitter (glutamate), showing that the postsynaptic re-
sponse to a puff is unchanged even after habituation. The molecular mechanisms un-
derlying this decreased transmitter release are still being worked out; the Castellucci lab
published a paper just last year about how transmitter release is modulated when Pro-
tein Kinase C (PKC) phosphorylates a protein called SNAP-25 (synaptosomal-
associated protein of size 25 kiloDaltons). The molecular tools that they used to show
this, not surprisingly, included GFP (green fluorescent protein).

It’s worth pointing out that in this series of experiments, they haven’t demonstrated that
homosynaptic depression at the sensory-neuron-to-L7 synapse is the only mechanism
underlying habituation. There could be numerous other mechanisms operating in paral-
lel, during normal habituation of the behavioral response. In fact this is likely, since biol-
ogy usually evolves multiple redundant mechanisms for important functions. Rather
than try and work out all the mechanisms that are involved, the strategy they used in
this paper was to try and get the most detailed explanation possible for one such
mechanism.
Neural mechanisms of habituation and sensitization in Aplysia 8

Facilitation caused by tail shock.

Sensitizing stimulus to tail

!
activation of modulatory interneurons
!
seratonin release
!
"Adenylate Cyclase
!
"cAMP
!
"PKA
!
phosphorylation of K+channel
!
!K+conductance
!
broadened spike in SN
!
"Ca++flux into terminal
!
"transmitter release
!
"EPSP in MN!
!
"motor neuron activity
!
"behavioral response
AfterAfter decades
decades of further
of further experiments
experiments (and prize),
(and a Nobel a Nobelthisprize), this
heterosynaptic facili-
heterosynaptic
tation facilitation
is now understood is now
in incredible understood
detail in incredible
at the molecular detail
and cellular at
level.
the molecular and cellular level. 31 /24
Neural mechanisms of habituation and sensitization in Aplysia 9

Facilitation
Although the sequence of experiments is not as impressive, Castellucci et al. also make
substantial progress in working out the mechanisms underlying facilitation in this paper.
They observed that the PSP in L7 is facilitated after tail shock, which is a neural corre-
late of behavioral facilitation. Input resistance is not changed after facilitation, so that
can be excluded as a possible explanation. With dual intracellular recordings from sen-
sory neurons and L7, they could evoke single spikes in the sensory neuron, and look at
the resulting PSPs in L7. After tail shock, these were facilitated, suggesting that the
strength of the synapse is actually increased. This suggests that spikes from the tail
shock pathway are somehow increasing the synaptic strength of the sensory-neuron-to-
L7 synapse. This most likely involves one synapse increasing the strength of another,
which would be an example of heterosynaptic facilitation. Castellucci et al. don’t prove
it, but they suggest that presynaptic facilitation is the most likely explanation. In later ex-
periments from the Kandel lab, the molecular and cellular mechanisms underlying this
heterosynaptic facilitation have been worked out in detail. It turns out that the modula-
tory information from the tail shock pathway is carried by the neuromodulator seratonin.
Seratonin acts at receptors on the presynaptic terminals of the sensory-neuron-to-L7
synapse. These receptors activate a second messenger cascade in which adenylate
cyclase produces cyclic AMP, which causes PKA to phosphorylate a potassium channel.
This decreases the conductance of the potassium channel during the repolarization
phase of the action potential. Since the action potential is less able to repolarize, it stays
depolarized longer. In other words, the action potential is broadened. This means that
voltage-gated calcium channels are activated for a longer time during the action poten-
tial, letting in more calcium, and thereby causing more transmitter release. This under-
lies the facilitation of the gill withdrawal response. This was one of the first synapses at
which learning was understood in such incredible detail, and which led to the Nobel
Prize being awarded to Eric Kandel.