Is LTP in the amygdala a synaptic mechanism for the acquisition of conditional fear?

Summary
The major amygdaloid nuclei have long been proposed as the loci of emotional processing. In the past dacade, intense research has been carried out on the lateral and basolateral amygdaloid nuclei. These structures are thought to be necessary for Pavlovian (classical) fear conditioning, that is learning about fearful events, believed to be carried out through glutamatergic LTP. The proposed experiment would (1) test for the presence of amygdalar LTP in the GluR-A-/- mice and then (2) establish whether the mice show fear conditioned responses. The study would thus investigate whether increases in synaptic strength in the appropriate neural circuit is indeed necessary for fear conditioning.

BACKGROUND

Anatomy
The basolateral amygdaloid complex (BLA) receives multiple inputs from a variety of structures. It receives, among others, auditory information (from the auditory thalamus), visual information (from the perirhinal cortex), spatial and contextual information (from the hippocampus) and somatosensory information (from the insular cortex). The BLA, therefore, is anatomically situated to integrate information from a variety of sensory domains.

Thus, the BLA is a locus of sensory convergence and a site for conditional stimulus-unconditional stimulus (CS-US) association in the amygdala. Intra-amygdaloid circuitry conveys the CS-US association to the central nucleus, where projections to the hypothalamus and brainstem mediate fear responses such as freezing (periaqueductal gray), potentiated acoustic startle (nucleus reticularis pontis caudalis), increased heart rate and blood pressure (lateral hypothalamus), increased respiration (parabrachial nucleus) and glucocorticoid release (paraventricular nucleus of the hypothalamus).

Fear conditioning
Fear conditioning is a paradigm that has been used as a model for emotional learning in animals. During fear conditioning, a neutral conditioned stimulus (the CS; for example, a tone) is followed by an unconditioned stimulus (the US; for example, a foot shock), which routinely elicits a stereotypic response - the behavioural correlates of fear. The lateral amygdala (LA) is thought to be the critical structure in which information from the CS and US converge (LeDoux, 1995). Commonly, researchers have tested versions of the following hypothesis: LTP strengthens the input from the auditory thalamus to the amygdala; this how the fear gets attached to the tone. Deletion of LTP should block learning about the relationship between CS and US.

The acquisition of conditioned fear behaviour is evaluated by measuring 'freezing', a characteristic defensive posture expressed in the presence of stimuli that predict danger. The amount of time accounted for by freezing was measured during the 20-s CS and also during the 20 s immediately before CS onset (pre-CS period). The latter is a measure of the acquisition of aversive conditioning to the experimental context in which the US is delivered (such as the conditioning chamber); freezing to the experimental context is independent of the presence or absence of an explicit CS, and is typically seen with both paired and unpaired training. In this experiment and pilot studies, the pattern of behaviour exhibited during the 'one tone per second' 20-s CS was in all respects similar to the behaviour exhibited by animals trained with a 20-s continuous tone CS; for example, rats did not respond to the individual tones that made up the CS, but rather behaved as though the 20-s CS period was a continuous tone.

As the CS-US relation is learned, innate physiological and behavioural responses come under the control of the CS (Figure 1). For example, if a rat is given a tone CS followed by an electric shock US, after a few tone-shock pairings (one is often sufficient), defensive responses (responses that typically occur in the presence of danger) will be elicited by the tone. Examples of species-typical defensive responses that are brought under the control of the CS include defensive behaviours (such as freezing) and autonomic (e.g. heart rate, blood pressure) and endocrine (hormone release) responses, as well as alterations in pain sensitivity (analgesia) and reflex expression (fear potentiated startle and eye blink responses).

Long term potentiation
There have been a number of studies of LTP in the amygdala, mostly involving in vitro brain slices and pathways carrying information from the cortex to LA and B (Chapman et al 1990, Chapman & Bellevance 1992, Gean et al 1993, Huang & Kandel 1998). These studies have led to mixed results regarding the possible role of NMDA receptors in cortico-amygdala LTP, with some studies finding effects (Huang & Kandel 1998) and some not (Chapman & Bellevance 1992). Recent in vitro studies indicate that LTP in the thalamo-amygdala pathway requires postsynaptic calcium but the calcium does not enter through NMDA receptors (Weisskopf et al 1999). Instead, calcium entry appears through L-type voltage-gated calcium channels. These channels have also been implicated in a form of LTP that occurs in the hippocampus (Cavus & Teyler 1996). It has also been shown that prior fear conditioning leads to an enhancement in synaptic responses recorded subsequently in vitro from amygdala slices (McKernan & Schinnick-Gallagher 1997). The receptor mechansisms underlying this form of plasticity have not been elucidated.

LTP has also been studied in vivo in the thalamo-amygdala pathway using recordings of extracellular field potentials (Clugnet & LeDoux 1990, Rogan & LeDoux 1995, Rogan et al 1997). These studies show that LTP occurs in fear processing pathways, that the processing of natural stimuli similar to those used as a CS in conditioning studies is facilitated following LTP induction, and that fear conditioning and LTP induction produce similar changes in the processing of CS-like stimuli (Figure 6). Although exploration of mechanisms are difficult in these in vivo studies, they nevertheless provide some of the strongest evidence to date in any brain system of a relation between natural learning and LTP (Barnes 1995, Eichenbaum 1995, Stevens 1998). LTP has been found in vivo in the hippocampal-amygdala pathway, which is believed to be involved in context conditioning (Maren & Fanselow 1995).

If the amygdala is involved in learning of associations between stimuli and fearful outcomes, is this process LTP-dependent? Studies along these lines are only just beginning but there is considerable support for the hypothesis. For example, it has been known for some time that the pathways into and out of the amygdala are very plastic, and readily show LTP. The basolateral nucleus is richly endowed with NMDA receptors, and drugs that block these receptors (such as AP5) block acquisition but not expression of fear responses, both to simple stimuli and to context. Furthermore, induction of LTP in the auditory inputs increases auditory evoked potentials in the amygdala, and auditory fear conditioning increases the size of these responses, as well as neural activity along this pathway.

Long term potentiation has been proposed as the synaptic mechanism underlying learning about fearful events. Several studies concentrate on the neurons in the LA and BLA, which receive inputs from the medial geniculate nucleus of the thalamus (aversive auditory cues, e.g. burst of white noise), and that these two structures, through learning, exhibit associative spiking. Glutamate receptors are thought to be essential for fear conditioning. LTP induction on amygdaloid interneurons appears to be mediated by AMPA rather than NMDA receptors (Mahanty and Sah, 1998).

Because the CS is a simple sensory stimulus, the afferents that carry the CS information into the LA can be defined. This connection between the auditory thalamus and LA can express LTP (Clugnet and LeDoux, 1990) which, when induced with electrical stimulation, causes an increase in the response of the LA to auditory stimulation (Rogan and LeDoux, 1995). The processing of natural stimuli can, therefore, use the mechanisms set up by artificially induced LTP.

By preparing in vitro slices of the LA from fear conditioned rats, McKernan and Shinnick-Gallagher (1997) examined the synaptic responses of LA neurons to stimulation of afferents from the auditory thalamus. These responses were consistently larger than those recorded from LA slices prepared from control animals that had undergone unpaired CS-US training. Furthermore, the synaptic responses of LA neurons to an independent input that was not involved in fear conditioning remained unaltered. The compelling conclusion is that fear conditioning caused an increase in synaptic efficacy (for example, LTP), specifically at the synapses that process the CS. The authors also suggest that this behaviourally induced increase in synaptic strength may be due, at least in part, to presynaptic modifications, because it was accompanied by a change in one form of short-term presynaptic plasticity.

To test for LTP in the LA, LeDoux and colleagues (Bauer et al, 2001) make use of whole-cell recordings from LA neurons in slice preparations. Stimulating electrodes were placed in the path of the thalamic afferents to LA. LTP is induced by pairing a train of 10 presynaptic stimuli (30 Hz) to the thalamic fibers with a train of 10 depolarizations (1 nA, 5 ms) of the postsynaptic cell given 5-10 ms after the onset of each EPSP to produce an action potential at the peak of each EPSP. Pairing is given 15 times at 20-second intervals. A second group of cells receives unpaired depolarizations (UDs) 10 seconds after each pairing. This arrangement reduces the overall probability with which depolarizations occurring in conjunction with presynaptic stimulation to 50%, but maintains the temporal contiguity between the EPSPs and depolarizations.

Is the LTP-memory connection now established enough to silence the sceptics? Rogan, Staubli and LeDoux (1997) point out several features common to fear conditioning and hippocampal LTP, including their dependence on NMDA receptors (Miserendino, Sananes, Melia and Davis, 1990; Gewirtz and Davis, 1997) Nevertheless, it remains to be shown that the mechanisms responsible for the behaviourally induced synaptic changes are the same as those underlying electrically induced LTP. But the new reports (Rogan, Staubli and LeDoux, 1997; McKernan and Shinnick-Gallagher, 1997) indicate that attempts to study LTP have not simply been an intellectual exercise, and that progress continues towards a comprehensive understanding of the mechanisms that underlie learning and memory.

In vivo recordings
We have previously shown that LTP induction in pathways that transmit auditory CS information to the lateral nucleus of the amygdala (LA) increases auditory-evoked field potentials in this nucleus7. Transmission of auditory information from the medial geniculate body to the lateral nucleus of the amygdala is believed to be involved in the conditioning of fear responses to acoustic stimuli. This pathway exhibits LTP of electrically evoked field potentials after high frequency stimulation of the medial geniculate body. High frequency stimulation of the medial geniculate body also results in a long-lasting potentiation of a field potential in the lateral amygdala elicited by a naturally transduced acoustic stimulus.

Rats were anaesthetized and implanted with a stainless-steel recording electrode (0.6 mW) in the LA, and a ground electrode in the skull, under aseptic surgical conditions.Now we show that fear conditioning alters auditory CS-evoked responses in LA in the same way as LTP induction. The changes parallel the acquisition of CS-elicited fear behaviour, are enduring, and do not occur if the CS and US remain unpaired. LTP-like associative processes thus occur during fear conditioning, and these may underlie the long-term associative plasticity that constitutes memory of the conditioning experience.v

To address this in vivo, Rogan, Staubli and LeDoux (1997) monitored the extracellular potential in the LA, in response to the CS tones while a rat was trained. As the CS and US were paired, and the animal learned to respond to the CS with a behavioural correlate of fear, the response in the LA to the CS alone grew, and remained at a high level. Further presentations of the CS alone extinguished the behavioural response (that is, the memory), and the auditory-evoked potential returned to baseline. Importantly, when the CS and US were unpaired - so no learning occurred - there was no significant growth in the auditory-evoked potential.

Genetic manipulations
Is it possible to impair fear conditioning by eliminating amygdaloid LTP genetically? Consistent with the deficit of titanic LTP in the amygdala, genetic KO models have been produced. The best one to date a RasGRF knockout mice engineered by Brambilla and colleagues (Brambilla et al, 1999). They found that mice that lack RasGRF exhibit deficits in both tetanic LTP in the amygdala and Pavlovian fear conditioning. Tetanic LTP in the BLA was characterized in vitro in brain slices obtained from wild-type mice and mice lacking RasGRF. Extracellular field potentials in BLA were elicited by electrical stimulation of the LA. High-frequency stimulation of LA induced a robust LTP of synaptic transmission in the BL of wild-type mice, but resulted in a rapidly decaying potentiation in mice that lack RasGRF.

Consistent with the deficit in tetanic LTP in the amygdala, mice lacking RasGRF exhibited deficits in long-term retention of Pavlovian fear conditioning. Conditional freezing to a tone that was paired with footshock and the context in which the tone-shock pairing occurred was impaired in RasGRF knockouts when they were tested 24 h after training. However, their short-term memory for the fear conditioning was intact when they were tested 30 min after training. This indicates that the deficit in long-term retention in mice lacking RasGRF was not the result of an inability of these mice to exhibit freezing behavior. Hippocampal LTP in RasGRF knockouts was normal. This study constitutes some of the most compelling evidence for the link between LTP and fear conditioning.

PROPOSED EXPERIMENT

The proposed study, to be conducted on the global and conditional knockout mice, would test the following hypothesis:

Hypothesis
If fear conditioning requires AMPA-mediated LTP, then LTP-deficient mice should fail to show fear conditioned responses.

Tests
1. Test for LTP in the thalamic medial geniculate nucleus (MGN) - basolateral amygdaloid complex (BLA) pathway.

2. Test for auditorily evoked fear responses in the potentiated startle paradigm.

3. Perform amygdalectomies to control for extra-amygdalar compensation

Possible outcomes if LTP is deficient:
1. Normal fear conditioning
2. Lack of fear conditioning

Both outcomes are compelling. If there is no fear conditioning or fear potentiated startle, this will almost conclusively be because of the lack of LTP. On the other hand, if fear conditioning is indistinguishable from the wild types, then some other mechanism and the argument put forward by LeDoux et al is demolished. Either way, the role of AMPA based LTP will be elucidated to a much greater extent than in previous studies.

Conditional knockouts
There is also a question about whether the amygdala is involved in induction or consolidation. With NMDA blockade there is no induction. Presumably there would be no learning either in the global KOs. However, with the inducible KOs, one could show that the same animal can learn the fear potentiated startle or the normal fear conditioning paradigms at one time, then omit doxycycline from their diet and see whether their ability is impaired.

Compensation by extra-amygdalar structures
In the event that fear conditioning is intact in the knockouts, an intriguing finding, one might attempt several explanations. One of these explanations would be compensation (especially in case of the global knockouts). In order to control for possible extra-hippocampal compensation, we would perform cytotoxic amygdalectomies on the GluR-A-/- mice and then test them fear conditioning. This would provide very strong evidence indeed for the LTP hypothesis whether the mice are able to learn about fearful events or not.