The neural circuits underlying classical fear conditioning are well characterized (LeDoux, 2000, 2007; Pare et al., 2004; Maren, 2005; Lang and Davis, 2006). In classical fear conditioning a previously neutral stimulus comes to trigger a behavioral fear response. For this to occur, two events become persistently associated through associative memory formation. Fear conditioning requires the coordinated presentation or pairing of two sensory inputs. One input is an initially neutral stimulus, which becomes the conditioned stimulus (CS). The other input is a biologically significant event, the unconditioned stimulus (US). The pairing of these stimuli results in a conditioned response (CR) such that the initially neutral stimulus (CS) now elicits the same behavioral response as the US.

The neurobiology of fear conditioning is best characterized for an auditory fear-conditioning task employing an acoustic CS and a nociceptive US. These stimuli converge in the lateral amygdala (LA; see Romanski et al., 1993; Johnson et al., 2009 for review). The acoustic signal conveying the tone CS enters the LA via the auditory thalamus (subcortical) and the auditory association cortex (Romanski et al., 1993; LeDoux, 2003). Both of these pathways enter the LA, where they converge with US signals. Additionally, both pathways converge onto single LA neurons on adjacent spines (Humeau et al., 2005) and both pathways undergo plasticity during fear conditioning (Mckernan and Shinnick-Gallagher, 1997; Doyere et al., 2003; Shin et al., 2006). The two auditory routes, are generally thought to provide different aspects of the CS to the LA, with the subcortical (thalamus direct to amygdala) input, coined the “low road,” providing a rudimentary version of the CS and the cortical input, coined the “high road,” providing more detail (LeDoux, 1994, 2000).

Stress and arousal are known to activate norepinephrine (NE) release in the brain (Aston-Jones and Bloom, 1981; Mcgaugh, 2000). NE is especially important to conditioned fear because fear learning occurs during times of threat and stress. Thus, NE release may modulate the neural substrates of conditioned fear (Huang et al., 2000; Tully et al., 2007). NE, especially its β receptors, has been implicated in the formation, reconsolidation, and extinction of fear memories, and has been proposed as a potential treatment for PTSD (Berlau and Mcgaugh, 2006; Debiec and LeDoux, 2006; Brunet et al., 2008). Recent data by Tully et al. (2007) shows that NE promotes the formation of long term potentiation (LTP), a cellular model of memory in the thalamo-amygdala pathway. Despite the psychopharmacological data in animals and the therapeutic potential in humans, relatively little is known about the underlying network and cellular mechanism in the LA. A key question is whether the two pathways are differentially regulated by NE. A differential regulation could shift the balance between the “high” and “low” roads of sensory input to the amygdala fear conditioning circuit.

In order to screen for possible network effects of NE and its receptors we used in vitro field potential recordings which would allow us to detect changes in the synaptic activity as well as excitability of the network as a whole. We find NE and its β receptors differentially regulate synaptic transmission between the cortical and subcortical pathways to the LA. Differences in synaptic transmission are dependent upon GABA receptors in the LA. Together these data provide a mechanism by which increased NE signaling in LA, activated by stress and arousal, may critically shift the balance in synaptic transmission toward the faster subcortical afferents and thus promote survival.

In order to record the electrophysiological, excitatory, or inhibitory, response of LA neurons to NE and its β receptors, we made extracellular field potential recordings. We measured the evoked field excitatory potentials synaptic potential (fEPSP) in response to evoked activation of the subcortical (or thalamic) and cortical inputs pathways to the LA (Weisskopf and LeDoux, 1997).

Physiological Recordings in vitro: Coronal brain slices containing the LA were prepared from Sprague-Dawley rats aged 4–7 weeks. Once deeply anesthetized with ketamine and xylazine the chest was exposed to reveal the heart. A sharp 26-gage needle was placed in the left ventricle, an incision made in the right atrium, and 5–10 ml of ice cold oxygenated artificial cerebro-spinal fluid (ACSF) was injected in less than 1 min. The animal was then quickly decapitated and the brain removed. Chilled brains were blocked into coronal sections containing the entire amygdala and then sliced in hemi-coronal sections on a vibratome to 400 μm. Brain slices were warmed to 32°C for 30 min and then slowly lowered to room temperature and maintained for several hours. Slices were transferred to a recording chamber, flow rate 2.5 ml/min, containing ACSF (in mM) 115 NaCl, 3.3 KCl, 1 MgSO₄, 2 CaCl₂, 25.5 NaHCO₃, 1.2 NaH₂PO₄, 5 lactic acid, and 25 glucose, equilibrated with 95% O₂/5% CO₂.

Slices and neurons were visualized with an upright fixed stage microscope equipped with infrared differential interference contrast optics. Recording electrodes of 5–10 MΩ (Lamprecht et al., 2006) filled with ACSF, were guided to dorsolateral amygdala (LAd) neurons. Two bipolar stimulating electrodes were placed in the brain slice to stimulate auditory subcortical (thalamic) and cortical afferents to the LA (Weisskopf et al., 1999). The subcortical afferents electrode was placed medial to the LA and lateral to stria terminalis, where dorsal to the central nucleus of the amygdala, thalamus to amygdala fibers course. The cortical afferents electrode was placed dorsal to the LA on the external capsule/cortical border (see LeDoux et al., 1990; Romanski et al., 1993 for details of anatomical input to LA). Current was passed at 0–0.2 mA to activate suprathreshold EPSPs and IPSPs. Stimulation was controlled by Axon software (see Weisskopf and LeDoux, 1997 for further description). Voltage recordings were amplified using an AxoClamp 2B amplifier, signals filtered at 3 kHz and digitized at 5 kHz using an Axon analog to digital converter and analyzed off line with pClamp. Drugs (Norepinephrine, Isoproterenol, PTX, Sigma, MO, USA) were added to the superfused ACSF. All drugs were superfused onto slice with a flow rate of 2.5 ml/min. Data was also analysis using Excel (Microsoft, WA, USA) and statistical tests were made with GraphPad, Prism. Data are presented as mean ± SEM. A probability level of <0.05 was considered significant.

We evoked extracellular field potentials in the LA in response to stimulation of both cortical and subcortical pathways (Weisskopf and LeDoux, 1999) in order to directly compare (thalamic) both afferent responses (Figure 1A) to NE in LA. Potentials were sequentially activated, with a suitable separation latency, which allowed for direct comparison of potentials. Extracellular evoked fEPSP amplitudes were measured as previously described (Lamprecht et al., 2006; Johnson et al., 2008, 2009; see also Huang et al., 2000). The fEPSP in both pathways ranged in size from 0.2 to 0.6 mV without picrotoxin (PTX) and from 0.5 to 1.2 mV in the presence of PTX (Figure 1B). Differences in amplitude in fEPSP between the pathways at baseline were not observed (Lamprecht et al., 2006). Stimulation of both pathways was reduced to bring the fEPSP to approximately 50% of maximum amplitude. Recordings were stabilized prior to applications of drugs. We found that NE differentially regulates synaptic transmission in the subcortical and cortical pathways to the LA. Five key findings are described below.

The amygdala has been extensively implicated in the neurobiology of conditioned fear (LeDoux, 2000; Sah et al., 2003; Pare et al., 2004). NE, especially its β receptors, has been implicated in formation, reconsolidation, and extinction of fear memories (Debiec and LeDoux, 2004; Berlau and Mcgaugh, 2006; Tully et al., 2007) and is known to be a mechanism of aspects of stress and arousal in the brain (Aston-Jones et al., 1996; Mcgaugh, 2000). The LA receives subcortical and cortical sensory afferents and both have been implicated in conditioned fear (Mckernan and Shinnick-Gallagher, 1997; Doyere et al., 2003; Shin et al., 2006). Here we asked whether NE mediates these effects at afferent input synapses, and if so, what mechanism is involved. We recorded evoked field potential (EFPs) following stimulation of subcortical and cortical afferents in vitro in order to screen for network effects. NE transiently inhibited EFPs in both pathways. However, inhibition was greater in the cortical path. In the presence of the GABA_A antagonist PTX, inhibition by NE remained in both pathways but was reduced in amplitude. Importantly the difference in the effects of NE on the two pathways was eliminated, indicating that GABA networks contribute to the NE effects in LA and that NE also regulates the cortical evoked GABA network to a greater extent. Application of the β adrenergic agonist ISO potentiated EFPs in both pathways. ISO produced a greater potentiation of the cortical path. Importantly this difference was removed by PTX, suggesting that an additional potentiation may occur via a separate cortical pathway activated GABA network in LA.

Previous studies have shown that NE has a potent effect on synaptic transmission in amygdala. Consistent with our findings, Braga et al. (2004) found that NE reduces the amplitude of the fEPSP in the external capsule input to basolateral amygdala (BLA). Additionally they found that IPSC frequency was potentiated by NE (Braga et al., 2004). Consistent with an apparent reduction in excitatory transmission, Delaney et al. (2007) found that NE dramatically reduced EPSC amplitude in the parabrachial nucleus inputs to central amygdala pathway. The mechanism involved presynaptic α₂ receptors, which inhibits transmitter release. Taken together, these studies in amygdala show multiple mechanisms by which NE can reduce local network excitability, acting at both glutamatergic and GABAergic transmitter systems (Figure 4). The response to NE and the response to the NE β receptor agonist (ISO) are different. NE results in an overall decrease in the field amplitude whereas ISO results in an increase. The NE response occurs only during the drug, whereas the response to ISO continues after the drug. This differential response between NE and the β receptor agonist has been previously described (for example see Huang et al., 2000). The implication from previous data is that NE has a short lasting (during drug) inhibitory effect, whereas the β agonist ISO has a long lasting (beyond drug) effect that is permissive on plasticity (Huang et al., 2000). The difference we observed between the subcortical and cortical pathways suggest that different networks of GABA neurons mediate NE regulation of the two pathways. We propose that an additional local GABA circuit is activated by the cortical pathway, which provides further inhibition to cortically derived sensory input during stress and arousal (Figure 5).

In order to directly compare cortical and subcortical afferent responses to NE, we evoked extracellular field potentials in the LA in response to stimulation of both the pathways. These pathways have previously been recognized to contain the subcortical and cortical pathways, including the auditory medial geniculate and auditory TE3 cortical paths to the LA (LeDoux et al., 1990; Doron and LeDoux, 1999; Weisskopf and LeDoux, 1999; Doyere et al., 2003; Humeau et al., 2005; Shin et al., 2006; Johnson et al., 2009). We measured extracellular evoked fEPSP as previously described (Huang et al., 2000; Lamprecht et al., 2006). These pathways have previously been shown to be independent from each other with respect to independently induced monosynaptic plasticity (Weisskopf and LeDoux, 1997; Lamprecht et al., 2006). By using the fEPSP we were able to ascertain overall excitability of the LA network evoked by each of the pathways (Huang and Kandel, 1996; Collins and Pare, 2000; Huang et al., 2000; Shinnick-Gallagher et al., 2003).

In vivo the input to the LA via the cortex, arrives some 30 ms later to provide more sensory detail (Armony et al., 1995, 1998; Quirk et al., 1995; Li et al., 1996; LeDoux, 2000). Either of thalamo-LA or cortico-LA pathways is sufficient for conditioning to simple auditory stimuli. However the cortical pathway appears to be required for learning about more complex stimuli (LeDoux, 2000, 2007). One interpretation of the basis for these two anatomically distinct pathways is that the direct thalamic (subcortical) route to the amygdala can provide a fast, and possibly life saving, sensory signal to warn the amygdala of potential danger (LeDoux, 2000). The two sensory pathways both arrive in the LA, and both synapse onto LA principal neurons where they contact different types of dendritic spines (Humeau et al., 2005). While there is spatial convergence of these sensory inputs, there is also temporal divergence with the two inputs being separated in time. One possible mechanism that may allow for the two temporally segregated sensory inputs to converge in time as well as in space is a recurrent network in the LA. This network may allow for thalamo-LA signals to feedback to the superior parts of the LA during conditioning where they will meet incoming cortical signals (Johnson et al., 2008). A reduction in excitability in the NE pathway could reduce feedback within the LA and create a disconnect with the cortical pathway. These effects may explain why activation of β receptors alone appears to promote plasticity in the cortical pathway.

In studies of the subcortical to LA pathway, Tully et al. (2007; Tully and Bolshakov, 2010) found NE reduced IPSP amplitude. They found that NE increases network excitability, which could then promote the induction of LTP. The NE β antagonist propranolol has previously been shown to inhibit the maintenance of LTP in the cortical pathway (Huang et al., 2000). While NE alone, acting on all its receptors, resulted in a decrease in excitability, activation of β receptors alone resulted in a greater potentiation of the cortical pathway. These β receptor data strongly suggest that plasticity is favored in the cortical pathway compared to the subcortical pathway. This interpretation is consistent with the known, effects of β receptors on synaptic plasticity in the cortical pathway (Huang et al., 2000). Differences in plasticity between the cortical and subcortical pathways is an important question for future studies.

In conclusion, NE may regulate memory formation in the LA with both temporal and synapse precision. These results may indicate that the differential cortical and subcortical regulation helps protect mammals under times of stress. Activation of the subcortical “low road” (LeDoux, 2000) is hypothesized to allow rapid sensory access to the LA in order to initiate defensive amygdala dependent fear responses. These data show that while NE inhibits synaptic transmission in both pathways the balance between the two is shifted toward the subcortical low road. The cortical “high road” (LeDoux, 2000) is hypothesized to allow refined sensory input to the LA in order to better identify threatening stimuli. Our findings that activation of the NE β receptor potentiates the cortical pathway more than the subcortical pathway may allow increase plasticity and hence learning about dangerous stimuli at the cortical pathway under times of stress and arousal, while at the same time preventing learning less accurate non-cortically processed sensory information in the subcortical pathway. Further work on differential plasticity and regulation between the subcortical and cortical pathways is needed to confirm these ideas.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.