Rats were trained on a lateral amygdala-dependent auditory fear conditioning paradigm as previously described (Sehgal et al., 2014). Following auditory fear learning, all animals (other than the immediate shock group) were tested using a CS presentation in a novel context to assess the strength of auditory fear memory. The immediate shock (IS) group was tested in the original training context to assess the lack of context-US association. Electrophysiological recordings were made immediately following the recall test (Figures 1A,B). During the conditioning session, rats from all groups displayed low levels of baseline freezing (0.41 ± 0.14%) with no significant differences observed between groups [one way ANOVA, F (5, 31) = 1.8, p = 0.13; Figure 1C]. Following the baseline period, rats in all the conditioning groups displayed comparable levels of increased freezing while the control group, CS-alone, that received CS presentations in the absence of the US displayed low levels of freezing (repeated measures ANOVA, significant effect of training trial [F(5.9, 181.4) = 16.2, p < 0.001; Greenhouse-Geisser corrected] as well as group [F(5, 31) =4.7, p < 0.005], Figure 1C). Pairwise comparisons demonstrated that rats in the CS-alone group froze less during CS presentations than rats in all other groups except the Poor learners (all p values < 0.05). While all other groups displayed higher freezing with the training trial (p < 0.05), the CS-alone group did not display a significant increase in CS freezing over the course of conditioning. These data indicate that all animals that received paired presentations of the CS and US displayed successful acquisition of fear learning relative to control animals receiving CS presentations alone.

Similarly, during the testing session, the average baseline freezing was low across all groups (6.0 ± 2.1%, Figure 1D) and there were no significant differences between the groups [One way ANOVA, F (5, 30) = 1.07, p = 0.40]. Specifically, the immediate shock (IS) group, which was tested in the original training context to assess for the lack of context-US association, also displayed low baseline or context freezing (6.6 ± 3.32%). Therefore, as expected, the IS rats did not form a fearful association between the training context and the US presentation and are thus an appropriate control for the footshock presentation in the absence of learning.

Following the baseline period, rats in the fear conditioned groups (i.e., Cond 24 h, Cond 4d, and Cond 10d) displayed robust freezing to the CS presentation. Some rats displayed low levels of fear to the CS presentation (less than 50% freezing) and were separated into a Poor learner group. A one-way ANOVA revealed a significant effect of group on CS freezing [F(5, 28) = 117.7, p < 0.001]. Post hoc tests indicated that the control groups (CS-alone and immediate shock) as well as Poor learners displayed significantly lower CS freezing than all the conditioned groups (p < 0.001) with no significant differences within control or conditioned groups. Thus, the rats in the conditioned groups displayed robust fear memory up to 10 days following fear conditioning while the rats exposed to CS or US presentations alone displayed no fear learning.

To determine the time course of intrinsic excitability changes following fear conditioning, we used patch clamp whole-cell recordings to assess LA neuronal excitability. LA neurons were injected with 1 s long current pulses ranging from 0 to 500 pA and the number of action potentials (APs) elicited during the pulse were counted (Figure 2). A repeated-measures ANOVA indicated a significant effect of current [F(1.83, 606.7) = 1984.4, p < 0.001; Greenhouse-Geisser corrected], behavioral condition [F(7, 332) = 5.3, p < 0.001] as well as a group by current interaction [F(12.8, 606.7) = 5.589, p < 0.001; Greenhouse-Geisser corrected] on the number of APs elicited. Post hoc analyses revealed that LA neurons from Cond-1 h, Cond-24 h, Cond-4d as well as the IS group fired significantly more APs relative to neurons from naïve rats (p < 0.05). In addition, the number of APs elicited did not differ between naïve, CS-alone, Poor learners, and Cond-10d groups. Follow-up one-way ANOVAs indicated a significant effect of behavioral condition on the number of APs elicited for current injections ranging from 100 to 500 pA (all p values < 0.05). Post hoc comparisons confirmed that relative to LA neurons from naïve rats, LA neurons from Cond-1 h, Cond-24 h, Cond-4d and IS group fired more APs following a 300–500 pA current injection. Furthermore, post hoc comparisons also confirmed that relative to LA neurons from CS alone rats, LA neurons from Cond-1 h, Cond-4d and IS groups fired more APs following a 450–500 pA current injection (all p values < 0.05; for Cond-24 h, p = 0.09 and p = 0.03 for injections of 450 pA and 500 pA respectively). Finally, LA neurons from Cond-10d rats had excitability measures that were similar to naïve animals (all p values > 0.05) but significantly lower than other conditioned animals (Cond-1 h, Cond-24 h, and Cond-4d, all p values < 0.05) indicating that intrinsic excitability was back to naïve values 10 days later. These data demonstrate that successful acquisition of fear conditioning leads to a transient increase in LA neuronal excitability that returns to naïve levels by 10 days following fear conditioning. These changes in excitability were not observed in LA neurons from rats that failed to acquire fear conditioning (i.e., Poor learners). However, LA neurons from rats in the IS group also displayed increased intrinsic excitability indicating that shock presentation alone could lead to alterations in LA intrinsic excitability.

Although the neuronal response to a stepwise increase in somatic depolarization is widely used as a measure of neuronal excitability, subtle changes in input resistance, and in certain cases the cessation of AP firing following very high somatic depolarizations, can influence this measure of excitability. To obtain additional indices of intrinsic excitability, we also measured the maximum number of APs fired during current injections ranging from 0 to 500 pA (Figure 2C) and spike frequency adaptation (Figure 2D). Once more, a one-way ANOVA revealed a significant effect of behavioral group on the maximum number of APs elicited [F(7, 335) = 6.95, p < 0.001]. Post hoc tests confirmed that relative to LA neurons from naïve rats, those from IS, Cond-1 h. Cond-24 h and Cond-4d rats exhibited an increase in the maximum number of APs elicited (all p values < 0.05). No significant differences were found between the naïve and other control groups (i.e., CS-alone and Poor learners) as well as the Cond-10d group. Similar to our previous findings (Figure 2B), post hoc tests confirmed that LA neurons from IS, Cond-1 h, Cond-24 h and Cond-4d rats had an increased number of maximum APs (all p values < 0.05) relative to those from CS-alone group. These data confirm our earlier observation of a transient increase in intrinsic excitability within LA neurons of conditioned animals. In addition, we also demonstrated an increase in the intrinsic excitability of LA neurons from IS animals in the absence of learning.

To obtain a measure of spike frequency adaptation, a current injection sufficient to elicit 3 APs in the first 100-ms was injected over a period of 1 s and the number of APs elicited were counted (Figure 1D). Again, a one-way ANOVA demonstrated a significant effect of behavioral condition on the number of APs elicited [F(7, 321) = 3.5, p < 0.001]. Post hoc tests confirmed that LA neurons from Cond-1 h, Cond-24 h, Cond-4d and IS rats fired significantly more APs relative to LA neurons from naïve rats (all p values < 0.05). As before, no significant differences were found between the number of APs elicited in LA neurons from Poor learners, CS-alone and Cond-10d relative to naïve rats. These data demonstrate that (1) rats that successfully acquire fear conditioning display an increase in LA neuronal excitability relative to rats that do not learn, (2) these intrinsic excitability changes are transient, lasting up to 4 days before returning back to naïve levels by 10 days, and (3) LA neuronal intrinsic excitability may also increase following footshock US-alone presentations.

Changes in neuronal excitability can also result from changes in certain passive membrane properties (Table 1). We controlled for any effect of variations in resting membrane potential (RMP) by obtaining all our intrinsic excitability measurements at −60 mV. In addition, we did not observe any significant differences in RMP between groups [F(7, 336) = 1.0, p = 0.4]. Input resistance of the neuron can also affect the resulting depolarization following a somatic current injection. Like RMP, input resistance was also unchanged between various groups [F(7, 336) = 0.367, p = 0.9]. These data indicate that any changes in the intrinsic excitability of LA neurons observed following behavioral training did not result from changes in passive membrane properties. Furthermore, since changes in action potential properties can also accompany changes in the intrinsic excitability of a neuron, we measured single AP properties and found no significant differences in AP amplitude [F(7, 305) = 0.583, p = 0.8] or AP halfwidth among the behavioral groups [F(7, 305) = 0.502, p = 0.8]. Therefore, fear conditioning did not alter action potential amplitude or duration.

To investigate the relationship between intrinsic and synaptic plasticity following fear conditioning, we obtained measurements of synaptic strength as well as intrinsic excitability from the same neurons. Rats were fear conditioned as described before and control groups included naïve, CS-alone, or IS rats. After obtaining intrinsic excitability measurements under current clamp mode, neurons were voltage clamped at −70 mV and measures of synaptic strength were obtained.

Fear learning results in synaptic plasticity within amygdala neurons (McKernan and Shinnick-Gallagher, 1997; Clem and Huganir, 2010) and a well-established measure of synaptic plasticity is the paired pulse ratio (PPR). To measure the PPR, pairs of EPSCs were generated by temporally close presynaptic stimulation of thalamic fibers synapsing onto the LA neurons. Stimulation intensity was adjusted to obtain a reliable EPSC ranging from 50 to 150 pA. Baseline measurements were obtained for 1 min to establish that the EPSC amplitude remained stable. The interstimulus interval varied from 25 ms to 150 ms (Figure 3A). A repeated measures ANOVA revealed a significant effect of interstimulus interval [F(4.0, 460.9) = 13.9, p < 0.001; Greenhouse-Geisser corrected], group [F(5, 116) = 3.62, p < 0.005] as well as a group by time interaction [F(19.9, 460.9) = 1.65, p < 0.05; Greenhouse-Geisser corrected]. A follow-up one-way ANOVA demonstrated a significant effect of behavioral training on PPR at the shortest interstimulus interval (ISI), i.e., 25 ms [F(5,121) = 4.5, p < 0.001]. Post hoc tests revealed that while LA neurons from naïve rats displayed enhancement of the second EPSC (i.e., paired pulse facilitation), this facilitation was significantly attenuated in LA neurons from the Cond-1 h as well as Cond-24 h rats (Figure 3B, p < 0.05). Interestingly, IS rats were also significantly different from the naïve rats but in the opposite direction than that of conditioned animals (p < 0.05). In other words, the PPR was significantly higher for LA neurons from IS rats compared to those from naïve rats. No significant differences were observed between naïve, CS-alone and Cond-4d rats. These data as well as various other published reports (e.g., McKernan and Shinnick-Gallagher, 1997) demonstrate that fear learning is accompanied by a reduction in paired pulse ratio, a well-accepted measure of synaptic plasticity.

In addition to a change in paired pulse ratio at a short ISI of 25-ms, follow up ANOVAs demonstrated a significant effect or strong trend of behavioral group on PPR at various other time points (all p values < 0.08). Post hoc tests revealed that these effects were driven by a significant increase in PPR in LA neurons from IS rats relative to those from naïve rats (p < 0.05). Interestingly, in contrast to all other behavioral groups varying the interstimulus interval between the pairs of EPSCs had no significant effect on the PPR for the IS group [repeated measure; F (5, 50) = 1.822, p = 0.13]. Therefore, the synaptic plasticity observed in the IS group is more robust and in the opposite direction to that observed in the conditioned animals.

The primary aim of this analysis was to investigate whether intrinsic and synaptic plasticity were co-localized within the same neurons following fear conditioning. To determine this, we measured the relationship between PPR at a 25-ms ISI (PPR-25) and various measures of intrinsic excitability. Notably, PPR-25 was found to be modestly but significantly correlated with spike frequency adaptation, the number of APs elicited following current injections ranging from 50 to 500 pA, as well as maximum number of APs elicited (Spearman’s correlations = −0.181, −0.231 (300 pA), and −0.217 respectively; all p values < 0.05, Figure 4). This correlation was not dependent on the behavioral group and was also observable in LA neurons from naïve animals (Spearman’s correlation = −0.36, p < 0.05). Thus, a significant correlation exists between synaptic efficacy and intrinsic excitability which we analyzed more thoroughly below.

To further analyze the relationship between these two forms of plasticity, we split the neurons from Cond-1 h and Cond-24 h group into high (unchanged) and low (changed) PPR based on the median PPR-25 value for these groups (Figure 5). This is especially important as correlations are highly sensitive to outliers and non-normal distribution of data, and thus a median split allows us to better analyze the presence or absence of co-localization of these two forms of plasticity. The neurons in the low PPR or changed groups should be the ones that display reduced PPR because of fear learning.

As predicted, a one-way ANOVA revealed a significant effect of group on PPR [F (4, 91) = 9.4, p < 0.001]. The PPR was significantly reduced for Cond-1 h changed as well as Cond-24 h changed neurons but not the Cond-1 h unchanged or the Cond-24 h unchanged neurons (p < 0.001). More importantly, we next analyzed the measures of intrinsic excitability based on these new groups. One-way ANOVA revealed that the number of APs elicited following a series of current pulses ranging from 0 to 500 pA were significantly different between the groups. This effect emerged at current injections of 100 pA and remained significant up to 500 pA (all p values < 0.05). Interestingly, these results were driven by a significant increase in AP firing in Cond-1 h changed neurons [F(4, 91) = 6.4, p < 0.001] relative to neurons from the naïve group. No significant differences were found between Cond-1 h unchanged, Cond-24 h changed, Cond-24 h unchanged and naïve groups. Relative to neurons from the naïve group, the maximum number of APs elicited during these current steps were also significantly increased for Cond-1 h changed but not any of the other groups (p < 0.001). A one-way ANOVA revealed a strong trend for a similar effect on spike frequency adaptation [F(4, 91) = 2.26, p = 0.07]. Post hoc tests confirmed that LA neurons from Cond-1 h changed group were more excitable relative to neurons from the other groups (p < 0.005). These data demonstrate that within LA neurons, potentiation of intrinsic excitability is co-localized with synaptic plasticity, immediately but not 24 h following fear conditioning.

We further substantiated this analysis by classifying neurons as “changed” when the PPR-25 was two standard deviations below the mean Naïve PPR-25. This method of detecting changed neurons resulted in similar conclusions where PPR was reduced for Cond-1 h changed as well as Cond-24 h changed neurons but not the Cond-1 h unchanged or the Cond-24 h unchanged neurons [F(4, 91) = 11.0, p < 0.001, post hoc p < 0.001]. Measures of intrinsic excitability were only increased in Cond-1 h changed group but not in other groups [Number of spikes for current injections: F_interaction (40, 910) = 4.1, for 100–400 pA all p values < 0.05; Maximum APs: F(4, 91) = 5.6, p < 0.001]; Adaptation: [F(4, 91) = 2.8, p < 0.05, all post hoc p values < 0.05]. Therefore, our conclusion that potentiation of intrinsic excitability, as well as synaptic efficacy, is co-localized immediately following fear conditioning is robust and does not depend on our method of analysis.

As previously described, a majority of studies investigating the time course of intrinsic plasticity demonstrate that it is transient, lasting less than 7–14 days (Moyer et al., 1996; Saar et al., 1998; Zelcer et al., 2006; Motanis et al., 2014). However, long lasting changes in the intrinsic excitability of cerebellar Purkinje neurons following eyeblink conditioning can last for up to 1 month (Schreurs et al., 1998). Therefore, it is possible that the persistence of learning-related intrinsic plasticity is different between structures such as the hippocampus and the amygdala especially given their differential role in memory processes over the longer term (Kim and Fanselow, 1992; Do-Monte et al., 2015).

Here we found that changes in LA neuronal excitability following fear conditioning were transient and were back to naïve levels by 10 days following fear conditioning. This is in accordance with data from olfactory discrimination learning studies where changes in neuronal excitability within basolateral amygdala started to reverse by 3 days following learning (Motanis et al., 2014). Thus, intrinsic plasticity following behavioral training within the amygdala appears to be transient. The transient nature of such learning-related plasticity can be informative. It follows that any effect of amygdala intrinsic plasticity on future plasticity, synaptic or behavioral is likely to be transient as well. Indeed, transient increases in intrinsic excitability can impact memory organization and linking via such time-dependent effects (Sehgal et al., 2013, 2018; Cai et al., 2016; Rashid et al., 2016; Lisman et al., 2018).

Retrieval of a context or contextual fear memory can result in a transient increase of neuronal excitability in dentate gyrus neurons (Pignatelli et al., 2019). Since a majority of conditioned animals (i.e., Cond-24 h, Cond-4d, and Cond-10d) in our study also received a retrieval session immediately before slice preparation, it is possible that the enhanced excitability in these groups can result from conditioning or the subsequent retrieval session. Incorporation of a retrieval session is an experimental design that is commonly employed to establish that changes in excitability do not reflect changes in memory recall and allows for the correlation of behavioral performance with observed intrinsic plasticity (for example, Moyer et al., 1996; Cohen-Matsliah et al., 2009; Kaczorowski and Disterhoft, 2009; Song et al., 2012, 2015; Sehgal et al., 2014). We previously demonstrated that LA neurons display increased intrinsic excitability 24-h post-fear conditioning with and without a retrieval session (Sehgal et al., 2014). Therefore, fear conditioning itself (without a retrieval session) is sufficient to increase intrinsic excitability up to 24 h following fear conditioning. In addition, the increase in excitability is no longer observed despite the use of a retrieval session at 10 days post-conditioning. We cannot rule out that the enhanced excitability observed in the Cond-4d group is a result of retrieval and not the conditioning session—an important limitation to our current findings. Therefore, we conclude that even though retrieval sessions may increase excitability up to 4 days post conditioning, intrinsic excitability is no longer enhanced 10 days post-conditioning, and retrieval sessions at this time point do not increase neuronal excitability. Encoding as well as retrieval of an auditory fear memory allows linking of a subsequently formed auditory fear memory to the orginal fear memory indicating that the retrieval event is sufficient to reengage plasticty mechanisms necessary for this process (Rashid et al., 2016). Our data indicate that any behavioral effects mediated by increased excitability following memory retrieval as described in previous studies (Rashid et al., 2016; Pignatelli et al., 2019) are also transient.

To control for the effect of US presentations on LA intrinsic excitability, we used an immediate shock deficit paradigm. The phenomenon of immediate shock deficit is based on the principle that if a rodent is shocked immediately after being placed into the chamber (e.g., within 6 s), the rat displays no fear of context on a subsequent day as there was no time for a context-shock association (Fanselow, 1986). It is possible to rescue this immediate shock deficit by pre-exposure to the context the day before shock presentation (Fanselow, 1990). This allows the rats to form a context representation, indicating that immediate shock deficit is a context processing deficit (but see Lattal and Abel, 2001; Landeira-Fernandez et al., 2006). In the current study, rats were presented with the same duration of footshock as other conditioned groups (10s) immediately upon being placed in the conditioning chamber and removed from the chamber as soon as the shock presentation terminated.

We found that IS rats displayed little freezing upon exposure to the conditioning context the following day. Surprisingly, LA neurons from the IS rats did show a robust increase in neuronal excitability. The lack of any other neurophysiological data using an IS group restricts our interpretation of these data. First, it is possible that our observed changes in neuronal excitability within LA neurons are due to shock presentation rather than learning itself. We think this is unlikely as the Poor learners (animals that received conditioning but displayed little or no fear to the tone the following day) display no change in excitability relative to LA neurons from naïve rats. This demonstrates that learning, irrespective of stimulus presentation, is necessary for increased excitability.

Furthermore, repeated footshock US presentations in the absence of learning are a rodent model for stress (Valenti et al., 2011) that can result in plasticity within LA neurons. Indeed, chronic but not acute stress leads to enhanced excitability of LA neurons (Rosenkranz et al., 2010). However, it is unlikely that our observed conditioning-related changes in LA neuronal excitability result from a stress response for several reasons. Most important among these are that a single session of fear conditioning or immediate shock presentations is akin to acute stress which does not lead to changes in LA excitability. Additionally, the degree of intrinsic plasticity following chronic stress (reduced RMP, increased input resistance, reduced sAHP and spike-frequency adaptation) was more extensive than that seen in the current study.

The second explanation for excitability changes in the IS animals is that even though IS rats do not associate conditioning context with shock presentation, animals do learn to associate other cues (e.g., transport, handler, etc.) with the shock (Rudy et al., 2002). It is possible that such context learning is driving our observed changes in neuronal excitability. The most popular control for US presentation, namely pseudo-conditioning (or explicitly unpaired CS and US presentations) leads to inhibitory conditioning to the CS, i.e., animals receiving unpaired CS and US presentations encode CS as a safety signal (Rescorla and Lolordo, 1965) and display distinct forms of plasticity within amygdala (Amano et al., 2010). In addition, unpaired animals show higher context fear than those receiving paired CS-US presentations (Phillips and LeDoux, 1994). These factors make pseudo-conditioning a poor control for studies investigating learning-related neurophysiological changes in amygdala. Together, US-alone presentations (context conditioning), pseudo-conditioning as well as immediate shock paradigm make for inadequate controls to establish the effect of US presentation in the absence of learning on amygdala neurophysiology. We believe that an extended investigation of the behavioral correlates as well as neurophysiological changes in the IS group is an important area of future research. It is likely that changes in excitatory (Figure 3) and inhibitory transmission (Saar et al., 2012; Kfir et al., 2014) together with excitability changes sculpt lateral amygdala microcircuits in distinct ways following fear conditioning and immediate shock.

The long-delay fear conditioning paradigm used here produces robust fear learning. However, we found that a few rats tested at different time points (2 for Cond-24 h, 1 each of Cond-4d and Cond-10d) displayed low levels of fear to the CS presentation indicating poor retention of memory. We found that these same rats also failed to modulate LA intrinsic excitability. It is possible that these rats had much lower basal levels of LA neuronal intrinsic excitability preventing the successful acquisition of fear learning. Several studies have demonstrated that learning performance and learning-induced changes in intrinsic excitability are correlated (Cohen-Matsliah et al., 2009; Song et al., 2012). Animals that fail to learn in trace fear conditioning or olfactory discrimination task fail to modulate neuronal excitability in the hippocampus and piriform cortex, respectively, (Cohen-Matsliah et al., 2009; Song et al., 2012). Aberrant intrinsic plasticity may also underlie aging-related learning deficits where aging animals that fail to acquire a task also fail to modulate neuronal intrinsic excitability (Moyer et al., 2000; Kaczorowski et al., 2012). Regardless, we found that intrinsic excitability levels are a good index of learning making modulation of intrinsic excitability a good target for therapeutic interventions.

It is well documented that memory formation during auditory fear conditioning involves synaptic plasticity within LA neurons (McKernan and Shinnick-Gallagher, 1997; Clem and Huganir, 2010). Paired pulse ratio (PPR) is one such way to measure experience-dependent synaptic plasticity. Although previously considered a form of short-term presynaptic plasticity that reflects changes in release probability, myriad other cellular mechanisms can influence PPR (Glasgow et al., 2019). In LA neurons from naïve animals, a short interstimulus interval (ISI; e.g., 25 ms) usually results in paired-pulse facilitation (McKernan and Shinnick-Gallagher, 1997). As the ISI between pairs of presynaptic stimulation increases, the PPR is reduced. We demonstrate that long-delay fear conditioning is accompanied by a reduction in paired pulse ratio immediately as well as 24 h following fear conditioning (Figure 3). These changes were no longer evident 4 days following fear conditioning and PPR was back to naïve levels at this time point. These data are in accordance with other reports that fear learning is accompanied by a reduction in PPR at the synapses carrying auditory information onto the LA neurons.

In addition to the reduced PPR observed in fear conditioned rats, LA neurons from IS animals exhibited a robust increase in PPR. This increase appears to be independent of the ISI as varying the ISI did not seem to affect the PPR (Figure 3A). Therefore, footshock US presentations alone results in plasticity at thalamo-amygdala synapses. As with the changes in intrinsic excitability, it is possible that these synaptic changes in the IS group could be due to the shock presentation itself or due to learning.

The current study sought to measure synaptic and intrinsic plasticity following fear learning within the same neurons. We found that fear conditioning is accompanied by a reduction in PPR at the thalamo-amygdala synapses immediately as well as 24 h following fear conditioning. These changes were no longer evident 4 days later indicating the transient nature of this form of synaptic plasticity. In addition, we observed robust changes in PPR in the immediate shock deficit group, but the direction of these changes was reversed such that LA neurons from the IS group had a larger mean PPR than LA neurons from naïve rats. More importantly, in the LA neurons from conditioned animals these changes were co-localized with the intrinsic excitability changes immediately following fear conditioning. These data support our hypothesis that intrinsic and synaptic plasticity is colocalized within lateral amygdala neurons following fear conditioning.

To investigate whether there is a relationship between intrinsic and synaptic plasticity, we looked to see whether these measures were correlated with each other. We found that intrinsic excitability and PPR were negatively correlated such that a lower PPR was associated with increased intrinsic excitability (Figure 4). In other words, synaptic changes that emerged following fear conditioning predicted intrinsic excitability changes. Specifically, neurons exhibiting the largest reduction in PPR immediately following conditioning were also the most excitable (Figure 5). This demonstrates that synaptic and intrinsic excitability following fear conditioning are co-localized to the same neurons. Interestingly, although both synaptic and intrinsic plasticity were significantly altered 24-h post conditioning, a significant correlation between these forms of plasticity was not observed at this time point. These data indicate a time-dependent shift in the relationship between intrinsic and synaptic plasticity within LA neurons following fear conditioning.

We have previously demonstrated that fear learning results in increased intrinsic excitability in ~32% of lateral amygdala neurons (Sehgal et al., 2014). These estimates are in agreement with the estimates of the proportion of lateral amygdala neurons recruited during fear conditioning (Rumpel et al., 2005; Han et al., 2007, 2009; Reijmers et al., 2007). In recent years, various genetic strategies have allowed a more precise understanding of electrophysiological changes in recently activated neurons during learning and memory formation (Cai et al., 2016; Crestani et al., 2019; Pignatelli et al., 2019; Chen et al., 2023). The use of similar genetic strategies would allow us to target neurons recruited into memory formation more precisely in future studies. However, there are some challenges to adapting these methods for our current experiments, such as the feasibility of using viral vectors in rats and the fact that neuronal labeling takes hours and therefore does not allow access to the neurons immediately after the learning event (e.g., Cond-1 h group). In addition, any particular labeling strategy (for example, those mediated by cFos, Arc, or Npas-4 expression) might not allow access to the complete neuronal ensemble underlying a particular memory (Sun et al., 2020) resulting in inaccurate conclusions. Nonetheless, the use of such genetic strategies will greatly accelerate our understanding of plasticity within neuronal ensembles underlying learning in the future.

Taken together, the current experiments demonstrate that intrinsic and synaptic plasticity in the amygdala following fear conditioning are transient and can be colocalized to the same neurons. Based on the current models of memory allocation, these data predict that future amygdala-dependent memories should be colocalized to the same neuron undergoing intrinsic plasticity albeit in a time-dependent manner.

Adult male Sprague Dawley rats (~3 months) were individually housed in clear plastic cages. Rats were maintained on a 14 h light/10 h dark cycle (lights on at 7 a.m.) with unlimited access to both water and standard laboratory rat chow (Harlan Laboratories). All procedures were conducted in accordance with the University of Wisconsin-Milwaukee animal care and use committee (ACUC) and NIH guidelines.

Brain slices were prepared by an individual blind to the training condition. Rats were deeply anesthetized with isoflurane, perfused with ice-cold oxygenated sucrose aCSF (composition in mM: 2.8 KCl, 1.25 NaH₂PO₄, 2 MgSO₄, 1 CaCl₂, 1 MgCl₂ 26 NaHCO₃, 248 Sucrose and 10 dextrose) and decapitated. The brain was quickly removed and placed in ice-cold oxygenated (95% O₂/5% CO₂) aCSF (composition in mM: 124 NaCl, 2.8 KCl, 1.25 NaH₂PO₄, 2 MgSO₄, 2 CaCl₂, 26 NaHCO₃, and 20 dextrose). The brain was then blocked and 400 μm-thick coronal brain slices were cut in aCSF at ~0°C using a vibrating microtome (VT1200, Leica). Only slices that were located between 1.88 and 3.30 mm posterior to bregma, thus containing lateral amygdala, were used (Paxinos and Watson, 1998). Slices were then transferred to a holding chamber (Moyer and Brown, 1998) containing oxygenated aCSF at 32–36°C and allowed to recover for 30 min. The slices were then kept at room temperature (21–23°C) for an additional 30 min before any electrophysiological recordings began.

For electrophysiological measurements, slices were transferred as needed to the recording chamber, where they were perfused with oxygenated aCSF at 32°C. The slices were held in place using nylon strands stretched across a U-shaped platinum wire. Visually-guided whole cell patch clamp recordings were made using infrared differential interference contrast optics. All recordings were obtained using a MultiClamp 700B amplifier system (Molecular Devices, Union City, CA). Experiments were controlled by PClamp 10 software running on a PC, and the data were acquired using the Digidata 1440A acquisition system. All recording electrodes (3–8 MΩ) were pulled from thin-walled capillary glass (A-M Systems, Carlsborg, WA) using a Sutter Instruments P97 puller. The patch pipettes were filled with internal solution containing (in mM) 110 K-gluconate, 20 KCl, 10 Di-Tris-P-Creatine, 10 HEPES, 2 MgCl₂, 2 Na₂ATP, 0.3 Na₂GTP, 0.2% Biocytin, with a pH of 7.3 and osmolarity of 290 mOsm. Only cells with a stable, uncorrected resting membrane potential (RMP) between −50 and −85 mV, overshooting action potentials, and an input resistance (R _N ) >100 MΩ were used. Figure 2A shows a photograph of a coronal slice containing lateral amygdala with the typical recording site noted.

Neurons were filled with biocytin to confirm the position and identity of pyramidal cells in LA. After obtaining electrophysiological measurements, slices were fixed in 10% neutral-buffered formalin at 4°C for 1 to 7 days before further processing. To visualize LA neurons, slices were incubated in 3% H₂O₂/10% methanol for 45 min, washed with PBS for 10 min (3X), followed by 0.25% Triton X-100/ 2% BSA for 60 min. The slices were then incubated with 1:500 streptavidin Alexa Fluor 488 (Invitrogen) for 120 min in the dark and washed with PBS for 10 min (3X). They were mounted onto slides, coverslipped with Ultra Cruz Mounting Medium (Santa Cruz Biotechnology, Santa Cruz, CA), and sealed with nail polish. The neurons were viewed and photographed using a fluorescence microscope (BX51WI, Olympus) at 20X or Olympus Fluoview FV1200 confocal system. Neurons were classified as pyramidal when either a prominent apical dendrite, large soma or dendritic spines were detected. Representative biocytin-filled LA pyramidal neurons are shown in Figure 2A.

The main effects and post hoc comparisons were analyzed using ANOVA. Additionally, Pearson’s and Spearman’s correlation coefficients were calculated to determine whether post-training intrinsic excitability was correlated with post-training synaptic plasticity. All data were expressed as mean ± SEM. All statistical analyses were performed on cells.