Abstract
Activity in the basolateral amygdala complex (BLA) is needed to encode fears acquired through contact with both innate sources of danger (i.e., things that are painful) and learned sources of danger (e.g., being threatened with a gun). However, within the BLA, the molecular processes required to consolidate the two types of fear are not the same: protein synthesis is needed to consolidate the first type of fear (so-called first-order fear) but not the latter (so-called second-order fear). The present study examined why first- and second-order fears differ in this respect. Specifically, it used a range of conditioning protocols in male and female rats, and assessed the effects of a BLA infusion of the protein synthesis inhibitor, cycloheximide, on first- and second-order conditioned fear. The results revealed that the differential protein synthesis requirements for consolidation of first- and second-order fears reflect differences in what is learned in each case. Protein synthesis in the BLA is needed to consolidate fears that result from encoding of relations between stimuli in the environment (stimulus–stimulus associations, typical for first-order fear) but is not needed to consolidate fears that form when environmental stimuli associate directly with fear responses emitted by the animal (stimulus–response associations, typical for second-order fear). Thus, the substrates of Pavlovian fear conditioning in the BLA depend on the way that the environment impinges upon the animal. This is discussed with respect to theories of amygdala function in Pavlovian fear conditioning, and ways in which stimulus–response associations might be consolidated in the brain.
Significance Statement
Protein synthesis in the basolateral amygdala complex (BLA) is widely considered to be essential for the consolidation of fear memories. The present findings are significant in showing that this is not the case. Specifically, our findings show that protein synthesis in the BLA is required for consolidating associations between stimuli that form part of a dangerous experience (i.e., stimulus–stimulus [S–S] associations). It is not, however, required for consolidating associations between environmental stimuli and fear responses emitted by an animal (i.e., stimulus–fear response [S–R] associations). More generally, these findings are significant in showing that the substrates of learning and memory in the mammalian brain are not fixed and immutable. Instead, these substrates vary depending on what is learned.
Introduction
Rats quickly learn about a stimulus (e.g., sound) paired with an aversive unconditioned stimulus (US), typically foot shock. They express this learning when re-exposed to the now conditioned auditory stimulus (S1) in defensive responses (e.g., freezing, potentiated startle) indicative of fear or threat in people (Davis, 1992; Fanselow, 1998; McNally and Westbrook, 2006; LeDoux, 2014). Rats also quickly learn about a stimulus (e.g., light) paired with the already-conditioned sound, again expressing this learning when re-exposed to the now conditioned light (S2) in freezing responses and potentiated startle (Rizley and Rescorla, 1972; Parkes and Westbrook, 2010; Holmes et al., 2014; Michalscheck et al., 2021). Because the light was never paired with the US but rather with the sound that had been so-paired, Pavlov (1927) thought of the sound and light as undergoing first- and second-order conditioning, respectively.
First- and second-order fears require activity in the basolateral amygdala complex (BLA) for their acquisition and retrieval/expression in defensive responses (Gewirtz and Davis, 1997; Parkes and Westbrook, 2010; Holmes et al., 2013). However, the two types of fear differ with respect to the processes required for their consolidation. For example, whereas consolidation of first-order fear requires the synthesis of new proteins in the BLA (Schafe and LeDoux, 2000; Maren et al., 2003), consolidation of second-order fear occurs independently of this synthesis (Lay et al., 2018; Leidl et al., 2018; Williams-Spooner et al., 2019). That is, rats that receive a BLA infusion of a protein synthesis inhibitor (e.g., cycloheximide) immediately after a session containing S1–shock pairings freeze less to the S1 at test than vehicle-infused controls. By contrast, rats that receive a BLA cycloheximide infusion immediately after a session containing pairings of S2 and an already-conditioned S1 freeze just as much to S2 at test as vehicle-infused controls. Such findings have been taken to suggest that consolidation of second-order fear to S2 exploits cellular changes that occur during the prior acquisition of first-order fear to S1: hence, de novo protein synthesis is not required in the BLA (Leidl et al., 2018).
First- and second-order fears also differ with respect to their contents, or what has been encoded. This has been demonstrated in studies which show that, subsequent to second-order conditioning: (1) habituating rats to the US undermines the expression of fear to the first-order S1 without affecting the expression of fear to the second-order S2; and (2) repeated S1 alone exposures extinguish the ability of S1 to elicit fear responses without affecting the expression of fear to S2 (Rizley and Rescorla, 1972; Rescorla, 1973; Holmes et al., 2014). These findings have led to the proposal that while first-order conditioning consists in an association between the sensory properties of the conditioned (S1) and unconditioned stimuli (i.e., a stimulus–stimulus [S–S] association), second-order conditioning involves an association between S2 and the defensive responses or central fear state elicited by the already-conditioned S1 (i.e., a stimulus–response [S–R] association). Thus, an alternative explanation of the finding that protein synthesis is required to consolidate first- but not second-order conditioned fear is that the protein synthesis requirement is specific to a particular type of learning. That is, protein synthesis in the BLA is needed to consolidate S–S associations of the sort that typically form in first-order conditioning but is not needed to consolidate S–R associations of the sort that typically form in second-order conditioning.
The present study examined why protein synthesis is required to consolidate first- but not second-order fear in the BLA. It tested two hypotheses. The first was that second-order fear conditioning exploits changes in the BLA that occur during the prior first-order conditioning. The second hypothesis was that the dissociable protein synthesis requirement reflects differences in the content of learning between first- and second-order conditioning. The two hypotheses were tested using a range of second-order conditioning protocols. In each case, we assessed the impact of a BLA cycloheximide infusion on consolidation of fear to the S1 and S2.
Materials and Methods
Subjects
Subjects were 257 experimentally naïve adult male (n = 162) and female (n = 95) rats, obtained from the Rat Breeding Facility maintained at the University of New South Wales. Experiments 1 and 2a used Sprague–Dawley rats while the remaining experiments used Long–Evans rats. Experiments 1, 6a, and 6b were control and/or demonstrational experiments. They were conducted using male rats but we note that the effects shown in these experiments have been previously reported in female rats (Rescorla, 1982; Michalscheck et al., 2021; Fam et al., 2023). The remaining experiments, including all cycloheximide experiments, contained both male and female rats. In each case, there were no statistically significant interactions between sex and drug condition on levels of freezing at test. The rats were housed in a temperature-controlled colony room (20–22 °C) and maintained on a 12 h light/dark cycle (lights on at 0700). Rats had ad libitum access to food and water and were handled daily for at least 5 d prior to the start of the experiment.
Apparatus
All procedures were conducted in a set of four identical chambers (30 cm high × 27 cm long × 30 cm wide). Their side walls and ceilings were made of aluminum, and the back and front walls were made of clear plastic. Their floors consisted of stainless-steel rods, 2 mm in diameter, spaced 13 mm apart, center to center. A shock could be delivered through the grid floor via a custom-built generator located in another room in the laboratory. Each chamber was enclosed in a light- and sound-attenuating wooden cabinet. A 2 × 3 array of white LEDs and a speaker were mounted on the back wall of each cabinet and used for presentation of the visual and auditory stimuli. Illumination of each chamber was provided by an infrared light source. This allowed each rat to be visualized and scored for freezing. A camera mounted on the back wall of each cabinet was used to record the behavior of each rat. Each camera was connected to a monitor and a DVD recorder located in another room in the laboratory. This room contained the computer that controlled stimulus presentations via appropriate software (MATLAB, MathWorks).
Stimuli
Each experiment used auditory and visual stimuli which served as the first-order (S1) and second-order (S2) conditioned stimuli. The auditory stimulus was a tone (620 Hz square-wave tone at 70 dB) and the visual stimulus was a light (57 lux measured at the center of the chamber). In each experiment, the US was a 0.8 mA, 0.5 s foot shock. In experiments 1–4, the light and tone were counterbalanced for their roles as S1 and S2, such that the tone was S2 when the light was S1, and the light was S2 when the tone was S1. In experiments 5–7, the S1 was always a light and the S2 was always a tone. The tone was selected as the S2 as unpublished data showed that simultaneous second-order conditioning is more effective with this particular arrangement. In experiments 1–4, 6a, and 7a, the duration of the S1 was 10 s and the duration of the S2 was 30 s. In experiments 5, 6b, and 7b, the duration of the S2 was shortened from 30 to 10 s so that S2 and S1 could be presented simultaneously (i.e., in an entirely overlapping manner).
Experimental design and statistical analyses
Freezing was the measure of conditioned fear. It was scored by two experimenters, one of whom was blind to the group allocation, using a time-sampling procedure in which each rat was scored as either “freezing” or “not freezing” every 2 s (Fanselow, 1980). The correlation between the scores by the two observers was high (>0.9) and any differences in the scores were resolved in favor of that by the blind observer. A percentage score was calculated for the proportion of the total observations each rat spent freezing.
Freezing to S2 and S1 was analyzed separately. Training data were analyzed using planned orthogonal contrasts with a between-subject factor of group and a within-subject factor of trial (Hays, 1963). Test data were averaged across all trials (as the trial variable provided no additional information) and analyzed using the same between-subject contrasts. The criterion for rejection of the null hypothesis (α) was set at 0.05. Effect sizes were calculated using partial eta squared for training data (η2p; 0.14 is a large effect size) and Cohen’s d (d) for test data (0.8 is considered a large effect size).
Surgery
All rats were surgically implanted with bilateral cannulas targeting the BLA. Rats were anesthetized with isoflurane gas (5% for induction, 2–2.5% for maintenance; Cenvet Australia) and mounted on a stereotaxic apparatus. Two 26-gauge guide cannulas were implanted into the brain of each rat through two holes drilled in the skull. The tips of the guide cannulas were aimed at the BLA in each hemisphere (2.4 mm posterior to bregma, 4.9 mm lateral to the midline, and 8.3 mm ventral to bregma). The guide cannulas were secured in position with dental cement and four jeweler screws. A dummy cannula was kept in each guide at all times other than during infusions. Immediately after surgery, rats received an intraperitoneal (i.p.) injection of penicillin (0.3 ml) and were placed on a heating mat until they had recovered from the effects of the anesthetic. They were then returned to their home cages. Rats were given a minimum of 7 d to recover from surgery. During this period, rats were handled and weighed daily.
Drug infusions
Cycloheximide is commonly used to block protein synthesis in discrete regions of the brain (Schneider-Poetsch et al., 2010), and its infusion into the BLA disrupts consolidation of first-order conditioned fear (e.g., Duvarci et al., 2005). Cycloheximide (Sigma-Aldrich, Missouri, USA) was dissolved in 70% ethanol to yield a stock solution with 200 µg/µL concentration. This was then diluted 1:4 with artificial cerebral spinal fluid (ACSF; Sigma-Aldrich) to a final concentration of 40 µg/µl. Vehicle was made by diluting 70% ethanol 1:4 with ACSF. This dose has been shown to reduce protein synthesis in the lateral amygdala by 95% relative to saline-infused controls (Kesner et al., 1981), is effective in disrupting consolidation of auditory fear conditioning in the BLA (Duvarci et al., 2005), and has the same effects on learning and memory as other protein synthesis inhibitors, such as anisomycin (Duvarci et al., 2005; Milekic et al., 2006; Lai et al., 2008). Rats received bilateral BLA infusions of cycloheximide (CHX) or vehicle (VEH; 0.5 µl per hemisphere). The infusion procedure for each rat commenced with removal of dummy caps from the two guide cannulas and insertion of 33-gauge internal cannulas. The two internal cannulas were each connected to separate 25 µl Hamilton syringes, which were driven by an infusion pump (Harvard Apparatus). Drug or vehicle was infused bilaterally into the BLA through the internal cannulas at a rate of 0.25 µl/min. Following infusion, the infusion cannulas were left in place for an additional 2 min to allow for diffusion of the drug away from the cannula tip.
Histology
Following behavioral testing, rats were euthanized with a lethal dose of sodium pentobarbital (injected i.p.). Rats were decapitated, their brains rapidly removed and frozen. Brains were sliced on a cryostat into coronal sections of 40 µm thickness. Every other section through the region of interest was mounted on a glass microscope slide and then stained with cresyl violet. Slides were observed under a microscope to confirm the location of cannulas using the brain atlas of Paxinos and Watson (2007) (Fig. 1). Rats with inaccurate cannula placements (one or both cannulas positioned outside the boundaries of the BLA) or extensive damage to the BLA were excluded from the statistical analysis.
Cannula placements for all rats included in the study. The black dot represents the most ventral position of the cannula.
Behavioral procedures
Experiment 1
Context pre-exposure
All rats were familiarized with the conditioning chambers (context) on days 1 and 2. There were two sessions of context pre-exposure on each of those days, spaced 3 h apart, with each session lasting 20 min. This was done to eliminate any context-elicited neophobic responses that could interfere with detection of conditioned responses.
First-order conditioning
On day 3, rats in groups paired–paired (PP) and paired–unpaired (PU) received four pairings of S1 and a 0.8 mA × 0.5 s foot shock in stage 1. The duration of S1 was 10 s and its presentation co-terminated with the foot shock. Onset of the first S1–shock pairing occurred 5 min after placement in the chamber, and the interval between subsequent pairings was 5 min (measured from the offset of the previous shock to the onset of the subsequent S1 presentation). Rats remained in the chamber for an additional 1 min after the final S1–shock pairing. Rats in group unpaired–paired (UP) received equivalent exposure to S1 and shock but these events were presented in an explicitly unpaired arrangement. S1 and shock were presented in the order S1, shock, shock, S1, S1, shock, shock, S1, with each event spaced 3 min apart and the first S1 presentation occurring 2 min after placement in the chamber. Rats remained in the chamber for 1 min following the final S1 presentation.
Second-order conditioning
Rats were returned to the chambers 10 min later for stage 2. Rats in groups PP and UP received four pairings of a 30 s S2 and the already conditioned, 10 s S1. After a 2 min baseline period, S2 was presented and followed immediately by S1. This pairing was repeated three times with a 5 min interval between pairings, yielding a total of four S2 → S1 pairings. Rats in group PU received four presentations each of the 30 s S2 and 10 s S1. The first stimulus presentation occurred 2 min after placement in the chamber and the interval between stimulus presentations was 2.5 min. S2 and S1 were presented in the order S2, S1, S1, S2, S2, S1, S1, S2. Rats remained in the chamber for 1 min following the final stimulus presentation.
Context extinction
Rats received context extinction on day 4 in order to reduce any context-elicited freezing, thereby ensuring that the subsequent measure of freezing to S2 was not confounded with freezing elicited by the context. Rats were placed in the chambers for two sessions in the absence of any stimuli. Each session was 20 min in duration and spaced approximately 3 h apart. Rats received a further 10 min extinction exposure to the chambers on the morning of day 5. This was done to extinguish any context-elicited freezing that had spontaneously recovered from the previous day and would have interfered with the detection of freezing to S2.
Test
Rats were tested with S2 on the afternoon of day 5. Rats were placed into the chambers and, following a 2 min baseline period, the S2 was presented. There were eight 30 s stimulus presentations in each test session, and the interval between the presentations was fixed at 3 min. Rats were removed from the chamber 1min after the final S2 presentation.
Experiments 2a–c
In each experiment, the procedures used in stages 1 and 2 and testing were identical to those described for group PP in experiment 1. Briefly, on days 1 and 2, rats were exposed to the context. On day 3, they were exposed to S1–shock pairings, removed from the chamber for approximately 10 min, returned to the chambers and exposed to S2 → S1 pairings. On day 4, rats received context extinction, and on days 5 and 6, they were tested for freezing to S2 and S1. The order of testing was counterbalanced such that half the rats in each group were tested with S1 on day 5 and S2 on day 6, while the remainder were tested with S2 on day 5 and S1 on day 6.
The three experiments differed only with respect to the timing of the intra-BLA infusion (CHX or VEH). Rats received the infusion either immediately after the session containing the S2 → S1 pairings (experiment 2a), immediately after the session containing the S1–shock pairings (experiment 2b) or immediately before the session containing the S1–shock pairings (experiment 2c).
Experiment 3
Behavioral procedures were identical to those used in experiment 1, except that stages 1 and 2 were reversed thereby allowing us to examine sensory preconditioned fear. On days 1 and 2, rats were familiarized with the chambers. On day 3, rats in groups PP and PU received four presentations of a 30 s S2 each of which co-terminated in the onset of a 10 s S1 (S2 → S1). The first pairing occurred 5 min after placement in the chambers, the interval between pairings was 5 min and rats were removed from the chambers 1 min after the final pairing. Rats in group UP received equivalent exposure to S2 and S1 but presented in an explicitly unpaired arrangement. Rats were returned to the chambers 10 min later for stage 2. Rats in groups PP and UP received four S1–shock pairings in the manner described previously. Rats in group PU received four presentations of the S1 and shock in an explicitly unpaired arrangement. Following a session of context extinction on day 4, rats were tested with S2 or S1 on days 5 and 6. The order of testing was counterbalanced, such that half of the rats in each group were tested with S2 on day 5 and S1 on day 6, and the remainder tested with S2 and S1 in the reverse order. Freezing was not scored across pairings of the affectively neutral S2 and S1 during stage 1, as rats do not freeze to these stimuli in advance of their pairing with shock.
Experiment 4
Behavioral procedures were identical to those used for group PP in experiment 3. On days 1–2, rats were exposed to the contexts. On day 3, they received four S2 → S1 pairings followed 10 min later by four S1–shock pairings. Immediately after the session of S1–shock pairings, rats were removed and administered a BLA infusion of cycloheximide (group CHX) or vehicle (group VEH). Following a session of context extinction on day 4, rats were tested with S2 on day 5 and S1 on day 6 in the manner previously described.
Experiment 5
In the remaining experiments, the interval between first-order conditioning of S1 in stage 1 and second-order conditioning of S2 in stage 2 was 48 h rather than 10 min. The aim of experiment 5 was to demonstrate second-order conditioned fear under these circumstances. On days 1 and 2, rats were familiarized to the contexts. On day 3, rats in groups PP and PU received four S1–shock pairings. Those in group UP received equivalent exposure to S1 and shock but presented in an explicitly unpaired arrangement. On day 4, rats received two sessions of context extinction, spaced 3 h apart, each lasting 20 min. On day 5, rats received second-order conditioning. Rats in groups PP and UP received four presentations of a 10 s simultaneous S2S1 compound. Those in group PU received four presentations each of the S2 and S1 in an explicitly unpaired arrangement. Rats were tested for fear to the S2 and S1 on days 6 and 7, respectively.
Experiments 6a–b
Rats were familiarized to the contexts on days 1 and 2, exposed to four S1–shock pairings on day 3 and extinguished to the context on day 4. On day 5, rats received second-order conditioning. In experiment 6a, this consisted of four presentations of the 30 s S2 each of which co-terminated in the onset of the 10 s S1 (S2 → S1). In experiment 6b, it consisted of four presentations of a 10 s S2S1 simultaneous compound. On day 6, in each experiment, rats in group EXT were exposed to 20 S1 alone presentations to extinguish its ability to elicit freezing, while those in group No EXT were placed in the chambers for an equivalent period of time, but no stimuli were presented. Rats were tested with the S2 and S1 on days 7 and 8, respectively.
Experiment 7a–b
Rats were trained in the manner described for the No EXT groups in the previous experiment. Briefly, they were exposed to the context alone on days 1 and 2, four S1–shock pairings on day 3, and two sessions of context extinction on day 4. On day 5, in experiment 7a, rats received four presentations of a 30 s S2 each of which terminated in the presentation of a 10 s S1 (S2 → S1). In experiment 7b, rats received four presentations of a 10 s simultaneous S2S1 compound. Immediately after the serial-order or simultaneous second-order conditioning session, rats received a BLA infusion of cycloheximide (group CHX) or vehicle (group VEH). Finally, all rats were tested with S2 on day 6 and S1 on day 7.
Results
Experiment 1
This experiment used a truncated protocol in which the S1–shock pairings and the S2 → S1 pairings were separated by 10 min rather than the 2 d used previously (Lay et al., 2018; Leidl et al. 2018). The aim was to show that rats acquired fear to the S2 in this protocol and that this fear was associatively mediated: due to the S2 → S1 pairings rather than generalization from the conditioned S1 or any intrinsic ability of S1 to condition fear (see Rizley and Rescorla, 1972). If fear to S2 is associatively mediated in this truncated protocol, it will be used in the subsequent experiments to test the hypothesis that second-order fear conditioning exploits changes in the BLA that occur during the prior first-order conditioning. The design, shown in Figure 2A, consisted of three groups of rats: those in group paired–paired (PP) were exposed to S1–shock pairings and then to S2 → S1 pairings; those in group paired–unpaired (PU) were exposed to S1–shock pairings and then to explicitly unpaired presentations of S2 and S1; finally, those in group unpaired–paired (UP) were exposed to explicitly unpaired presentations of S1 and shock and then to S2 → S1 pairings. All rats were tested with S2 and with S1.
Freezing to S2 in the truncated protocol is due to second-order conditioning. A, Experimental groups for conditioning and testing. A (→) indicates that the stimuli were presented in serial compound, whereas (/) indicates that the stimuli were explicitly unpaired. B, Mean (±SEM) levels of freezing to presentations of the S1 during stage 1. C, Mean (±SEM) levels of freezing to presentations of the S2 and the S1 during stage 2. D, Mean (+SEM) levels of freezing across test presentations of the S2. Subject information: ns = 8 for groups PP and PU, and n = 7 for group UP.
Figure 2B shows levels of freezing during S1 presentations in stage 1. Averaged across groups, there was a significant increase in freezing across the four trials (F(1, 20) = 26.52, p < 0.001, η2p = 0.57). There was no significant difference in freezing levels between rats that received S1–shock pairings (groups PP and PU) and rats that received unpaired S1 and shock presentations (group UP) (F(1, 20) = 0.15, p = 0.70). Further, there were no significant group × trend interactions (Fs < 2.8, p > 0.05). The failure to detect a difference in freezing between the paired and unpaired groups likely reflects context conditioning by the shock in the unpaired group. This suggestion was confirmed when S1 was presented subsequent to context extinction in stage 2.
Figure 2C shows levels of freezing during S2 presentations in stage 2. Averaged across groups, there was a significant increase in freezing to S2 across the four trials (F(1, 20) = 12.53, p < 0.005, η2p = 0.39), indicating that rats acquired freezing to S2 across the session. Averaged across trials, rats in group PP froze more to S2 that those in groups PU and UP (F(1, 20) = 14.10, p < 0.005, η2p = 0.41). Likewise, there was a difference in the rate of acquisition between rats in group PP and those in groups PU and UP (significant group × linear interaction, F(1, 20) = 10.40, p < 0.005, η2p = 0.34). During stage 2, freezing to the S1 remained stable across the four trials (F < 1.0, p > 0.05). Averaged across trials, rats in groups PP and PU froze significantly more than those in group UP (F(1, 20) = 4.74, p < 0.05, η2p = 0.19), confirming that responding to S1 in the former two groups was due to its pairings with shock in stage 1.
At S2 test (Fig. 2D), rats in group PP froze significantly more than those in the other two groups (PP vs PU, F(1, 20) = 31.82, p < 0.0001, d = 2.51, PP vs UP, F(1, 20) = 16.68, p < 0.0001, d = 1.78). There was no significant difference in freezing between groups PU and UP (F(1, 20) = 1.87, p = 0.19). Together, these results indicate that freezing to S2 in the truncated protocol is conditional on its pairing with the conditioned S1 in stage 2, as well as the prior pairing of S1 with shock in stage 1. That is, this freezing to S2 is the result of second-order conditioning rather than generalization from the conditioned S1, or any unconditioned ability of S1 to condition freezing.
Experiments 2a–c
The next experiments used the truncated protocol to test the hypothesis that proteins synthesized in the BLA during (or shortly after) first-order conditioning are used to consolidate second-order conditioned fear. In each experiment, rats received S1–shock pairings in stage 1, followed 10 min later by S2 → S1 pairings in stage 2. Finally, rats received separate tests of S2 and S1. The experiments differed with respect to the timing of the intra-BLA infusion of the protein synthesis inhibitor, cycloheximide (CHX). In experiment 2a, CHX or vehicle was infused immediately after the session containing the S2 → S1 pairings, consistent with previous studies in which CHX was infused immediately after such pairings (Lay et al., 2018). However, there is some evidence that proteins synthesized during a training session play a critical role in consolidating information acquired across the session (Pearce et al., 2017). In order to examine the possibility that proteins synthesized during first- or second-order conditioning were contributing to the consolidation of fear to S2, two additional experiments were conducted. In experiment 2b, rats were infused with CHX or vehicle immediately after the session containing the S1–shock pairings, while in experiment 2c, rats were infused with CHX or vehicle immediately prior to the session containing the S1–shock pairings.
The results of experiments 2a–c are shown in Figure 3. Across all experiments, rats acquired fear to S1 during first-order conditioning and to S2 across second-order conditioning. Analysis of freezing across S1–shock pairings revealed a significant linear trend (Fig. 3B, F(1, 46) = 119.70, p < 0.001, η2p = 0.72, Fig. 3D, F(1, 19) = 69.96, p < 0.001, η2p = 0.79, Fig. 3G, F(1, 23) = 97.89, p < 0.001, η2p = 0.81). There was no significant difference between groups in any of the experiments in the rate of increase in freezing to S1 or in the overall level of freezing to S1 (Fs < 2.5, p > 0.05).
Protein synthesis in the BLA is required for consolidation of fear to S1 but not S2 in the truncated second-order conditioning protocol. A, Experimental timeline of conditioning and testing for rats that received intra-BLA infusions of CHX (experiment 2a, n = 25; experiment 2b, n = 12; experiment 2c, n = 15) or VEH (experiment 2a, n = 23; experiment 2b, n = 9; experiment 2c, n = 10). B, D, F, Mean (±SEM) levels of freezing to presentations of the S1 and S2 during first-order and second-order conditioning. C, E, G, Mean (+SEM) levels of freezing across test presentations of the S2 and the S1.
Freezing to S2 increased across its pairings with the conditioned S1 in all experiments, as evidenced by a significant linear trend (Fig. 3B, F(1, 46) = 119.70, p < 0.001, η2p = 0.72, Fig. 3D, F(1, 19) = 69.96, p < 0.001, η2p = 0.79, Fig. 3G, F(1, 23) = 97.89, p < 0.001, η2p = 0.81). There were no significant differences between groups in the rate of increase in freezing to S2 or overall level of freezing to S2 in any of the experiments (Fs < 3, p > 0.05). In experiment 2b, there was no significant change in freezing to S1 across second-order conditioning (Fig. 3F, F(1, 23) = 2.12, p > 0.05). However, in experiments 2a and 2c, freezing to the S1 increased across S2 → S1 pairings (Fig. 3B, F(1, 46) = 13.02, p < 0.001, η2p = 0.22, Fig. 3D, F(1, 19) = 8.41, p < 0.001, η2p = 0.31), perhaps indicating a transient disruption by the prior presentation of the S2. There were no significant differences between groups in the rate of increase in freezing to S1 or overall level of freezing to S1 in any of the experiments (Fs < 2.5, p > 0.05).
In each experiment, rats in group CHX showed significantly less freezing when tested with S1 than those in group VEH (experiment 2a: Fig. 3C, F(1, 46) = 17.77, p < 0.001, d = 1.18; experiment 3b: Fig. 3E, F(1, 19) = 7.24, p < 0.05, d = 1.17; experiment 2c: Fig. 3G, F(1, 23) = 5.30, p < 0.05, d = 0.89). Thus, the BLA infusion of CHX impaired consolidation of first-order fear to S1 regardless of whether it was administered before (experiment 2c) or after (experiment 2b) the session containing the S1–shock pairings; or after the session containing the S2 → S1 pairings (experiment 2a). By contrast, in each experiment, rats in groups CHX and VEH did not differ in their levels of freezing when tested with S2 (Fs < 1, p > 0.05). Importantly, the absence of a cycloheximide effect on freezing to S2 was not due to a floor effect in the data: the levels of freezing in groups CHX and VEH were equivalent across the entire range of performance, which started at 40% and ended at 10% as the test was conducted under conditions of extinction. Cycloheximide effects are also reported in experiment 4 where the level of freezing to S2 is the same as that observed here. As such, these findings are inconsistent with the hypothesis that proteins synthesized in the BLA during (or shortly after) first-order conditioning are used to consolidate second-order conditioned fear.
Experiment 3
The previous experiment found no evidence to support the hypothesis that second-order conditioned fear of the S2 is consolidated via proteins synthesized during first-order conditioning of the S1. Infusion of the protein synthesis inhibitor cycloheximide into the BLA consistently disrupted consolidation of fear to the S1 while leaving intact fear to the S2. The following experiments examined whether the contrasting effect of cycloheximide on consolidation of fear to the S2 and S1 is specific to the truncated second-order protocol. In experiments 3 and 4, the order of exposure to the S1–shock and S2 → S1 pairings in stages 1 and 2 were reversed generating a truncated sensory preconditioning protocol. Experiment 3 aimed to show that rats acquire fear to S2 in this protocol and that this fear is associatively mediated. The design consisted of three groups (Fig. 4A). Group PP received S2 → S1 pairings in stage 1 and S1–shock pairings in stage 2. Group PU received S2 → S1 pairings in stage 1 and explicitly unpaired presentations of S1 and shock in stage 2. Group UP received unpaired presentations of S2 and S1 in stage 1 and S1–shock pairings in stage 2.
Freezing to S2 in the truncated sensory preconditioning protocol is associatively mediated. A, Experimental groups for conditioning and testing (ns = 8 for all groups). A (/) indicates that the stimuli were explicitly unpaired. B, Mean (±SEM) levels of freezing to presentations of the S1 during stage 1. C, Mean (±SEM) test levels of freezing to S2. D, Mean (+SEM) test levels of freezing to S1.
Figure 4B shows levels of freezing to the S1 across its pairing with shock in stage 2. Overall, there was a significant linear increase in freezing across trials (F(1, 21) = 20.20, p < 0.001, η2p = 0.49). The groups that received S1–shock pairings (group PP and group UP) showed significantly higher freezing to the S1 than group PU (F(1, 21) = 160.01, p < 0.001, η2p = 0.88) but did not differ significantly from each other (F(1, 21) = 0.01, p = 0.94). There was a significant trial × group interaction (F(1, 21) = 8.83, p < 0.01, η2p = 0.30) which was due to rats in groups PP and UP displaying a more rapid increase in freezing to S1 compared to rats in group PU (F(1, 21) = 8.829, p < 0.001, η2p = 0.88).
Figure 4C and D shows levels of freezing to the S2 and S1 at test. During the S2 test, rats in group PP froze significantly more than rats in groups PU and UP (F(1, 21) = 11.93, p < 0.01, d = 1.05), who did not differ significantly from one another (F(1, 21) = 0.63, p = 0.44). During the S1 test, rats that received paired presentations of S1 and shock in stage 2 (group PP and group UP) showed significantly higher freezing to the S1 than those that received unpaired presentations of the S1 and shock (group PU) (F(1, 21) = 81.91, p < 0.001, d = 4.18). Levels of freezing to the S1 did not differ significantly between group PP and group UP (F(1, 21) = 0.59, p = 0.452). Together, these results indicate that freezing to the S2 is associative in nature, due to the S2 → S1 pairings in stage 1 and S1–shock pairings in stage 2, rather than due to generalization of freezing from the conditioned S1 or to a long-delay association between the S2 presentations in stage 1 and the shock presentations in stage 2.
Experiment 4
Experiment 4 used the truncated sensory preconditioning protocol to examine the effect of a BLA cycloheximide infusion (administered immediately after the session containing the S1–shock pairings) on consolidation of first-order fear to the S1 and sensory preconditioned fear to the S2.
Conditioning to the S1 was successful. There was a significant linear increase in freezing to the S1 across S1–shock pairings (Fig. 5B, F(1, 16) = 35.83, p < 0.001, η2p = 0.69). There were no between-group differences in the development of freezing across the pairings or in the overall levels of freezing (Fs < 1).
Protein synthesis in the BLA is required for consolidation of fear to S1 and S2 in the truncated sensory preconditioning protocol. A, Experimental timeline of conditioning and testing for rats that received intra-BLA infusions of CHX (n = 7) or VEH (n = 11). B, Mean (±SEM) levels of freezing to presentations of the S1 during first-order conditioning. C, D, Mean (+SEM) levels of freezing across test presentations of the S2 (C) and the S1 (D).
At test, rats in group CHX froze significantly less than control rats to both the S1 (Fig. 5D, F(1, 16) = 6.35, p < 0.05, d = 1.24) and the S2 (Fig. 5C, F(1, 16) = 7.42, p < 0.05, d = 1.33). This contrasts with the previous experiments where cycloheximide failed to disrupt consolidation of the S2. Taken together, these results indicate that absence of a cycloheximide effect in the previous experiments was not simply due to S2 being subjected to higher-order conditioning. Rather, it is a consequence of something specific to second-order conditioning.
Experiment 5
The next experiments examined whether what is learned in second-order conditioning determines the involvement of protein synthesis in its consolidation. Past research has shown that the content of the association formed during second-order conditioning depends on the temporal relation between S2 and S1. When these two stimuli are presented serially (S2 → S1), as in the standard protocol, responding to the S2 is unaffected by extinction of the fear conditioned S1 or habituation of the fear-eliciting US (Rizley and Rescorla, 1972; Rescorla, 1973; Holmes et al., 2014). These results have been taken to mean that responding to the S2 is mediated by an association between S2 and the fear responses or the central emotional state elicited by the S1. Effectively, the animals remember that S2 elicited fear but not the source of this fear. In contrast, when S2 and S1 are presented in a simultaneous compound (S2S1), fear to the S2 is eliminated by extinction of the S1 (Rescorla, 1982). That is, responding to S2 is contingent on the current value of the S1, indicating that S2 had associated with the stimulus features of S1 across the S2S1 simultaneous compound presentations. That is, the animals remember that S2 elicited fear as well as the source of this fear.
The next experiments used the simultaneous and serial protocols to test the hypothesis that protein synthesis in the BLA is required to consolidate fear mediated by the S–S association in the former protocol but is not required to consolidate fear mediated by the S–R association in the latter protocol. The initial experiment was designed to show that fear of the S2 in the simultaneous protocol is associative in nature, contingent upon its presentation in compound with S1 and on prior pairings of S1 and shock. Three groups of rats were used to provide such a demonstration. Rats in group PP were exposed to S1–shock pairings in stage 1 and to presentations of the simultaneous S2S1 compound in stage 2. Those in group PU were exposed to S1–shock pairings in stage 1 and to explicitly unpaired presentations of S2 and S1 in stage 2. Finally, rats in group UP were exposed to unpaired presentations of S1 and shock in stage 1 and to presentations of the simultaneous S2S1 compound in stage 2.
Figure 6B shows the mean (±SEM) levels of freezing to S1 across its pairings with shock. Freezing significantly increased across the pairings (F(1, 21) = 129.45, p < 0.001, η2p = 0.86). There was a significant trial × group interaction (F(1, 21) = 23.97, p < 0.001, η2p = 0.53) which, from inspection of the figure, was due to rats in groups PP and PU but not those in group UP increasing freezing to S1 across its pairings with shock. Finally, rats in groups PP and PU froze significantly more to S1 overall than those in group UP (F(1, 21) = 31.68, p < 0.001, η2p = 0.60) but did not differ from each other (F(1, 21) < 1.3).
Freezing to S2 in the simultaneous protocol is associatively mediated. A, Experimental groups for conditioning and testing. A (/) indicates that the stimuli were explicitly unpaired. B, Mean (±SEM) levels of freezing to presentations of the S1 during stage 1. C, Mean (±SEM) levels of freezing to presentations of the S2 and S1 during stage 2. D, E, Mean (+SEM) levels of freezing across test presentations of the S2 (D) and S1 (E). Subject information: ns = 8 for all groups.
Figure 6C shows the levels of freezing to S2 and S1 during stage 2. S2 and S1 were presented simultaneously for groups PP and UP: hence, for these groups there is only one data set for freezing to the S2S1 compound. For group PU, the S2 and S1 were presented separately in stage 2 and, hence, separate datasets are presented for freezing to S2 and S1. Freezing to the S2 and S1 was analyzed separately, with the data for rats in groups PP and UP being duplicated across the two analyses. To take this duplication into account, the criterion for rejection of the null hypothesis (α) was reduced to 0.025 for analysis of stage 2 training only.
Freezing to the S2S1 compound was significantly greater in group PP than to S2 alone in group PU or to the compound in group UP (F(1, 21) = 13.62, p = 0.001, η2p = 0.39), while rats in groups PU and UP did not differ significantly (F(1, 21) = 4.54, p = 0.45). Freezing to the S2S1 compound by rats in group PP and to S1 by rats in group PU was significantly greater than freezing to the S2S1 compound by rats in group UP (F(1, 21) = 47.68, p < 0.001, η2p = 0.82). Further, rats in group PU froze significantly more to S1 than rats in group PP, who were exposed to S1 in compound with S2 (F(1, 21) = 41.58, p < 0.001, η2p = 0.66). This is likely due to external inhibition of freezing to S1 by the novel S2, or a generalization decrement from the conditioned S1 to its presentation in compound with S2.
The test of S2 (Fig. 6D) revealed that rats in group PP froze significantly more than those in groups PU and UP (F(1, 21) = 18.79, p < 0.001, d = 1.59) who did not differ significantly from each other (F(1, 21) < 1.8). The test of S1 (Fig. 6E) revealed that rats who had been exposed to S1–shock pairings in stage 1 (groups PP and PU) froze significantly more than rats who had been exposed to unpaired presentations of the S1 and shock (group UP; F(1, 21) = 40.74, p < 0.001, d = 2.95); and group PP froze significantly more than rats in group PU (F(1, 21) = 4.81, p < 0.05, d = 0.90). These results show that freezing to S2 in the simultaneous protocol is associative in nature, due to the presentation of S2 in compound with the already conditioned S1, rather than generalization of fear from S1, or any unconditioned ability of S1 to condition freezing to its S2 associate in the compound.
It is worth noting here that freezing levels to S2 differed between the experimental protocols used in our study. In experiments 1–4, first- and second-order conditioning were separated by 10 min and we consistently observed around 20% freezing to S2. In experiments 5 6–7, first- and second-order conditioning are separated by 48 h and freezing to the S2 is between 25 and 40%. Importantly, while the two protocols produce different levels of freezing to the S2, in both cases learning is contingent on pairings between the S2 and the S1 (rather than stimulus generalization) and in both cases freezing to the S2 is resistant to disruption by intra-BLA CHX infusion. We additionally note that freezing levels are typically lower in animals that have undergone surgery and drug infusions compared to animals that have undergone behavior-only protocols.
Experiments 6a and 6b
Experiments 6a and 6b sought to replicate previous demonstrations (Rescorla, 1982; Holmes et al., 2014) that extinction of the previously conditioned S1 differentially affects responding to S2 at test depending on whether S2 and S1 had been presented serially (S2 → S1) or in a simultaneous (S2S1) compound during second-order conditioning. Rats received S1–shock pairings in stage 1 and pairings of S2 and S1 in stage 2. The experiments differed with respect to the presentation of the stimuli during second-order conditioning in stage 2. In experiment 6a, rats received serial S2 → S1 pairings, whereas in experiment 6b, rats received simultaneous presentations of S2 and S1. The following day, rats in group EXT received 20 S1 alone exposures to extinguish its ability to elicit freezing, while rats in group No EXT received an equivalent amount of context exposure. Finally, both groups in each experiment were tested with the S2 and then with the S1.
The results of experiments 6a and 6b are shown in Figure 7. In stage 1, rats in experiment 6a developed freezing to S1 across its pairings with shock (Fig. 7B, F(1, 14) = 376.45, p < 0.001, η2p = 0.96) as did rats in experiment 6b (Fig. 7G, F(1, 16) = 871.76, p < 0.001, η2p = 0.98). There were no significant between-group or group × trend interactions in either experiment (Fs < 2, ps > 0.05). In stage 2 of experiment 6a, freezing to S2 increased across the serial-order (S2 → S1) presentations (Fig. 7C, F(1, 14) = 46.70, p < 0.001, η2p = 0.77), while freezing to the S1 remained stable (Fs < 1.44, p > 0.05). In stage 2 of experiment 6b, freezing also increased across presentations of the simultaneous (S2S1) compound presentations (Fig. 7H, F(1, 16) = 6.16, p < 0.05, η2p = 0.28). There were no significant between-group differences or group × trend interactions in either experiment (Fs < 1.20, p > 0.05).
Extinction of S1 reduces responding to S2 in the simultaneous second-order conditioning protocol (S2S1), but not in the serial second-order conditioning protocol (S2 → S1). A, Experimental groups for conditioning and testing. B, G, Mean (±SEM) levels of freezing to presentations of the S1 during first-order conditioning in rats in the serial (B) and simultaneous (G) second-order conditioning groups. C, Mean (±SEM) levels of freezing to presentations of the S2 and the S1 during serial second-order conditioning. H, Mean (±SEM) levels of freezing to simultaneous presentations of the S2 and S1. D, I, Mean (±SEM) levels of freezing across extinction of S1 after serial and simultaneous second-order conditioning. E, F, J, K, Mean (+SEM) levels of freezing across test presentations of the S2 (E, J) and S1 (F, K). Subject information: ns = 8 for groups Serial No EXT and Serial EXT; ns = 9 for groups Simultaneous No EXT and Simultaneous EXT.
In the extinction stage, there was a significant decrease in freezing across the S1 alone presentations in each experiment (Fig. 7D, F(1, 7) = 91.56, p < 0.001, η2p = 0.93; Fig. 7I, F(1, 8) = 23.132, p < 0.001, η2p = 0.74). At test, rats in group Simultaneous EXT showed significantly lower freezing to both the S2 and S1 than rats in group Simultaneous No EXT (S1, F(1, 16) = 4.66, p < 0.05, d = 1.01; S2, F(1, 16) = 16.84, p < 0.001, d = 1.93). In contrast, levels of freezing to the S2 did not differ between groups Serial EXT and Serial No EXT (F(1, 14) = 0.85, p = 0.37). This was not due to a failure to extinguish the S1, as test levels of freezing to the S1 were significantly lower in group Serial EXT than group Serial No EXT (F(1, 14) = 5.8, p < 0.05, d = 1.20). These results replicate previous findings demonstrating that conditioned responding to the S2 is sensitive to extinction of its first-order associate, S1, when S2 and S1 are presented in a simultaneous compound (S2S1) during second-order conditioning, but not when they are presented serially (S2 → S1).
Experiments 7a and 7b
These experiments examined whether protein synthesis in the BLA is needed to consolidate fear of the S2 in the serial and simultaneous second-order conditioning protocols. In stage 1 of both experiments, rats were exposed to S1–shock pairings. In stage 2, rats were exposed to serial (S2 → S1, experiment 7a) or simultaneous (S2S1, experiment 7b) presentations of S2 and S1. Immediately after stage 2 in both experiments, rats received an intra-BLA infusion of cycloheximide or vehicle. Rats were subsequently tested for fear of S2 and then of S1.
Experiment 7a
Figure 8 shows the results from stages 1 and 2, and the tests of S2 and S1. In stage 1, levels of freezing to S1 increased across its pairings with shock (Fig. 8B). Averaged across groups, there was a significant increase in freezing across the S1–shock pairings (F(1, 16) = 21.29, p < 0.001, η2p = 0.57). There was no significant difference in freezing levels between rats assigned to the CHX and VEH groups (F(1, 16) = 0.031, p = 0.86) and there was no significant group × trend interaction (F(1, 16) = 0.5, p = 0.49).
Consolidation of fear to S2 in the serial second-order conditioning protocol does not require de novo protein synthesis in the BLA. A, Experimental timeline of conditioning and testing for rats that received intra-BLA infusions of CHX (n = 9) or VEH (n = 9). B, Mean (±SEM) levels of freezing to presentations of the S1 during first-order conditioning. C, Mean (±SEM) levels of freezing to presentations of the S2 and the S1 during second-order conditioning. D, E, Mean (+SEM) levels of freezing across test presentations of the S2 (D) and S1 (E). F, G, Mean (±SEM) levels of freezing to presentations of the S2 during its reconditioning with shock (F) and testing (G).
In stage 2, freezing increased to the S2 across serial second-order conditioning (Fig. 8C, F(1, 16) = 52.00, p < 0.001, η2p = 0.76). There was no significant difference between CHX and VEH groups (F(1, 16) = 0.01, p = 0.91) and no significant group × trend interaction (F(1, 16) = 0.30, p = 0.59). Unexpectedly, freezing to the S1 across stage 2 was significantly higher in the rats assigned to the CHX group (F(1, 16) = 5.46, p < 0.05, η2p = 0.25). There was no significant effect of trial (F(1, 16) = 0.66, p = 0.43) or group × trial interaction (F(1, 16) = 0.07, p = 0.79) on levels of freezing to the S1.
At test, both groups froze at similar levels to S2 and also to S1. Statistical analyses showed that there was no significant effect of CHX on levels of freezing to either the S2 (F(1, 16) = 0.002, p = 0.97) or the S1 (F(1, 16) = 1.14, p = 0.30). These results are consistent with our previous findings in which consolidation of the second-order S2 in the serial protocol is insensitive to the effects of CHX (Lay et al., 2018; Leidl et al., 2018). One might expect that administration of CHX during stage 2 would disrupt reconsolidation of fear to S1 when it is infused into the BLA 48 h after S1–shock pairings. The absence of such an effect could have been due to the presence of the novel S2, which may have protected the S1 from disruption by CHX. It may have also been due to the fact that the first- and second-order fear conditioning occurred in the same context, as it has been shown that protein synthesis inhibitors do not disrupt reconsolidation of first-order conditioned fear when rats are re-exposed to a CS in the context in which it had been conditioned (Jarome et al., 2015). However, it is important to note that the absence of a CHX effect on reconsolidation of fear to S1 replicates results previously reported using very similar parameters to those used in the present study (Lay et al, 2018; Leidl et al, 2018; Williams-Spooner et al., 2019) and the question of why S1 was not disrupted remains to be addressed in future research.
In order to confirm that the BLA infusion of CHX was effective in our experiment, we sought to replicate, within the same animals, the well-documented effect of CHX on consolidation of first-order conditioned fear. Thus, S2 was completely extinguished after initial S1 and S2 testing and then retrained as a first-order stimulus by exposing rats to four S2–shock pairings. Immediately after first-order retraining, rats were infused with CHX or VEH and then tested for fear to the S2 the following day. Figure 8F and G shows levels of freezing to the S2 across its pairings with shock and subsequent testing of the S2. There was a significant increase in freezing to the S2 across its pairings with shock (F(1, 16) = 46.53, p < 0.001, η2p = 0.74) and no significant effect of group (F(1, 16) = 1.55, p = 0.23) or group × trend interaction (F(1, 16) = 2.39, p = 0.14). At test, there was significantly higher freezing in the VEH group compared to the CHX group (F(1, 16) = 6.14, p < 0.05, d = 1.17), indicating that the drug had successfully disrupted consolidation of first-order conditioned fear to the S2. Hence, that the failure of the drug to affect consolidation of the second-order S2 was not due to its ineffectiveness when infused into the BLA.
Experiment 7b
Figure 9 shows the results of stages 1 and 2, and the tests of S2 and S1. In stage 1, freezing to S1 increased across its pairings with shock (Fig. 9B, F(1, 20) = 201.31, p < 0.001, η2p = 0.91). There were no significant between-group differences in the overall levels of freezing (F(1, 20) = 0.03, p = 0.86), nor a significant group × trials interaction (F(1, 20) = 0.04, p = 0.84). In stage 2, freezing remained relatively stable across presentations of the S2S1 simultaneous compound and there was no between-group difference or group × trials interaction (Fig. 9C, Fs < 1.96, p > 0.05).
Consolidation of fear to S2 in the simultaneous second-order conditioning protocol requires de novo protein synthesis in the BLA. A, Experimental timeline of conditioning and testing for rats that received intra-BLA infusions of CHX (n = 12) or VEH (n = 10). B, Mean (±SEM) levels of freezing to presentations of the S1 during first-order conditioning. C, Mean (±SEM) levels of freezing to simultaneous presentations of S2 and S1 during second-order conditioning. D, E, Mean (+SEM) levels of freezing across test presentations of the S2 (D) and the S1 (E).
At S2 test, rats in group CHX froze significantly less than those in group VEH (Fig. 9D, F(1, 20) = 6.99, p < 0.05, d = 1.12). Freezing to the S1 did not differ significantly between groups VEH and CHX (Fig. 9E, F(1, 20) = 0.82, p = 0.38). Thus, consolidation of fear to S2 in the simultaneous second-order protocol does require protein synthesis in the BLA.
A note on the type of associations formed during second-order conditioning
We (Holmes et al., 2014) and others (Rizley and Rescorla, 1972; Rescorla, 1973) have argued that, when S2 and S1 are stimuli from different sensory modalities (e.g., auditory and visual, counterbalanced) and paired in a forward serial arrangement (S2 → S1), S2 predominantly associates with the central state of fear (or the defensive responses) elicited by the S1. This is based on two important findings. The first is that, once established, second-order fear of the S2 is largely unaffected by extinction of the S1, thus ruling-out the possibility that fear to S2 is mediated by S1, as would be the case if rats formed an S2–S1 association and chained it with the initially established S1–US association (Rizley and Rescorla, 1972). The second finding is that, once established, second-order fear of the S2 is unaffected by treatments that diminish the effectiveness of the aversive US. As noted previously, Rescorla (1973) conditioned first-order fear to S1 through its pairing with a burst of loud noise from a klaxon (US), and second-order fear to S2 through its pairing with the already conditioned S1. He then habituated rats to the loud noise US until it ceased to elicit an unconditioned startle response and then tested rats with the S1 alone and S2 alone. These tests revealed that the habituation treatment undermined the expression of first-order fear to S1 but completely spared the expression of second-order fear to S2. Hence, what is learned about the S1 and S2 in this protocol is not the same: whereas fear to the S1 is mediated by some association with the US, fear to the S2 is not. Instead, the fact that fear to S2 is unaffected by extinction of S1 or US habituation implies that, in this protocol, fear to S2 is largely supported by a direct association between S2 and the central state of fear elicited by the S1, or defensive responses that animals express in this state. It is for these reasons that, in the protocol where S2 and S1 are paired in a forward-serial arrangement, we refer to the association produced by these pairings as S–R in nature.
This being said, we acknowledge that fear conditioning in any protocol is likely to involve formation of both S–R and S–S associations, and that different procedures and/or parameters can alter the degree to which such conditioning is supported by each type of association. Thus, rather than the serial second-order protocol producing an S–R association and the simultaneous second-order protocol producing an S–S association, it is more accurate to say that responding to S2 in the serial protocol is predominantly supported by an S–R association, and that responding to S2 in the simultaneous protocol is predominantly supported by an S–S association. Indeed, this acknowledgement may help to contextualize some of the present findings. For example, in experiments 2a–c, rats were exposed to S1–shock pairings in stage 1 and, 10 min later, to serial-order S2 → S1 pairings in stage 2. The critical result was that a BLA infusion of cycloheximide disrupted conditioning to S1 but had no effect on conditioning to S2, as evidenced by the final levels of responding to these stimuli in drug-free tests. However, the drug did not completely abolish performance to the S1: rather, it reduced that performance from ∼60% (in the controls) to ∼40% (which was a highly reliable effect), suggesting that at least some of the conditioning remained intact. One explanation for this pattern of results is that, even for the first-order conditioned S1, responding is supported by both S–S and S–R associations; and the cycloheximide infusion disrupted the portion of responding controlled by the S–S association but completely spared the portion of responding controlled by the S–R association. We cannot, of course, know a priori how much responding to the S1 is controlled by S–S versus S–R associations. We can, however, note that second-order conditioning procedures that have been independently shown to promote S–R associations were largely unaffected by the BLA infusion of cycloheximide, whereas second-order conditioning procedures that have been independently shown to promote S–S associations consistently revealed effects of the BLA cycloheximide infusion. Hence, we conclude that, across all types of conditioning protocols, protein synthesis in the BLA is required to consolidate associations of a particular sort: S–S associations but not S–R associations.
Across the series of experiments, the effects of cycloheximide on fear to S2 and S1 were expected according to our “type of learning” hypothesis and previous research (Lay et al, 2018; Leidl et al, 2018; Williams-Spooner et al., 2019). In experiment 2, second-order conditioning was established using the serial protocol in which fear to S2 is supported by an S–R association (Rizley and Rescorla, 1972; Rescorla, 1973; Holmes et al., 2014); and the interval between S1–shock pairings in stage 1 and S2 → S1 pairings in stage 2 was just 10 min. Under these circumstances, the BLA infusion of cycloheximide disrupted consolidation of first-order fear to S1 (which had not yet been consolidated) but spared consolidation of second-order fear to S2. In experiment 7a, second-order conditioning was again established using the serial protocol; however, in contrast to experiment 2, the interval between S1–shock pairings in stage 1 and S2 → S1 pairings in stage 2 was 48 h and, hence, fear to S1 had already been consolidated. Here, the BLA infusion of cycloheximide in stage 2 spared consolidation of second-order fear to S2 as well as any potential reconsolidation of first-order fear to S1. Importantly, the absence of a cycloheximide effect on fear to S2 and S1 was not due to a failure of the infusion per se as, within this experiment, we again replicated the finding that a BLA infusion of cycloheximide disrupts consolidation of first-order conditioned fear. The fact that cycloheximide had dissociable effects on consolidation of first- and second-order conditioned fear in each of these experiments rules out trivial explanations of the results in terms of differences in statistical power and/or counterbalancing of the S2 and S1 tests. Instead, we conclude that the effects of cycloheximide on consolidation of Pavlovian conditioned fear vary with the type of conditioning and, therefore, what is learned. Protein synthesis in the BLA is required to consolidate Pavlovian conditioned fear when it is supported by an S–S association, as occurs during first-order conditioning, and second-order conditioning in the simultaneous protocol. It is not, however, required to consolidate Pavlovian conditioned fear when it is supported by an S–R association, as occurs during second-order conditioning in the serial protocol.
Discussion
This study originated in findings that protein synthesis in the BLA is required to consolidate first- but not second-order conditioned fear (Lay et al., 2018; Leidl et al., 2018). Its aim was to determine why this is the case and, thereby, the circumstances under which protein synthesis is needed to consolidate new information in the BLA. In each experiment, rats were exposed to a session of S1–shock pairings in stage 1 and a session of S2–S1 pairings in stage 2. The latter session was preceded or followed by an intra-BLA infusion of the protein synthesis inhibitor, cycloheximide. Finally, rats were tested for fear to test presentations of S2 alone and S1 alone.
The first set of experiments tested the hypothesis that consolidation of second-order fear can be supported by proteins that were translated during first-order conditioning: hence, it does not require de novo protein synthesis in the BLA. To test this hypothesis, we examined the effect of a BLA cycloheximide infusion on consolidation of second-order fear using a protocol in which the interval between S1–shock pairings in stage 1 and S2 → S1 pairings in stage 2 was just 10 min (as opposed to the 48 h used previously). Under these circumstances, consolidation of second-order fear could not be supported by proteins synthesized during first-order conditioning as these would also be disrupted by the cycloheximide infusion: thus, if the hypothesis is correct, cycloheximide should disrupt fear to both S1 and S2 at test. The initial experiment confirmed that freezing to S2 in this “truncated” protocol is indeed due to second-order conditioning (experiment 1): rats exposed to paired stimulus presentations in both stages of training froze more when tested with S2 than rats exposed to unpaired stimulus presentations in either stage of training. Subsequent experiments then showed that, contrary to the hypothesis, the effects of cycloheximide on the test levels of freezing to S1 and S2 were dissociable: the drug disrupted freezing to S1 without affecting freezing to S2; and this was true regardless of whether it was infused into the BLA before/after the session of S1–shock pairings in stage 1 (experiments 2b, 2c), or after the session of S2 → S1 pairings in stage 2 (experiment 2a). Together, these findings confirm that consolidation of first-order fear is a time- and protein synthesis-dependent process in the BLA (note the absence of any acute drug effects on freezing to S1 in Fig. 3F [experiment 2c]); and show that consolidation of second-order fear does not depend on proteins synthesized during first-order conditioning. That is, second-order fear to S2 was intact even through first-order fear to S1 was disrupted, showing that consolidation of fear to S2 simply does not require protein synthesis in the BLA.
The second set of experiments examined whether the differential protein synthesis requirement for consolidation of first- and second-order fear is related to differences in the contents of such fears – in what has been learned. Specifically, we tested the hypothesis that protein synthesis in the BLA is needed to consolidate S–S associations of the sort that form in first-order conditioning but is not needed to consolidate S–R associations of the sort that form in second-order conditioning, at least as it is standardly conducted. To do so, we used two second-order conditioning protocols that have been shown to produce different associations: the standard protocol in which S2 and S1 are presented serially in stage 2 (S2 → S1); and a protocol in which S2 and S1 are presented simultaneously in stage 2 (S2S1). The initial experiments confirmed that the two protocols generate different associations. Fear to S2 in the simultaneous protocol was reduced by extinction of the S1 indicating that, in this case, S2 associates with S1 across the S2S1 pairings (experiment 6b) and comes to elicit fear through integration of S2–S1 and S1–shock associations at test. By contrast, fear to S2 in the serial protocol was unaffected by extinction of the S1 (experiment 6a). This is consistent with the well-established view that, in this case, S2 associates with the fear state or fear responses elicited by S1 across the S2 → S1 pairings and, thereby, elicits fear directly at test (Rizley and Rescorla, 1972; Rescorla, 1973).
Subsequent experiments examined the protein synthesis requirement for consolidation of fear to S2 in the simultaneous and serial protocols. They showed that a BLA infusion of cycloheximide disrupted consolidation of fear to S2 in the simultaneous protocol: relative to controls, rats that received this infusion after the session of S2S1 pairings froze less when tested with S2 but just as much when tested with the S1, suggesting a specific disruption of the S2–S1 association (experiment 7b). By contrast, a BLA cycloheximide infusion had no effect on consolidation of the S2–fear association in the serial protocol: rats that received this infusion after the session of S2 → S1 pairings froze just as much as the controls when tested with both S2 and S1 (experiment 7a). Thus, within the BLA, the protein synthesis requirement for consolidation of Pavlovian conditioned fear does indeed depend on what is learned in conditioning. Protein synthesis is needed to consolidate fear to S2 when it is supported by integration of S2–S1 and S1–shock associations in the simultaneous protocol; but is not needed to consolidate fear to S2 when it is supported by an S2–fear association in the serial protocol.
Why is protein synthesis in the BLA needed to consolidate fear to S2 in the simultaneous protocol but not in the serial protocol? While the BLA is the site where protein synthesis is needed to integrate/consolidate the S2–S1 and S1–shock associations in the simultaneous protocol, the protein synthesis requirement for consolidation of the S2–fear association in the serial protocol may lie elsewhere in the circuitry engaged for fear expression. Candidate regions include the central nucleus of the amygdala (CeA), periaqueductal grey (PAG), and prelimbic region of the medial prefrontal cortex (PL). Each of these regions is strongly connected with the BLA (which plays a role in consolidating the S2–fear association) and each has been implicated in maintaining/expressing different components of a fear response (Di Scala et al., 1987; Walker and Carrive, 2003; Corcoran and Quirk, 2007; Ciocchi et al., 2010; Tovote et al., 2015). However, at present, only the CeA has been examined for its involvement in second-order conditioned fear (Holmes et al., 2022). Lay et al (2018) showed that protein synthesis in the CeA is not required to consolidate fear of S2 in the serial protocol used here: rats that received a CeA infusion of cycloheximide immediately after the session of S2 → S1 pairings exhibited just as much freezing to S2 at test as the vehicle-infused controls. As such, future work will examine the involvement of the PAG and/or PL in consolidating second-order fear to S2: specifically, whether this consolidation requires protein synthesis-dependent changes in the PAG and/or PL; and, if so, how these changes might be regulated by the BLA. We hypothesize that the BLA coordinates consolidation of the S2–fear association by regulating protein synthesis-dependent changes in the PAG and/or PL; and will seek to determine how this is achieved via the known requirements for calcium-calmodulin protein kinases (CaMK) signaling, gene transcription, and DNA methylation in the BLA (Lay et al., 2018). This being said, it is important to acknowledge that consolidation of the S2–fear association in the serial second-order protocol may not require protein synthesis at all: this could be tested by administering the protein synthesis inhibitor systemically or via an intracerebroventricular infusion after the session of S2 → S1 pairings. Alternatively, fear to S2 could be supported by a later wave of protein synthesis in the BLA, consistent with evidence that learning is sometimes supported by multiple phases of consolidation between 1 and 6 h post-training (Grecksch and Matthies, 1980; Bourtchouladze et al., 1998; Quevedo et al., 1999). Finally, we acknowledge that the use of broad-spectrum pharmacological inhibitors of transcription (e.g., actinomycin-D; Lay et al., 2018) and translation (e.g., cycloheximide) precludes the identification of specific genes, proteins, and epigenetic changes underlining second-order conditioning in the BLA. Future experiments will employ techniques such as PCR and proteomics to identify these changes.
In summary, the present study has shown that, within the BLA, the protein synthesis requirement for consolidation of second-order fear is determined by what is learned when S2 is paired with the already-conditioned S1. Protein synthesis is required to consolidate fear to S2 when it is supported by integration of S2–S1 and S1–shock associations formed in training; but is not required to consolidate fear to S2 when it is supported by a direct association between S2 and the central state of fear elicited by the S1 (S2-fear). These findings are significant as they show that the substrates of learning and memory in the mammalian brain are not fixed and immutable; instead, these substrates vary with the information that animals extract from their environments. Future research will examine how direct stimulus–fear associations are consolidated in the brain: specifically, whether circuitry that underlies expression of fear responses is involved in acquisition of stimulus–fear associations; and how the BLA interacts with this circuitry to consolidate the stimulus–fear associations.
Footnotes
This work was supported by an Australian Research Council (ARC) Discovery Grant to N.M.H. and R.F.W. (DP200102969), an ARC Future Fellowship to N.M.H. (FT190100697), an ARC Discovery Early Career Research Award to J.P.F. (DE200100856), and Australian Government Research Training Fellowships to D.M.L. and O.A.Q.
The authors have no conflicts of interest to declare.
- Correspondence should be addressed to Nathan M. Holmes at n.holmes{at}unsw.edu.au.
