|Home | About | Journals | Submit | Contact Us | Français|
Fear conditioning associated with inescapable shock training (ST) and fearful context re-exposure (CR) alone can produce significant behavioral fear, a stress response and alterations in subsequent REM sleep. These alterations may vary among animals and are mediated by the basolateral nucleus of the amygdala (BLA). Here, we used the GABAA agonist, muscimol (Mus), to inactivate BLA prior to CR and examined the effects on sleep, freezing and stress-induced hyperthermia (SIH). Wistar rats (n=28) were implanted with electrodes for recording sleep, data loggers for recording core body temperature, and with cannulae aimed bilaterally into BLA. After recovery, the animals were habituated to the injection procedure and baseline sleep was recorded. On experimental day 1, rats received ST (20 footshocks, 0.8mA, 0.5s duration, 60s interstimulus interval). On experimental day 7, the rats received microinjections (0.5ul) into BLA of either Mus (1.0uM; n = 13) or vehicle (Veh; n = 15) prior to CR (CR1). On experimental day 21, the animals experienced a second CR (CR2) without Mus. For analysis, the rats were separated into 4 groups: (Veh-vulnerable (Veh-Vul; n=8), Veh-resilient (Veh-Res; n=7), Mus-vulnerable (Mus-Vul; n=7), and Mus-resilient (Mus-Res; n=6)) based on whether or not REM was decreased, compared to baseline, during the first 4 h following ST. Pre-CR1 inactivation of BLA did not alter freezing or SIH, but did block the reduction in REM in the Mus-Vul group compared to the Veh-Vul group. These data indicate that BLA is an important region for mediating the effects of fearful memories on sleep.
The conditioned fear paradigm is a powerful classical conditioning procedure in which an association is formed between an explicit neutral stimulus (generally a light or auditory stimulus) or situational context and an aversive stimulus (usually footshock) (Davis, 1992; Davis, 1992 ). After training, the previously neutral explicit stimulus or context assumes fear-inducing qualities similar to the aversive stimulus and produces similar behavioral and physiologic outcomes (Misslin, 2003; Nijsen, Croiset, Diamant, Stam, Delsing, de Wied, and Wiegant, 1998; Stiedl, Tovote, Ogren, and Meyer, 2004). Changes in sleep also can be fear-conditioned; i.e., evoking fearful memories produce changes in sleep in the period after fear was evoked that are similar to those that occur after the initial fearful stressor. However, the relationship of fear conditioning to sleep is complex. The best evidence of this complexity is that fear conditioning and the stress response are not predictive of subsequent alterations in sleep. For example, extensive training using inescapable shock (IS) as the aversive stimulus can significantly reduce rapid eye movement (REM) sleep and training with escapable shock (ES) can produce significant increases in REM sleep (Sanford, Yang, Wellman, Liu, and Tang, 2010; Yang, Wellman, Ambrozewicz, and Sanford, 2011) whereas indices of fear (freezing) and stress (stress-induced hyperthermia (SIH)) are similar for both conditions (Yang et al., 2011). Given increasing evidence that REM is important for the processing of emotional (Walker and van der Helm, 2009) and traumatic memories (Mellman, Bustamante, Fins, Pigeon, and Nolan, 2002; Mellman, Pigeon, Nowell, and Nolan, 2007), understanding the neural processes by which fear and stress can produce directionally different alterations in sleep is likely key to understanding sleep disturbances in disorders such as posttraumatic stress disorder (PTSD), which is viewed as arising from abnormal functioning of the brain’s fear system (Shvil, Rusch, Sullivan, and Neria, 2013).
The amygdala is central in current concepts of fear conditioning (e.g., (Myers and Davis, 2007)), it is hyperactive in PTSD (Bremner, Vermetten, Schmahl, Vaccarino, Vythilingam, Afzal, Grillon, and Charney, 2005), and it has an established role in regulating fear- and stress-induced alterations in sleep, especially REM sleep (Liu, Wellman, Yang, Ambrozewicz, Tang, and Sanford, 2011; Liu, Yang, Wellman, Tang, and Sanford, 2009; Wellman, Yang, Ambrozewicz, Machida, and Sanford, 2013). The central nucleus of the amygdala (CNA) (Inagaki, Kawai, Matsuzaki, Shiosaka, and Tohyama, 1983; Peyron, Petit, Rampon, Jouvet, and Luppi, 1998; Price, Russchen, and Amaral, 1987; Semba and Fibiger, 1992), along with the lateral division of the bed nucleus of the stria terminalis (BNST) (Amaral, Price, Pitkanen, and Carmichael, 1992; Davis and Whalen, 2001), projects to brainstem REM sleep regulatory regions. The basolateral nucleus of the amygdala (BLA) has output to both CNA and BNST (Amaral et al., 1992; Davis and Whalen, 2001) and likely regulates the influence of fearful experiences and memories on REM sleep via these descending pathways. Several studies have reported that damage to, or inactivation of, BLA prior to or after fear conditioning (e.g., (Cousens and Otto, 1998; Koo, Han, and Kim, 2004; Maren, 1998; Maren, Aharonov, and Fanselow, 1996; Sacchetti, Lorenzini, Baldi, Tassoni, and Bucherelli, 1999)) or prior to context re-exposure (Helmstetter and Bellgowan, 1994; Muller, Corodimas, Fridel, and LeDoux, 1997) attenuates freezing in the fearful context. These studies have been taken to support a role for BLA in the acquisition and consolidation of fear conditioning. However, functional inactivation of BLA using the GABAA agonist muscimol (Mus) after single trial fear conditioning did not prevent learning, thereby suggesting that BLA is important for fear acquisition, but not fear memory consolidation (Wilensky, Schafe, and LeDoux, 2000). Thus, there is still some question as to the putative role of BLA in the acquisition and consolidation of fear memory.
BLA appears to be critical for the formation of fear memories that can impact sleep. Microinjections of the corticotropin releasing factor antagonist, antalarmin (ANT) into BLA of rats prior to shock training (ST) blocked both IS-induced reductions in REM sleep and the formation of memories that alter sleep without blocking fear memory as indicated by contextual freezing (Wellman et al., 2013). By comparison, global inactivation of BLA with microinjections of Mus, prior to ST blocked the post-training reduction in REM sleep seen in vehicle treated rats (Wellman, Fitzpatrick, Machida, and Sanford, 2014). Furthermore, in Mus treated rats, REM sleep after re-exposures to the fearful context was at baseline levels and freezing was significantly attenuated. Together, these data indicated that BLA is an important regulator of stress- and fear-induced alterations in sleep and that it is critical for the acquisition of fear memories that can impact sleep.
We recently found that outbred Wistar rats can show different REM responses to IS that are independent of freezing and SIH. Some show pronounced decreases in post-ST REM whereas others do not show reductions compared to baseline levels, thereby suggesting individual differences in the sleep response to stress (Wellman, Fitzpatrick, Hallum, Sutton, Williams, and Sanford, 2016). Differences in REM also were observed with post-ST inactivation of BLA. However, post-ST inactivation of BLA blocked the conditioned reduction in REM without blocking freezing or reducing SIH (Wellman et al., 2016). These data suggest that activity in BLA during memory consolidation is important for determining the subsequent effects of fearful memories on REM, but not for forming memories of fearful events. The role that BLA may play in the recall of fearful memories that impact REM is not known.
In this study, we trained rats with ST and inactivated BLA with microinjections of Mus prior to context re-exposure and examined the relationship between fear behavior and sleep on two exposures to the fearful context alone. Our goal was to assess whether pre-recall inactivation of BLA could alter fear memory as assessed by fear behavior and fear-induced alterations in sleep. We also recorded core body temperature in order to assess SIH as an index of the stress response.
The subjects were 28 ninety-day-old Wistar rats obtained from Harlan Laboratories (Frederick, MD). Upon arrival, the rats were individually housed in polycarbonate cages and given ad lib access to food and water. The rooms were kept on a 12:12 light:dark cycle with lights on from 07:00 to 19:00 h. Light intensity during the light period was 100–110 lux and less than 1 lux during the dark period. Ambient room temperature was maintained at 24.5 ± 0.5 °C.
Beginning one week following arrival, the rats were anesthetized with isoflurane (5% induction; 2% maintenance) and implanted with skull screw electrodes for recording their electroencephalogram (EEG) and stainless steel wire electrodes sutured to the dorsal neck musculature for recording their electromyogram (EMG). Leads from the recording electrodes were routed to a 9-pin miniature plug that mated to one attached to a recording cable. Bilateral guide cannulae (26 ga.) for microinjections into BLA were implanted with their tips aimed 1.0 mm above BLA (A 2.6, ML ±4.8, DV 8.0 (Kruger, Saporta, and Swanson, 1995)). The recording plug and cannulae were affixed to the skull with dental acrylic and stainless steel anchor screws. During the same surgery, temperature recorders (SubCue Standard Dataloggers, Canadian Analytical Technologies Inc. Calgary, Alberta, Canada) were implanted intraperitoneally. Ibuprofen (15 mg/kg) was made available in their water supply for relief of post-operative pain. All procedures were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Experimental Animals and were approved by Eastern Virginia Medical School’s Animal Care and Use Committee (Protocol # 13-003).
Mus (muscimol hydrobromide, 5-aminomethyl-3-hydroxyisoxazole) was obtained from Sigma–Aldrich, St. Louis, MO, USA. It was prepared in pyrogen-free distilled water as a vehicle (Veh; 1.0 μM)) and was sonicated for 20 min to ensure that the drug was dissolved completely. A fresh solution was prepared for each experimental day.
All experimental manipulations were conducted during the fourth h of the light period such that sleep recording would begin at the start of the fifth h. This resulted in 8 h of light period recording on each experimental day.
Home cages were changed at least 3 days prior to injection day. The same room was used for animal housing and sleep recording. The microinjections and behavioral testing were conducted in a separate room from that used for recording.
For recording sleep, each animal in its home cage, was placed on a rack outfitted for electrophysiological recording and a lightweight, shielded cable was connected to the miniature plug on the rat’s head. The cable was attached to a commutator that permitted free movement of the rat within its cage. EEG and EMG signals were processed by a Grass, Model 12 polygraph equipped with model 12A5 amplifiers and routed to an A/D board (Model USB-2533, Measurement Computing) housed in a personal computer. The signals were digitized at 256 Hz and collected in 10 s epochs using the SleepWave™ (Biosoft Studio) data collection program.
The rats were allowed a post-surgery recovery period of 14 days prior to beginning the experiment. Once recovered, the animals were randomly assigned to one of two groups: Mus prior to context (Mus; n=13) or Veh prior to context (Veh; n=15) for studies of its effects on fear memory and sleep. All rats were habituated to the recording cable and chamber over 3 consecutive days. Then the rats were habituated to the 5 min handling procedure necessary for microinjections over 2 consecutive days and a baseline following handling (BH) was recorded.
All microinjections were given 30 minutes prior to the start of the first context reexposure (CR1). Injection cannulae (33 ga.) were secured in place within the guide cannulae, and projecting 1.0 mm beyond the tip of the guide cannulae for delivery of drug into the target region. The injection cannulae were connected to one end of a section of polyethylene tubing that had the other end connected to 5.0 μl Hamilton syringes. The injection cannulae and tubing were prefilled with the solution to be injected. Once the cannulae were in place, 0.5 μl of either Mus or Veh alone was bilaterally infused over 3 min. The cannulae were left in place one min pre- and post-injection to allow for maximal absorption of the solution.
Each shock-training session (ST) lasted 30 min. During this procedure, individual rats were placed in shock chambers (Coulbourn Habitest cages equipped with grid floors (Model E10-18RF) that were housed in Coulbourn Isolation Cubicles (Model H10–23)) and were allowed to freely explore for 5 min. Over the next 20 min, they were presented with 20 footshocks (0.8 mA, 0.5 s duration) at 1.0 min intervals. Shock was produced by Coulbourn Precision Regulated Animal Shockers (Model E13–14) and presented via the grid floor of the shock chamber. Five min after the last shock, the rats were returned to their home cages.
On days 7 and 21 after training, the rats were placed back in the shock chambers and allowed to explore freely for 30 min (no shock presented) before being returned to their home cage. These two context re-exposures (CR1 and CR2) were used to test for fear memory (assessed by behavioral freezing) and for post-exposure alterations in sleep.
The shock chamber was thoroughly cleaned with diluted alcohol following each session. Each session was videotaped using mini video cameras (Weldex, WDH-2500BS, 3.6 mm lens) attached to the center of the ceiling of the shock chamber for subsequent visual scoring of freezing.
Computerized EEG and EMG records were visually scored by trained observers blind to drug condition in 10 s epochs to determine wakefulness, NREM and REM. Wakefulness was scored based on the presence of low-voltage, fast EEG and high amplitude, tonic EMG levels. NREM was characterized by the presence of spindles interspersed with slow waves, lower muscle tone and no gross body movements. REM was scored continuously during the presence of low voltage, fast EEG, theta rhythm and muscle atonia. Data were collapsed into 4 h blocks (B1 & B2). The following sleep parameters were examined in the data analyses: total NREM (min), total REM (min); total sleep (TST; REM + NREM), and number and average durations of NREM and REM episodes (defined as contiguous 10 s epochs of a given state).
After an initial assessment that revealed two distinct sleep responses after ST and ascertaining that baseline sleep was not significantly different among groups, the rats were separated into 4 groups: Veh-vulnerable (Veh-Vul; n=8), Veh-resilient (Veh-Res; n=7), Mus-vulnerable (Mus-Vul; n=7), and Mus-resilient (Mus-Res; n=6). The groups were formed based on whether, compared to baseline, the individual rats showed a decrease in REM or either no decrease or an increase in REM during B1 following ST. Grouping as Vul or Res was based on current hypotheses that REM is important for processing emotional memories (Mellman et al., 2002; Mellman et al., 2007; Walker and van der Helm, 2009).
Videos of the ST and CR sessions were scored for freezing, defined as the absence of body movement except for respiration (Blanchard and Blanchard, 1969; Doyere, Gisquet-Verrier, de Marsanich, and Ammassari-Teule, 2000). Freezing was scored by a trained observer blind to condition in 5 s intervals during 1.0 min observation periods. The percentage time spent freezing was calculated (FT%: freezing time/observed time × 100) for each animal for each observation period.
Freezing videos were scored during the five min Pre-ST period to obtain baseline levels prior to ST. Freezing data for the CR days were analyzed for the entire 30 min exposure and compared to the Pre-ST period on the ST day and across drug treatment groups on the CR test days.
The Subcue Dataloggers were programmed to record an animal’s temperature every 15 min over the course of the experiment. To determine the effect of fear and shock on SIH and its relationship to sleep, temperature data for the time in the shock chamber and for the first 4 h of the sleep recording period were compared to the 30 min period immediately prior to ST, CR1 or CR2 and across treatment conditions (ST, CR1 and CR2).
The data were analyzed with two-way mixed factors Group X Treatment ANOVAs with repeated measures on Treatment. The Group factor for all ANOVAs were Veh-Vul, Veh-Res, Mus-Vul, and Mus-Res. For the analyses of sleep, the Treatment factor was recording days (Bas; ST; CR1; CR2) with separate analyses conducted for B1 and B2. The Treatment factor for freezing was assessments periods as described above (Pre-ST; CR1; CR2). Treatment factors for temperature included separate analyses for measurements across days for period in the shock chamber and for the first 4 h after sleep recording had begun. Multiple comparisons among means, when appropriate, were conducted using the Holm-Sidak method to maintain error rate at p < .05.
Pearson Product Moment Correlation was used to examine the relationship between freezing on the first and second exposure to the fearful context and amounts of REM and NREM during B1 and B2 on ST, CR1 and CR2. All animals were used for the analysis (n=28) regardless of whether they showed reduced REM or had received Mus prior to CR1.
The temperature data were analyzed with two-way mixed factors (Group (Veh-Vul; Veh-Res; Mus-Vul; Mus-Res) X Treatment (days)) ANOVAs with repeated measures on Treatment. The Holm-Sidak method was used to determine differences among means as appropriate.
To localize the microinjection sites in BLA, brain slices (40 μm) were made through the amygdala and the sections were mounted on slides and stained with cresyl violet. The sections were then examined in conjunction with a stereotaxic atlas (Kruger et al., 1995) to confirm cannulae placements. Though there were rostral-caudal variations in the placements among animals, the histology indicated that Mus or Veh would have been infused into BLA and adjacent areas in all the rats (Supplemental Data, Figure 1), and all animals were used in the data analyses.
The groups did not significantly differ in baseline REM amounts (Veh-Res: 16.85 ± 2.85; Veh-Vul: 17.04 ± 2.35; Mus-Res: 19.86 ± 2.72; Mus-Vul: 17.29 ± 2.45) during B1 prior to separation based on the effects of ST on REM (Supplemental Data, Figure 2). There also were no significant differences in baseline NREM amounts (Supplemental Data, Figure 3).
After ST was completed, the groups were sorted based on amounts of REM. Comparisons of the REM resilient and vulnerable groups are provided in (Supplemental Data, Figure 4)
The ANOVA for REM amounts (Figure 1A, D, G) during B1 across days revealed a significant Group X Treatment interaction (F(9,71) = 2.98, p < .004) and subsequent comparisons were made within groups across days. The Veh-Vul group showed significant reductions in REM on ST, CR1 and CR2 compared to Baseline and on ST compared to CR2.
The Mus-Vul group showed significant reductions in REM only on the ST day whereas REM in the Veh-Res and Mus-Res groups did not vary across days. Both Veh-Vul and Mus-Vul rats showed less REM on the ST day than either Veh-Res and Mus-Res rats whereas on CR1 only the Veh-Vul rats showed reduced REM. The only difference in B2 was a significant group effect (F(3,25) = 3.85, p < .05) where the Mus-Res rats showed enhanced REM over the Veh-Vul (p<.05) irrespective of days.
The ANOVA for number of REM episodes (Figure 1B, E, H) for B1 was significant for Group (F(3,25) = 10.28, p < .001) and Day (F(3,71) = 7.84, p < .001) Main Effects. The number of REM episodes was greater in Veh-Res (p<.001) and Mus-Res (p<.002) compared to Veh-Vul, and in Veh-Res (p<.02) compared to Mus-Vul. The difference between the Mus-Res and Mus-Vul animals did not reach significance (p >.06). Across days, the number of REM episodes was greater during baseline than on ST and CR1 and CR2. No other comparisons were significant for B1. There also were no significant differences for REM episodes in B2, and REM duration (Figure 1C, F, I) did not differ in either B1 or B2.
There were no significant group effects during B1 or B2 for any measure of NREM that we examined (Supplemental Data, Figure 5) or for TST (not shown).
Freezing (Figure 3) was compared between groups and across days using a mixed factor ANOVA with repeated measures across context exposures (Pre-ST, CR1 and CR2). The ANOVA revealed a significant main effect for context exposure (F(2,52) = 101.72, p < .001). Freezing was significantly elevated relative to the Pre-ST period on CR1 (p <.001) and CR2 (p <.005), and freezing was greater on CR1 than on CR2 (p <.001). There were no significant differences between groups.
Freezing was also examined in 15 min intervals across context exposures (Pre-ST, CR1a, CR1b, CR2a and CR2b). The repeated measures ANOVA revealed a significant main effect for context exposure (F(4,100) = 40.17, p < .001). Rankings for freezing across time periods were CR1a > CR1b > CR2a = CR2b > Pre-ST (not shown). The lack of a difference between CR2a and CR2b suggest that freezing did not fully extinguish on CR2.
The results of the correlation analysis are demonstrated in Table 1. There were no significant correlations between either NREM or REM amounts on any recording day and level of freezing on either CR1 or CR2.
Core body temperature was examined in 15 min intervals during the period the rats were in the shock chamber (ST [with shock presented], CR1 and CR2 [no shock presented]) and in 1 h blocks for 4 h after sleep recording had started. Temperature for two rats (1 Veh-Vul and 1 Mus-Vul) were not obtained due to inability to retrieve the data from the Dataloggers.
The ANOVA for comparisons of temperature in the shock chamber across conditions revealed significant Treatment main effects for the first (F(3,64) = 110.63, p < .001) and second (F(3,45) = 241.55, p < .001) 15 min interval during the period the rats were in the shock chamber, but there were no significant differences between groups. In the first 15 min interval, temperature was significantly increased relative to the pre-treatment period on ST, CR1 and CR2, and on ST compared to CR1 and CR2, and on CR1 compared to CR2 (Figure 4). In the second 15 min interval, temperature also was significantly increased relative to the pre-treatment period on all context exposure days and on ST relative to CR1 and CR2, whereas temperature on CR1 and CR2 did not differ significantly.
The ANOVAs for comparisons of temperature across time after sleep recording had begun (Figure 5) revealed significant effects for treatment during H1 (F(3,65) = 162.01, p < .001), H2 (F(3,65) = 56.43, p < .001), H3 (F(3,65) = 19.30, p < .001) and H4 (F(3,65) = 23.76, p < .001). Temperature was elevated relative to the pre-treatment period during H1–4. In H1, temperature was significantly elevated after ST relative to CR1 and CR2, and after CR2 relative to CR1. In H2, the Veh-Vul group showed greater temperature on ST and CR2 relative to CR1.
The only group differences was found in H2 of the ST day where the Veh-Vul group (37.91 ± 0.11 degree C) showed small but significant increases in temperature compared to the Veh-Res group (37.55 ± 0. 0.8 degree C, p < .05), Mus-Vul group (37.53 ± 0.06 degree C, p < .05) and Mus-Res group (37.45 ± 0.06 degree C, p < .05) and in H4 of the RT day where the Veh-Vul group (37.46 ± 0.11 degree C) showed small but significant increases in temperature compared to the Veh-Res group (37.09 ± 0. 0.7 degree C, p < .05).
The current results support our previous finding that differences in stress- and fear-induced changes in sleep can be observed in individual, outbred Wistar strain rats that have virtually identical stress responses and fear behavior. They also support a role for BLA in mediating these differences and demonstrate that pre-context inhibition of BLA with the GABAA agonist, Mus, attenuates conditioned reductions in REM in Vul rats without significantly attenuating the stress response (as indicated by SIH), and without otherwise significantly altering fear memory or fear extinction (as indicated by freezing). By comparison, Res rats continued to show baseline or enhanced levels of REM on context re-exposure regardless of whether they received microinjections of Veh or Mus in BLA. Together, these data suggest that BLA is a significant mediator of the effects of fear memory on arousal and sleep, but that it may play a reduced role in mediating fear behavior once fear memories are consolidated.
Several studies have reported that lesions of BLA prior to or after fear conditioning (e.g., (Cousens and Otto, 1998; Koo et al., 2004; Maren, 1998; Maren et al., 1996; Sacchetti et al., 1999)) attenuates subsequent freezing in the fearful context. Our prior study inactivating BLA with Mus prior to ST also found an attenuation of freezing upon re-exposure to the fearful context (Wellman et al., 2014) whereas inactivating BLA after ST did not (Wellman et al., 2016). Previous work from other labs has also reported that inactivation of BLA prior to context re-exposure attenuates freezing in the fearful context (Helmstetter and Bellgowan, 1994; Muller et al., 1997). In contrast, we found no significant alteration in freezing when BLA was inactivated prior to context re-exposure, and there was no significant difference between treatment groups on a second exposure to the shock context. One obvious difference across studies was the number of training trials, with 3 to 5 shocks presented in the previous studies (Helmstetter and Bellgowan, 1994; Muller et al., 1997) and 20 shocks presented in our study. There is evidence that overtraining can ameliorate fear learning deficits produced by BLA lesions and involve regions outside BLA in some conditions (Maren, 1999). In particular, the hippocampus is necessary for contextual fear (LeDoux, 2000). Another difference is that we tested the animals at 7 and 21 days after ST whereas the previous studies tested the animals at 24 h after training. The increased ST trials, the additional time between training and testing and the potential involvement of additional brain regions may have made some aspects of the fear memories more resistant to disruption by inactivation of BLA; however, they did not prevent attenuation of fear-conditioned reductions in sleep in Vul animals.
Fear conditioning is stressful and in addition to overt behavioral manifestations (e.g., freezing), it induces stress-related physiological responses including increased body temperature (SIH, (van Bogaert, Groenink, Oosting, Westphal, van der Gugten, and Olivier, 2006; Vinckers, van Oorschot, Olivier, and Groenink, 2009)), hypothalamic–pituitary–adrenal axis (HPA) activation and corticosteroid release as well as increases in respiration and heart rate (Reviewed in (Vinckers et al., 2009)). SIH is produced by both physiological and psychological stress, is rapidly induced (Clement, Mills, and Brockway, 1989; Krarup, Chattopadhyay, Bhattacharjee, Burge, and Ruble, 1999), is stable across repeated presentations of a stressor (Vinckers et al., 2009), and its time course parallels that of HPA activation (Groenink, van der Gugten, Zethof, van der Heyden, and Olivier, 1994; Veening, Bouwknecht, Joosten, Dederen, Zethof, Groenink, van der Gugten, and Olivier, 2004). In general, SIH was higher on the ST days, but increased temperature was also observed on CR1 and CR2. There also were minimal differences between Mus and Veh treated animals, indicating that inactivation of BLA prior to CR1 did not alter the stress response (as indicated by SIH) associated with fear memory recall.
In previous work, we have found that similar SIH responses occur with stressors that either decrease or increase subsequent REM (Yang et al., 2011). In the present study, SIH was induced by ST and by both context re-exposures (CR1 and CR2). SIH also was similar across drug conditions and was similar across treatments even though sleep could significantly vary. Thus, similar to freezing, SIH also does not predict subsequent changes in sleep (Yang et al., 2011).
One of the most compelling findings of our work on conditioned fear and sleep is that strong fear memory associated with multi-trial conditioning sessions can give rise to either decreases or increases in REM sleep. As noted above, extensive training using IS as the aversive stimulus can significantly reduce REM sleep whereas training with ES can produce significant increases in REM sleep (Sanford et al., 2010; Yang et al., 2011). These directionally different changes in sleep can occur with similar freezing and SIH for both ES and IS (Yang et al., 2011) and are regulated by centrally acting corticotrophin releasing factor (Liu et al., 2011; Wellman et al., 2013). Recent work (Wellman et al., 2016) and the current study also demonstrate that individual outbred rats can show either decreases or increases in REM in response to IS, also without differences in freezing or SIH. Thus, while several studies have demonstrated that sleep may play a role in the consolidation of contextual fear memory associated with brief or mild fearful experiences (e.g., Graves, Heller, Pack, and Abel, 2003; Greenwood, Thompson, Opp, and Fleshner, 2014; Hagewoud, Bultsma, Barf, Koolhaas, and Meerlo, 2011; Hellman and Abel, 2007; Kumar and Jha, 2012; Menz, Rihm, and Büchel, 2016; Menz, Rihm, Salari, Born, Kalisch, Pape, Marshall, and Buchel, 2013; Rossi, Tiba, Moreira, Ferreira, Oliveira, and Suchecki, 2014; Silvestri, 2005; for recent review see Havekes, Meerlo, and Abel, 2015), there is no evidence that REM sleep is necessary for the formation of contextual fear memory associated with relatively intense stressful experiences. Interestingly, we found no correlation between freezing and either REM or NREM sleep. However, NREM was much less altered by ST or CR and, at present, one cannot rule out a role for NREM in consolidating even intense fear memory.
Blocking the formation of contextual fear memory that produces behavioral freezing also blocks fear memory that alters sleep (Wellman et al., 2014). Sleep also appears to be important for fear extinction. For example, extinction of contextual fear arising from extensive fear training normalizes subsequent sleep, including increased REM compared to that of animals that continue to show fear (Wellman, Yang, Tang, and Sanford, 2008). Additional evidence comes from recent work in humans that suggests that REM is important for effective fear extinction (Menz et al., 2016). Thus, the factors that mediate the relationships between stress and fearful memories and their impact on sleep are quite complex and appear to involve processes beyond those required simply for forming fear memories, at least as typically defined and measured. Unfortunately, these processes currently are not well-defined nor understood at the neurobiological level, though they appear to involve the amygdala.
In a series of studies, we have examined the effects of inactivation of BLA on fear acquisition (Wellman et al., 2014), fear consolidation (Wellman et al., 2016) and fear retrieval (present results). Collectively, these studies indicate that only inactivation prior to ST (acquisition) can block both freezing and conditioned changes in sleep whereas inactivation after ST (consolidation) and prior to CR (retrieval) can block fear-induced reductions in sleep, but not fear behaviors. Thus, the amygdala appears to be a critical region for mediating the effects of fear memory on sleep, and suggests that targeting processes in BLA may be fruitful for developing therapies to treat fear-associated disturbances in sleep and arousal.
The ability of pre-context inactivation of BLA to significantly change the ability of fearful memories to alter sleep presumably would involve modification of fear memory which can be destabilized and subject to reconsolidation, and modification, upon retrieval (Nader, Schafe, and Le Doux, 2000). Indeed, previous work has demonstrated that reactivation of fear memories makes them susceptible to modification, e.g., local inhibition of protein synthesis in the amygdala after retrieval abolished fear memory associated with a single trial auditory-shock training session (Nader et al., 2000). A similar finding has been reported for fear memories associated with mild shock training (3–4 trials) and reactivated during sleep. The presentation during sleep of odor previously associated with footshock may strengthen fear responses in subsequent waking whereas pairing odor presentation during sleep with inhibition of protein synthesis in BLA can reduce subsequent fear (Rolls, Makam, Kroeger, Colas, de Lecea, and Heller, 2013). In our study, inactivation of BLA with Mus was able to alter the ability of fear memory to reduce REM, but did not abolish or significantly decrease fear memory, as indicated by freezing and SIH. Intriguingly, inactivating BLA did not alter REM in rats that showed post-stress and post-fear increases in REM suggesting that activity in BLA is responsible for mediating the effects of fear on REM.
Experimental fear conditioning is one of our most important research models related to PTSD as well as of other anxiety disorders (Davis, 1990; Davis, 1992; Foa, Zinbarg, and Rothbaum, 1992; Grillon, Southwick, and Charney, 1996; Shalev, Ragel-Fuchs, and Pitman, 1992), and fear training and fear conditioned memories can produce significant alterations in sleep. However, current conceptions of fear conditioning do not explain its relationship to sleep or how it can give rise to clinically significant changes in the regulation of sleep and arousal. The relationship of sleep to PTSD is similarly unclear. While subjective sleep disturbances that persist after a traumatic experience have been linked to the genesis of PTSD (Koren, Arnon, Lavie, and Klein, 2002), and continued sleep disturbances have been suggested to be a hallmark symptom of PTSD (Hagewoud et al., 2011), the exact nature of these disturbances and their association to PTSD have not been fully clarified.
It is important to note that fear conditioning is normally an adaptive process that extinguishes quite readily when the fear-inducing stimulus is withdrawn (Breslau, Kessler, Chilcoat, Schultz, Davis, and Andreski, 1998) and that most individuals who experience traumatic stress do not develop PTSD (Breslau et al., 1998; Weiss, Marmar, Schlenger, Fairbank, Jordan, Hough, and al., 1992). Thus, understanding how traumatic events can differentially impact the fear, stress and arousal systems to produce persisting symptomology will be critical for improved understanding of stress-related psychopathology. Given the dissociations that can occur between standard measures of fear memory, stress responses and subsequent sleep, this will likely require models that incorporate measures of multiple systems and a determination of how, or whether, they predict persisting behavioral and physiological changes.
Sleep, which appears to have value as a predictor of, and symptom of, stress-related psychopathology, may be a critical measure in these models. Its putative role, in particular REM sleep, in “decoupling” memory from its emotional charge (Walker and van der Helm, 2009) and in the processing of traumatic memories (Mellman et al., 2002; Mellman et al., 2007) also suggest that sleep can be a marker of adaptive or non-adaptive processing of fearful experiences. Mellman et al. have suggested that PTSD may be linked to reductions in REM following trauma and to subsequent increases over time as additional processes promote REM in ways that may assist recovery (Mellman, Kobayashi, Lavela, Wilson, and Hall Brown, 2014). Others have suggested that the lack of a strong phasic REM sleep response after a stressful event could predispose to the later increase in REM sleep phasic activity that has been observed in humans with chronic PTSD (DaSilva, Lei, Madan, Mann, Ross, Tejani-Butt, and Morrison, 2011). All of these suggestions, as well as findings from our work in animals, are consistent with REM playing a significant role in mediating stress-related emotion.
The current work demonstrates that alterations in sleep associated with fear memories can differ even though fear behavior and the stress response are similar. Additionally, microinjections of Mus into BLA can attenuate the fear-induced reductions in REM that occur in some animals, without altering behavioral fear or stress response, thereby indicating that these differences in sleep are mediated by the BLA. The differences in sleep demonstrate that fear memories can have complex effects that are not fully assessed by assays of overt behavioral fear and indicate the need for delineating neural processes in the amygdala, and potentially elsewhere, that enable fear memories to disturb or promote sleep. This could allow improved assessment of dysfunctional fear memory and improve modeling of disorders such as PTSD.
This work was supported by NIH research grant MH64827.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.