Periaqueductal Grey differential modulation of Nucleus Accumbens and Basolateral Amygdala plasticity under controllable and uncontrollable stress

Resilience has been conceptualized in part as a dynamic process that includes the ability to adapt to stressful conditions. As such it encompasses the extent to which neural plasticity may be promoted. The current study examined metaplasticity by referring to the “plasticity of synaptic plasticity” in a neural circuit composed of the basolateral amygdala (BLA) and the nucleus accumbens (NAcc), using behavioural stress controllability with or without preceding stimulation of the dorsal periaqueductal gray (i.e. dPAG priming). A tendency for increased plasticity in the controllable versus the uncontrollable group was found in both the BLA and NAcc. dPAG priming suppressed NAcc LTP in all groups, but it suppressed BLA LTP only in the uncontrollable group, demonstrating dissociation between either controllable and uncontrollable groups or the NAcc and BLA. Thus, metaplasticity in the dPAG-BLA-NAcc circuit regulated differentially by controllable or uncontrollable stress may underlie stress coping, and thus contribute to stress-related psychopathologies.

of the controllable group. Thus, the only difference between the groups is the controllability over the aversive stimuli being experienced 11 . Since the pioneering work of Seligman & Maier 10 on maladaptive responses in the face of uncontrollable stress, an entire body of research has been dedicated to elucidate the mechanisms associated with the effects of stressors controllability or the lack of it 8 . Profound behavioural, physiological, neural, and immunological deficits as a result of exposure to uncontrollable stress have been reported, affecting activity 8,[12][13][14] , aggression 15,16 , affectivity 8 , motivation 17 , and learning and memory capabilities 10,13,18 . Furthermore, there has been progress in identifying the neural mechanisms associated with these behavioural alterations. For example, the prefrontal cortex and neuro-modulation by serotonin from the dorsal raphe nucleus and by corticotrophin-releasing factor have been implicated with these alternations 19 .
Many of the studies of controllable/uncontrollable stress were performed implementing multiple exposures to aversive stimuli within a single day. Repeated exposure to cycles of tone-shock enables animals to acquire the avoidance task and towards the end of such a training day, animals may display clear levels of escape and avoidance responses, indicating that they have gained a level of controllability over the challenge 12 . However, if these animals are brought back to the TWSA, they still exhibit high levels of contextual freezing, indicating that while they did gain a level of operational controllability they are still afraid of the context 12 . We found that prolonged training in the TWSA may lead to the development not only of operational controllability, but also of emotional controllability, i.e., the suppression also of the contextual fear response 12 . Interestingly, the development of emotional controllability was found to be associated with the development of stress resilience, while the prolonged uncontrollable stress exposure was found to lead to a lasting state of learned helplessness, with symptoms of depression 12,20 . In the present study we will focus on the impact of prolonged exposure to controllable or uncontrollable stress on behaviours and synaptic plasticity.
Learning and memory processes in the brain have a close association with the activity of positive/negative affective neural circuits [21][22][23] . Typically however, positive neural circuits such as those involving the striatum and specifically the Nucleus Accumbens (NAcc) have been mainly studied in appetitive/addictive learning conditions 24,25 while negative circuits such as those involving the amygdala have been traditionally studied separately and in relation to aversive/withdrawal learning 26,27 . One study indicated on increased ERK2 and CREB activation in the BLA in the uncontrollable group compared with the controllable and naïve groups 12 . In line with this finding, other studies have shown that lesioning the BLA reverses the previously escape-malfunctioning of uncontrollable rats 28 . Similarly, different controllability levels modulated activity and plasticity in the hippocampus and the BLA simultaneously 14 . Briefly, the authors reported that uncontrollable stress enhances neural plasticity in the hippocampus and increases baseline responses in the amygdala 14 . Fewer studies have examined the effects of behavioural controllability on positive affect related brain structures. Among these, some studies have reported an interaction between the NAcc and levels of behavioural controllability through highlighting opposite responses of mesolimbic dopamine system to controllable and uncontrollable aversive experiences 29 . Other studies have documented a modulation of serotonin efflux but not of dopamine efflux in the NAcc shell following different stressor controllability levels 30,31 . It is worth noting that NAcc functions are not restricted to appetitive stimuli only, as these functions have been shown to also to aversive stimuli (for a review, please see: ref. 32). Recent evidence also suggests an analogous picture with regards to the BLA function and appetitive stimuli (for example, see: ref. 33).
Modulation of activity of positive/negative affective neural circuits by higher limbic regions such as the prefrontal cortex (PFC) -(Appetitive learning: ref. 34 37) has been indicated. Behavioural controllability was shown to have a significant effect on these regions as well. For example, an immunohistochemistry study indicated that stressor controllability modulates stress-induced decreases in neurogenesis and increases in fibroblast growth factor-2 in a rats hippocampus 30,31 . A sex dependent effect for controllability with a modulation of hippocampal neurogenesis in males but not in females was also reported 38 . A follow-up study attributed the effects of different levels of stressor controllability on the NAcc to a modulation by the dorsal raphe nucleus 39 . However, the possibility of brain stem modulation of positive/negative affective neural circuits has so far gained less attention.
Furthermore, so far behavioural controllability effects on brain regions associated with either positive or negative affect have been studied separately. The simultaneous effects of behavioural controllability on both positive and negative affect circuits have not been addressed. Recently, dorsal periaqueductal gray (dPAG) simultaneous modulation of ventral subiculum induced-plasticity in both the BLA and NAcc was reported 40,41 . This places the dPAG as a candidate for simultaneously modulating both positive and negative affect circuits.
An additional repeated measures ANOVA was conducted for the learning curves during the training. Violation of the sphericity assumption was found in Mauchly's test for avoidance responses [χ 2 (14) = 28.41, p = 0.014], and degrees of freedom were therefore corrected with Greenhouse-Geisser estimates of sphericity (ε = 0.58). A main effect was found for avoidance responses [F (2.91,40.75) = 6.55, p < 0.001, η p 2 = 0.32]. Sphericity was assumed for escape responses and a main effect was found [F (5,70) = 10.10, p < 0.001, η p 2 = 0.42]. Animals made too few escape-failure responses and therefore no statistical comparisons could be conducted. Bonfferoni corrected post-hoc comparisons revealed that over the 6 days of training, animals increased their avoidance responses while reducing their escape responses [p < 0.05; Fig. 1 (lines)].
Uncontrollable rats were exposed to tones and foot-shocks similar to those of the controllable group but their behaviour had no effect on the outcome in the TWSA (described in the methods). Naïve animals were handled for 10 min outside the vivarium during this time and were then returned to their home-cages.
Performances in the test following the TWSA task. Two weeks following the exposure to the TWSA task, animals' activity was assessed in the conditioned context (i.e. TWSA box) through a reminder of the CS (i.e., tone) but without the US (i.e., footshock). A one-way ANOVA for the effects of the TWSA exposure on activity in context revealed significant differences between the groups [F (2,38) = 14.98, p < 0.001]. Figure 2-A depicts Bonfferoni corrected post-hoc comparisons, which revealed that both controllable (27.33 ± 3.87, n = 18) and naïve (14.91 ± 1.37, n = 12) animals performed significantly more side to side transitions during the 10 min test, compared to uncontrollable rats (4.16 ± 1.23; n = 13). Controllable rats also differed from naïve rats, [p < 0.05].
Activity during the different stages of the test was analysed by using a repeated measure ANOVA. Assumption of sphericity was violated as indicated by Mauchly's test [χ 2 (5) = 24.52, p = 0.000], and Greenhouse-Geisser estimates were therefore used to correct degrees of freedom (ε = 0.71). Figure 2 Electrophysiological results. Within one week following the last behavioural assessment (i.e., PND 75), in vivo electrophysiology measurements were performed to explore whether levels of behavioural controllability would influence plasticity in the NAcc and BLA simultaneously 41 . In a previous study by our group, no effects were found for the dPAG HFS by itself 41 . Therefore, in the current study only the ability of dPAG priming + vSub HFS to differentially modulate BLA and NAcc plasticity was assessed.
Measurements of input-output (IO) curve responses were collected to determine the stimulation intensity needed for recording stable baseline responses. During IO-curves, stimulation range started at 0.2 mA and up to 1.8 mA with steps of 0.2 mA between one to the next. Although the NAcc and the BLA were recorded simultaneously, their statistical analysis is presented separately.
In the NAcc, an ANOVA with repeated measures was conducted on a within-subject factor (stimulation intensities) and a between-subject factor (groups). Mauchly's test indicated that the assumption of sphericity had been violated [χ 2 (35) = 424.64, p = 0.000], and degrees of freedom were corrected using Greenhouse-Geisser estimates of sphericity (ε = 0.19). A main effect was found for stimulation intensities [F (1.52,54.75) = 60.69, p < 0.001, η p 2 = 0.628] but no effect was found for groups [F (2,36) = 2.88, n.s., η p 2 = 0.138] or for the interaction of stimulation intensities X groups [F (3.04,54.75) = 2.59, n.s., η p 2 = 0.039]. Likewise, in the BLA, a repeated measures ANOVA was used to assess a within-subject factor (stimulation intensities) and a between-subject factor (groups). Mauchly's test indicated that the assumption of sphericity had been violated [χ 2 (35) = 311.19, p = 0.000], and degrees of freedom were corrected using Greenhouse-Geisser estimates of sphericity (ε = 0.32). A main effect was found for stimulation intensities [F (2.53 The effects of the ventral Subiculum (vSub) HFS on plasticity in the BLA and NAcc under controllable or uncontrollable conditions. NAcc. A repeated measures ANOVA was conducted for a within-subject factor (time points during baseline before the application of HFS) and a between-subject factor (groups). Mauchly's test indicated that the assumption of sphericity had been violated [χ 2 (5) = 27.84, p = 0.000], and degrees of freedom were therefore corrected using Greenhouse-Geisser estimates of sphericity (ε = 0.59 Similarly, an additional repeated measure ANOVA was conducted for a within-subject factor (time points) and a between-subject factor (groups) after applying HFS stimulation. Sphericity was assumed. A main effect was found for time points [F (15,240) = 13.19, p < 0.001, η p 2 = 0.452] but not for groups [F (2,16) = 0.692, n.s., η p 2 = 0.080] or for the interaction between time points X groups [F (30,240) = 0.773, n.s., η p 2 = 0.088]. As depicted in Fig. 3-B, all groups exhibited plasticity in the form of LTP following vSub HFS [p < 0.05].

BLA.
A repeated measures ANOVA was conducted for a within-subject factor (time points during baseline before the application of HFS) and a between-subject factor (groups). Mauchly's test indicated that the assumption of sphericity had been violated [χ 2 (5) = 15.31, p = 0.000], and degrees of freedom were therefore corrected using Greenhouse-Geisser estimates of sphericity (ε = 0.62). No effects were found for time points [F (1.87,31.74) = 4.92, n.s., η p 2 = 0.013], groups [F (2,17) = 0.531, n.s., η p 2 = 0.059] or for the interaction of time points X groups [F (3.73,31.74) = 18.33, n.s., η p 2 = 0.091]. After applying HFS stimulation, an additional repeated measures ANOVA was conducted for a within-subject factor (time points) and a between-subject factor (groups). Again, Mauchly's test indicated that the assumption of sphericity had been violated [χ 2 (119) = 636.37, p = 0.000], and degrees of freedom were therefore corrected using Greenhouse-Geisser estimates of sphericity (ε = 0.09  Fig. 4-A, all groups exhibited LTP following vSub HFS, and a LSD post-hoc comparison revealed a trend for increased plasticity in controllable rats (n = 8) compared to naïve (n = 6) and uncontrollable rats (n = 6), The effects of 0.5 mA dPAG priming on plasticity in the BLA and NAcc under controllable or uncontrollable conditions. NAcc. A repeated measures ANOVA was conducted for a within-subject factor (time points during baseline before the application of HFS) and a between-subject factor (groups). Sphericity assumed [χ 2 (5) = 10.00, p = 0.076]. No effects were found during baseline under the 0.5 mA dPAG priming condition for time points [F (3,48) = 2.74, n.s., η p 2 = 0.146], groups [F (2,16) = 0.50, n.s., η p 2 = 0.0.059] or for the interaction of time points X groups [F (6,48) = 0.692, n.s., η p 2 = 0.080]. After 0.5 mA dPAG priming + vSub HFS, an additional repeated measures ANOVA was conducted for a within-subject factor (time points) and a between-subject factor (groups). Mauchly's test indicated that the assumption of sphericity had been violated [χ 2 (119) = 364.93, p = 0.000], and degrees of freedom were therefore corrected using Greenhouse-Geisser estimates of sphericity (ε = 0. 15 Fig. 3-B, 0.5 mA dPAG priming before application of vSub HFS resulted in a failure to induce plasticity in all the experimental groups. The interactive effects of group X stimulation protocols. Plasticity Indices (i.e., the ratio between the last 10 min baseline to the last 10 min following theta stimulation) were calculated for all groups. A two-way ANOVA for testing the interaction between groups (i.e., controllable, uncontrollable) and stimulation protocol (i.e., HFS, Priming + HFS) was insignificant, but two major main effects emerged.
Between groups comparisons. Figure 5 depicts the results of a one way ANOVA for testing the effects of group allocation on BLA and NAcc plasticity indices, which was significant for the BLA [F (2,38)   priming + HFS stimulation when compared to HFS stimulation only [p < 0.05]. A trend was also found for the NAcc plasticity index in this group [p = 0.07]. A significant reduction in the NAcc plasticity index following priming + HFS stimulation, as compared to HFS stimulation, was also found in naïve animals [p < 0.05]. In contrast, no changes in plasticity indices were observed in the controllable group, whether under HFS stimulation or under the priming + HFS stimulation, indicating that the impact of dPAG priming was reduced in this group as compared to the naïve and in particular the uncontrollable group (Fig. 6).

Discussion
Previous research has shown that an individual's level of controllability over a stressor is critical for the behavioural and neural processing of the experience 6,7,9,10 . Behavioural controllability has been linked to activation of both positive and negative-affect related brain regions 21-23 but traditionally their modulation has been studied separately. In the current study behavioural controllability or the lack there of (i.e., controllable vs. uncontrollable conditions) were used to assess simultaneous potential shifts in plasticity of positive and negative-affect related brain regions. Further, previous studies indicated that inescapable stress exposure induced ΔFosB in the PAG in mice 42 . Others reported that escapable shock in rats, in contrast to inescapable shock, increased activation in the dPAG (i.e. extracellular 5-HT) 43 . In accordance with the latter hypothesis, our group has recently found total activation (c-Fos-expressing cells) in the dPAG in stress-exposed animals 44 . Collectively, these results point to the PAG's activation under exposure to different types of stressors. Thus, the ability of behavioural controllability to affect metaplasticity in these regions through dPAG priming was also examined.
Over prolonged training in the TWSA, controllable rats gradually learned to gain controllability over the appearance of the US. Beginning from the 2 nd day of training, animals increasingly performed more avoidance responses compared to escape responses and escape-failures. These results are in line with previous studies regarding gaining controllability over the course of prolonged exposure to the TWSA task 12, 14, 20 . Prolonged TWSA exposure induced different behavioural states as was evident in the test performed two weeks following the initial exposure phase. During this test, behavioural differences were conspicuous before, throughout and after an exposure to a reminder of the CS (within the conditioned context). A significant reduction in the number of 'side to side' transitions was found in the uncontrollable group, compared to the controllable and naïve animals. These differences were evident in each stage within the test. The current results along with previous studies 12,20 indicate that even after the cessation of exposure to the CS-reminder during the test (i.e., the 2 nd and last 3 min free exploration), uncontrollable animals continued to exhibit higher levels of immobility, indicating that uncontrollable rats did not show extinction of fear.
Further, it has been reported elsewhere that control over a shock blocked behavioural effects of later social defeat 15 . Others reported that fear-potentiation in the plus-maze was dependent on stressor controllability and contextual conditioning 42 . Taken together, gaining controllability over a stressor is a developing process that entails differences in behaviour through the course of learning. We previously have addressed the gradual development of emotional controllability following mastering controllability 12,20 . In particular, it was demonstrated that while both controllable and uncontrollable groups have demonstrated similar levels of fear when brought Figure 5. The interactive effects of group X stimulation protocols: between groups effects: under HFS stimulation a border-line significance for higher BLA plasticity index in controllable rats was evident compared to naïve rats (p = 0.08). Under 0.5 mA priming + HFS, both controllable and naïve rats exhibited higher BLA plasticity indices compared to uncontrollable rats (** < 0.001).
back to the context of exposure after one day of training, this fear response was maintained only in the uncontrollable group following 6 days of training 12 . These results demonstrated the profound temporal differences in behaviour. Yet, the time passing from training to testing might in fact serve as a confounding factor. It was already shown previously that over the course of time, even a single stress exposure might augment long term responses (for example: ref. 45). Hence, this might also contribute to the results presented here. To what extent does such 'incubation' effect contribute under the current setting should be further examined in future studies. Nevertheless, the main aim of the current study was to assess dPAG ability to modulate a previously induced plasticity in the BLA and NAcc in well-developed controllability or uncontrollability. Thus, our main focus was on testing animals after these emotional states were fully established. This was verified by the marked behavioural differences observed between the two groups in the test.
Considering the accepted view that the BLA is associated more with responses to negative experiences and with negative affect 26,27 and that the NAcc is more involved in positive affect and in appetitive reinforcement learning 24,25,46 , it could be expected that the prolonged uncontrollable stress exposure would be associated with reduced plasticity in the NAcc and increased plasticity in the BLA. In addition, it could be hypothesized that the controllable state would be associated with the mirror effect, i.e., reduced plasticity in the BLA and increased plasticity in the NAcc. However, the emerging picture of the above associations is more complex. There was a tendency for a higher level of plasticity in the controllable versus the uncontrollable group, but this was not significant and furthermore was present both in the BLA and the NAcc [Fig. 5].
We have previously reported that the dPAG HFS fails to independently induce plasticity. Yet, HFS to the vSub enabled metaplasticity effects of dPAG priming to be revealed 40,41 . Using 0.5 mA dPAG priming in the current study induced metaplasticity that suppressed the level of LTP that could be induced. This effect was also seen both in the BLA and the NAcc [Fig. 6]. Of note, the most significant suppression of LTP following dPAG priming was found in the uncontrollable group, and was observed in the BLA as opposed to the the NAcc [Fig. 5]. These findings may hint that the simplistic view presented above is inaccurate; the assumption that prolonged controllable stress would be associated with increased plasticity in the NAcc while uncontrollable stress would be associated with increased plasticity in the BLA, does not hold. Instead, a more complex mode of operation is suggested, which requires separating the analysis to two scenarios: without or with dPAG priming.
Without dPAG priming there is a coordinated impact of a prolonged training regimen, with a tendency for a facilitation of LTP level in the BLA among controllable animals. It should be noted that this tendency was not statistically significant and should be explored further to carefully verify whether this tendency is due to a small but evident effect, whether LTP is not consistently facilitated following prolonged training in the TWSA, or rather Figure 6. The interactive effects of group X stimulation protocols: within groups effects: a significant reduction in BLA plasticity index of uncontrollable rats following priming + HFS stimulation, compared to HFS stimulation was found (**p < 0.001). A trend was found for the NAcc plasticity index in this group [p = 0.07]. A significant reduction in NAcc plasticity index following priming + HFS stimulation, compared to HFS stimulation was found also in naïve animals [p < 0.05]. No changes in plasticity indices were observed in the controllable group, neither under HFS stimulation nor under priming + HFS stimulation.
occurs as a result of large variances in recordings. Regardless, in relation to the current discussion the results imply that under these conditions there is no clear bias towards plasticity in either the BLA or the NAcc, and that a significant level of LTP was evident in both brain regions under both prolonged controllable and uncontrollable conditions. It can thus be concluded that under stressful conditions that do not involve the activation of the dPAG, no clear imbalance of plasticity is induced whether stress is controllable or uncontrollable. dPAG priming reveals a significant difference between animals who have been exposed to either prolonged controllable or uncontrollable stress. dPAG priming suppressed NAcc plasticity in all groups, but this suppression was not statistically significant for the controllable stress group. However, only in the uncontrollable stress group did dPAG priming significantly suppress LTP in the BLA, forming a clear dissociation between the controllable and uncontrollable exposure conditioning.
This may indicate that under dPAG priming, with regards to the balance between the BLA and the NAcc, plasticity is differentially modulated by controllable or uncontrollable stress; under uncontrollable conditions dPAG priming suppresses LTP both in the BLA and in the NAcc, while under controllable stress conditions there is a bias for plasticity in the BLA. This stems from the fact that dPAG priming suppresses LTP in the NAcc but not in the BLA. According to the simplistic 'BLA-Negative affect/NAcc-positive affect' view, such a bias of plasticity in the BLA relative to the NAcc could be expected in the uncontrollable group but not in the controllable group. Thus, in line with recent additional studies 32, 33 , the results presented in the current study support a more complex model in which both the BLA and the NAcc systems contribute to both positive and negative affect.
We have previously demonstrated that in naïve animals the effect of dPAG priming was dependent on the intensity of stimulation 41 . A relatively strong priming intensity (1.0 mA) resulted in suppression of LTP in both the BLA and the NAcc, and even the appearance of long-term depression of the response. Milder intensity priming (0.5 mA, as was used in the current study) resulted in suppression of LTP in the the NAcc but not in the BLA 41 . The current results reverberate well with our previous findings. The effects of dPAG priming in the controllable group are similar to those of the milder intensity priming in the previous study, while the effects of dPAG priming in the uncontrollable group resemble those of the stronger intensity priming (1.0 mA) in the previous study despite the use of the milder priming intensity (0.5 mA). Indeed, classically, the PAGs' functions are mainly attributed to negative-affect related behaviours (e.g., defensive responses such as flight, fight, and freezing; submissive postures; tonic immobilization; and autonomic arousal; for a review see: ref. 47). It is thus plausible to assume that prolonged uncontrollable stress exposure results in hypersensitivity to dPAG priming, which leads to priming effects with milder priming intensity that are similar to those observed with stronger intensity in naïve and controllable animals. The result is modulation of plasticity in the BLA and NAcc, which reflects more severe experiences even under mild conditions. This could underlie in part the greater difficulty of uncontrollable animals to cope with stressful challenges 20 .
The current exploratory study adds to the growing body of research on the effects of prolonged controllable exposure, which is associated with resilience to stress, and to the effects of prolonged uncontrollable exposure, which is associated with impaired coping abilities and depressive symptoms 20 . The study's results indicate that prolonged controllable exposure is associated with a tendency for greater plasticity in both the BLA and NAcc. In contrast, prolonged uncontrollable exposure was found to lead to hypersensitivity of the NAcc and mainly the BLA to priming of the dPAG. These results suggest that both the BLA and the NAcc systems are involved in attempts to cope with stressful challenges, and that a differentiated function of their activation is associated with the end outcome. Metaplasticity in the dPAG-BLA-NAcc circuit induced by prolonged exposure to uncontrollable stress may underlie the related impaired ability to cope with emotional challenges 20 , and may contribute to psychopathologies associated with prolonged exposure to stress. Experimental groups. Levels of controllability were obtained by using the TWSA task. Rats were randomly assigned to one of the following: Animals were handled and weighted once a week and during the other groups experiments [N total = 12; vSub HFS, n = 6; 0.5 mA dPAG priming + vSub HFS, n = 6].

Methods
Experimental design. Starting on the 6 th day and continuing for six consecutive days, animals underwent the TWSA task according to their groups. Two weeks later, animals were tested in the training context while exposed to the CS tone without US presentations. Within a week from the last behavioral assessment, animals were anesthetized and underwent the electrophysiological assessments by using either vSub HFS or 0.5 mA dPAG priming + vSub HFS protocols (detailed below). Immediately following the electrophysiological assessments, animals were decapitated and their brains were harvested for testing electrodes positioning by histology (i.e. Cresyl Violet staining).

Behavioral manipulations and assessment
Novel-setting exploration. Rats were placed in the TWSA apparatus (Model LE 916, Panlab S.L., Barcelona, Spain) while it was in an inoperative mode and were allowed to explore both compartments for a total of 10 min. Crossing over between compartments provided a basic index for the rat's exploratory tendency.
Two-way shuttle avoidance (TWSA) task. Immediately after the exploratory behaviour assessment a training session began.
Apparatus. the TWSA box, placed in a dimly lit, ventilated, sound attenuated room, is a rectangular chamber (51 × 25 × 24(h) cm − internal size) divided by an opaque partition with a small passage (8 × 8 cm) that connects two equal sized, side-by-side, cube shaped compartments. Both metal grid floors of the compartments are weight sensitive and electrifiable. Micro-switches transmit information about the location of the rat to a computer control and data collection program. This program controls both conditioned stimulus (CS) presentations (a tone produced by loud speakers located on the distal walls of the compartments) and unconditioned stimulus (US)electric shock deliveries (to the animals' feet through the compartment floor, by a Solid State Shocker/Distributor; Model LE 10026; Letica, Barcelona, Spain). Both CS and US deliveries were regulated by a Shutavoid software (Panlab S.L., Barcelona, Spain).
1. Controllable training: A single daily training session of 75 trials (on days 1 and 2), 50 trials (on days 3 and 4), and 25 trials (on days 5 and 6) was implanted. Each trial consisted of a 10 sec tone followed by a 10 sec electrical foot shock (0.7 mA) overlapping on the 9 th sec of the CS presentation (i.e., a trace conditioning protocol). The inter-trial interval (ITI) was 30 sec with variation of 25% from one ITI to another. Rats could perform one of the following responses: (1) avoidance response: shuttling to the adjacent chamber of the apparatus while the tone (CS) was on, thus avoiding the shock; (2) escape response: shuttling to the other compartment after the shock began (US), thus reducing exposure to the shock; (3) no escape: not shuttling to the adjacent chamber, thus receiving the full length of the shock (US). 2. Uncontrollable exposure: This group was subjected to the same trial schedule as the controllable group but the animals had no control over the stressor. A computerized program delivered the averaged protocols of tone/shock durations based on the performance of the controllable group in each day.
Test. Two weeks after the end of the training, a test was performed: 3 min of exposure to the training chamber (1 st test exploration), followed by 10 trials consisting of 10 CS presentations (tone). Each CS was presented for 10 sec and was separated by an ITI of 30 sec (ITI) with variation of 25% from one to another; no foot shocks were presented during the test. Following the CSs' presentations, an additional 3 min free exploration in the TWSA box was measured (2 nd test exploration). During the test, all groups were given the opportunity to make an instrumental (i.e., avoidance) response. The number of 'side to side' transitions during the 1 st exploration was referred to as the animals' affective state, mainly for expressing the first responsiveness phenotype while re-encountering the previously exposed context. In addition, 'side to side' transitions were calculated for all stages within the test (i.e. 1 st test exploration, tone, ITI's and 2 nd test exploration).

Electrophysiological manipulations and assessments
Surgical procedure. Rats were anesthetized (Urethane (0.5 mg/kg body weight), ip) and mounted in a stereotaxic apparatus (Stoelting Co. Illinois, USA). The scalp was incised and retracted, and head position was adjusted to place Bregma and Lambda in the same horizontal plane. Small burr holes (2 mm diameter) were drilled unilaterally in the skull for the placement of stimulating and recording electrodes. A 125 μm coated wire reference electrode was affixed to the skull in the area overlapping the nasal sinus. Placement of the stimulating electrodes was done according to the stereotaxic criteria and was based on our previous publications on these pathways 40,41 .
Stimulating electrodes were implanted in the dPAG and the vSub, and recording electrodes were implanted in the NAcc and BLA. During the course of experiments, body temperature was maintained at 36.5-37.4 °C with a feedback regulated temperature controller (FHC, Bowdoinham, ME, USA). Electrodes characteristics. Bipolar concentric stimulating electrodes (125 μm; Kopf, Tujunga, CA) were used for stimulating the dPAG and the vSub. For recordings in the BLA and NAcc, we used stainless steel recording electrodes (tip diameter, 2 μm; 20 mm length; Plastic One Inc., model: E363/2/SPC ELEC.008-SS).

Electrodes positioning.
Electrophysiological recording protocols. For testing vSub ability to induce plasticity in the BLA and NAcc, High Frequency Stimulation (HFS) train consisted of stimulating (the vSub) for 10 brief bursts (200 ms) of 100 Hz stimulation delivered at 1 Hz (a total of 200 pulses). A 20 min pre-HFS baseline was collected at stimulation intensity that elicited a field potential response that reached 35-40% of the maximum response collected during input-output recordings for both the BLA and the NAcc. Immediately following baseline, rats received 4 HFS trains separated by 5 min (i.e., ISI). Responses were collected (once every 20 sec) during the baseline session and for 60 min following the last stimulation session 41 . For testing dPAG priming on BLA and NAcc plasticity following vSub HFS, priming stimulation was composed of a single (0.5 mA) HFS train to the dPAG, delivered 10 sec before the application of HFS to the vSub.
Calculating ratio peak height (PH). In both the BLA and the NAcc, the principal measure of size of the averaged evoked field potentials was 'peak to peak' amplitude. Peak height amplitude was defined from the highest peak before a trough to the lowest peak.

Calculating plasticity Index
Plasticity index refer to the ratio between the last 10 min baseline to the last 10 min following theta stimulation. Indices were calculated for the NAcc and the BLA separately in order to simply assess the plasticity during the electrophysiological recording with regard to the animal's behavioural manifestations.
Histology. At the completion of the electrophysiological assessment, rats were deeply anesthetized with a lethal dose of Urethane (0.5 mg/kg, into the heart), their brain was removed from their skulls. Brains were maintained in −80 °C. Serial 40 μm brain coronal sections were cut using a cryostat (−21 °C), mounted on gelatin-coated slides and stained with cresyl violet (5%, Sigma-Aldrich) to localize the electrodes sites by microscopic examination according to the atlas of Paxinos and Watson 48 .
Statistical analyses. All statistics were conducted in SPSS 20.0. Initial tests were conducted using one way or repeated measures analysis of variance (ANOVA) All post hoc comparisons were made using the least significant difference multiple comparison tests. The results are expressed as means ± SEM, unless stated otherwise.