Anxiolytic reversal of classically conditioned / chronic stress-induced gene expression and learning in the Stress Alternatives Model

Abstract

Social decision-making is critically influenced by neurocircuitries that regulate stress responsiveness. Adaptive choices, therefore, are altered by stress-related neuromodulatory peptide systems, such as corticotropin releasing factor (CRF). Experimental designs that take advantage of ecologically salient fear-inducing stimuli allow for revelation of neural mechanisms that regulate the balance between pro- and anti-stress responsiveness. To accomplish this, we developed a social stress and conditioning protocol, the Stress Alternatives Model (SAM), that utilizes a simple dichotomous choice, and produces distinctive behavioral phenotypes (Escape or Stay). The experiments involve repeated social aggression, a potent unconditioned stimulus (US), from a novel larger conspecific (a 3X larger Rainbow trout). Prior to the social interaction, the smaller test fish is presented with an auditory conditioning stimulus (water off = CS). During the social aggression, an escape route is available, but is only large enough for the smaller test animal. Surprisingly, although the new aggressor provides vigorous attacks each day, only 50% of the test fish choose Escape. Stay fish, treated with the CRF₁ antagonist antalarmin, a potent anxiolytic drug, on day 4, promotes Escape behavior for the last 4 days of the SAM protocol. The results suggest that the decision to Escape, required a reduction in stress reactivity. The Stay fish that chose Escape following anxiolytic treatment, learned how to use the escape route prior to stress reduction, as the Escape latency in these fish was significantly faster than first time escapers. In Escape fish, the use of the escape route is learned over several days, reducing the Escape latency over time in the SAM. Fear conditioning (water off + aggression) resulted in elevated hippocampal (DL) Bdnf mRNA levels, with coincident reduction in the AMPA receptor subunit Glua1 expression, a result that is reversed following a one-time treatment (during SAM aggression on day 4) with the anxiolytic CRF₁ receptor antagonist antalarmin.

1. Introduction

Fear learning (associative learning that produces adaptive behavior to aversive stimuli) is often examined using artificial experimental protocols, such as foot shock [1, 2], but is also an important component of natural systems, resulting from such diverse stimuli as predators [3], prey, and aggressive conspecifics, as well as the smells [4, 5] and sounds that come from them [6–9]. Significant emotional involvement adds salience to memories and drives learning mechanisms to encode important contextual and social cues for future reference [10]. Acute fearful stimuli activate specific central, sympathetic, and neuroendocrine (HPA) responses, in humans and other vertebrates, which together mobilize bodily resources for stress responses [6, 11]. These responses are adaptive, acting in the limbic brain to strengthen memory formation of fearful conditions [12], but may also be maladaptive when the stressor is chronic or traumatic, leading to hippocampal atrophy, compromised immune function and development of post-traumatic stress disorder (PTSD) [13, 14].

Social aggression is among the most potent fear-inducing stimuli [15], due to its unpredictable nature and potential payoffs, which may include resources, territory and mating [16, 17]. We have developed a model of fear learning that utilizes a social defeat protocol to produce stress, and offers a simple dichotomous behavioral choice, escaping or not, to facilitate examination of neural regulation of fear learning and decision-making. The Stress Alternatives Model (SAM) [18], was designed for rodents [19, 20] and rainbow trout (Oncorhynchus mykiss) [6, 21], and takes into account ecologically and evolutionarily relevant stimuli and adaptive behavioral responses [22, 23]., Fear-eliciting stimulus in the SAM is social aggression from a much larger, novel conspecific territorial intruder, from which the test animal has the choice to utilize escape routes, thus two distinct, evenly distributed, and stable behavioral phenotypes emerge: Escape (active avoidance) or Stay (acceptance of subordination). Pavlovian Fear Conditioning is added by applying an auditory cue, the conditioned stimulus (CS, cessation of water flow) prior to the interaction, is paired with the unconditioned stimulus (US, aggression) over several days of training. However, in fish, only individuals of the Stay phenotype demonstrate the conditioned responses (CR; defined here as a physiological response: elevated cortisol stress hormone secretion, but also includes increased monoaminergic activity in amygdala, hypothalamus and raphé [6]) elicited in the absence of the US, to the CS alone. Understanding the mechanisms of fear learning in relation to real-world scenarios elucidates how appropriate behavioral and physiological responses are selected. These ecologically relevant mechanisms may have important clinical implications for a wide variety of psychiatric maladies, including anxiety disorders, depression, and PTSD [24–26].

Thus, Escape and Stay phenotypes differentially take advantage of social avoidance learning (Escape) or Pavlovian fear conditioning (Stay) [6, 21], as stress coping strategies similar to those of rodents and numerous other species [27–30]. During stress, the neuropeptide corticotrophin releasing factor (CRF), initiates central stress actions and the Hypothalamic-Pituitary-Adrenal stress cascade (HPA; HPI in fish) via CRF₁ receptors in the pituitary [31–37]. These CRF₁ receptors are also found in limbic brain regions (including preoptic area, amygdala (Dm and Vc/Vl), hippocampus (Dl), and raphé) [32, 38], such that intracerebroventricular (icv) CRF treatment in rainbow trout stimulates locomotion, anxiogenic behavior, and influences the nature and outcome of aggressive interactions [33, 39]. Additionally, while CRF has been demonstrated to elevate anxiety and indecision [21, 33, 39, 40], and to induce a stereotypical behavior, snap shake [21, 33, 40, 41], that is associated with indecision regarding escape, the selective anxiolytic (anxiety reducing) CRF type 1 (CRF₁) receptor antagonist Antalarmin not only blocks snap shake, but in doing so appears to facilitate decision-making and faster escape [21]. To investigate the role of CRF₁ antagonism directly on learned escape in our Rainbow trout model, both Escape and Stay fish were administered Antalarmin (or saline) orally on day 4 of the 7-day training period. Antalarmin was selected for its specificity to the CRF₁ receptor, its proven efficacy in teleost fish and that orally administered drug crosses the blood-brain barrier to influence central CRF binding [42, 43]. Furthermore, formation of long-lasting memories requires hippocampal gene transcription, synthesis of new proteins, and long-term potentiation [44–46]. Included among genes necessary for specific types of learning are brain-derived neurotrophic factor (BDNF, Bdnf gene), and the NMDA and AMPA ionotropic glutamate receptors, specifically the AMPA receptor subunits GluA 1–4 (Glua1, Glua2, Glua3, Glua3 genes)[47–52]. The Bdnf gene is rapidly and selectively induced in response to external stimuli [53] via region specific promoters in the hippocampus [54, 55] during contextual learning [56, 57], and promotes the synaptic incorporation of GluA₁ subunits increasing molecular plasticity [58]. Additionally, when Glua1 translation is repressed, BDNF-induced neuronal activity is blocked [58], and when GluA_1–4 subunits are deleted from the CA₁ region of the hippocampus, spatial learning tasks are inhibited [59–61]. These genes and their products play an important role in synaptic plasticity, modulation of neurotransmitter release and, when absent, produce deficits in learning and memory [48]. Importantly for our model, hippocampal BDNF levels in mice are positively correlated with ability to learn a spatial task [62, 63].

To test whether stress-resilient behavior, such as learned Escape [64–66], is regulated by social stress via CRF₁ receptor activity, and concomitantly produces enhanced Bdnf plus Glua1 expression, we used the trout SAM paradigm coupled with Antalarmin treatment to examine behavioral and gene expression effects following learned escape and classical conditioning [20, 64]. Learned escape is a complex spatial task that includes finding (and using) a previously unknown escape route during aggressive social conflict, plus learning and remembering the location of this escape route, to limit vulnerability to future attacks. Conversely, Stay fish that never use the escape hole are subject to daily social aggression and exhibit robust fear conditioning [6]. More advanced analysis of fear conditioning in mouse SAM experiments demonstrates that both Stay and Escape mice exhibit enhanced freezing in response to CS (tone) plus US (aggression) pairing [64, 66, 67]. We hypothesized that hippocampal Bdnf gene expression would be differentially elevated in response to the spatial “Learned Escape” task compared to Stay fish, which experience more robust fear conditioning. Similarly, we hypothesized that commensurately elevated expression of Glua1 in Escape and Stay phenotypes. Finally, we hypothesize that pharmacologically induced (CRF₁) receptor antagonism will induce behavioral and gene expression reversal.

2. Methods

2.1. Animals

Rainbow trout (Oncorhynchus mykiss) raised from eggs at Gavins Point National Fish Hatchery were reared in 6 foot circular fiberglass tanks under natural light conditions [6, 21]. Prior to experimentation, small (125–150 g) test fish (N = 40) were netted out of the group tank and placed individually into one of three separate compartments (3 opaque plexiglass dividers, one with an escape hole) of glass aquaria, which were lit and aerated, and held 75 gallons; where each was fed and allowed to acclimate (home space) for 10 days (Figure 1A). On day seven of acclimation, a blood sample was taken from the caudal vein of each fish to determine baseline levels of circulating cortisol (F). Following testing on experimental day 8, fish were anaesthetized by placement in a 10L bucket of water treated with MS-222 (Sigma) at a concentration of 500mg/L until loss of equilibrium (~12–15 seconds), and an additional blood sample and intact brains were collected. Plasma cortisol and regional monoamine data from this SAM + fear conditioning protocol have been published previously [6]. Large (350–450 g) adult hatchery brood-stock (N = 8) were used as the aggressive social stimulus; a novel aggressor presented as unconditioned stimulus (US) each day. These large fish were housed separately prior to experimentation, rotated and rested throughout the experiment to insure a high level of aggression towards the test fish. One larger aggressor was added to an empty compartment of each aquarium 1h prior to social interaction. All experiments were conducted in a manner that minimized suffering and the number of animals used, in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (NIH Publications No. 80–23), under a protocol approved by University of South Dakota IACUC.

Fig. 1.

Experimental design and SAM aquatic social fear conditioning and behavioral arena. (A) The investigation design timeline included a week of acclimation to experimental tanks, followed by pretest blood sample 3 days of recovery, and then SAM conditioning. (B) Pulling the opaque dividers obscuring combatants and escape route, following CS presentation (water off), allowed for (C) social interactions with a novel large aggressor (US) occurred daily in the SAM over the next week, with CR testing on the 8^th day. Fifteen minutes after the initiation of testing terminal blood and brain samples were collected for analysis. Individual trout exhibiting (D) Stay or (E) Escape phenotypes were reliably stable across the training period.

2.2. Experimental design - SAM

Following the 10 day acclimation period, on experimental day 1, water inflow to the tank was turned off (CS), and 15 seconds later, the barrier separating the large and small fish was removed, along with the barrier that covered the escape hole, making an empty chamber available to the test fish (Figure 1B). Importantly, the escape hole was large enough for only the test fish to pass through. Following the CS presentation for fear conditioning, fish were allowed to interact for 15 minutes (Figure 1C), and latency to first attack, aggressive/submissive behavior, and escape time (if applicable) were recorded. If the test fish did not escape within 15 minutes following the first attack (Figure 1D), the fish were separated, barriers re-inserted, water inflow was turned back on, and the interaction was over. If the test fish did escape (Figure 1E), the water was turned back on, the big fish moved back to his chamber, and the small fish was allowed to remain in the empty compartment for five minutes before being ushered back into the home chamber. Pairings of CS:US occurred once daily over seven days, and two stable phenotypes developed by day 2 during social interaction: active avoidance (Escape) or accepting confrontation (Stay) [29, 68]. On day 4 of training, both Escape (n = 10) and Stay (non-escaping; n = 8) individuals were orally administered either saline or the CRF₁ receptor antagonist Antalarmin (2mg/Kg) in food, to produce 4 treatment groups: 1. Saline Control-Escape, 2. Saline Control-Stay, 3. Antalarmin-Escape, 4. Antalarmin-Stay. Antalarmin crosses the blood brain barrier. On day eight, test fish were presented with the CS only (water inflow turned off and barriers removed) with no large fish present, and fish were observed for 15 minutes. Immediately following this observation, test fish tissues were rapidly collected for analysis.

For purposes of determining CR, two other groups were included in the design that did not experience CS:US pairing. In the first of those groups, test fish (n = 12) were exposed to the aggression (US) alone, over 7 days of SAM interaction, with no access to the escape hole. These fish were included as an aggression only controls, to compare the effects of daily aggression on CR, without pairing to the water-off CS. The interactions and timeline for these fish were the same as described above. The final group of test fish (n = 12) was included as a CS only control group and was exposed only to the water off CS for 15 minutes once a day. These fish participated in no aggressive interactions over the course of the experiment but followed the same timeline as that described above. These two control groups, either US alone, or CS alone, were compared for hormonal and gene expression to that of CS:US paired Escape and Stay fish.

2.3. Sequencing of rainbow trout Bdnf, Glua1, Glua2 and Gapdh

Total RNA was isolated from the DL of the Rainbow trout Oncorhynchus mykiss using the RNeasy Mini kit System (Qiagen). During the process of isolation, samples were treated by RNase-free DNase (Qiagen) to eliminate DNA contamination of the samples according to the manufacturer’s protocol. Reverse transcribed RNA, was made using the SuperScript^® First-Strand Synthesis System for RT-PCR (Invitrogen) with random hexamer according to the manufacturer’s instructions.

Degenerated primers were designed for the conserved regions of Bdnf, Glua1 and Glua2 and Gapdh using the known sequences from Mus musculus, Xenopus tropicalis, Gallus gallus, and Danio rerio. The following primers were used to amplify the coding sequences of Bdnf forward primer (5’- CCTKTTCSTTACTATGGTTATYTCAT −3’) and reverse primer (5’- CTGCCCCTCTTAA TGGTYARTGTRC −3’); AMPA receptor subunits Glua1 and Glua2: Glua1 forward primer (5’- YATMGTYGGHGGYGTCTGGTGGTTCTT −3’) and reverse primer (5’- YTTCAT GGTGTCACARGGYTTBC −3’), and Glua2 forward primer (5’- TCTAGAGGAGT TTTTGCCATTTTTG −3’) and reverse primer (5’- TYCTRACTAGGAAYARGACCAC −3’), and Gapdh: forward primer (5’- CCYTTCATCGACCTGSASTA −3’) and reverse primer (5’- GGATGACCTTGCCSACAG −3’). Fragments for PCR were purified from agarose gels with Zymoclean Gel DNA Recovery kit (Zymo Research), cloned into pGEM^®-T System plasmid (Promega) according to instructions from the manufacturer and sequenced (Iowa State University Sequencing Facility). All sequences were submitted to and published in GenBank, and the Accession numbers for Bdnf, Gapdh, Glua1 and Glua2 are: bankit1276240 GU108573, bankit1276246 GU108574, bankit1276252 GU108575 and bankit1276255 GU108576 respectively.

2.4. Total RNA isolation and quantitative PCR (qPCR)

Fish crania were sliced coronally at 300 μm in a temperature controlled (−12°C) cryostat (Leica-Jung 1800). Brain slices were thaw-mounted onto glass microscope slides and individual tissue samples were microdissected with a 500 μm diameter punch [6, 33, 39]. The dorsolateral pallium (DL) was chosen for analysis as this region is homologous to the hippocampus in higher vertebrates, including mammals and birds [69–73]. As our model utilizes trace conditioning, and learning to use the escape hole during aggressive social interaction is a spatial learning task, both hippocampally dependent tasks, the dorsolateral pallium represents the brain region most likely involved.

Total RNA was isolated from the DL of the Rainbow trout Oncorhynchus mykiss using the RNeasy Mini kit System (Qiagen), and RNA was reverse transcribed with the SuperScript^® First-Strand Synthesis System for RT-PCR (Invitrogen) with random hexamer according to the manufacturer’s instructions. Water was used instead of RNA in the negative control no-template reaction. Water was used instead of enzyme in the negative control no-enzyme reactions. Gene-specific primers and probes were created for Oncorhynchus mykiss Bdnf, Glua1, Glua2 and Gapdh mRNAs using the Primer Express Software (Applied Biosystems, Inc.; Table 1). Specificity of designed qPCR assays was tested by agarose gel electrophoresis. Assays for efficiency for qPCR were tested by serial dilutions of templates, and was 97.8% for Bdnf, 96.3% for Glua1, 98.2% for Glua2 and 102.3% for Gapdh.

Table 1.

Sequences of primers and probes used for analysis of Rainbow trout mRNA expression by quantitative real-time PCR. Polymerase chain reaction fragments from degenerate primer reactions were purified from agarose gels, cloned into plasmids and sequenced. From these sequences, gene-specific primers and probes were created for genes of interest using Primer Express Software (Applied Biosystems, Inc.).

Target	Forward primer	Reverse primer	MGB Probe
Gapdh	ATGTATGAAGCCCCATGAATCC	GAAATGGGAAGAGGCCTTGTC	CCGGTGCTGATTACGTCGTTGAGTCC
Bdnf	GATCAGCAACCAAGTGCCTTTA	GCCGCCGTACCCTCATG	CCACCGCTGCTTTTTCTCCT
Glua1	AAGGAGTTTTTCAGGAGGTCCAA	CCCCTCCGCTGTGGTTTT	CCGTATTTGAGAAGATGTGGTCTTACA
Glua2	ATACGAGGGCTACTGTGTCGATTT	CTCGGGCTCCATATTTTCCA	CCGCTGAGATAGCCAAGCACTGTGG

Real time RT-PCR was performed using 50 ng of cDNA per reaction combined with primer/probe sets and TaqMan^® Gold RT-PCR Master Mix (Applied Biosystems, Inc.). Real time assays were run on an ABI 7000 (Applied Biosystems, Inc.). The real-time PCR profile consisted of one cycle at 50°C for 2 min and 95°C for 10 min, followed by 45 cycles at 95°C for 15 s and 60°C for 1 min. All reactions were run in duplicate and results from individuals averaged. Water was used instead of cDNA in the negative no-template control reactions. No-template and no-enzyme negative controls from the reverse transcription step were used to eliminate false-positive results.

Gene expression was normalized to Gapdh, and the relative quantity of mRNAs was calculated based on the comparative Ct method. TaqMan RT-PCR values for Gapdh were subtracted from Bdnf [74], Glua1 and Glua2 values, and fold expression calculated by the 2^−ΔΔCT method [75]. The resulting corrected values were used for comparisons across the experimental groups.

2.5. Statistical Analyses

Statistical analyses and experimental designs were based on a priori hypotheses, for the purpose of avoiding combinatorial exponential expansion of error from multiple tests [76]. This statistical pre-planning allows for a wider range of multiple comparison analyses across hypothetical designs. A level of 0.05 was set as the limit for statistical significance. Two-way ANOVA (Drug [Antalarmin or saline] × Phenotype [Escape, Stay] design) was used to examine the contribution of drug effects relative to behavioral phenotype expression (Stay × Escape). For behavioral results, comparison of latency to escape was made using repeated measures one-way ANOVA. Evaluations of gene expression among un-treated groups were assessed by one-way ANOVA. Comparisons between two conditions within a given phenotype (Escape or Stay) were investigated by Student’s t-tests.

Each animal was a singular sample source, from which multiple measures and analyses were taken. Five assumptions of parametric statistics were applied to the data (Random and Equal samples, Normal distribution of data, Homogeneity of variance [similar homoscedasticity], Independence of data for different groups, Interval level data – linearity), which were transformed, when necessary, but also compared to non-parametric analyses, and graphed in their raw form. Analyses with both non-parametric and parametric statistics were performed along with examination for multiple comparisons using the Holm-Sidak method, and when the statistical analyses match, as they do for the data herein, we report the parametric results without α adjustment [77–82] based on a priori hypothesis driven exclusion from combinatorial effects [76]. Significant effects between groups for one-way analyses were examined with Student–Newman–Keuls post hoc analyses (to minimize Type I error) and Duncan’s Multiple Range Test (to minimize Type II error).

3. Results

3.1. Expression of Escape and Stay phenotypes

As with previous experiments in trout and mice, the division of behavioral phenotypes, Escape and Stay, was near 50% for each [6, 18, 19] (Figure 2A). However, when considering just Stay trout after treatment with the CRF₁ antagonist, which on days prior to Antalarmin exhibited 0% Escape behavior, now include 67% Escape animals (Day 4, χ2: p < 0.022; Figure 2B). There were 33% of Antalarmin-treated Stay trout that did not make use of the escape route.

Fig. 2.

The CRF₁ antagonist Antalarmin shifts pro-stress behavior and neuroendocrine responsiveness toward anti-stress phenotypes and outcomes. (A) Individual trout originally exhibiting Stay or Escape phenotypes were roughly equivalent: 45% Escape, 55% Stay. While antalarmin treatment did not affect the behavioral phenotype of Escape fish (data not shown), (B) most Stay animals (67%) began, and continued, using the escape route following CRF₁ antagonist treatment on day 4. A minority (33%) of Stay fish continued to Stay following antalarmin treatment.

3.2. The effect of CRF₁ antagonist on plasma cortisol concentrations

Measuring blood samples taken 3 days prior to SAM social interaction (left bars, Figure 3) all fish had similar plasma cortisol concentrations ([F], ~15 ng/ml), without significant differences between future group status (Escape vs Stay: F_1,20 = 0.057, p > 0.81; Saline Control vs Antalarmin: F_1,20 = 2.63, p ≥ 0.12; Interaction: F_1,20 = 0.19, p > 0.73; Figure 3). After 7 days of social interaction, and blood samples taken again on day 8, when no large aggressor was present (US⁻), but the cue (CS⁺, water off) was given, plasma cortisol was significantly elevated (~50 ng/ml; Escape vs Stay: F_1,20 = 25.6, p < 0.001) in Stay trout (right bars, Figure 3). This conditioned response (CR) is not due to elevated cortisol remaining from social aggression on day 7, but has been previously demonstrated to result from Pavlovian Conditioning, a response to the CS on day 8 (when the US is absent) [6]. After treatment with the CRF₁ antagonist, Escape fish had cortisol concentrations on day 8 that were not significantly greater than in fish measured 3 days prior to social interaction. What is more, Stay fish that became Escape fish following Antalarmin treatment, also did not show elevated plasma cortisol concentrations (~15 ng/ml). However, Stay trout that remained Stay, did have statistically elevated plasma cortisol (~50 ng/ml) compared with pretreatment Stay saline control [F] (t₁₀ = 4.73, p < 0.0008), compared to all controls (t₃₁ = 4.73, p < 0.000001) and compared to trout treated with Antalarmin that Escape (t₇ = 3.64, p < 0.008; Figure 3).

Fig. 3.

Plasma cortisol concentration [F] was similar in all fish before treatments or group placement (blood drawn 3 days prior to SAM exposure and 7 days before drug treatment; left bars). In the same animals, following all 7 days of SAM exposure, saline-treated control Stay fish without antalarmin treatment (hatched white bar) produced a significant physiological conditioned response (CR), elevated plasma [F], in response to the CS only (water off; no aggression) on day 8 (from a second blood draw), compared to saline-treated control Escape fish (white bar). Day 4 CRF₁ receptor antagonist treatment (gray bars) eliminated increased [F] in Stay fish that Escape from day 4 onward (St/Esc; gray right-down hatched bar), but not in Stay fish that retained the Stay phenotype (St/St), which had elevated plasma [F] compared to fish treated with Antalarmin that Escape (*), and fish that began as Stay trout then Escape after Antalarmin pretreatment (*), Pre-SAM Stay controls [F] (+), and compared to all controls (#).

3.3. Latency to escape

In fish initially learning to escape, latency to escape dramatically improves with each trial, regardless of treatment with CRF₁ antagonist (Figure 4 solid line, F_6,57 = 24.37, p < 0.001; compared to dashed line, F_6,22 = 13.25, p < 0.001). However, a significant portion (66%) of Stay trout showed Escape behavior following CRF₁ antagonist exposure (Figure 4 dotted line). Importantly, initial Antalarmin-induced Escape (on day 5) in Stay fish was more rapid than any Escape on day 1, with or without drug treatment. Day 5 Escape latencies are identical in Stay Antalarmin-treated, Escape Antalarmin-treated, and saline-treated control Escapers (Figure 4 compare dotted, dashed, and solid lines). This suggests that non-escaping fish (Stay) learn the location of the escape hole during the first 4 training sessions, but do not utilize the option. Treatment with the CRF₁ antagonist allowed expression of escape behavior. Importantly, treatment with this anxiolytic drug also allowed a trade-off in learning modalities, with fish switching from submission to learned escape.

Fig. 4.

Latency to escape (mean ± SEM) over 7 days SAM social interaction with CS:US pairings. Black line with circles indicates escape latency in seconds for saline-treated control fish that escaped during social interactions. Black hashed line with squares indicates fish that escaped during the first 4 training sessions and were treated with CRF₁ receptor antagonist on day 4. Diamonds with dotted line indicates fish that did not escape during the first 4 training sessions, were treated with CRF₁ receptor antagonist on day 4 and began escaping.

3.4. Gene expression

After 7 days of SAM social interaction, on day 8 following presentation of the CS alone on test day, hippocampal levels of Bdnf mRNA were significantly higher in escaping fish (two way ANOVA-Phenotype: F_1,17 = 6.64, p ≤ 0.02; -saline vs CRF₁ antagonist treatment: F_1,17 = 64.98, p < 0.001; -phenotype X saline or CRF₁ antagonist interaction: F_1,17 = 6.13, p ≤ 0.024; one way ANOVA-CS & US controls vs saline-treated: F_3,14 = 20.09, p < 0.001; one way ANOVA- CS & US controls vs Antalarmin-treated: F_3,17 = 0.55, p > 0.65; Figure 5A). In addition, because control groups on day 8 for the CS (water off) and US also exhibited unelevated Bdnf expression relative to the escaping fish, aggression (experienced by US only, Escape and Stay animals) did not promote Bdnf transcription. Pairing of the CS with aggression (US) occurred only in saline- or Antalarmin-treated Stay and Escape groups; the only groups for which elevated Bdnf mRNA are elevated are saline-treated (Figure 5A). Thus, conditioning is evident for elevated Bdnf expression in both Escape and Stay phenotypes [64]. It is important to note, that elevated Bdnf gene expression was blocked by anxiolytic treatment by means of CRF₁ antagonism.

Fig. 5.

(A) Hippocampal Bdnf gene expression (mean fold ± SEM) is significantly but differentially enhanced in saline-treated control Escape (Esc; white tilt-hatched bar) vs saline-treated control Stay (white flat-hatched bar) fish following CS only presentation on test day (day 8). Escape and Stay fish expressed significantly elevated (*) Bdnf gene expression compared to individuals that received the conditioned stimulus only (CS; water off; white bar), to individuals receiving daily exposure to the unconditioned stimulus only (US; aggression from larger, novel conspecific; white dotted bar), and to Escape and Stay fish treated (day 4) with the CRF₁ antagonist antalarmin (gray bars). Importantly, Escape Bdnf mRNA levels were also significantly (+) elevated compared to Stay fish; and stimulated hippocampal Bdnf gene expression is reversed by treatment with the CRF₁ antagonist antalarmin. (B) Hippocampal Glua1 gene expression is significantly (*) reduced in saline-treated control Stay fish following presentation of the CS alone on test day compared to all other groups. Importantly Reduced hippocampal Glua1 mRNA expression is reversed following treatment with the CRF₁ antagonist antalarmin. (C) Hippocampal Glua2 gene expression is not influenced by presentation of the CS only on test day. Following presentation of the CS only on test day (day 8), no differences in Glua2 mRNA were detected between groups. Additionally, treatment with the CRF₁ antagonist did not influence Hippocampal GluA2 mRNA expression following presentation of the CS only on test day (day 8).

Levels of the AMPA receptor subunit Glua1 expression were significantly lower in Stay (non-escaping) fish compared to Escape fish (two way ANOVA-Phenotype: F_1,21 = 14.0, p < 0.001; -saline or CRF₁ drug treatment: F_1,21 = 1.9, p > 0.18; -phenotype X saline or CRF₁ antagonist interaction: F_1,21 = 0.69, p > 0.41; one way ANOVA- CS & US controls vs saline-treated: F_3,22 = 3.77, p ≤ 0.025; one way ANOVA- CS & US controls vs Antalarmin-treated: F_3,19 = 1.42, p > 0.28; Figure 5B), while no difference was recorded in Glua2 subunit gene expression (two way ANOVA-Phenotype: F_1,18 = 0.000, p > 0.99; - saline or CRF₁ drug treatment: F_1,18 = 2.13, p > 0.16; -phenotype X saline or CRF₁ antagonist interaction: F_1,18 = 0.012, p > 0.91; one way ANOVA- CS & US controls vs saline-treated: F_3,20 = 0.27, p > 0.84; one way ANOVA- CS & US controls vs saline-treated: F_3,20 = 0.73, p > 0.54; Figure 5C). Classical conditioning also appears to be also true for inhibited expression of Glua1 (Figure 5B). Additionally, inhibited Glua1 gene expression was blocked by CRF₁ antagonism, an anxiolytic treatment.

4. Discussion

The Stress Alternatives Model is chronic-repetitive social stress paradigm that produces two distinctive phenotypes, Escape and Stay, through behavioral, genetic, and signaling neuroplasticity. Behaviorally, active avoidance (Escape) or passively accepting confrontation (Stay) fits well with stress coping strategies present numerous species [27–30, 83]. The Escape and Stay phenotypes appear to be evolutionarily conserved behavioral representations of a balance between pro-stress and anti-stress circuitries in the limbic regions of the brain [21, 64, 66, 67]. Similar outcomes for behavior and neuroendocrine regulation derived from SAM experimentation occur in fish (Figures 1–5)[6, 21], hamsters [84], mice [19, 20, 64–66, 85, 86], and rats [18]. The similarities in these outcomes include: early behavioral phenotypic stability, even distribution of Stay and Escape phenotypes, classically conditioned stress-related responses in both Stay and Escape animals, but with demonstrably greater conditioning, both cued and contextual, in the Stay phenotype, and numerous learned responses underlying specific phenotypic behavior. Another important evolutionarily conserved quality of outcomes derives from plasticity in the stress-related neurocircuitry that produces the phenotypes, such that anxiolytic treatments (including CRF₁ or Orx₁ receptor antagonism, Orx₂ receptor stimulation, Neuropeptide S, exercise) transforms Stay into Escape animals, or modifies behavior toward pro-Escape consequences (such as reducing escape latency) in mice [20, 64, 66] and trout (see Figures 2–4) [6, 21]. In mice, anxiogenic treatments (including the α₂-adrenoreceptor antagonist yohimbine, and Orx₁ receptor stimulation) transforms Escape into Stay animals, or modifies behavior to be more Stay-oriented [20, 64]. We posit that the critical elements producing the evolutionarily conservative phenotypes, Escape and Stay, devolve from neuroplasticity of stress circuitry signaling and gene expression.

Here we show that both saline-treated stress-resilient Escape and stress-susceptible Stay fish exhibit elevated Bdnf gene expression in response to socially stressful interaction, and both groups show learning. Classical conditioning and spatial learning are clearly evident for both groups, although Stay fish only express the spatial Escape task upon relief of social aggression-induced stress responsiveness, via the CRF₁ receptor antagonist antalarmin [20]. We have previously also demonstrated social learning in fish using this model [21]. Additionally, SAM social stress plus classical conditioning modifies hippocampal gene expression in a phenotype dependent manner (Figure 5). Similar results from recent work on mice from our lab demonstrates phenotype-dependent gene expression neuroplasticity in the basolateral amygdala (BLA) and hippocampus following SAM + conditioning [64, 86]. In submissive-Stay trout conditioned to an aggressive social interaction, hippocampal BDNF gene expression increases while GluA₁ expression is reduced, and both effects disappear following CRF₁ receptor antagonism-induced behavioral reversal: when Stay fish become Escape fish. At the same time, while saline-treated Escape fish have significantly elevated Bdnf expression, these mRNA levels are also statistically less than those of saline-treated Stay fish. Additionally, Escape fish show no change in GluA₁ mRNA. Interestingly, the concentration of the stress hormone cortisol ([F]), low in all groups prior to SAM interaction (Figure 3), and elevated by social aggression [39] in the SAM, is only increased in plasma concentration on day 8 following the CS alone [= conditioned response, CR [6]] in saline-treated Stay trout. After CRF₁ receptor antagonist treatment, Escape animals [F] remains low, and Stay animals that become Escape fish, also have un-stimulated [F]. However, plasma cortisol concentrations in Stay fish treated with Antalarmin that do not switch to Escape behavior are also significantly elevated (Figure 3).

As CRF₁ antagonism has a response dependent on phenotype × behavior with respect to plasma [F], we are left to surmise that the stress neurocircuitry that regulates behavior and that controls the neuroendocrine stress response are tightly linked and synchronized. Trout that do not show a behaviorally anxiolytic response to Antalarmin (Escape), and remain Stay animals, appear to be unable to limit stress hormone responsiveness in response to the CS (Figure 3). On the other hand, fish that were stress-susceptible Stay animals, who in response to the CRF₁ antagonist treatment switched phenotypes to Escape, also no longer required elevated plasma [F], and did not exhibit this stress response (Figure 3). This distinction in [F] response, suggests that some neuroendocrine trigger, perhaps [CRF], CRF₂ receptors, or arginine vasotocin (AVT), must have remained capable of stimulating a neuroendocrine stress response [87–89] despite the inhibition of CRF₁ receptors. The results also suggest that a subset of the trout Stay phenotype population remains somewhat insensitive to anxiolytic or antidepressive treatments, similar to human populations [90–96].

It is important that elevation of BDNF expression is measured in both groups of saline-treated and conditioned Escape and Stay fish. However, only saline-treated Stay fish exhibit a significant decrease in GluA₁ gene expression. Similarly, the plasma stress hormone cortisol is also only elevated in Stay trout. In all cases the anti-stress CRF₁ receptor antagonist Antalarmin blocked the effect (Figures 3, 5). For plasma [F], we have verified a CR in elevated concentrations [6], but the elevation in BDNF and reduction in GluA₁ gene expression resulted from the same protocol. This protocol, however, did not affect GluA₂, at the times measured. The data suggest that conditioned responses in BDNF and GluA₁ mRNA are coincident with the CR for [F]. The role for BDNF and GluA₁ in classical conditioning have been demonstrated in other species [50, 51, 97–104]. During turtle eye blink conditioning, BDNF and extracellular signal-regulated kinases (ERK) are necessary for early and late acquisition, leading to upregulation, trafficking, and synaptic localization of GluA₁ subunits, followed by GluA₄ subunits [50, 51, 97, 103–105]. By the time of retention and memory expression, GluA₁ subunit expression is decreasing, while GluA₄ expression remains elevated [50, 51, 97, 103–105]. These results fit well with our experiments, since BDNF and GluA₁ gene expression was measured at the time of the conditioned response (day 8), but also suggests that future experiments should examine trafficking and expression of GluA₄ subunits. Our own recent work in mice demonstrates behavioral conditioned responses in both Stay and Escape mice, with BDNF expression elevated in both phenotypes, but regulated specifically by phenotype through orexin 1 or orexin 2 receptor mechanisms [64]. Our data in trout suggest that similar mechanisms for BDNF-regulated neuroplasticity and AMPA receptor trafficking may occur during fear conditioning.

Since both Escape and Stay groups learned the position of the escape hole, and experienced significantly greater elevation of BDNF coincident with expression of conditioning, potentially, the anxiolytic effect of the CRF₁ antagonist treatment inhibits the conditioned stimulation of BDNF expression and inhibition of GluA₁ expression (Figure 5). The CRF₁ antagonism can release behavioral expression of learned escape, even while this treatment blocks the conditioned responses of increasing BDNF and reduced GluA₁ expression levels in newly escaping fish. This suggests that the development of fear conditioning both promoted the expression of BDNF and inhibited expression of GluA₁.

Initially, we hypothesized that reduced, or smaller increase in, hippocampal BDNF expression was directly related to the absence of learned escape. However, following treatment with the CRF₁ antagonist on day 4, most treated Stay fish begin escaping, with latencies equivalent to the original escaping group after 5 days of learning (Figure 4 dotted line). This suggests Stay (non-escapers) fish learn the location of the escape hole, but do not express escape behavior until after treatment with the CRF₁ antagonist. By the time these antalarmin-treated fish would potentially express the BDNF-related CR, following presentation of the CS only on day 8, BDNF mRNA is not different from controls. In all Stay fish treated with the CRF₁ antagonist, both BDNF and GluA₁ mRNA levels are similar to those in CS only controls (Figure 5). Therefore, inhibiting this stress neuromodulator at CRF₁ receptors reverses not only behavior, but also the conditioned stimulation or inhibition of genes associated with neural plasticity and learning. We have argued that inhibitory modulation of the pro-stress elements of emotional neurocircuitry, via CRF or orexin receptor agents, has implications for affective disorders such as PTSD [64, 66, 67, 85]. In our model, fish [or mice [20]] effectively learn an adaptive spatial task that allows for relief from aggressive trauma, but stress-susceptible Stay individuals are impaired from accessing this relief due to unremitting stress responsiveness, similar to what occurs in post-traumatic stress disorder (PTSD) patients [106, 107]. Anxiolytic drugs, that act directly on specific CRF or orexinergic elements of the stress neurocircuitry involved, have been demonstrated in our SAM protocol to allow stress-vulnerable animals to avoid stressful conditions and responses [20, 64–66], which may prove useful for patients with PTSD, anxiety, and depression.

It is important to note that BDNF expression is rapidly and selectively induced by learning dependent on the hippocampus [45], as is synaptic GluA₁, which are influenced by glucocorticoid stress hormones and their receptors [108, 109]. Additionally, hippocampal BDNF controls synaptic increases in GluA₁ subunits necessary for memory consolidation in inhibitory avoidance [110]. In our study, elevated [F] is measured only in saline-treated, non-escaping Stay fish and reversed by CRF₁ antagonism in those that begin to Escape, but not by those that don’t. Concurrent with increased [F], BDNF mRNA levels are elevated by social stress, more in Escape than Stay fish, and also reversed by CRF₁ antagonism in both groups. Finally, stress induced by social aggression alone (which does promote elevated glucocorticoid levels) does not influence BDNF mRNA levels (Fig. 4A; US only, dotted bar).

5. Conclusions

Utilizing the behavioral strategy of “learned escape” circumvents conditioned increases in plasma cortisol and regionally specific central monoamine activity seen in non-escaping individuals [6]. Concurrent conditioned stress-responsive cortisol concentrations, as well as stimulation of BDNF and inhibition of GluA₁ gene expression in the hippocampus, suggest a suite of molecular, neural, and hormonal changes specific to expression of particular learning modalities. It is significant that this distinctive gene expression of both BDNF and GluA₁, but not GluA₂, is coincident with adoption of mutually exclusive, but equally adaptive, behavioral coping strategies. Additionally, CRF₁ antagonist treatment reverses these conditioned responses, while stimulating this shift in behavioral strategy, from submissive non-escape (Stay), to learned Escape. Taken together, these data suggest that the CRF system, via the CRF₁ receptor, mediates behavioral response to aggressive social interaction [39] and influences gene expression in the hippocampus during conditioning to a socially relevant and powerful fear-inducing stimulus. Moreover, the conditioned responses to a social context in this aquatic vertebrate suggest evolutionary conservation of fear-learning systems and strategies.