Despite extensive research on the impact of emotional stressors on brain function using immediate-early genes (e.g., c-fos), there are still important questions that remain unanswered such as the reason for the progressive decline of c-fos expression in response to prolonged stress and the neuronal populations activated by different stressors. This study tackles these 2 questions by evaluating c-fos expression in response to 2 different emotional stressors applied sequentially, and performing a fluorescent double labeling of c-Fos protein and c-fos mRNA on stress-related brain areas. Results were complemented with the assessment of the hypothalamic–pituitary–adrenal axis activation. We showed that the progressive decline of c-fos expression could be related to 2 differing mechanisms involving either transcriptional repression or changes in stimulatory inputs. Moreover, the neuronal populations that respond to the different stressors appear to be predominantly separated in high-level processing areas (e.g., medial prefrontal cortex). However, in low-hierarchy areas (e.g., paraventricular nucleus of the hypothalamus) neuronal populations appear to respond unspecifically. The data suggest that the distinct physiological and behavioral consequences of emotional stressors, and their implication in the development of psychopathologies, are likely to be closely associated with neuronal populations specifically activated by each stressor.
Exposure to purely emotional (psychological or psychogenic) or predominantly emotional stressors (e.g., restraint, immobilization, forced swim [FS] in warm water) triggers behavioral and physiological responses that can be considered as anticipatory toward a possible serious challenge to homeostasis. These responses can be generated by an innate adaptive predisposition or by classical conditioning processes. Information relative to emotional stressors, including contextual information or emotional significance is processed in a wide range of brain areas along the forebrain and the brainstem (Herman et al. 2003, 2005). All this information converges at different effector brain areas including the paraventricular nucleus of the hypothalamus (PVN), which is a complex structure that initiates endocrine and autonomic responses to stress.
The expression of immediate-early gene (IEG), namely c-fos, has greatly contributed to our understanding of brain processing of stressors, but there are some relevant questions that remain unanswered. First, it is well-known that c-fos gene expression progressively declines in response to prolonged exposure (a few hours) to acute emotional stressors (Trnecková et al. 2007; Weinberg et al. 2007). The precise mechanisms involved are unknown, but there are 2 possible explanations: (1) c-fos gene expression could be affected by transcriptional repression mechanisms or (2) inputs arriving to neurons progressively decline because no apparent actual threat to homeostasis appears. Second, a widespread and roughly similar brain activation pattern is observed in response to different emotional stressors (Armario 2006). Considering this pattern, it appears plausible that some of the activated neurons are common to the different stimuli (not-stressor specific), reflecting general arousal. However, other neurons could be specifically activated (stressor-specific), thus explaining the clearly distinct behavioral consequences of exposure to different emotional stressors.
To our knowledge, there is no previous study aimed to characterize the stressor-specificity of neurons activated by emotional stressors, but some recent optogenetic studies have demonstrated that different types of neurons in the cortical (Root et al. 2014) and basolateral amygdala (Gore et al. 2015; Namburi et al. 2015) mediate fearful and rewarding stimuli. Classical IEGs experiments do not allow to distinguish neurons specifically activated by different stimuli. However, the time-course of subcellular localization of IEG RNA and the temporal dynamics of IEG mRNA and protein can be exploited for this purpose. The “cellular compartment analysis of temporal activity by fluorescence in situ hybridization” (catFISH), developed by Guzowski et al. (1999, 2005), is based on the temporal localization of the RNA of the IEG Arc (activity-regulated cytoskeleton-associated protein). This molecule remains into the nucleus up to 16 min emerging to the cytoplasm 20–45 min after a neuronal activation. Thus, fluorescent cytoplasmic signal indicates a response to a first experience, whereas intranuclear signal corresponds to the most recent stimulus applied about 30 min later. An alternative approach is based on the different temporal dynamics of c-fos mRNA and c-Fos protein. A brief exposure to a stimulus increases c-fos mRNA levels with a maximum at about 30 min that markedly decreases about 2 h later when the protein is at its maximum (Kovács and Sawchenko 1996; Zangenehpour and Chaudhuri 2002; Kovács 2008). Therefore, those neurons showing both mRNA and protein after a second stimulus are responsive to both stimuli, whereas those only showing one marker are specifically responsive to the first (protein) or to the second (mRNA). In the present work, we followed the latter approach performing double labeling of c-Fos protein and c-fos mRNA combining immunofluorescence and fluorescent in situ hybridization (IF–FISH), respectively.
Animals were exposed to prolonged (4 h) immobilization on board (IMO), a severe stressor, and then to a relatively brief (20 min) exposure to a milder novel stressor (FS). By using this particular approach, we wanted to test 3 main hypotheses: (1) the progressive decline in c-fos expression during a prolonged exposure to a stressor could involve either transcriptional repression and/or changes in neuronal inputs, (2) response to particular emotional stressors probably involves activation of specific neuronal populations that would define their different behavioral and cognitive consequences, and (3) the degree of specificity of these neurons could be dependent on the brain area and its hierarchical position in the processing of stressors.
The experimental design is based on the dynamics of c-fos mRNA and c-Fos protein showed in Figure 1. This allows us to analyze, by radioactive in situ hybridization (rISH), the c-fos response to a different stimuli applied after prolonged IMO. Additionally, if such exposure causes an increase in c-fos expression, we can distinguish by IF–FISH: the neurons activated in response to the first stressor (IMO), ideally IF+/FISH–; those exclusively activated by the second stressor (FS), ideally IF–/FISH+; and those activated by both, ideally IF+/FISH–. The c-fos expression after prolonged stress was also compared with the peripheral stress response (activation of the hypothalamic–pituitary–adrenal [HPA] axis). Thus, we assessed the expression of the corticotrophin releasing hormone (CRH) gene (CRH heteronuclear RNA [hnRNA]) and the levels of HPA hormones (adrenocorticotropic hormone [ACTH] and corticosterone).
The present work advances our knowledge in regard to processes affecting the decrease of c-fos expression in response to prolonged stress. Moreover, we also demonstrate that the use of IF–FISH double labeling is an excellent tool to find stimulus-specific neuronal populations. We believe that the existence of stressor-specific neuronal populations is a clear step forward in the study of the pathogenesis of stress-related disorders and their successful treatment.
In all experiments, 2- to 3-month-old Sprague-Dawley male rats bred at the Animal Facility Service of the Universitat Autònoma de Barcelona were used. Animals were housed in pairs in polypropylene opaque wide-topped, solid-bottom cages (21.5 × 46.5 × 14.5 cm; “Type 1000 cm2”, Panlab) containing wood shaving bedding (Lignocel ¾, Harlan Interfauna Ibérica). They were kept understandard conditions of temperature (21 ± 1 °C) and a 12:12 h light/dark schedule (lights on at 8:00 am). Food (SAFE-diet A04, Panlab) and water were available ad libitum. The experimental protocol was approved by the Committee of Ethics of the Universitat Autònoma de Barcelona and the Generalitat de Catalunya. It was also carried out in accordance with the European Communities Council Directive (2010-63-UE) and Spanish legislation (RD 53/2013).
Rationale for the Selection of the Stressors and the Brain Areas Studied
We studied changes in c-fos expression after the exposure to 2 predominantly emotional stressors: IMO and FS. These 2 stressors differ in intensity, measured by the magnitude of the physiological changes they elicit (Rabasa et al. 2015), and also in their qualitative features, for example, emotional impact, different coping possibilities. For the IMO procedure, the limbs of the animals were taped to 4 metal mounts attached to a board (Muñoz-Abellán et al. 2011). Lateral head movements were restricted with 2 plastic pieces. For FS, rats were placed on methacrylate cylinders (40 cm height × 19 cm diameter) filled with water at 36 °C (Ons et al. 2004).
We chose to study several brain areas that are critically involved in the control of the stress response (Herman et al. 2003): medial prefrontal cortex (mPFC), ventral subdivision of the lateral septum (LSv), medial amygdala (MeA), and medial parvocellular dorsal subdivision of the PVN (mpdPVN). Some of these areas are known to be sensitive to the intensity of stressors in terms of c-fos expression (MeA, LSv, and PVN; Campeau et al. 2002; Ons et al. 2004; Pace et al. 2005). Only the mpdPVN subdivision was studied in the PVN due to its relevance in the stress response as the main CRH producing region (Ulrich-Lai and Herman 2009).
The aim of the Experiment 1 was to evaluate if there are neurons capable of responding to a novel stressor (FS) in terms of c-fos expression after prior exposure to prolonged IMO. Rats were randomly assigned to 5 groups: BASAL, that received no treatment (n = 8); IMO 20′, immobilized for 20 min (n = 8); FS 20′, exposed to FS for 20 min (n = 8); IMO 4 h, immobilized for 4 h (n = 8); and IMO–FS, immobilized for 3 h and 40 min and exposed to FS during 20 additional minutes (n = 8). Immediately after the treatments, animals were anesthetized and fixed by transcardial perfusion as described below. Brains were extracted for the analysis of c-fos mRNA and CRH hnRNA levels by rISH.
The aim of the Experiment 2 was to demonstrate whether or not animals previously exposed to a prolonged IMO are able to release ACTH when they are exposed to a novel stressor (FS). One group of animals were exposed to IMO for 4 h (n = 8), whereas the other group remained in the home-cage (n = 8). After this period, both control and IMO groups were exposed to FS for 20 min. Blood samples were taken in both groups by the tail-nick procedure (Belda et al. 2008) at 20 min; 4 h after initial exposure to IMO (or the corresponding time in controls); and immediately after FS (see Fig. 3C). About 300 μL of blood per sample were collected into ice-cold EDTA capillary tubes (Sarsted, Spain). The plasma obtained after centrifugation was stored at −20 °C until further analysis of ACTH and corticosterone levels.
The previous experiment showed c-fos induction in response to FS after prolonged IMO. Thus, the aim of the Experiment 3 was to distinguish between 2 main possibilities: (1) the response was restored due to reactivation of the same neurons that responded to prolonged IMO or (2) novel, FS-specific neuronal populations were activated. Groups were similar to the Experiment 1: BASAL, that received no treatment (n = 4); IMO 20′, immobilized for 20 min (n = 8); FS 20′, exposed to FS for 20 min (n = 8); IMO 4 h, immobilized for 4 h (n = 6); and IMO–FS, immobilized for 3 h and 40 min and exposed to FS during 20 additional minutes (n = 8). An additional group was included in this experiment (IMO–IMO [n = 6]) to distinguish the activation caused specifically by FS from a possible nonspecific activation due to the mere manipulation of animals when they are released from the IMO board. These animals were immobilized for 3 h and 40 min, released and immediately re-exposed to IMO for 20 additional minutes. Animals were anesthetized and perfused after the treatments. Brains were extracted for the simultaneous detection of c-fos mRNA and c-Fos protein by IF–FISH double labeling.
Plasma ACTH and costicosterone levels were determined by double antibody radioimmunoassay (RIA), as previously described (Muñoz-Abellán et al. 2011). In brief, ACTH RIA used 125I-ACTH (GE Healthcare) as tracer, rat synthetic ACTH 1-39 (Sigma) as standard, and an antibody raised against rat ACTH kindly provided by Dr William Engeland (Department of Neuroscience, University of Minnesota, MN, USA). Corticosterone RIA used 125I-corticosterone-carboximethyloxime-tyrosine-methyl ester (ICN-Biolink 2000), synthetic corticosterone (Sigma) as the standard, and an antibody raised in rabbits against corticosterone–carboximethyloxime-BSA kindly provided by Dr Gábor Makara (Institute of Experimental Medicine, Budapest, Hungary). We followed the RIA protocol recommended by Dr Makara (plasma corticosteroid-binding globulin was inactivated by low pH). All samples to be statistically compared were run in the same assay to avoid interassay variability. The intraassay coefficient of variation was 6.0% for ACTH and corticosterone. The sensitivity was 17 pg/mL for ACTH assay and 0.1 μg/dL for corticosterone assay.
Perfusion and Histological Processing
Animals were anesthetized by isoflurane inhalation (Esteve Laboratories) and fixed by transcardial perfusion, introducing first a physiological saline solution (0.4% NaCl) and then a fixative solution (4% PFA + 3.8% Borax). After perfusion, brains were removed, postfixed in the same solution and embedded in a cryoprotectant solution containing 30% sucrose in potassium phosphate-buffered saline (0.2 M NaCl, 43 mM potassium phosphate, KPBS). Brains were then frozen in dry ice cooled isopentane at and preserved at −80 °C until cryosection. About 14 μm thick coronal sections were collected in cryoprotectant solution (0.05 M sodium phosphate buffer pH 7.3, 30% ethylene glycol, 20% glycerol) and stored at −20 °C. All histological techniques were performed on at least 6 sections per area.
In Situ Hybridization
Preparation of probes
C-fos probe was generated from an EcoRI fragment of rat c-fos cDNA (Dr Inder Verma, The Salk Institute, CA, USA; van Beveren et al. 1983), subcloned into pBluescript SK-1 (Stratagene) and linearized with SmaI. CRH hnRNA probe was generated from a HindIII/EcoRI fragment of rat CRH intron 1 (Dr Lisa Bain, University of Michigan, MI, USA) subcloned into pGem-3 plasmid (Stratagene) and linearized with HindIII. Both plasmids were generously provided by Dr Serge Rivest (Laval University, Quebec, Canada). Radioactive antisense cRNA copies were generated using a transcription kit (SP6/T7 Transcription Kit, Roche). α-35S-UTP (1250 Ci/mmol, PerkinElmer) or Digoxigenin-11-UTP (DIG RNA Labeling Mix, Roche) were used as labeled ribonucleotides for the radioactive and nonradioactive probes, respectively. The transcription process was stopped by adding 40 μL of a sodium chloride–Tris–EDTA buffer solution (STE: 0.1 M NaCl, 10 mM Tris–HCl pH 8.0, 1 mM EDTA). Then, the product was heated during 5 min at 65 °C. The probes were isolated through gel filtration columns (mini Quick Spin RNA Columns, Roche) and stored at −20 °C.
The protocol used for rISH was adapted from Simmons et al. (1989). All the solutions used before the hybridization with the probe were pretreated with diethylpyrocarbonate (DEPC) and sterilized. Sections were previously mounted on positive-charged slides (Superfrost Plus, Thermo Scientific). Tissue was postfixed in 4% PFA + 3.8% Borax, washed in KPBS and incubated in the presence of 0.01 mg/mL of proteinase K (Roche) in an appropriate buffer (0.1 M Tris–HCl pH 8.0, 50 mM EDTA pH 8.0) for 15 min. After digestion, sections were rinsed in DEPC-treated water and acetylated during 10 min in 0.25% acetic anhydrous in 0.1 M TEA pH 8.0. Finally, they were washed in 2× saline-sodium citrate (SSC) solution, dehydrated through graded concentrations of ethanol and air-dried. Thereafter, 100 μL of hybridization buffer (50% formamide, 0.3 M NaCl, 10 mM Tris–Cl pH 8.0, 1 mM EDTA pH 8.0, 1× Denhardts, 10% dextranesulphate, yeast tRNA 500 μg/mL, and 10 mM DTT) containing the 35S labeled probe (1 × 106 cpm/100 μL) or the DIG labeled probe (1:2000) were added onto each slide and sealed with a coverslip. Sections were incubated for 16–18 h in a humid chamber at 60 °C. After this time, sections were washed in 4× SSC and RNA digested with RNase A (GE Healthcare) at 0.02 mg/mL in an appropriate buffer (0.5 M NaCl, 10 mM Tris–HCl pH 8.0, 1 mM EDTA pH 8.0). After RNA digestion, sections were washed in descending concentrations of SSC (2× to 0.5×) containing 1 mM DTT and heated at 60 °C in 0.1× SSC during 30 min.
Autoradiographic detection (rISH)
After the last wash in 0.1× SSC at 60 °C sections were dehydrated through graded concentrations of ethanol and air-dried. Then, slides were exposed to XAR-5 Kodak Biomax MR autoradiography films during an appropriate time depending on the probe used and the brain area studied. In order to avoid signal saturation, optimal exposure time was chosen comparing densitometric values of images of exposed sections that showed low, medium, and high intensity with a signal saturation curve obtained from films exposed to 14C microscales (GE Healthcare). Slides were counterstained in 0.25% thionine as histological control.
IF–FISH Double Labeling
Free-floating sections were washed with DEPC-treated KPBS to remove the cryoprotectant solution and then incubated directly with polyclonal rabbit antiserum (sc-52; Santa Cruz Biotechnology) at 1:500 in 1% Blocking Reagent (Roche) in KPBS, for 16 h at 4 °C. Sections were then washed in KPBS and incubated with an fluorochrome conjugated secondary antibody (Alexa Fluor 568 Goat Anti-Rabbit IgG [H+L], Invitrogen) at 1:500 during 2 h at room temperature (RT). Sections were then washed in KPBS and mounted on positive-charged slides. FISH protocol was the same as the previously described for ISH but reducing the proteinase K digestion time from 15 to 8 min to avoid damage to the Ag–Ab complex. After the last wash in 0.1× SSC at 60 °C sections were equilibrated in Tris–buffered saline with Tween20 (T-TBS; 0.1 M Tris–HCl pH 7.5, 0.15 M NaCl, 0.05% Tween20).
Sections were previously incubated for 30 min in blocking buffer (2% bovine serum albumin [BSA] in T-TBS) for immunodetection of DIG. Then, blocking buffer was removed and a peroxidase-conjugated anti-DIG antibody (anti-DIG-POD, Roche) was added at 1:500 in 1% fetal calf serum, 0.1% acetylated BSA, and 0.1% Tween20 in TBS using incubation chambers (CoverWell, Grace Bio-Labs). After incubation during 2 h at RT, incubation chambers were removed and slides were washed in T-TBS. Signal was amplified using a tyramide signal amplification kit (TSA-plus Fluorescein, PerkinElmer). After amplification, nuclei were counterstained with Hoechst 33 258 pentahydrate (Invitrogen) at 1:10 000 in TBS. After a last rinse in deionized water, slides were covered with an aqueous mounting medium (Fluoromount, Sigma) and their edges sealed with rapid mounting medium (Entellan, Merck). Slides were stored at 4 °C until confocal image capture.
Image Capture and Analysis
The same coordinates were used for each particular area in all histological analysis. The reference was the stereotaxic atlas by Paxinos and Watson (2014). The mPFC subdivisions were: cingulate (Cg), prelimbic (PrL), and infralimbic (IL). These areas were quantified between Bregma 3.20 and 2.70 mm. The other areas analyzed were: LSv, between Bregma 1.20 and 0.20 mm; MeA, between Bregma −2.56 and −2.80 mm; and mpdPVN, between Bregma −1.70 and −1.90 mm. For the processing and the analysis of images, ImageJ public domain image processing software (FIJI v1.47f) was used. Images were previously coded before counting.
Images of autoradiographic films were taken using a bright-field microscope (NIKON, Eclipse E400) coupled to a digital camera (NIKON, DMX 1200). Radioactive signal was semiquantitatively determined measuring the product of the labeled area and the mean gray value (integrated optical density, IOD) in defined areas. The background signal was determined by the IOD of an unlabeled brain region (corpus callosum). Average of at least 6 fields (1 field per hemisphere of 3 slices) per brain area and animal was used for the statistical analysis.
IF–FISH Double Labeling Images
Fields were selected by epifluorescence with the minimal light intensity before capturing the confocal images to avoid photobleaching. A stack of 5 sections per field was taken using a Leica TCS SP5 confocal imaging system (Servei de Microscòpia, Institut de Neurociències, UAB). Sections were assembled obtaining a 6 μm projection. Background signal was determined by the mean integrated density (ID) value of 20 unlabeled regions of interest (ROIs) per projection. IF and FISH particles were considered positive cells when ID value of ROIs exceeded 3 standard deviations over the average background. This resulted in 3 different types of cells: positive only for protein (IF+/FISH–), positive only for mRNA (IF–/FISH+), and double labeled (IF+/FISH+). Total ID of each group of neurons (IF+/FISH+ and IF–/FISH+) was also determined as an index of their degree of activation. Average of at least 4 fields (1 field per hemisphere of 2 slices) per brain area and animal was used for the statistical analysis.
The statistical package for social science (SPSS, version 24 for Windows) software was used for the analysis. Statistical analyses were performed using the generalized linear model (GzLM; McCulloch and Searle 2001) when only between-subjects factors were included, and generalized estimating equations (GEEs) when within-subjects factors were present in the analysis (Hardin and Hilbe 2003). These models are a more flexible statistical tool than the standard general linear model because several types of distribution and different covariance structures of repeated measures can be chosen. In addition, the GzLM does not require homogeneity of variances. The significance of the effects was determined with the Wald chi-square statistic (χ2). After the main analysis, appropriate pair-wise comparisons were carried out and corrected by the sequential Bonferroni procedure (Holm 1979). The criterion for significance was set at P < 0.05. Particular comparisons were planned (see Results) to avoid or reduce corrections for multiple comparisons in all tests. All samples to be statistically compared were processed in the same assay to avoid interassay variability.
Experiments 1 and 2
In the Experiment 1, c-fos mRNA and CRH hnRNA rISH data were analyzed by a GzLM with one between-subjects factor (GROUP, 5 levels). The planned comparisons were as follows: (1) IMO 20′ versus FS 20′ to compare response to both stressors; (2) IMO 4 h versus IMO 20′ to study the temporal dynamics of the response; (3) IMO–FS versus IMO 4 h to evaluate the response to the novel stressor; and (4) IMO–FS versus FS 20′ to compare the response to the novel stressor with or without previous exposure to the prolonged IMO.
A low or nondetectable signal for c-fos mRNA was obtained in basal conditions. However, exposure to both IMO and FS for 20 min (short) resulted in a widespread induction of c-fos mRNA in all studied regions (representative images in Fig. 2A,C,E,G). Analysis by GzLM of total IOD showed a significant effect of GROUP in all studied regions: Cg1 [Wald χ2(4) = 77.2, P < 0.001], PrL [χ2(4) = 82.98, P < 0.001], IL [χ2(4) = 76.5, P < 0.001], LSv [χ2(4) = 113.1, P < 0.001], MeA [χ2(4) = 251.2, P < 0.001], and mpdPVN [χ2(4) = 204.8, P < 0.001]. Comparison of short exposure to IMO and FS showed similar responses in mPFC and LSv (Fig. 2B,D), but the levels were higher after IMO in MeA and mpdPVN (Fig. 2F,H). Exposure to prolonged IMO (4 h) resulted in decreased c-fos mRNA levels when compared with short IMO in all areas. The IMO–FS group presented higher c-fos mRNA levels than prolonged IMO in all areas. c-fos mRNA levels reached in the IMO–FS group were equivalent to those of FS group in all areas except in the mpdPVN, where lower levels were observed (P < 0.05).
CRH hnRNA levels of stressed groups were always higher than the basal group (representative images and quantification in Fig. 3A,B). GzLM analysis showed a significant effect of GROUP [χ2(4) = 52.83, P < 0.001]. Further comparisons indicated that CRH hnRNA levels were lower in the FS and prolonged IMO groups when compared with the short IMO group (P < 0.05). Additionally, the IMO–FS and the prolonged IMO group presented similar levels of CRH hnRNA suggesting a blockage in the response to a novel stressor.
In the Experiment 2, plasma ACTH and corticosterone data (Fig. 3D,E) were analyzed by GEE, with treatment (CONTROL, IMO) as the between-subjects factor, and blood sampling time (TIME) as the within-subjects factor. Planned comparisons were (1) IMO 20′ versus FS 20′ to compare the response to both stressors; (2) IMO 20′ versus IMO 4 h to demonstrate changes after prolonged exposure; (3) IMO–FS versus IMO 4 h to evaluate responsiveness to the novel stressor; and (4) IMO–FS versus FS 20′ to compare the response to the novel stressor with or without previous exposure to the prolonged IMO. GEE analysis of ACTH showed a significant effect of GROUP [χ2(1) = 219.1, P < 0.001], TIME [χ2(2) = 208.8, P < 0.001], and the interaction GROUP × TIME [χ2(2) = 190.6, P < 0.001]. The programmed comparisons revealed that exposure to FS resulted in lower levels of ACTH than short IMO, which is concordant with the results observed in the mpdPVN (CRH hnRNA and c-fos mRNA). After prolonged IMO, ACTH levels decreased to 30% of short IMO levels, but they were still above basal levels. Exposure to FS after prolonged IMO did not result in any additional increase of ACTH levels, suggesting that its release is completely blocked. GEE analysis of corticosterone showed a significant effect of GROUP [χ2(1) = 276.0, P < 0.001], TIME [χ2(2) = 543.6, P < 0.001], and the interaction GROUP × TIME [χ2(2) = 358.3, P < 0.001]. The programmed comparisons showed not significant differences except that corticosterone levels were higher after FS than after short IMO.
In the Experiment 3, different analyses were performed:
The responses to IMO 20′ and FS 20′ were compared (t-test) regarding the total number of activated neurons (FISH+) and total ID values.
To evaluate reactivation of neurons in response to the same or a novel stressor after prolonged IMO, the number of double labeled cells (IF+/FISH+), and their ID were analyzed by GzLM with one between-subjects factor (GROUP, with 3 categories [IMO 4 h, IMO–IMO, and IMO–FS]). Planned comparisons were: IMO–FS versus IMO–IMO, IMO–FS versus IMO 4 h, and IMO–IMO versus IMO 4 h.
To evaluate activation of new neurons in response to a second stressor, the number of cells positive only for mRNA (IF–/FISH+) and their ID were analyzed in the same way as for IF+/FISH+ neurons.
Comparison of the total number of activated neurons (FISH+ cells) between IMO 20′ and FS 20′ groups showed a higher response after IMO in MeA [t(6) = 3.665, P < 0.05] and mpdPVN [t(6) = 2.678, P < 0.05]. In the latter nucleus, both groups also differed in total ID value [t(6) = 2.948, P < 0.05].
Statistical analysis by GzLM of IF+/FISH+ cells showed a significant effect of GROUP (groups IMO 4 h, IMO–IMO, and IMO–FS) in all studied regions (representative images in Fig. 4A,C,E,G): PrL [χ2(2) = 23.97, P < 0.001], LSv [χ2(2) = 27.48, P < 0.001], MeA [χ2(2) = 9.64, P < 0.01], and mpdPVN [χ2(2) = 20.46, P < 0.001], suggesting that a second exposure to a short duration stressor induces reactivation of neurons already activated during prolonged IMO (Fig. 4B1,D1,F1,H1). Equivalent results were obtained for the ID of IF+/FISH+ cells: PrL [χ2(2) = 29.77, P < 0.001], LSv [χ2(2) = 25.25, P < 0.001], MeA [χ2(2) = 14.79, P = 0.001], and mpdPVN [χ2(2) = 25.76, P < 0.001], indicating that exposure to a second stressor increases the degree of activation of neurons activated during prolonged IMO (Fig. 4B2,D2,F2,H2).
GzLM analysis of the number of IF–/FISH+ cells showed a significant effect of GROUP (groups IMO 4 h, IMO–IMO, and IMO–FS) in PrL [χ2(2) = 57.60, P < 0.001], LSv [χ2(2) = 237.11, P < 0.001], and MeA [χ2(2) = 14.35, P = 0.001], but not in the mpdPVN. This suggests that in the 3 former regions a second exposure to a short duration stressor induces activation of new neurons (Fig. 4B1,D1,F1,H1). Equivalent results were obtained for ID of IF–/FISH+ cells: PrL [χ2(2) = 49.25, P < 0.001], LSv [χ2(2) = 190.59, P < 0.001], and MeA [χ2(2) = 16.51, P < 0.001], supporting also the activation of new neurons in the above mentioned regions (Fig. 4B2,D2,F2,H2).
In the mPFC, ~50% of neurons already activated during prolonged IMO were reactivated (IF+/FISH+) after re-exposure to IMO or exposure to FS (P < 0.001 in both cases). Both re-exposure to IMO and exposure to FS-induced activation of new neurons (IF–/FISH+; P < 0.001). However, a higher number of new neurons was measured after exposure to FS when compared with re-exposure to IMO (P = 0.001; Fig. 4B1). The analysis of ID values reflected the same results (Fig. 4B2).
In the LSv, ~30% of neurons already activated during prolonged IMO were reactivated after re-exposure to IMO (P = 0.001) and exposure to FS (P < 0.001). Only the exposure to FS-induced activation of new neurons (P < 0.001; Fig. 4D1). Analysis of ID values reflected the same results (Fig. 4D2).
In the MeA, ∼30% of neurons already activated during prolonged IMO were reactivated after re-exposure to IMO (P < 0.05) and exposure to FS (P < 0.01). Only the exposure to FS-induced activation of new neurons (P < 0.001). The number of IF+/FISH+ and IF–/FISH+ cells did not differ from those obtained after re-exposure to IMO (Fig. 4F1); however, greater ID values were observed in those exposed to FS when compared with those re-exposed to IMO (both P < 0.05; Fig. 4F2).
In the mpdPVN, ~30% of neurons already activated during prolonged IMO were reactivated after re-exposure to IMO (P < 0.01) and exposure to FS (P < 0.001). No activation of new neurons was observed (Fig. 4H1). Analysis of ID values reflected the same results (Fig. 4H2).
The aims of the present work were to explore the processes involved in the progressive decrease of c-fos transcription after prolonged exposure to a severe—predominantly emotional—stressor (IMO) and determine the specificity of neuronal populations activated by 2 different predominantly emotional stressors (IMO and FS). The study was performed in some relevant stress-related brain areas including those whose activation is positively related to the intensity of stressors (e.g., LSv or MeA) and those not sensitive to this factor (e.g., mPFC).
The above 2 aims are closely related. Brain c-fos expression declines strongly after prolonged exposure to emotional stressors (Imaki et al. 1992; Senba et al. 1994; Trnecková et al. 2007). There are 2 main hypotheses to explain this reduction: (1) c-fos expression could be affected by transcriptional repression mechanisms, including repression by Fos protein probably through serum response element (Morgan and Curran 1991) and (2) prolonged exposure to an emotional stressor without any physical injury would lead to a reduction of synaptic inputs with the consequent reduction of c-fos expression. According to the first hypothesis, if most neurons were not specifically activated by all types of emotional stressors, they would not be able to respond to a novel stressor after prolonged exposure to a first stressor. Under the second hypothesis, neurons could be sensitive to a novel stressor despite activation by a previous stressor. Moreover, c-fos response to a novel stressor could also be observed if a population of neurons is specifically activated by this stimulus. Therefore, the response to a novel stressor could involve 2 clearly different but not mutually exclusive processes: the reactivation of previously activated neurons or the activation of a different neuronal population.
We performed a first experiment using rISH to discern between the 2 main hypotheses. The strong decrease of c-fos expression after prolonged IMO was confirmed, although c-fos mRNA levels were still above those of unstressed controls likely due to the high intensity of the stressor. Moreover, short-term exposure to IMO and FS caused a roughly similar activation in the different subdivisions of the mPFC and the LSv. However, a greater induction was observed after IMO in the MeA and the mpdPVN. Similarly, CRH gene expression in the mpdPVN was greater after IMO than after FS. These data fully agree with the plasma ACTH response to both stressors, supporting that activation of the HPA axis is very sensitive to the intensity of stressors at all its levels (García et al. 2000; Martí et al. 2001; Burow et al. 2005; Pace et al. 2005).
Several previous reports are in agreement with our present data in which c-fos expression in the mpdPVN and MeA reflects the intensity of stressors (Campeau and Watson 1997; Ons et al. 2004; Pace et al. 2005). In contrast, no differences were noted in the present work in the LSv although c-fos expression in this area has been previously reported to discriminate between stressors differing in intensity (Campeau and Watson 1997; Ons et al. 2004). It is possible that among those areas potentially sensitive to the intensity of stressors differences might exist in the precise relationship between c-fos expression and stressor intensity. Thus, some areas could discriminate better between low and intermediate intensity stressors rather than between intermediate and high intensity stressors. In this study, the expression of c-fos in the mPFC was similar after IMO and FS. This is compatible with several previous reports supporting that there is no relationship between the intensity of predominantly emotional stressors and c-fos expression in this area (Campeau and Watson 1997; Ons et al. 2004).
Rats previously exposed to prolonged IMO and exposed later to FS presented an additional increase of c-fos expression in all regions. In the mPFC, LSv, and MeA, c-fos mRNA reached levels equivalent to those observed in animals only exposed to FS. However, in the mpdPVN, levels were below those achieved after FS suggesting impaired responsiveness of neurons to novel stressors. Despite this partial c-fos activation, CRH gene expression in mpdPVN was completely blocked when rats were exposed to FS after prolonged IMO. Dissociation between c-fos and CRH expression in mpdPVN neurons has been sometimes found (e.g., Ginsberg et al. 2003; Vallès et al. 2003) as they are 2 parallel rather than causally related processes. In the present study, repression of CRH gene through the inducible cAMP early repressor is likely to be responsible for the lack of response to FS (Shepard et al. 2005; Misund et al. 2007).
Importantly, exposure to FS after prolonged exposure to IMO did not further elevate plasma ACTH levels in contrast with the marked increase observed in stress-naive rats. This result indicates that prolonged IMO completely blocked the response of corticotropic cells to novel stressors. A similar study using prolonged exposure to a less severe method of immobilization (immobilization with mess) reported a partial ACTH response to the novel stressor (restraint) (Dhabhar et al. 1997). Thus, the lack of responsiveness of the corticotropic cells observed in the present study could be related to the severity of the IMO procedure. Therefore, the progressive decline of CRH gene expression and ACTH release after prolonged exposure to a severe stressor appears to reflect an incapability of PVN neurons and corticotropic cells to maintain these functions. On the basis of some previous reports, both familiarization with a stressor, negative glucocorticoid feedback and exhaustion of corticotropes might be involved in such HPA repression (Rivier and Vale 1987; Martí et al. 1999). Plasma corticosterone levels were in some instances not parallel to those of ACTH. However, the results can be easily reconciled considering the following: (1) adrenocortical secretion was probably at its maximum with the plasma levels of ACTH obtained after prolonged IMO (Keller-Wood et al. 1983), which explains that there are no differences when compared with short IMO and (2) the higher levels of corticosterone after short FS than after short IMO could be explained by an increase in adrenocortical responsiveness to stress related to circadian rythm (not associated with changes in ACTH) (Dunn et al. 1972; Engeland et al. 1977). In fact, no differences were found between the IMO–FS and the FS groups, in accordance with the lack of response of ACTH to the novel stressor.
The c-fos expression data do not support the dual hypothesis that most neurons are not specifically activated by all types of emotional stressors and that prolonged exposure represses c-fos expression. However, there are 2 possible explanations for the results: first, neurons are not-stressor specific but the progressive reduction of the response to prolonged IMO is due to diminished synaptic inputs that recover after exposure to a novel stressor; second, a novel population of FS-specific neurons was activated. To distinguish between these 2 possibilities, we designed a new experiment combining IF and FISH and introducing an additional group: rats exposed to prolonged IMO that were released from the board and immobilized again for 20 min. This group was included to detect any effect of the procedures not directly related to the novel stressor.
IF–FISH allows us to count the number of neurons activated and also the overall activation degree (ID signal). Results were similar using both indexes and only the number of neurons will be discussed. Regarding the mPFC, only PrL cortex was analyzed by IF–FISH double labeling because a similar response pattern was observed in the previous experiment in the 3 mPFC subregions studied (Fig. 2B). Thus, it can be expected that the conclusions on stressor-specific neuronal populations obtained in PrL can be applied to the other subregions of the mPFC.
As expected, prolonged exposure to IMO strongly decreased the number of FISH+ neurons in all areas, suggesting that most of neurons activated shortly after the stressors were unable to maintain sustained activation. In PrL, the number of IF+/FISH+ neurons—which reflect reactivation of neurons already activated by prolonged IMO—increased to the same extent in the IMO–IMO and IMO–FS groups. This suggests that some neurons activated by IMO were reactivated when animals were released from the stressor regardless of whether or not they were further exposed to the same stressor or to a novel one. This neuronal population activated when animals were released from IMO is likely to be nonspecifically activated by multiple stressors. The fact that it was activated for a second time suggests that the decline in c-fos expression in response to a prolonged emotional stressor is not due to transcriptional repression. In the IMO–IMO group, the number of IF–/FISH+ neurons—which reflect the newly activated neurons—increased, suggesting that some signals associated to the release of the animals from the board activated a novel population of neurons even if the animals were immobilized again. There are at least 2 possible explanations for these findings. First, the mere manipulation of the rats could represent a situation of unknown consequences and some stress-sensitive neuronal populations are activated; second, there could be a population of PrL cortex neurons that is specifically activated when animals are released from the stressful situation. To our knowledge, there is no previous study on this subject, but there is some evidence that stress termination can activate dopamine release (Imperato et al. 1992) and has rewarding properties (Shen et al. 2010, 2011). Importantly, an additional recruitment of new neurons was observed in the IMO–FS group strongly suggesting that exposure to FS activate a population of neurons not previously activated by IMO. We can then conclude that the majority of PrL neurons are commonly activated in response to both stressors, but certain neuronal population appears to be FS-specific.
As in the PrL, in LSv, MeA and mpdPVN, the number of IF+/FISH+ neurons increased to the same extent in the IMO–IMO and IMO–FS groups. These results indicate that some of the neurons activated by IMO were reactivated by procedures associated with the release from the board regardless of whether the rats were re-exposed to IMO or exposed to FS. This suggests that exposure to the novel stressor contributed to the reactivation of some neurons already activated by IMO. Overall, these data indicate that the decline in c-fos expression after prolonged IMO is unlikely to be due to transcriptional repression.
Despite the small discrepancies between the different brain regions regarding the reactivation of neurons after an additional short exposure to stress, striking differences were observed regarding activation of new neurons (IF–/FISH+). With the exception of the PrL, no significant increase was found in IF–/FISH+ neurons after re-exposure to IMO. This fact rules out the activation of new neurons in the rest of areas by procedures associated with the release from the board. However, exposure to FS after prolonged IMO did activate new neurons in the LSv and the MeA. In the MeA, the differences between the IMO–IMO and IMO–FS groups were not statistically significant but they were different when evaluating the ID of IF–/FISH+ neurons. It thus appears that in both the LSv and the MeA, there are some neurons that appear to be specifically recruited by FS. Interestingly, no recruitment of new neurons was observed in the mpdPVN, a low-order stress processing area.
One important gap concerning the interpretation of results in the LSv and the mpdPVN is the marked increase in the number of IF+ neurons observed after 20 min of IMO and to a lower extent after 20 min of FS. This means that some of the IF+/FISH+ neurons could be newly activated neurons rather than previously activated neurons in rats exposed to FS after prolonged IMO. However, it is unlikely that they represent neurons which specifically respond to FS as the total number of IF+ neurons did not increase in those areas in the IMO–IMO or the IMO–FS versus the IMO 4 h group (not shown). All these data can be reconciled if most of the neurons showing a prompt increase in the protein would belong to the commonly activated population.
Collectively, the present data tentatively suggest that in high-order brain areas, processing of emotional stressors involves at least 2 populations of neurons: one that appears to be common to different types of emotional stressors, and another one specific for each particular stressor. In our case, some of the commonly activated neurons can be reactivated by procedures associated with the release of the rats from the stressor regardless of whether they are later re-exposed to IMO or to a novel stressor. This suggests that the progressive decline in c-fos expression after the prolonged IMO was due to a reduction in inputs to neurons rather than to the transcriptional repression of c-fos gene. However, another population of neurons was unable to further respond to the acute procedures (those IF+/FISH–) and might represent neurons in which c-fos is affected by transcriptional repression. In low-order brain areas such as the mpdPVN, there was no evidence for stress-specific neuronal populations and the majority of neurons belongs to those showing c-fos repression. This conclusion is in concordance with the lack of induction of the CRH gene in response to the novel stressor. If high levels of glucocorticoids are present when animals are confronted to a novel stressor, as it is the case of prolonged IMO, an additional activation could not have adaptive value and the HPA axis appears to be programmed to impede it.
To our knowledge, this is the first work specifically addressing the specificity of neuronal populations activated by (predominantly) emotional stressors. Although the present data are robust and consistent the study has some limitations that need to be acknowledged. First, identification of positive neurons using IF or FISH can be strongly dependent on the sensitivity of each technique, which could affect the interpretation of the results. Second, combinations of different emotional stressors using short exposures with a resting period in between can contribute to more precisely define stressor-specific versus not-stressor-specific neuronal populations.
In conclusion, the present results suggest that the progressive decline in c-fos expression observed in response to prolonged emotional stress is mediated by different processes probably involving both transcriptional repression and changes in inputs that arrive to neurons. On the other hand, although this study is limited to the 2 employed stressors, it appears that in high-order brain areas, such as the mPFC, the LSv, and the MeA, there are both stressor-specific and not-stressor-specific neuronal populations, whereas in low-order areas, such as the PVN, all neurons appear to respond unspecifically. The existence of a majority of neurons commonly activated by different emotional stressors is compatible with the contribution of arousal centers (e.g., locus coeruleus, nonspecific thalamic nuclei) to brain activation under situations requiring optimum cognitive processing and appraisal processes. Besides, the existence of stressor-specific neuronal populations can explain the extremely different functional consequences of exposure to stressful situations, such as novel environments, IMO, FS, or aversive stimuli inducing fear conditioning. The present work is also in line with very recent efforts to identify and selectively manipulate neurons responding to aversive versus appetitive stimuli in the cortical (Root et al. 2014) and basolateral amygdala (Gore et al. 2015; Namburi et al. 2015). While the task of identifying and manipulating stressor-specific neuronal populations is presumably more complex than merely distinguishing between the valence of stimuli, optogenetic and allied techniques are opening new avenues to explore this possibility.
Spanish grants to A.A. and/or R.N.: “Plan Nacional sobre Drogas, Ministerio de Sanidad, Servicios Sociales e Igualdad” (grant number 2011/021); “Ministerio de Economía y Competitividad” (grant number SAF2014-53876R); “Redes Temáticas de Investigación Cooperativa en Salud (RETICS), Instituto de Salud Carlos III” (grant number RD12/0028/0014); and “Generalitat de Catalunya” (grant number SGR2014-1020). R.A. is supported by a NARSAD Young Investigator Grant (grant number 22434) and “Ramón y Cajal” programme (grant number RYC2014-15784). R.N. is the recipient of an ICREA-ACADEMIA award (2015–19) from “Generalitat de Catalunya”. I.M.-B. was a recipient of a predoctoral fellowship from the Basque Government (grant number 2008-AE).
We thank Dr Núria Daviu, Dr Humberto Gagliano, Dr Xavier Belda, Dr Javier Carrasco and the rest of members of the HPA laboratory (Institut de Neurociències, UAB) for their help in the experiments. We also appreciate the technical support of Mar Castillo [Histology Service (Institut de Neurociències, UAB)], Martí de Cabo (Microscopy Service, UAB), and Núria Barba [Microscopy Service (Institut de Neurociències, UAB)]. Conflict of Interest: None declared.