Address correspondence and reprint requests to Antonio Armario, Animal Physiology Unit, School of Biosciences, Universitat Autònoma de Barcelona, Bellaterra 08193, Barcelona, Spain. E-mail: firstname.lastname@example.org
Expression of c-fos is used for the characterization of brain areas activated by stressors. Recently, some epigenetic markers associated with enhanced transcription have been identified that may be also useful to detect neuronal populations important for the processing of stressors: phosphorylation of histone H3 in serine 10 or 28 (pH3S10 or pH3S28). Then, we compared in rats the response to stress of c-fos and these epigenetic changes. More specifically, we studied the influence of the type of stressor (novel environment vs. immobilization, IMO) and the dynamics of the response to IMO. Stress increased pH3S10 positive neurons, with a more restricted pattern than that of c-fos, both in terms of brain areas activated and number of positive neurons. Changes in pH3S10 showed a maximum at 30 min, then progressively declining in most areas in spite of the persistence of IMO. Moreover, the decline was in general more sensitive than c-fos to the termination of IMO. The pattern of pH3S28 was even more restricted that of pH3S10, but they showed co-localization. The present data demonstrate a more selective pattern of stress-induced histone H3 phosphorylation than c-fos. The factors determining such a selectivity and its biological meaning remain to be studied.
paraventricular nucleus of the hypothalamus, medial parvocellular dorsal division
The induction of immediate early genes (IEGs), the most extensively used being c-fos, has greatly contributed to the identification of brain areas activated by drugs and stressful stimuli (Herrera and Robertson 1996; Pacak and Palkovits 2001; Hoffman and Lyo 2002). Such identification is the first step and it may be followed by other complementary experimental approaches (i.e., lesions or inactivation of particular brain areas) for the characterization of the putative role of the activated areas in the processing of each stressor and its functional consequences.
Systemic stressors induce c-fos expression and other IEGs in a restricted set of brain areas, with a pattern that clearly differs from one particular stressor to another (i.e., Pacak and Palkovits 2001), thus allowing to focus further studies in a few number of brain areas. Emotional or predominantly emotional stressors, on the other hand, elicit widespread brain activation, involving for instance, the medial prefrontal cortex (mPFC), lateral septum (LS), amygdala (mainly medial amygdala, MeA), several thalamic nuclei, nucleus accumbens (Acb), the paraventricular nucleus of the hypothalamus (PVN), the dorsal raphe and the locus coeruleus (Cullinan et al. 1995; Kovacs 1998; Pacak and Palkovits 2001; Armario 2006). In addition to this widespread activation, no sensitivity to the intensity of the stressors was observed in most brain areas evaluated, including the mPFC and the thalamic nuclei, as quantified by c-fos and other IEGs (Campeau and Watson 1997; Ons et al. 2004; Pace et al. 2005). In this respect, only a restricted set of brain areas that includes the LS, the MeA, the PVN, the dorsal raphe and the locus coeruleus are sensitive at least to certain ranges of stressor intensities (Ons et al. 2004). Thus, it might seem that c-fos quantification after emotional stressors reflect to a high extent processes of general arousal, making it difficult to use this approach alone to identify brain areas and neuronal populations that are actually important for the specific processing of these stressors and for their consequences (Armario 2006).
Recently, great attention is being devoted to epigenetic processes, because of the fact that they can either enhance or reduce expression of particular genes for long periods of time. These changes in gene expression are possible because of certain modifications in the DNA (methylation) and histones that alter the structure of chromatin and the accessibility of transcriptional machinery to DNA. Evidence shows that epigenetic mechanisms could play a critical role in the long-term effects of exposure to stressors and addictive drugs, including their associated learning processes (i.e., Meaney et al. 2007; Reul and Chandramohan 2007; Tsankova et al. 2007). We hypothesized that epigenetic changes, as some particular histone modifications, may mainly affect those neurons most critically involved in the impact of stress, and they could be identified by immunohistochemistry.
Phosphorylation of histone H3 and acetylation of histone H3 and H4 are two types of modifications that have been repeatedly associated with enhanced gene expression (Berger 2001). Thus, immunohistochemistry procedures have been successful to detect increases in the number of pH3S10 (phosphorylation of histone H3 in serine 10) and pAcH3S10L14 (phosphorylation of histone H3 in serine 10 and acetylation in lysine 14) positive neurons in response to stress in the dentate gyrus (Bilang-Bleuel et al. 2005; Chandramohan et al. 2007, 2008). Although the latter authors informally reported increases in the number pH3S10 and/or pAcH3S10L14 positive neurons in some brain regions other than the dentate gyrus, there is no precise evidence at present for widespread induction of these changes after stress.
On the basis of the above, we initially wanted to demonstrate the brain pattern of modifications in the number of pH3S10 and pAcH3S10L14 positive neurons after exposure to stress. Both pH3S10 and pAcH3S10L14 show similar patterns after stress (Bilang-Bleuel et al. 2005) or cocaine administration (Brami-Cherrier et al. 2005). However, we have demonstrated increases in the number of pH3S10 positive neurons after amphetamine (Rotllant and Armario 2012), whereas we were unable to obtain satisfactory immunohistochemistry results using available commercial antibodies against pAcH3S10L14 modifications. Moreover, we observed that all pH3S10 positive neurons were also c-fos positive, while the opposite was not true, suggesting that only a subset of c-fos neurons are able to show pH3S10 immunoreactivity (pH3S10-IR). This fact reinforces the possibility that some c-fos positive neurons may manifest other types of phosphorylation changes in H3. We then decided to study two other histone H3 modifications that have been associated with enhanced gene transcription in non-brain tissue: phosphorylation of histone H3 in serine 28 and threonine 11 (Perez-Cadahia et al. 2009). Our hypothesis is that the pattern of histone H3 phosphorylation after stress would be more restricted than that of c-fos and that such pattern may help delineating possible areas of interest to the study of long-term stress-induced functional and epigenetic changes.
In pilot studies, we mapped changes in the number of pH3S10, pH3S28, and pH3Th11 positive neurons under basal conditions and after exposure to immobilization (IMO) stress. We found very low number or absence of positive neurons under basal conditions in all cases. However, in response to IMO, increases in the number of pH3S10 and pH3S28 positive neurons were observed in a restricted set of brain areas. Maximum increases in pH3S10 positive neurons were usually observed around 20-30 min of stress exposure. Then, we designed an experiment with three main purposes: (i) to characterize the dynamics of the changes in histone H3 in relation to the time of exposure to acute IMO and the release of the animals from the stressful situation; (ii) to compare the response of two stressors differing qualitatively and in terms of their intensity: exposure to a novel environment and IMO (Rotllant et al. 2007); and (iii) to demonstrate co-localization of these histone H3 changes with c-fos expression.
Male, 60 days old, Sprague-Dawley rats bred in the Animal Facility Service of the Universitat Autònoma de Barcelona were used. Animals were maintained two per cage under standard conditions of temperature (22 ± 1°C) and a 12 h-12 h light–dark schedule (lights on at 7:00), with ad libitum access to food and water. Before the experimental procedures, animals were acclimated to the housing conditions for one week and received two handling sessions. All experimental treatments were performed in the morning. This study has been carried out in accordance with the Guide for the Care and Use of Laboratory Animals (National Institutes of Health) and approved by Ethical Committee for Animal Experimentation of the Universitat Autònoma de Barcelona.
Animals were randomly assigned to the following groups (n = 6 per group): BASAL, rats remained undisturbed in the animal room until perfusion; OPEN-FIELD, rats exposed to an open field for 30 min; IMO30, immobilization on boards (IMO) for 30 min, followed immediately by perfusion; IMO90, IMO for 90 min followed immediately by perfusion; and REC60, rats exposed to IMO for 30 min but then returned to the home cage for an additional period of 60 min before perfusion. The latter group would allow us to study the specific contribution of the release of the animals from the board. Rats were anesthetized with isoflurane and perfused with saline solution (4°C) for 2 min and with 4% paraformaldehyde and 3.8% borax (4°C) for 12 min. Then brains were removed, post-fixed overnight at 4°C and cryoprotected [0.2M NaCl, 43 mM potassium phosphate (KPBS) containing 30% sucrose] for 48 h at 4°C. Then, each brain was frozen on dry ice and four series of 20 μm sections were obtained and stored at −20°C in an anti-freeze solution (30% ethylene glycol, 20% glycerol in 0.25 mM phosphate buffer at pH 7.3).
Pilot studies were performed to characterize several commercial antibodies raised against different histone modifications (pH3S10, pH3S28, pH3Th11, pAcH3S10L14, AcH3L9, AcH3L9/14, AcH4L5, AcH4L8, AcH4S1L5/8/12) in brain tissue of basal or 30 min IMO rats (n = 2 per group). Only modifications that showed low basal levels where considered for further analysis (pH3S10, pH3S28, pH3Th11, pAcH3S10L14). Among these, only pH3S10 and pH3S28 where studied as pH3Th11 showed no detectable immunoreactivity levels in response to IMO and the results with the pAcH3S10L14 antibody showed high variability between lots. When different antibodies against the same epitope were available we chose the antibody with the best signal to background ratio. To check the specificity of anti-pH3S10 (sc-8656R; Santa Cruz Biotech, Santa Cruz, CA, USA) and anti-pH3S28 (sc-12927R; Santa Cruz Biotech) antibodies, both were incubated with the blocking peptide (sc-8656-p and sc-12927-p, respectively, Santa Cruz Biotech) for 1 h at 22°C prior to incubation with brain sections (Figure S1).
pH3S10 and pH3S28 immunohistochemistry using the biotin–streptavidin method was performed as described previously (Rotllant et al. 2010). The anti-pH3S10 and anti-pH3S28 antibody-peroxidase complex was revealed using diaminobenzidine (Sigma, Barcelona, Spain). Different brain areas were analyzed at the same coordinates for each animal, using as a reference the stereotaxic atlas of Paxinos and Watson (2007). Quantification of pH3S10 and pH3S28 immunopositive nuclei was performed over sample gray scale images captured by an experimenter blind to the treatment groups. To select pH3S10 positive and pH3S28 positive immunopositive nuclei as targets for quantification, PC-based software (Scion Image; Scion Corporation, Frederick, MD, USA) was used and targets were subsequently identified in the captured images by gray level thresholding (Rotllant et al. 2010). The pH3S10 and pH3S28 nuclei per mm2 were calculated for each particular area and animal as the average number obtained quantifying four sections, taken at 80 μm intervals and including both hemispheres.
Appropriate antibodies against either pH3S10 or pH3S28 are only available in rabbits. The use of primary antibodies raised in the same specie entails cross-reactivity interferences by two mechanisms: first, binding of the second step primary antibody to free binding sites of the first step secondary antibody; second, recognition of residual IgG constant fragments (Fc) from the first step primary antibody by the second step secondary antibody. The use of polyclonal monovalent F(ab) fragments of a secondary antibody instead of the complete molecule in the first step have been proposed to avoid such cross-reactivity/interferences (Negoescu et al. 1994). Then, this procedure was followed to study pH3S28 and pH3S10 co-localization.
The qualitative study was performed in selected animals (IMO30; n = 2, four sections per animal per region). The sections were sequentially immunostained for pH3S28 and thereafter for pH3S10. Non-specific binding was reduced by incubating for 90 min with a blocking solution containing 0.4% Triton-X 100, 2% bovine serum albumin , and 2% fetal calf serum in TBS (Trizma base 0.02 M, Saline 0.9%, pH 7.5). Sections were incubated first with the rabbit polyclonal antibody against pH3S28 overnight at 4°C. To cover all the rabbit Fc site, goat F(ab) fragments against rabbit IgG labeled with Alexa Fluor-594 (Jackson Immunoresearch, Madrid, Spain, 111-578-003) were used at 1 : 200 final dilution, incubating for 2 h at 22°C. Thereafter, the non-specific binding was again reduced incubating for 90 min with the same blocking solution but fetal calf serum at 1.5% was used. Then, sections were incubated with the rabbit anti pH3S10 antibody overnight at 4°C. The second step primary antibody was visualized with a secondary goat anti-rabbit antibody labeled with Alexa Fluor-488 (Invitrogen, Madrid, Spain, A-11008) at 1 : 500 final dilution, incubating for 2 h at 22°C. The sections were counterstained with Hoechst, mounted on slides and cover slipped with fluoromount (Sigma). Additional cross-reactivity controls were included by omitting the pH3S10 primary antibody (Figure S2). The intensity threshold used to determine pH3S10 positive nuclei was set above the maximum intensities observed in the controls using ImageJ software (National Institutes of Health, Bethesda, MD, USA). Images were captured using confocal microscopy (Zeiss LSM 700, Zeiss, Barcelona, Spain).
Combined immunohistochemistry and in situ hybridization
Combined immunoperoxidase and isotopic hybridization histochemical localization for pH3S10 and c-fos mRNA, respectively, was performed as described previously (Rotllant and Armario 2012). Sections were first processed for pH3S10 detection, using the biotin–streptavidin method but all solutions were DEPC treated and blocking solutions included heparin to prevent RNA degradation. Then, sections were processed for c-fos mRNA detection by in situ hybridization. Radioactive signal was visualized exposing the slides to a XAR-5 Kodak Biomax MR auto-radiography film (Kodak, Madrid, Spain) and finally dipped in LM-1 nuclear emulsion (GE Healthcare, Cerdanyola, Spain). Emulsion was exposed for 15 days and developed.
Plasmid and probe preparation
The c-fos probe was generated from EcoRI fragment of rat c-fos cDNA (Dr. I. Verma, The Salk Institute, La Jolla, CA, USA), subcloned into pBluescript SK-1 (Stratagene, La Jolla, CA, USA) and linearized with SmaI. Radioactive antisense cRNA copies were generated using α−35S UTP (GE Healthcare) and a transcription kit (Roche, Sant Cugat del Vallès, Spain) following the protocol recommended by the supplier.
Densitometry analyses of the c-fos mRNA were done on the autoradiography films or the emulsion slides in function of the intensity of the signal, avoiding signal saturation (Ons et al. 2010). The RNA levels were semi-quantitatively determined in three sections per brain area (including both hemispheres) and animal according with the stereotaxic atlas of Paxinos and Watson (2007). Digitalized images were quantified using Image software (Scion Corporation) by gray level thresholding. Measures were obtained in arbitrary units (pixel area × average sum gray).
Double labeling was quantified as described previously (Rotllant and Armario 2012). Different brain areas were analyzed at the same coordinates for each animal, using as a reference the stereotaxic atlas of Paxinos and Watson (2007). Quantification of cell nuclei showing pH3S10-IR, neurons expressing c-fos mRNA and the number of neurons showing both pH3S10-IR and c-fos mRNA expression was performed over sample color images captured by an experimenter blind to the treatment groups. Measurements were performed placing individual pH3S10 nuclei within a circle of fixed diameter and counting the number of silver grains inside the area. Neurons were considered double-labeled when they showed both brown nuclear deposits and a silver grain density exceeding three times the background (estimated in the corpus callosum). Isolated silver grain clusters not showing a pH3S10+ nucleus were identified by relative silver grain density above the surrounding area and counted as c-fos mRNA single labeled neurons when silver grain density exceeds three times the background. The single and double labeled neurons per mm2 were calculated for each particular area and animal as the average number obtained quantifying three sections, taken at 80 μm intervals and including both hemispheres.
The ‘Statistical Package for Social Sciences’ (SPSS, version 17, IBM, Barcelona, Spain) was used to perform the statistical analysis. Inter-assay variability was avoided by processing all statistically compared samples in the same assay. Log-transformation of data was applied when necessary to achieve homogeneity of variances. Group differences in histone phosphorylation and c-fos mRNA expression were analyzed by one-way anova with treatment (BASAL, OPEN FIELD, IMO30, IMO90, REC60) as the main factor. Additional post hoc comparisons were done with the Student-Newman-Keuls test (SNK, p < 0.05). Double-labeling histochemistry results comparing the effect of the treatment (BASAL, IMO30) were analyzed by the Student's t-test.
Dynamics of histone phosphorylation and c-fos mRNA expression
Quantification of histone phosphorylation was performed in all brain areas showing positive immunoreactivity (Fig. 1). pH3S10 positive neurons were observed in all the subdivisions of the medial prefrontal cortex, and we chose the prelimbic cortex (PrL; from Bregma 3.24 to 2.76 mm) as representative of this area. In the Acb, pH3S10 positive neurons were observed in both the shell and the core, but quantification was done in the vertex of the shell where maximum c-fos induction was observed (AcbSh; from Bregma 2.28 to 1.56 mm). pH3S10 positive neurons were also observed and quantified in the ventral division of the LS (LSV; from Bregma 1.56 to 0.24 mm), the medial division of the amygdala (MeA; from Bregma −2.28 to −3.00 mm), and the medial parvocellular dorsal division of the PVN (PVNmpd; from Bregma −1.44 to −2.04 mm). Positive immunoreactivity for pH3S28 (pH3S28-IR) was observed only in the AcbSh and LSV.
The one-way anovas of the number of pH3S10 positive neurons revealed statistically significant effects for treatment in all selected areas (Fig. 2): PrL [F(4,29) = 36.449, p < 0.001], AcbSh [F(4,29) = 61.781, p < 0.001], LSV [F(4,29) = 44.052, p < 0.001], MeA [F(4,29) = 9.960, p < 0.001] and PVNmpd [F(4,29) = 43.855, p < 0.001]. Post hoc analysis indicated that whereas no differences between IMO and OPEN-FIELD were observed in the PrL and the AcbSh, in the LSV, the MeA and the PVNmpd, greater effect of IMO as compared to OPEN-FIELD was observed. In fact, in the LSV and the PVNmpd, the OPEN-FIELD did not increase the number of pH3S10 positive neurons as compared to the BASAL group. Regarding the dynamics of the response to IMO, the number of pH3S10 positive neurons was maintained at the same level after 30 and 90 min of IMO only in the AcbSh, whereas in the other areas a decrease was found after 90 min. Moreover, the release of the animals from IMO markedly decrease the number of pH3S10 positive neurons in all areas studied as compared to the IMO90 group, indicating a specific contribution of the release of the animals from the situation to the decrease in the number of pH3S10 positive neurons.
The one-way anovas of the number of pH3S28 positive neurons revealed statistically significant effects for treatment in the number of pH3S28 positive neurons in the AcbSh [F(4,29) = 59.835, p < 0.001] and the LSV [F(4,29) = 28.827, p < 0.001] (Fig. 2). Post-hoc analysis revealed that the behavior of pH3S28 positive neurons regarding the type of stressor and the dynamics of the response to IMO was similar as that of pH3S10 positive neurons in the same area. In the AcbSh and LSV qualitative analysis of the pH3S28 and pH3S10 double immunofluorescence study showed that virtually all pH3S28 positive nuclei were also positive for pH3S10 (Fig. 3).
In agreement with previous studies, we observed stress-induced c-fos mRNA expression in numerous brain areas, including the cortex (i.e., medial prefrontal, orbitofrontal, piriform), the nucleus accumbens, the septum, the bed nucleus of the stria terminalis, the thalamus (i.e., paraventricular nucleus), and the hypothalamus (i.e., preoptic area, supraoptic nucleus, PVN, ventromedial nucleus). Nevertheless, quantification of c-fos mRNA levels was performed only in brain areas also showing positive pH3S10 immunoreactivity (Fig. 1).
The one-way anovas revealed statistically significant effect for treatment in c-fos mRNA levels in the PrL [F(4,29) = 12.702, p < 0.001], the AcbSh [F(4,29) = 6.896, p < 0.001], the LSV [F(4,29) = 13.772, p < 0.001], the MeA [F(4,29) = 5.140, p < 0.001] and the PVNmpd [F(4,29) = 94.801, p < 0.001] (Fig. 2). Post hoc analysis showed that Levels of c-fos mRNA were similar after OPEN-FIELD and IMO in the PrL, the AcbSh and the MeA, whereas they were higher after IMO in the LSV and the PVNmpd. After prolonged (90 min) exposure to IMO a decline of c-fos mRNA levels was found in the PrL, whereas the (not significant) trend was observed in the AcbSh and the LSV. In the MeA and the PVNmpd the response was maintained. A specific contribution of the release of the animals from IMO was clearly observed only in the PVNmpd with a decrease in the c-fos mRNA levels as compared to the IMO90 group. The same pattern was found in the LSV.
pH3S10 and c-fos mRNA double labeling
The previous analysis of the dynamics data indicated that maximum stress-induced pH3S10-IR was observed at 30 min. Since maximum levels of c-fos mRNA are usually found at about 30 min after stress exposure, 30 min is the optimal time point for double labeling of c-fos mRNA levels and pH3S10-IR. In addition, in most areas immobilization stress induced higher pH3S10-IR and c-fos mRNA levels than open field exposure. Therefore, double labeling analysis was performed in the BASAL and IMO30 groups (Fig. 4). IMO increased the number of double labeled neurons in the PrL (t(10) = 11.725; p < 0.001), the AcbSh (t(10) = 7.081; p < 0.001), the LSV (t(10) = 24.258; p < 0.001), the MeA (t(10) = 3.132; p < 0.001) and the PVNmpd (t(10) = 9.137; p < 0.001) (Fig. 5). In all these areas, basically all pH3S10 positive neurons also showed c-fos mRNA expression. IMO also increased the number of neurons positive for c-fos mRNA but negative for pH3S10 signal in the PrL (t(10) = 1.538; p < 0.01), the LSV (t(10) = 12.233; p < 0.01), the MeA (t(10) = 0.344; p < 0.001) and the PVNmpd (t(10) = 2.721; p < 0.001), whereas such neurons were not observed in the AcbSh. These data indicate that in most brain areas studied only a subset of c-fos positive neurons were also pH3S10 positive, with no evidence for pH3S10-IR in absence of c-fos expression. In the particular case of the AcbSh, basically all c-fos positive neurons were also pH3S10 positive.
The present results demonstrate that exposure to a severe stressor (IMO) resulted in an increase in the number of pH3S10 positive neurons in several brain areas important for the processing of stressors (medial prefrontal cortex, AcbSh, LSV, MeA and PVN), although these changes were usually observed in a lower number of neurons than those expressing c-fos. In addition, an increase in the number of pH3S28 positive neurons was also observed, but the pattern was clearly more restricted than that of pH3S10-IR and was only detected in those neurons also showing pH3S10-IR. No evidence of pH3Th11 positive neurons was found after stress. Only a subset of neurons activated during stress in terms of c-fos expression showed marked changes in pH3S10 and those neurons showing pH3S28-IR were subsumed within the population showing pH3S10-IR.
Phosphorylation of histone H3 appears to be a fast process with peak levels around 30 min in response to psychostimulants (Brami-Cherrier et al. 2005; Bertran-Gonzalez et al. 2008; Sanchis-Segura et al. 2009; Rotllant and Armario 2012). However, it remains unclear the extent to which the same dynamics could apply to stress. Therefore, to study the dynamics of pH3S10-IR and pH3S28-IR we did an experiment using exposure to IMO for 30 and 90 min. The first period was chosen as it is the time point of maximum levels of c-fos mRNA levels (i.e., Trneckova et al. 2007) and also of pH3S10-IR on the basis of previous exploratory experiments. Moreover, we have previously reported that in most brain areas studied c-fos expression did not show a faster decline when animals were released from IMO than when the animals were maintained in the stressful situation (Trneckova et al. 2007). These data suggest that in most brain areas there are no specific signals to stop c-fos expression associated with the release from the board. To show whether or not pH3S10-IR and pH3S28-IR followed the same pattern as c-fos, we added a group exposed to IMO for 30 min and studied 60 min after the termination of IMO, to directly compare it with the response to 90 min of continuous IMO. Finally, we wanted to study whether changes in histone H3 phosphorylation are more sensitive to the intensity of stressors than c-fos expression. To this end, a group of rats exposed for 30 min to a novel environment was included.
We found an increase in the number of pH3S10 positive neurons after 30 min of IMO in the PrL subdivision of the medial prefrontal cortex, the AcbSh and the AcbC (not shown), the LSV, the MeA, and the PVNmpd. Induction of c-fos was observed in a wider range of brain areas, but no area showing increases in pH3S10 positive neurons was devoid of c-fos signal. Double labeling combining pH3S10-IR and c-fos ISH after 30 min IMO indicated that virtually all pH3S10 positive neurons were also c-fos positive, but the number of c-fos positive neurons clearly exceeded that of pH3S10 positive neurons in most areas. Interestingly, in the AcbSh most c-fos positive neurons were also pH3S10 positive. The absence of pH3S10–IR in those neurons not expressing c-fos confirms previous reports with stress and psychostimulants (Chandramohan et al. 2007; Rotllant and Armario 2012).
It is noteworthy that the relationship between c-fos expression and the increase in the number of pH3S10 positive neurons widely differ among the brain areas. Thus, in most areas c-fos induction was not associated with any increase in the number of pH3S10 positive neurons. In addition, in the few areas specifically studied in the present work, the ratio pH3S10 positive/c-fos positive neurons changes from one area to another: in the AbcSh, almost all c-fos positive neurons were also pH3S10 positive; in the LSV the majority, in the PNVmpd more than one half, in the PrL about one half and in the MeA about one third. Marked regional heterogeneity has been also observed after amphetamine administration (Rotllant and Armario 2012). Such a dissociation has also been observed after administration of cocaine that increased both c-fos and the number of pH3S10 positive nuclei in neurons of the striatum expressing dopamine-1 receptor (D1R), whereas in those neurons expressing the D2R only induction of c-fos was observed (Bertran-Gonzalez et al. 2008). The ultimate reasons for this heterogeneity remain unknown, but it seems likely that in order for a neuron to become pH3S10 positive, they should receive additional specific synaptic signals or, alternatively, stronger synaptic signals as compared to those required for c-fos expression. It should be noted that the absence of pH3S10-IR may ultimately relies on the lack of sensitivity of the antibody, but this possibility does not negate the fact that some neurons showed much more stress-induced pH3S10 changes than others and the former ones would be the most activated by the stressors.
This study also has relevance on the relationship between the intensity of stressors and the changes in pH3S10-IR. In accordance with the results previously obtained with c-fos and other IEGs (Campeau and Watson 1997; Ons et al. 2004; Pace et al. 2005), activation of the medial prefrontal cortex did not discriminate between stressors of different intensity. A similar pattern was observed in the AcbSh. In contrast, in other areas (LSV, PVNmpd) huge differences were found between the two stressors, also in accordance with previous conclusions using IEGs (Dielenberg et al. 2001; Ons et al. 2004; Burow et al. 2005). However, in the MeA, c-fos expression did not discriminate between the two conditions, whereas the number of pH3S10 positive neurons was higher after IMO than after OPEN-FIELD in the MeA. As exposure to a novel environment markedly differed from IMO in terms of intensity, the lack of differences between the two stressors regarding c-fos expression and pH3S10-IR in some areas is somewhat puzzling. A deeper understanding of this paradox would require novel approaches, but we have suggested that this can be at least in part explained if most neurons activated by predominantly emotional stressors reflect arousal processes rather than activation of stressor-specific circuits (Armario 2006). In fact, it is at present unclear whether or not we are activating the same neuronal populations with both stressors, but it appears likely considering that simultaneous exposure to two predominantly emotional stressors (cat odor and IMO) did not result in the recruitment of more Fos positive or pH3S10 positive neurons in this region (Muñoz-Abellán, Rotllant, Nadal and Armario, unpublished). The present results indicate that at least in certain areas such as the MeA, changes in pH3S10-IR may better discriminate between stressors than c-fos expression.
In all areas where 30 min of IMO induced increases in pH3S10 positive neurons, pH3S10-IR declined after 90 min of exposure to the stressor, with the exception of the AcbSh, where pH3S10-IR was maintained at the same level as at 30 min. In contrast, in those animals released from IMO at 30 min and maintained in resting conditions for 60 min before the animals were killed (post-IMO period), all brain areas studied (including the AcbSh) showed a strong decline in pH3S10-IR. Consequently, when comparing directly the two groups of animals (those exposed to 90 min IMO with those released from the situation at 30 min) significantly lower levels of pH3S10-IR were found in AcbSh, LSV, and PVNmpd. The maintenance of c-fos response and pH3S10-IR in the AcbSh after sustained IMO is not because of an incapability of these neurons to shut off pH3S10 once activation is triggered as a strong decline of the signal was observed during the post-IMO period. The sustained activation of pH3S10 in the AcbSh as compared to all other brain areas during relatively prolonged exposure to IMO is noteworthy and may reflect a very important role of AcbSh in short-term adaptation to prolonged stress. In fact, there is evidence for a critical role of the AcbSh in behavioral consequences of exposure to certain stressors (Barrot et al. 2002; Renthal et al. 2007; Wilkinson et al. 2009).
In some areas, the post-stress dynamics of pH3S10-IR differed from that observed with c-fos. In this latter case, a significant influence of releasing the animals from IMO was only observed in the PVNmpd and to a lower extent in the LSV. Whereas some of the present data agree with previous results on this subject, other data did not (Valles et al. 2006; Trneckova et al. 2007; Ons et al. 2010), probably because of the different times of exposure to the stressors. In any case, our present results clearly demonstrate that changes in the number of pH3S10 positive neurons are much more sensitive than changes in c-fos expression to putative signals associated to the termination of stressor exposure. Considering that the half-life of c-fos mRNA levels is of 10–15 min (Shyu et al. 1989; Zangenehpour and Chaudhuri 2002) and the behavior of c-fos in the LSV, it is unlikely that c-fos transcription stopped after releasing the animals from the board. In contrast, in the present experiment, we demonstrate that the release of the rats from IMO did specifically contribute to the decline in the number of pH3S10 positive neurons in all brain areas studied. Dynamics of pH3S10-IR may be more useful than that of c-fos to characterized post-stress effects.
The present results showing that exposure to stress can induce an important increase in the number of pH3S10 positive neurons in some brain areas important for processing of emotional or predominantly emotional stressors, are not apparently in accordance with some previous reports that observed increases in the number of pH3S10 positive and pAcH3S10L14 positive neurons restricted to the dentate gyrus of the hippocampal formation (i.e., Bilang-Bleuel et al. 2005; Chandramohan et al. 2007, 2008). The reasons for the discrepancies are unclear, although it is unlikely that our results are explained by the lack of specificity of our antibodies for the reasons outlined in Methods. Moreover, the finding that no positive pH3S10 signals were observed in c-fos negative neurons, neither in the present experiment nor in a previous report after amphetamine administration (Rotllant and Armario 2012) argues against non-specificity. Another possibility is that the above authors firstly studied changes in histones more than 8 h after exposure (Bilang-Bleuel et al. 2005), which may not be the appropriate times. However, in further reports the times studied (30 min to 24 h) were within the time frame of the present experiment, with maximum changes between 30 min and 2 h after initial exposure (i.e., Chandramohan et al. 2007, 2008). Methodological aspects such that the characteristics of antibodies and particularly the lack of in vivo perfusion in the above report, that may decrease pH3S10-IR. Moreover, the latter authors indicated that scattered pH3S10-IR was observed in some brain regions, including neorcortex and amygdala (Bilang-Bleuel et al. 2005), although no attempt was apparently made to quantify these pH3S10 positive nuclei.
Given that only a subset of c-fos positive neurons was pH3S10 positive, the possibility remains that c-fos positive neurons may manifest other alternative phosphorylation patterns. Then, we studied the effects of stress on pH3S28-IR. We only observed increased number of pH3S28 positive neurons in the LSV and the AcbSh. In these regions, the dynamics was similar to that of pH3S10-IR, suggesting that similar mechanisms may underlie both types of changes. We observed, using immunofluorescence, in two selected IMO rats complete co-localization of pH3S28-IR with pH3S10-IR, suggesting that only a subset of pH3S10 positive neurons also showed pH3S28-IR. Surprisingly, some areas where an important degree of pH3S10-IR was observed (e.g., the PVNmpd) were devoid of pH3S28-IR. The reason for this differential pattern remains to be elucidated, but there is evidence in other tissues for differential phosphorylation of histone H3 at these two sites (Choi et al. 2005) each perhaps affecting different histone tails (Dyson et al. 2005). Both pH3S10 and pH3S28 are associated with enhanced transcription, so we expect that histone H3 phosphorylation could act as a trigger of other more persistent changes in gene expression that would mediate synaptic plasticity and long-lasting stress-induced changes (Day and Sweatt 2011).
We have not explored the neurochemical mechanisms putatively involved in the stress-induced increases in pH3S10-IR and pH3S28-IR. Nevertheless, there is evidence that the above mentioned stress-induced changes in pH3S10 and pAcH3S10L14 in the DG appears to require simultaneous activation of glutamate NMDA and glucocorticoid receptors (Bilang-Bleuel et al. 2005; Chandramohan et al. 2007) and to involve mitogen- and stress-activated kinases (MSK) 1/2 and Elk-1 (Chandramohan et al. 2007; Gutierrez-Mecinas et al. 2011). These results are essentially in accordance with those observed after cocaine administration (Brami-Cherrier et al. 2005, 2007, 2009; Bertran-Gonzalez et al. 2008; Stipanovich et al. 2008; Besnard et al. 2011).
In summary, the present data indicate that exposure to stress is able to induce an increase in the number of pH3S10 positive neurons in a pattern that is much more restricted, both regionally an in terms of number of neurons, that the typical activation observed with c-fos. The increase in the number of pH3S28 positive neurons was still more restricted than that of pH3S10 positive neurons and both histone modifications showed complete co-localization. Changes in histone H3 phosphorylation are not in general different from c-fos to discriminate between stressor intensities, but are more sensitive to the release of the animals from the stressful situation. Although the functional impact of c-fos positive/pH3 positive versus c-fos positive/pH3 negative neurons are not known at present, we hypothesize that the former ones may be more important for brain processing of stressors.
DR and JP contributed to the acquisition of data. DR and AA both contributed to conception, design and analysis of the data, drafting, and revising the manuscript. All the authors declare that there are no personal financial holdings that could be perceived as constituting a potential conflict of interest. This study was supported by grants from Ministerio de Ciencia e Innovación (SAF2008-01175, SAF2011-28313), Instituto de Salud Carlos III (RD06/0001/0015, Redes Temáticas de Investigación Cooperativa en Salud, Ministerio de Sanidad y Consumo) and Generalitat de Catalunya (SGR2009-16).