Dynamic Proteomics of Nucleus Accumbens in Response to Acute Psychological Stress in Environmentally Enriched and Isolated Rats

Our prior research has shown that environmental enrichment (i.e. rats reared in an environment with novel objects, social contact with conspecifics) produces a protective antidepressant-like phenotype in rats and decreases neurobiological effects of acute psychological stress. Although CREB activity has been identified as a major player, the downstream molecular mechanisms remain largely unexplored. Thus, the current study investigates proteomic differences in the accumbens of rats raised in an enriched condition (EC) versus those raised in an isolated control condition (IC) under basal conditions and after 30 min of acute restraint stress. Results showed that under basal conditions, EC rats generally expressed less mitochondria-related proteins, particularly those involved in TCA cycle and electron transport compared to IC rats. After 30 min of acute stress, EC rats displayed increased expression of energy metabolism enzymes (among others) while IC rats exhibited decreased expression of similar proteins. Further, network and pathway analyses also identified links to AKT signaling proteins, 14-3-3 family proteins, heat-shock proteins, and ubiquitin-interacting proteins. The protein ENO1 showed marked differential expression and regulation; EC rats expressed higher levels under basal conditions that increased subsequent to stress, while the basal IC expression was lower and decreased further still after stress. The results of this study define differential protein expression in a protective rat model for major depression and additionally identify a dynamic and coordinated differential response to acute stress between the two groups. These results provide new avenues for exploration of the molecular determinants of depression and the response to acute stress.


Introduction
Major Depression is one of the most widespread medical conditions in the United States. Lifetime prevalence rates are 16.5% (1 in 6 Americans; 50 million people) and last 12 month prevalence is about 6.7% (20 million Americans) [1]. Although pharmacotherapeutics have revolutionized the treatment of depression, many patients are refractory to pharmacotherapy [2]. Thus, a better understanding of the molecular mechanisms causing major depression is crucial for developing novel pharmacotherapeutic strategies to effectively treat this condition.
Although animal models cannot fully recapitulate the complex cognitive and social aspects of human depression, several models have proven useful both for developing pharmacotherapies and also furthering basic science into the mechanisms of depression [3]. Among these are the Forced Swim Test (FST) as a model of ''behavioral despair'', the sucrose preference paradigm as a model of ''anhedonia'' and the social interaction test as a model of ''social withdrawal''. These three tests are easy to perform in rodents and model the most common symptoms of depression.
The environmental enrichment paradigm is a non-drug and non-surgical manipulation of individual differences in rat behavior. Rats in the control condition, termed the Isolated Condition (IC) are singly housed in standard polycarbonate cages with no access to novelty or social contact. Rats in the Enriched Condition (EC) are housed 12 per group in a large cage with novel objects (hard plastic children's toys). The EC rats experience social contact with conspecifics, exercise and novelty. The toys are changed daily to maximize novelty. After the toys are changed, the rats exhibit an incredibly high level of activity for at least 30 min as they explore the new toys, play and generally run around the cage. The activity they exhibit is much higher than locomotor stimulation seen with administration of any dose of cocaine or amphetamine (personal observation). After 30 min the activity levels wane and the rats eventually go back to sleep. However, as the light cycle transitions from light to dark, the rats show another robust burst of exploratory activity during their normal circadian activity peak. Interestingly, novelty, social contact and exercise are all rewarding to rats and all release dopamine in the mesolimbic pathway leading from the ventral tegmentum to the nucleus accumbens (NAcc) [4,5,6,7,8]. This ''natural reward'' pathway has received an incredible amount of attention in drug addiction studies [9], but has increasingly been studied for depression-related behavior [10]. After a month in these differential rearing conditions EC and IC rats show robust differences in behavior. Specifically, EC rats exhibit a protective phenotype in depression-related behavioral paradigms, meaning that they behave as if they were injected with antidepressants [11]. Thus, the innovative aspect of this study is the use of a rat model of a well described protective depression phenotype (i.e. EC/IC rats) to investigate the dynamic proteomics of differential responses to acute stress.
Our investigations into the molecular mechanisms underlying this protective phenotype has yielded evidence that EC rats have lower steady-state levels of the active phosphorylated form of the transcription factor cAMP response element binding protein (CREB) in the NAcc [11,12]. The hypothesis that low CREB activity is the underlying mechanism of the EC phenotype has ample support by the fact that blocking CREB activity in the NAcc produces an EC-like phenotype in 14 different behavioral paradigms, including increased stimulant sensitivity, decreased stimulant self-administration, increased anxiety-like behavior and, of importance to this paper, decreased depression-like behavior [11,13,14,15,16,17,18]. Because of these prior reports, differentially-regulated CREB target genes are of highest importance in placing the current results in a proper context.

Animal Treatment
Twenty four male Sprague Dawley rats (Harlan laboratories Inc, Houston), 21 days of age, were divided to two conditions (isolated control condition and enriched condition), twelve of each. For the IC group, the rats were separated one rat per cage with no access to social contact or novelty, whereas EC rats were housed together 12 per cage (77L678W660H cm) and with novel toys changed every day. The food and water were freely available for rats and all the rats were maintained in a controlled room environment (temperature, 22uC; relative humidity, 50%; and 12 h light/dark cycles) for 40 days (from post-natal days 21 to 61) prior to stress experimentation. The enrichment manipulation is a compound manipulation where rats experience social contact, exercise and novelty. All three of these aspects release dopamine in the nucleus accumbens and all are rewarding to rats [4,5,6,7,8]. Isolation rats were chosen as the control group rather than pairhoused rats because isolation is the absence of enrichment, and a pair-housed group would constitute an intermediately-enriched condition. Although it may appear that isolation produces a condition of chronic stress (as seen by elevated basal plasma corticosterone levels), enriched rats exhibit a blunted corticosterone response to stimulation. Chronic stress produces blunted corticosterone responses and increased sensitivity to stimulantinduced locomotor activity consistent with the idea that the life of an IC rat is less stressful than that of an EC rat [19,20].
Restraint stress: Six rats from each group were placed individually into plastic conical sleeves (Decapi-Cone; model DC200; Braintree Scientific, Braintree, MA) for 30 min [11,13,14]. The other six rats in each group remained undisturbed as controls for stress. Stress rats were then decapitated immediately at the end of stress. The brain was chilled in cold phosphatebuffered saline (PBS) before being placed into a metal matrix (ASI Instruments) for dissection. A 2 mm thick coronal slice was taken using the olfactory tubercles as a guide. NAcc was dissected from the slice on an ice-cold platform using a razor blade. The olfactory tubercles and medial septal areas were removed and a triangular dissection of the NAcc was stored at 280uC until further analysis [21].
The experiments were performed in accordance with the guidelines of National Institutes of Health and approved by the Institutional Animal Care and Use Committee of The University of Texas Medical Branch at Galveston.

Protein Extraction
The NAcc from two rats of the same group were pooled together in one tube for protein extraction due to its limited size. The tissue from controls and stress rats of both groups were first washed with ice cold tris buffered saline (TBS) and then homogenized in a buffer [TBS pH 7.4, 1% Igepal-CA630 (NP-40), 16protease inhibitor cocktail, 20 mM NaF, 1 mM Na 3 VO 4 , 10 mM DTT and 5 mM ethylenediaminetetraacetic acid (EDTA)] on ice and then centrifuged at 7506g for 20 min at 4uC to remove the cell debris. The top fraction was transferred to a new tube and centrifuged at 20,0006g for 20 min at 4uC. The resulting supernatant was removed and mixed with 1% streptomycin sulfate to remove DNA contaminants [22], then, four volumes methanol and 1 volume of chloroform (V protein :V methanol :V chloroform = 1:4: 1) were added to the protein supernatant and incubated at room temperature 15 to 30 min., vortexing every 5 min to remove lipids. The sample was centrifuged at 160006g, at 4uC for 20 min. The pellet was then washed with 3% HCI/ acetone to remove the methanol and chloroform (cytosolic fraction) [23,24,25]; the procedure was similar for the resulting pellet fraction to obtain a crude membrane fraction. Both fractions were then dissolved in a buffer containing 6M urea, 1% NP-40, 20 mM,tris-HCl pH 7.4, 16protease inhibitor cocktail, 10 mM dithiothreitol (DTT) and 5 mM EDTA. The protein extracts of both fractions were then subjected to further proteomic analysis.

2D Gel Electrophoresis
Four hundred micrograms of cytosolic protein from control and stress rats of both groups were dissolved in PSB 200 ul containing 100 mM DTT, a trace amount of bromophenol blue and 1% Immobilized pH Gradient (IPG) buffer, pH 3-11 NL, and incubated at 21uC for 1 hr. Three hundred fifty micrograms of protein from the membrane fraction were dissolved in 200 ml of 2D protein extraction buffer III containing 100 mM DTT, a trace amount of bromophenol blue and 1% IPG buffer, pH 3-11 NL, and incubated at 21uC for 1 hr. The proteins were then rehydrated to the DryStrip (11 cm, pH 3-11 NL) overnight at the same temperature. For the first dimension, isoelectric focusing (IEF) was performed at 20uC using an Ettan IPGphor3 (GE Healthcare, Sweden) in the following steps: 200 V for 30 min., 500 V for 1.5 hrs, 1000 V for 1.5 hrs, 8000 V for 2.0 hrs and 8000 V for 24000 Vhr. The strips were then equilibrated for 1 hr in equilibration buffer (50 mM Tris-HCl pH 8.8, 6 M urea, 20 mM iodoacetamide, 2% sodium dodecyl sulfate (SDS), 20% glycerol). After rinsing two times with SDS-polyacrylamide gel electrophoresis (PAGE) running buffer, the strips were loaded on to 10-20% SDS-Tris-glycine gradient gel (13.368.7616cm) and were then subjected to 150 V for 140 min at room temperature for the 2nd dimension separation [26]. Following the electrophoresis, the gel for each sample (N = 3 samples per group [i.e. 6 rats]) was stained with Coomassie blue G250 (CBB G250).

SameSpots Analysis of Protein Expression Variation
2D gel images (CBB G250 stained) were acquired by Typhoon Trio imaging system (GE Healthcare). The scanning resolution was 50 mm with excitation l 633 nm. The 2D gel images were then analyzed by the software Progenesis SameSpots (nonlinear dynamics, version 4.0) which has been judged to be much improved in reproducibility compared to previous generations of 2D gel analysis software [27]. To align the gel images, one of the six gels was chosen as a reference gel. With the help of manually drawn and automated vectors from each image to the reference, images were aligned at the pixel level. The program then performed automatic spot detection and background subtraction. The software assigns the same spot on every gel in the analysis with identical shape (spot outline) and spot number. Spot volumes were normalized to those of the reference gel to obtain normalized volumes that are comparable across gels. Protein expression fold changes between controls and stress rats in isolation and enrichment groups were determined as described below in the statistical analysis.

Trypsin Digestion and MALDI-TOF-TOF MS Analysis
The regulated spots in both groups were manually excised from the 2D gel. The protein was digested with trypsin (0.1 mg per spot, Promega) in 10 ml of 25 mM ammonium bicarbonate, pH 8.0, for 6 hrs at 37uC [26]. One ml of digested sample solution was used for matrix-assisted laser desorption time of flight tandem mass spectrometry (MALDI TOF/TOF MS) analysis using an Applied Biosystems 4800 MALDI TOF/TOF proteomics analyzer. The instrument was operated in a positive ion reflection mode with mass range from 850 to 3000 Da. The focus mass was set at 1700 Da. For MS data, 2000-4000 laser shots were acquired and averaged from each sample spot. Following MALDI MS analysis, MALDI MS/MS was performed on several (,5-10) abundant ions from each sample spot. A 1 kV positive ion MS/MS method was used to acquire data under post-source decay (PSD) conditions. The instrument precursor selection window was 63 Da. For MS/MS data, 2000 laser shots were acquired and averaged from each sample spot. Automatic external calibration was performed using reference fragment masses 175.120, 480.257, 684.347, 1056.475, and 1441.635 (from precursor mass 1570.700).

Protein Identification
Applied Biosystems GPS ExplorerTM (Version 3.6) software was employed for searching the respective protein database using both MS and MS/MS spectral data for protein identification. Protein match probabilities were determined by using MASCOT scores. A MASCOT score of more than 64 was considered significant (i.e. P,0.05). MS peak filtering included the following parameters: mass range 800 Da to 4000 Da, minimum S/N filter = 10, mass exclusion list tolerance = 0.5 Da, and mass exclusion list (for some trypsin and keratin-containing compounds) included masses 842. 51

Statistical Analysis for Proteins
The intensities of same protein spots in 2D gels of basal expression and stress-evoked EC and IC protein changes were compared on log-transformed normalized volumes by Student's t-test. P values ,0.05 were considered significant. However, one spot was added at a P value of 0.06.

Bioinformatics
First, basal and stress-evoked changes in proteins were analyzed via a gene/protein ontology procedure using the Database for Annotation, Visualization and Integrated Discovery (DAVID; http://david.abcc.ncifcrf.gov/) before being used in an Ingenuity Pathways Analysis (Ingenuity Systems; www.ingenuity.com).
Gene/protein ontology. Separate lists of proteins from Basal EC/IC Differences, IC Stress, and EC Stress were analyzed using the DAVID Functional Annotation Clustering tool. In this procedure, an algorithm clusters redundant and overlapping functional classifications to provide a clearer picture of protein functional regulation.
Ingenuity pathways analysis. The IPA suite was used to further analyze functional associations, canonical pathways and networks of interconnected proteins. Each of these analyses uses a Fisher's exact test to identify overrepresentation of protein groups with a P value of less than 0.05. The Function analysis uses a similar method of classifying functions as the gene/protein ontology analysis, yet with a different set of comparison data. The Canonical Pathways feature searches for statistical overrepresentation of targets among curated pathways of direct and wellaccepted biological systems (e.g. signaling cascades). The Network analysis feature is a more broad analysis of interactions among identified proteins mined from various literature sources.

Western Blot Validation of Expression Changes
The same protein extracts used in 2D gel analyses (50 ug for ENO1 or 70 ug for HSPA8 and PKM2 per sample) were loaded in 10-20% SDS-Tris-glycine gradient gel (13.368.7 cm), and run at room temperature until the dye marker reached the end of the gel. The separated proteins were then transferred to PVDF membrane, and sequentially incubated with TBS with tween (TBST [pH 7.4, 0.1% Tween20 (sigma-Aldrich)]) with 5% nonfat milk for 1 hr; primary rabbit to rat specific antibody (ENO1, 1:500; PKM2, 1:1000; Cell signaling. HSPA8, 1:1000; Epitomics) in TBST with 5% nonfat milk for 2 hrs; washing the membrane with TBST 15 min for three times; secondary dyelight550 conjugated goat to rabbit antibody (1:15,000) in TBST with 5% nonfat milk for one hr; washing the membrane with TBST 15 min three times; then captured the image by Typhoon Trio (EX/EM, 532/580 nm, 50 um resolution) and analyzed by software ImageQuant TL 7.0. All these steps were performed at 21uC.

Adenosine Triphosphate (ATP) Synthase Enzyme Activity
The ATP synthase activity and quantification were performed with an ATP synthase Specific Activity Microplate Assay Kit from ABCAM (ab109716). The STR tissue (with samples pooled as before) was homogenized in a buffer as described in the literature [28] to obtain the mitochondrial fraction. The F1F0 ATP synthase extraction from mitochondrial fraction was according to manufacturer's instructions. Briefly, 10 ug per well of 10% detergentextracts of protein was incubated in a 96-well plate at room temperature for 3 hrs to immuno-immobilize F1F0-ATP synthase, the enzyme hydrolyzing ATP to ADP and phosphate, and the ADP product is ultimately coupled to oxidized NADH to NAD + which led to absorbance decreases at 340 nm. The reaction was monitored by Spectromax M2 microplate reader (Molecular Devices). Subsequently, ATP synthase protein (amount, not activity) was quantified using an antibody directed at ATP5B and coupled to alkaline phosphatase, which converts the pNPP substrate to a yellow color assayed at 405 nm.

Possible Role for CREB in Protein Regulation
Because our prior research suggests EC/IC behavioral differences are due in large part to decreases in CREB activity in the NAcc of EC rats [11] and that blocking CREB activity produces an EC like behavioral phenotype in 14 different behavioral paradigms relating to addiction, depression, anxiety and natural reward [11,12,13,14,15], one might hypothesize many of the regulated proteins in this study would be CREB target genes. Accordingly, we utilized data from a published chromatin immunoprecipitation (ChIP) study [29] to identify which of our regulated proteins are CREB target genes. For our purposes, a CREB target gene is defined as P,0.005 in two or more ChIP experiments as listed in Table S4 of the Zhang et al. paper.

Results
In this study, we investigated dynamic protein changes in the NAcc of EC and IC rats. We first analyzed differences in basal protein levels between EC and IC rats and then analyzed the rapid proteomic responses to restraint stress separately for EC and for IC rats. We employed gene/protein ontology analyses as well as Ingenuity Pathway Analyses to identify coordinated responses to stress.

Basal Proteomic Differences between EC and IC Rats
Separate 2-DE gels were used for cytosolic and membrane fractions. The cytosolic fraction yielded 4 discrete proteins from 5 spots (see Figure 1; Table 1) and the membrane fraction yielded 10 separate proteins from 11 differentially expressed spots (see Figure 2; Table 2). It is notable that ENO1 was identified in two different spots with similar apparent molecular weight but significantly different pI. A similar situation occurred with VDAC1 except that the pI shift was not as dramatic.
When proteins of the cytosolic and membrane fractions were added together, the gene/protein ontology analysis produced 5 main clusters of related features, specifically relating to mitochon-drial membrane localization, oxidative phosphorylation, mitochondrial electron transport, nucleotide binding and mitochondrial matrix (see Table 3). The effect was such that the EC group had less expression of the proteins comprising the categories above. The overwhelming over-representation of mitochondriarelated proteins is a theme repeated in the IC stress analysis below.
The IPA analysis yielded top biological functions (with more than one protein) including cell death of neuroblastoma cell line, biosynthesis of purine ribonucleotide, refolding of protein, transport of synaptic vesicles and synthesis of ATP ( Figure 3a; Table 4). Note that the protein folding function included heat shock proteins HSPA8 and HSPD1. An additional heat-shock protein (HSP90AB1) was induced by stress in IC rats (see below). Top canonical pathways include oxidative phosphorylation and mitochondrial dysfunction (Figure 4a; Table 5).
Using the Network Analysis feature, a high scoring network (IPA network score of 40) involving 13 of the 14 identified proteins was detected (Figure 5a). The one feature all 13 proteins shared was binding with ubiquitin. Of these, the two heat-shock proteins HSPA8 and HSPD1, along with ENO1, TUBA1A were upregulated in EC rats. The rest were down-regulated.

Effect of 30 Min Restraint Stress on Protein Expression in IC Rats
Analysis of the gels for the cytosolic fraction yielded 5 unique proteins from 8 individual differentially regulated spots ( Figure 1; Table 1). The membrane fraction yielded 13 proteins from 16 spots ( Figure 2; Table 2). One additional spot (YWHAZ) with a Pvalue of 0.06 was included in the analysis. Proteins identified in two or more discrete spots were ATP5A1 (four spots), GAPDH (four spots) and DPYSL2 (two spots). It is noteworthy that the identified VDAC1 spot was not one of the two VDAC1 spots identified in the analysis of basal EC/IC protein differences as this spot had a lower pI than either of the other two. Of all regulated spots, 23 of 25 were downregulated after stress. A protein involved in glycolysis, ENO1, was lower in IC rats compared to EC rats  under basal conditions and decreased further in IC rats subsequent to stress. When analyzed together, the gene/protein ontology analysis derived 4 clusters of related attributes, namely those relating to glycolysis, mitochondria, 14-3-3 protein family and pigment granule ( Table 3). Note that the mitochondrial proteins were lower after stress, bringing them down in the neighborhood of basal EC expression.
The IPA analysis produced top biological functions of death of tumor cell line, cholangiocarcinoma, glycolysis, synthesis of reactive oxygen species, and targeting of protein ( Figure 3b; Table 4). A large number of high-scoring canonical pathways were identified, among them being glycolysis, AKT signaling, mitochondrial dysfunction, 14-3-3 signaling, and oxidative phosphorylation (Table 5; Figure 4b). All of the proteins comprising glycolysis and synthesis of reactive oxygen species functions were decreased in IC rats after stress.

Effect of 30 Min Restraint Stress on Protein Expression in EC Rats
The 2-DE gel analysis for the cytosolic fraction yielded 6 proteins across 9 individual spots ( Figure 1; Table 1) and the membrane fraction produced 7 identifiable proteins from 8 spots (one spot was unidentifiable; Figure 2; Table 2). In stark contrast to IC rats, all 16 identifiable spots were upregulated subsequent to stress for EC rats. Of these proteins, PKM2 was identified from 2 separate spots and ENO1 produced a train of 3 regulated spots, including the spot identified in the basal EC/IC and the IC stress analyses. As mentioned above, IC rats had lower basal levels of this protein and sank lower still after stress, while EC rats had higher basal levels and these levels increased even more with stress.
The gene/protein ontology analysis of all 13 spots only derived one cluster of classifications, that cluster being related to glycolysis (Table 3). Although this gene/protein ontology cluster is the same as in IC stressed rats, in this case the glycolysis proteins increased in EC rats rather than decreasing as seen with IC rats.
Top unique biological functions were catabolism of acidic amino acids, metabolism of nucleic acid component or derivative, and glycolysis of cells (Figure 3c; Table 4). The IPA analysis top canonical pathways were as follows: phenylalanine, tyrosine and tryptophan biosynthesis, glycolysis, and glutamate metabolism (Figure 4c; Table 5).
As with the basal EC/IC and the IC stress analyses, the network analysis for EC stress identified another network (score = 40) with ubiquitin c as a major node, only in this case, all proteins were upregulated subsequent to stress (Figure 5d). All 13 proteins regulated by stress in EC rats exhibit protein-protein interactions with ubiquitin c.

Orthogonal Validation of Protein Regulation Via Western Blot
Because CBB stains all protein, and because multiple different proteins can migrate to the same position on a 2D gel, it is important to provide orthogonal validation to verify that the 2D gel procedures coupled with mass spectrometry are correctly  identifying regulated proteins. Accordingly, three of the most critical regulated proteins were validated via 1D Western blot. The antibody specificity of Western blots provides additional confidence in the identification of regulated proteins. Figure S1 shows that in the cases of ENO1 (F(1,10) = 13.9, P,0.005), HSPA8 (F(1,5) = 19.8, P,0.05) and PKM2 (F(1,10) = 12.2, P,0.01), the CBB spot intensity correlated well with antibody-labeled signal on a 1D Western blot.

ATP Synthase Activity
Validation of differences in protein expression is good practice, but more important than protein expression is protein function. Accordingly, because energy producing proteins were most regulated by enrichment and stress, we performed a functional assay of ATP synthase activity. Results of the functional assay demonstrated that EC rat tissue exhibited more ATP synthase activity than IC rats (main effect: F(1,12) = 15.8, P,0.005; Figure 6a) and that, overall, stress increased ATP synthase activity (main effect: F(1,12) = 5.4, P,0.05). Further, there was a significant interaction (F(1,12) = 7.9, P,0.05). The nature of this interaction was such that the Stress main effect increase in activity was driven entirely by EC rats. Regarding quantity (Figure 6b), EC rats had slightly higher levels of ATP synthase (ATP5B) than IC rats (main effect: F(1,12) = 7.0, P,0.05; Figure 6a) but no significant Stress or interaction effect.
Looking more closely at these data (Figure 6c), one can see that there was a significant correlation of ATP synthase activity and ATP5B expression in EC rats (Pearson coefficient = 0.78, P,0.05) but not IC rats (Pearson coefficient = 0.58, n.s.). Similar results were found for ATP synthase activity and ATP5C1 expression in EC (Pearson coefficient = 0.81, P = 0.05; Figure 6d) but not IC rats (Pearson coefficient = 0.19, n.s.). No such relationship was found with expression of the non-CREB target ATP5A1 (data not shown).

Regulated Energy Metabolism Proteins are CREB Target Genes
Less than 30% of promoters for known protein-coding genes are bound by CREB [29]. For the energy metabolism proteins regulated in this study (Figure 7, large symbols), 11 of 13 proteins (84%) are CREB target genes (red symbols). Of the non-energy metabolism proteins identified in this study, only 38% are CREB target genes. This suggests that the low CREB phenotype that mimics the EC protective depression phenotype might be a function of differential energy metabolism.

Discussion
The current study utilizes the environmental enrichment paradigm, an animal model producing an antidepressant-like behavioral phenotype [11] to investigate differential protein regulation subsequent to an acute psychological stressor in rats. Environmentally enriched rats have been shown previously to be much less responsive with regard to mRNA induction of immediate-early genes 30 min post stress compared to rats raised in isolation [12]. The current experiments further demonstrate robust differences in protein expression patterns between EC and IC rats subsequent to the same stressor. All of the proteins regulated by stress in EC rats were increased in expression while the vast majority of regulated proteins were decreased in expression for IC rats after stress. These results, after only 30 min of stress, highlight the rapid and dynamic nature of protein regulation in response to a changing environment.

Energy Production
There is increasing evidence from the human literature that energy metabolism plays an important role in major depression [30,31] and animal studies corroborate these results in the responses to chronic psychological stress [32,33,34] and antidepressants [34,35]. The most striking result of this study was the rapid and robust regulation of proteins involved with glycolysis, the citric acid cycle and mitochondrial respiration. The major product of glycolysis is pyruvate, which is shuttled into the mitochondria to be used in the citric acid cycle which in turn feeds the electron transport chain to produce ATP. Fifteen of the total 35 proteins regulated (42%) in this study were related to energy production. This number also includes creatine kinase, an enzyme that rapidly creates ATP from phosphocreatine storage reservoirs in neurons in response to rapidly increased energy demands [36].
Under basal conditions, EC rats showed lower expression of seven proteins related to energy production and higher expression of only one protein (ENO1). After 30 min of restraint stress, five energy production-related proteins were induced in EC rats, including the ATP synthase protein ATP5C1. A functional ATP synthase assay confirmed that EC rats exhibited an increase in the ability to produce ATP within 30 min of restraint stress. In stark contrast to EC rats, IC rats exhibited a coordinated decrease in expression nine energy-related proteins and no increase in the ability to synthesize ATP after stress.
The ENO1 protein is unique among the proteins identified in this study. This protein was identified as 4 discrete spots, each with a high expression level. All had similar molecular weight but different pI. One spot in particular showed higher basal expression in EC rats compared to IC rats and was increased by stress in EC rats yet decreased by stress in IC rats. This opposite modulation of a highly expressed protein draws interest, particularly since this protein has also been identified in proteomic studies of chronic stress [33] and after antidepressant treatment [35].
Creatine kinase is a protein that stores energy as phosphocreatine for a rapid response to increases in demand for ATP [36]. More specifically, this reversible mechanism stores energy when the cell is not being challenged and then quickly releases the stored energy on demand independently from the glycolysis/citric acid/ electron transport system. This could be particularly important in light of the opposite modulation of glycolysis and electron transport by stress in EC versus IC rats. In this study, EC rats expressed less mitochondrial creatine kinase under basal conditions, but IC rats exhibited a nearly 3-fold decrease in expression of this protein after 30 min of restraint stress.

Mitochondria
The citric acid cycle and electron transport functions described above take place in the mitochondria. With such coordinated regulation of these proteins, one must ask the question of whether the ATP synthesis pathways are being regulated individually or if the mitochondria themselves are being regulated. The rapid timecourse of changes in protein levels might make one skeptical, but mitochondrial proteins not directly involved with the citric acid cycle or electron transport are regulated in concert with the energy metabolism proteins. For example, VDAC1 is not directly involved in energy metabolism but is lower in EC basal conditions and is decreased in IC tissue after stress in an identical fashion to the energy metabolism proteins. Similarly, SAMM50, TUFM and GLUD1 are non-energy mitochondrial proteins induced in EC tissue subsequent to stress in a similar fashion to the energy metabolism proteins. Although this idea is speculative at this point, the available evidence here is intriguing. The idea that major depression in humans might be a function of mitochondrial dysfunction is not novel [37]. There is increasing evidence showing that depressed patients are more likely to have mitochondrial disorders [38], that patients with mitochondrial disorders are more likely to be depressed [39] and that postmortem brain tissue from patients with major depression exhibit alterations in mitochondrial proteins [31]. It is important to note that these studies focus specifically on genetic perturbations of mitochondrial genomes leading to mitochondrial dysfunction. The current study complements these prior studies by demonstrating that acute psychological stress produces a rapid change in mitochondrial protein expression and that the direction of protein changes are dependent upon the living environment of the subject (EC vs. IC).

Heat-shock Proteins
Under basal conditions, EC rats had higher expression of the constitutively expressed heat-shock proteins HSPA8 (a.k.a. HSC70 or constitutive heat shock protein 70) and HSPD1 (a.k.a. chaparonin 60). It is interesting to note that the latter is a mitochondrial-associated protein and that a majority of differentially expressed mitochondrial proteins were decreased in EC rats  compared to IC rats. However, this protein is also found in the cytosol and the differential expression in this study was found in the cytosolic fraction. In addition to the two constitutive proteins differentially expressed under basal conditions, an additional heatshock protein, the inducible HSP90AB1 was decreased by psychological stress in IC rats. Interestingly, this protein is implicated in signal transduction subsequent to its role in refolding denatured proteins [40,41]. Another protein likely pulling double duty is ENO1. There is evidence that the Saccharomyces cervesiae homolog of the ENO1 protein (HSP48) also functions as a heat shock protein [42].
Although a role for these particular heat-shock proteins in human depression has yet to be identified, preclinical models have identified HSPA8 and HSPD1 as proteins regulated subsequent to chronic stress [32,33]. In addition, HSPD1 is a protein shown to be induced by antidepressant treatment [34]. The fact that the current study found higher expression of both proteins under basal conditions in EC versus IC rats, one might hypothesize that these proteins might underlie the EC diminished response to an acute stressor.

14-3-3 Proteins
The 14-3-3 family of proteins has received considerable attention in brain and in relation to neurodegenerative diseases [43]. These proteins can have a multitude of effects on other proteins, particularly those that are phosphorylated at Ser or Thr [44]. Specifically these proteins can play a role in changing the conformation of target proteins, protecting proteins from phosphatases or proteasome degradation, binding of two target proteins, and targeting/sequestering proteins in specific cellular compartments [45]. With such high expression and varied functions, these proteins are very important to the functioning of the cell. Indeed, a recent paper identified YWHAZ as a ''switchboard'' gene for neuropsychiatric disorders [30]. Thus, although the functional consequences have yet to be elucidated, the dynamic regulation of YWHAG, YWHAE and YWHAZ after stress in IC rats is intriguing.

DPYSL2 (CRMP2)
The Allen Brain Atlas shows that the mRNA for DPYSL2 exhibits targeted expression specifically to the NAcc compared to that of the dorsal striatum in mouse brain [46]. The protein DPYSL2 was found to be differentially regulated in EC and IC rats, with EC rats showing increased expression after psychological stress and IC rats showing decreased expression. Although the dominant function of this protein in adult brain has not been conclusively identified, it is clear that this protein is somehow involved with Alzheimer's disease [47,48] and possibly schizophrenia [31]. Relating to depression, this protein has appeared in preclinical proteomic studies of chronic stress and also antidepressant treatment [32,33,34].
The protein itself plays an important role in axon guidance and neurite outgrowth in developing neurons, as well as vesicle trafficking functions [48,49]. Interestingly, this protein has also been shown to affect current density of N-type voltage-gated Ca 2+ channels, presumably by controlling localization [50]. This effect has been shown to be functional, as evidenced by a decrease in vesicular neurotransmitter release [51]. Further, this protein has also been shown to regulate trafficking of NMDA receptors, providing an attractive target for preventing excitotoxicity [52].

Ubiquitin c
The utility of IPA's Network analysis is to identify proteins and systems that may not be directly identified as being regulated but still play an important role in regulating other proteins. Three of the four identified networks ( Figure 5) were centered on ubiquitin protein-protein interactions (26 of the total 35 proteins). Ubiquitin conjugation has been shown to be involved with proteasomal degradation, but an estimated 50% of ubiquitin sites are not proteasomal and presumably represent signaling modifications [53]. It is interesting that all 4 ubiquitin targets in the IC Stress network (Figure 5c) were downregulated by stress and all 13 ubiquitin target proteins in the EC Stress network were upregulated. Although ubiquitin c itself is too small to be detected on these gels, these networks suggest that EC and IC rats may show differential ubiquitin regulation. A recent bioinformatics paper has identified ubiquitin c as a possible ''switchboard'' gene for psychiatric disorders including depression [30].

Other Findings
Interestingly, one recent investigation of protein-protein interactions in human schizophrenic, bipolar and major depression prefrontal cortex tissue identified a very tight network of 12 proteins [30]. Of the 12 proteins identified by Lee et al., 6 were identified in the current study (HSPA8, HSPD1, HSP90AB1, GAPDH, TUBA1A and YWHAZ). Another protein in the network, ubiquitin c, is a protein too small to be seen on the 2DE gels used here but this protein was identified as a hub in multiple networks of the current study. For the UBC-centric network of EC/IC basal differences in protein expression (Figure 5a), EC rats had higher expression of only 4 of the 13 proteins. Of these 4 proteins, 3 were members of the network identified by Lee et al. (HSPA8, HSPD1 and TUBA1A).
Three of the networks identified in Figure 5 (a, b and d) were very high-scoring networks (scored 36 or better), each comprising a clear majority of regulated proteins in each analysis. These extensively-regulated networks suggest coordinated regulation of protein degradation (see ubiquitin discussion above) or kinase signaling cascades. The kinases themselves are typically regulated by phosphorylation, a modification not easily discerned when looking at total protein expression. As a result, the kinases themselves are not revealed, but network analyses can point to likely kinase involvement. Figure 5a shows ERK as a hub likely to be involved in EC/IC basal differences. Prior research provides Figure 6. Functional assay of ATP synthase activity and ATP5B quantity. Bars represent mean (6 SEM) amount of ATP synthase activity in mM/min (a) and quantity in OD/min (b) of ATP synthase. Asterisk (*) represents statistically significant difference from control. Bottom panels depict correlation of activity and quantity as measured from the kit (c) and activity and quantity of ATP5C1 as measured from the 2D gel (d). Circles represent IC rats and triangles represent EC rats. Controls are white symbols and Stress groups are red. Solid line depicts a significant correlation for EC rats and dashed line depicts a lack of correlation in IC rats. doi:10.1371/journal.pone.0073689.g006 ample evidence that ERK plays a role in depression [54,55]. Figure 5b also identifies ERK as a likely player in IC control vs. stress along with other known key signaling cascades P38 MAPK and PI3K/AKT. These results are also supported by literature demonstrating the importance of these cascades in depression [56,57]. Additionally, histone H3 was identified as a possible player.

Role for CREB
Our prior research has shown a clear role for CREB activity in producing the protective depression and addiction phenotypes seen with environmental enrichment [11,12,13,14,15,16,58]. Specifically, EC rats exhibit a very unique behavioral phenotype: they show less spontaneous locomotor activity than IC rats, show enhanced responsiveness to cocaine-and amphetamine-stimulated locomotor activity and conditioned place preference [11,59,60], exhibit an antidepressant-like phenotype in sucrose preference, social interaction and forced swim test paradigms [11], yet EC rats exhibit an anxiogenic-like phenotype in sucrose neophobia and cold-stress-induced defecation paradigms [11]. This very specific phenotype can be recapitulated by viral-mediated overexpression of a dominant-negative mutant CREB [13,16,18], an endogenous dominant-negative inhibitor of CREB transcription (i.e. ICER) [13], or by knocking down CREB expression with an shRNA vector [11] in the NAcc. Additionally, we showed that EC rats have less basal pCREB (i.e. active CREB) in the NAcc [11], an effect replicated by the Zhu laboratory [61].
Because a clear majority of nuclear-encoded mitochondrial genes are CREB target genes [29], and we know that CREB is an important mediator of the enrichment phenotype [11], it is not surprising that approximately half of the differentially-expressed proteins from EC and IC NAcc are mitochondrial proteins. When examining glycolysis proteins with TCA cycle proteins and electron transport proteins, 11 of the 13 regulated proteins are CREB target genes. Although further investigation is necessary to determine if these are the specific CREB target genes mediating the EC phenotype, the current data are consistent with the CREB hypothesis of environmental enrichment [11] and point to energy metabolism as a possible proximal mechanism of this phenotype.

Conclusions
Although other studies have examined the effects of chronic stress in other brain regions [32,33,34], this is the first such study to assess rapid and dynamic changes in protein levels after acute stress in the NAcc in a rat model of resistance to depression-like behavior. Our results identify rapid and opposite changes in proteins of EC and IC rats. The most robust findings are proteins related to energy metabolism (e.g. glycolysis, citric acid cycle and electron transport). Concomitant changes in other mitochondrial proteins not directly related to these functions suggest that the regulation might be at the mitochondrial level rather than individual proteins. These results further support the mitochondrial hypothesis of major depression [37]. In addition, the differences in energy metabolism, along with changes other proteins, provide possible new avenues for pharmacotherapeutic strategies for the treatment of stress and/or depression. Figure S1 Orthogonal validation of expression changes via correlation of 2D gel CBB normalized intensity and 1D antibodylabeled Western blot normalized intensity for (a) ENO1 (R 2 = 0.58), (b) HSPA8 (R 2 = 0.83) and PKM2 (R 2 = 0.55). (TIF) Figure 7. Regulated energy metabolism proteins are CREB targets. The diagram depicts significantly regulated proteins in the glycolysis, TCA and electron transport pathways. The thirteen significantly regulated proteins are shown as large symbols. In addition, 12 of the 13 regulated proteins are from CREB target genes (shown as red symbols). doi:10.1371/journal.pone.0073689.g007