Glucocorticoids Regulation of FosB/ΔFosB Expression Induced by Chronic Opiate Exposure in the Brain Stress System

Chronic use of drugs of abuse profoundly alters stress-responsive system. Repeated exposure to morphine leads to accumulation of the transcription factor ΔFosB, particularly in brain areas associated with reward and stress. The persistent effects of ΔFosB on target genes may play an important role in the plasticity induced by drugs of abuse. Recent evidence suggests that stress-related hormones (e.g., glucocorticoids, GC) may induce adaptations in the brain stress system that is likely to involve alteration in gene expression and transcription factors. This study examined the role of GC in regulation of FosB/ΔFosB in both hypothalamic and extrahypothalamic brain stress systems during morphine dependence. For that, expression of FosB/ΔFosB was measured in control (sham-operated) and adrenalectomized (ADX) rats that were made opiate dependent after ten days of morphine treatment. In sham-operated rats, FosB/ΔFosB was induced after chronic morphine administration in all the brain stress areas investigated: nucleus accumbens(shell) (NAc), bed nucleus of the stria terminalis (BNST), central amygdala (CeA), hypothalamic paraventricular nucleus (PVN) and nucleus of the solitary tract noradrenergic cell group (NTS-A2). Adrenalectomy attenuated the increased production of FosB/ΔFosB observed after chronic morphine exposure in NAc, CeA, and NTS. Furthermore, ADX decreased expression of FosB/ΔFosB within CRH-positive neurons of the BNST, PVN and CeA. Similar results were obtained in NTS-A2 TH-positive neurons and NAc pro-dynorphin-positive neurons. These data suggest that neuroadaptation (estimated as accumulation of FosB/ΔFosB) to opiates in brain areas associated with stress is modulated by GC, supporting the evidence of a link between brain stress hormones and addiction.


Introduction
Opiate drugs, such as morphine, are effective analgesic agents that are used for treating many forms of acute and chronic pain. However, serious adverse effects such as tolerance and withdrawal contribute to opiate dependence and limit their use. Further, the non-medical use of opiates (heroin, morphine) has increased during the past few years. Increasing evidence implicates various mechanisms of gene regulation (including epigenetic, molecular, cellular and circuit level effects) in the changes that drugs of abuse induce in the brain, indicating a potential therapeutic strategy for addiction therapy [1][2][3][4].
A central question in the drug abuse field has been to identify proteins that mediate the transition from acute to long-term effects of those drugs. Of particular interest in the study of addiction is the Fos family of transcription factors. This family includes c-Fos, Fra-1 and Fra-2, FosB and DFosB, a truncated splice variant of fulllength FosB [5]. In contrast to other members of the Fos family, DFosB is modestly induced in the brain after acute drug administration, but because of its unusual long half-life it persists for weeks, even months, after the cessation of drug use. As a result, DFosB levels gradually accumulate with repeated drug exposure [6,7], suggesting that DFosB could represent a mechanism by which drugs of abuse produce lasting changes in gene expression pattern long after the drug is withdrawn [8].
It has been reported that repeated exposure to cocaine, amphetamine, cannabinoids or morphine leads to an increase in DFosB in brain areas related to the positive reinforcing effects of drugs, such as nucleus accumbens (NAc), prefrontal cortex and dorsal striatum. This increase has been proposed to be a neuroadaptation that leads to increased sensitivity to drugs of abuse and vulnerability to develop characteristic behaviours of addiction [9][10][11][12][13]. We have recently shown that the enhancement of FosB/DFosB levels after chronic morphine administration is not only restricted to the reward system, but also occurs in the brain stress system (which has been related to the negative reinforcing effects of drugs), as well as in the nucleus of the solitary tract-A 2 noradrenergic cell group (NTS-A 2 , the main noradrenergic system innervating the stress neurocircuitry) [14]. In consonance with these findings, several forms of chronic stress also induce DFosB in the NAc and other brain regions [15,16].
Addiction is a complex disorder because many factors contribute to the development and maintenance of this neurological disorder. One factor is stress, which has been involved in specific aspects of drug addiction [17][18][19][20][21]. Both the hypothalamuspituitary-adrenal (HPA, the primary endocrine stress pathway) axis and the extrahypothalamic stress system (which comprises the extended amygdala and the NTS-A 2 ) are dysregulated by chronic administration of drugs with dependence potential [14,22]. In addition, HPA axis response is similar after both stressful stimuli and acute exposure to drugs of abuse [23,24], with elevated corticotropin-releasing hormone (CRH), adrenocorticotropic hormone (ACTH), and glucocorticoids (GC) release. This response facilitates adaptation to acute environmental changes, but also may lead to behavioural pathologies during chronic stress conditions, such as addiction and depression [25]. Studies have not yet examined the relationship between GC and morphine dependence-induced FosB/DFosB induction in the brain stress system. We therefore assessed the influence of GC on FosB/DFosB expression after continuous administration of morphine in brain stress-related areas. To address this question, we first examined the effects of bilateral adrenalectomy (ADX) on FosB/DFosB immunoreactivity (IR) in the main nuclei of the stress system in morphine-dependent rats.
The activity of the brain stress system is mediated by a number of neurotransmitters/neuromodulators. CRH is the main neuropeptide regulating the stress system activity and it has been postulated its contribution in both a pre-existing vulnerability to use drugs addictively and later vulnerability to relapse [26]. In addition, abundant work supports the importance of the NTS-A 2 innervating the brain stress system in drug addiction and the pivotal role of noradrenaline (NA) as a neurotransmitter modulating this neurocircuitry [27]. Finally, substantial evidence suggests that dynorphin expression is activated in the striatum and amygdala during acute and chronic drug administration [28]. Given these facts, the following aim of this study was to identify the role of GC on FosB/DFosB expression in specific populations of the brain stress system during morphine dependence.

Effects of Adrenalectomy on Body Weight Gain and Plasma ACTH and Corticosterone Concentration in Morphine-dependent Rats
Before performing the immunodetection assays, we assessed the efficacy of chronic treatment with morphine. For this purpose, the weight of animals was recorded on the day of pellets implantation and on the day of killing (day 10).  [29,30], post hoc analysis indicated that both sham and ADX groups rendered dependent on morphine showed a significantly (p,0.001) lower weight gain (213.7565.0 g, n = 12; 3.8462.45 g, n = 13, respectively) than that observed in sham and ADX animals receiving placebo pellets (44.5861.7 g, n = 12; 49.5762.4 g, n = 12, respectively), which has been attributed to the reduced food intake observed in these animals [29].
To check the efficacy of the adrenalectomy, hormone concentrations were measured in plasma. Two-way ANOVA examining effects of adrenalectomy and morphine on ACTH and corticosterone plasma concentration showed significant main effects of adrenalectomy [ACTH: F(1,18) = 68.12, p,0.0001; corticosterone: F(1,45) = 10.42, p = 0.0023). As expected, Newman's post hoc test showed (Fig. S1) that in placebo (n = 6)-and morphine (n = 4)-ADX rats plasma concentration of ACTH was higher (p,0.001) compared with sham-operated animals for the two treatment evaluated (placebo, n = 6; and morphine, n = 6). No modifications were observed in plasma corticosterone concentration between ADX-placebo (n = 14) and ADX-morphine (n = 13) treated rats. Concentration of corticosterone in ADX-morphine treated rats were significantly (p,0.01) lower than those seen in sham-morphine (n = 10) treated rats. No significant changes were seen in sham-morphine group compared with sham-placebo (n = 12).

Adrenalectomy Differentially Attenuates Morphine Dependence-induced FosB/DFosB in the Brain Stress System Sub Regions
In control animals (placebo-implanted rats), weak constitutive expression of FosB/DFosB-IR was identified in all the areas of the stress-related brain system. Chronic morphine treatment resulted in the appearance of FosB/DFosB in all the brain areas investigated. In the NAc(shell), two-way ANOVA showed  Table 1, Newman-Keuls post hoc test revealed that ADX placebo-and morphine-pelleted rats showed lower CRH neurons (p,0.001) than the shamplacebo or morphine-dependent groups. At the PVN, two-way ANOVA showed a significant effect of adrenalectomy  Table 1 depicts that there was a significant decrease in CRH neurons at the CeA from ADX placebo-or morphine-treated rats. Table 1 also shows that chronic morphine exposure induces an elevation (p,0.05) of CRH-positive neurons at BNST and CeA levels.

Effects of Adrenalectomy on FosB/DFosB into DYNpositive Neurons in the NAc(Shell)
We observed no significant changes in the number of pro-DYNpositive cells that co-express FosB/DFosB in the NAc(shell) after chronic morphine exposure (Fig. 7) regarding to the placebo group. Two-way ANOVA for FosB/DFosB-positive/pro-DYNpositive neurons in the NAc(shell) showed main effect of ADX [F(1,19) = 10.11, p = 0.0049]. As shown in Fig. 7C, the number of FosB/DFosB-positive neurons that co-express pro-DYN in ADX morphine-dependent rats was significantly (p,0.05) lower regarding to the corresponding treatment in sham animals. As shown in Table 1, ADX induced no changes in total pro-DYN cell from the NAc(shell).  Table 1, ADX induced no changes in total TH-positive neurons in the NTS.

Discussion
The current results indicated, for the first time, that brain GC signalling modulated chronic morphine administration-induced FosB/DFosB expression in the brain stress system in a regionspecific manner.
Converged lines of evidence indicate that stress increases risk of addictive behaviours [18]. Persistent stressful stimuli alter synthesis, expression and signalling in stress-related pathways (e.g. CRH, GC, NA, etc). In addition, drugs of abuse affect the stress pathways, which results in alteration of in gene expression, with signalling effects on reward and stress-related molecules [31]. Stress and abused drugs share the ability to trigger overlapping patterns of neuronal activation within the central nervous system, resulting in the activation of immediate gene expression. It has been extensively described that addictive substances and chronic stressful stimuli increase the expression of the transcription factor DFosB in the main nuclei involved in their positive reinforcing effects [12,13,32,33], and it has been proposed that the persistent effects of DFosB on target genes might play an important role in the development of adaptations that characterize addiction [9,34]. However, little is known about either the expression of DFosB in the brain stress system after chronic administration of drugs of abuse or the molecular mechanisms of morphine-evoked DFosB accumulation in stress-related areas. In the present study, we investigated the involvement of GC on the morphine-induced FosB/DFosB expression in the hormonal stress system (HPA axis), which is controlled by CRH in the PVN, as well as in the extrahypothalamic stress systems (which include the extended amygdala; [35]), mediated by CRH and other stress-related systems (including NA, and dynorphin; [19]).
We recently demonstrated that chronic morphine administration for seven days increased FosB/DFosB expression in the extended amygdala, PVN and NTS-A 2 [14]. Consistent with these data, present results show that chronic morphine administration evoked an increase in FosB/DFosB in CeA, BNST and NAc(shell) as well as in the PVN and NTS-A 2 . The extended amygdala has been associated with drug reward. In fact, all major drugs of abuse activate dopaminergic transmission from the VTA to the NAc(shell) and CeA. This activation and the consequent positive reinforcing effects of substances of abuse have been shown to be related with GC actions on GR located in the VTA [28]. Supporting this hypothesis, our results show that FosB/DFosB-IR in the NAc(shell) and the CeA was attenuated in ADX animals, which might indicate a cross-talk between AP-1 transcription factors and GR during morphine dependence [36]. In contrast to the observed effects of ADX on FosB expression in NAc and CeA, present results suggest that chronic morphine increased Fos/ DFosB expression in the PVN and BNST in a GC-independent manner.
Several neurotransmitters of the brain stress system, such as CRH, NA and DYN, have been related with the aversive states characteristic of the addiction process [35,[37][38][39]. Present findings showed that chronic morphine exposure evoked a significant increase in Fos/DFosB expression within CRH-containing neurons in the BNST, CeA, and PVN. The Fos family of transcription factors can act at cyclic AMP-response element (CRE) sites [40]. Considerable evidence indicates that DFosB can act as either a transcriptional repressor or activator [40,41]. Given that CRH gene has a CRE motif in its promoter sequence, it might be proposed that FosB/DFosB accumulation in CRH neurons can mediate the morphine-induced changes in CRH levels, as reported for cocaine effects [42], especially in the CeA and the BNST, where, for the first time, we reported an enhancement in the number of CRH-positive neurons during morphine dependence. Supporting this hypothesis, an increase in CRH mRNA levels has been described in the CeA after chronic administration of morphine [43]. However, the number of CRH-positive neurons was unchanged in the PVN after chronic morphine treatment. These results are in accordance with previous findings showing that chronic morphine exposure does not induce changes in the PVN CRH hnRNA [44]. Since PVN CRH expression is downregulated by GC [45], and given that present work and others [29,44] showed that chronic opiate exposure did not modify corticosterone release, it seems logical that chronic morphine administration did not change the number of CRH cells.
Peripheral administration of corticosterone increases CRH mRNA expression in the CeA and BNST [46,47]. Additionally, ADX decreased CRH expression in the CeA [48,49]. Accordingly, we have observed that the increase in the number of CRH neurons in the BNST and CeA during morphine dependence was abolished after adrenalectomy. Given that the increase of Fos/DFosB within CRH neurons during morphine dependence was attenuated in ADX animals, it might be suggested a role for Fos/DFosB in the regulation of CRH expression by GC in the extended amygdala. On the other hand, it is well known that GC negatively regulate the expression of CRH gene in the PVN [45]. In agreement, present data clearly showed that adrenalectomy abolished the increase in Fos/DFosB-IR in CRH-containing neurons during morphine dependence in this nucleus.
In this study, immunochemical data revealed that chronic morphine treatment significantly increased staining for Fos/ DFosB in the NTS-A 2 , which was decreased following ADX. When we examined the specific neural populations that expressed FosB/DFosB, we found a robust increase of Fos/ DFosB-IR within TH (the rate-limiting enzyme in catecholamine synthesis)-positive neurons. NA has been shown to play a main role in addiction [50]. In prior work, chronic morphine exposure evoked an enhancement in TH levels in the NTS-A 2 [14,29]. It is known that TH gene has an AP-1 site in its promoter [51]. Our results might suggest that FosB/DFosB is involved in opiate-induced increase of TH levels in the NTS-A 2 . Brainstem noradrenergic cell group express high levels of GC receptors (GR). It has been shown that GC have a permissive role in noradrenergic neurotransmission [29,30,52]. Accordingly, administration of GR antagonists affected several aspects of NA activity, including TH activation and neuronal activity [30]. Present results showed that, following ADX there was a decrease in TH-positive cells expressing FosB/DFosB in the NTS-A 2 . Taken into account that adrenalectomy blocked the increase in TH protein levels in the NTS-A 2 during morphine dependence [29], and that a GRE/AP-1 site has been described in the TH gene promoter [53], our data might point out DFosB as a mediator in the effects of GC on noradrenergic activity in the NTS-A 2 during opiate dependence.
DYN has been postulated as a possible target gene of DFosB [54,55]. Our results show that chronic morphine exposure did not significantly altered FosB/DFosB expression within pro-DYNexpressing neurons in the NAc(shell). Regarding DFosB regulation of DYN expression, Zachariou et al [33] reported a small but significant decrease of DYN mRNA levels in the NAc from DFosB over-expressing mice, thus postulating that this transcription factor inhibits DYN expression. From our data, it cannot be concluded that chronic morphine modifies DYN expression through FosB/ DFosB activity, given that our results have been obtained at the protein level.
It has been postulated that DYN/kappa opioid system seems to induce pro-depressive-like states that involve elements of aversion. This aversive response may involve reciprocal interactions with NAc, DA and the extrahypothalamic CRH system [35]. However, little is known about a possible GC regulation of DYN expression Figure 2. Adrenalectomy differentially regulates FosB/DFosB protein expression in the brain stress systems. Photographs represent immunohistochemical detection of FosB/DFosB in the NAc(shell) (A-D), BNST (F-I) and CeA (K-N) from sham-operated and adrenalectomized (ADX) rats pretreated with placebo (pla) or morphine (mor) pellets for 10 days. B', D'. G', I', L' and N' are high magnifications. Scale bar: 100 mm (30X, low magnification); 20 mm (100X, high magnification). LV, lateral ventricle; ac, anterior comisure; BLA, basolateral amygdala. E, J, O, quantitative analysis of FosB/DFosB immunoreactivity in the three nuclei from sham and ADX animal. Data correspond to the mean 6 SEM (percent of control). Post hoc comparisons revealed a significant increase of FosB/DFosB protein expression in sham animals after chronic morphine exposure in the NAc(shell), BNST and CeA (**p,0.01; ***p,0.001 versus sham-placebo animals). In ADX-morphine dependent rats it was observed an attenuation of FosB/DFosB expression in the NAc(shell) and CeA compared with sham-dependent rats ( + p,0.05; ++ p,0.01 versus sham+morphine). In addition, ADX decreased basal FosB/DFosB expression in the CeA ( ** p,0.01 versus sham-placebo group. ## p,0.01; ### p,0.001 versus ADX+placebo). doi:10.1371/journal.pone.0050264.g002 in the NAc. It has been proposed that DFosB in the NAc, partly through the repression of DYN expression, increases the sensitivity to the rewarding effects of morphine and cocaine and leads to resilience to stress [9,33]. Our data support a role for GC in regulating the positive reinforcing effects of morphine mediated by the mesocorticolimbic DA system, given that the number of FosB/ DFosB/pro-DYN-positive neurons decrease in the NAc from ADX rats. These effects could be mediated directly by GR, which are present throughout the mesolimbic reward pathway, or indirectly via CRH projections arising from the CeA and/or the BNST to the VTA and NAc, which are underactivated during morphine dependence in ADX rats.
In summary, this study provides evidence that GC are critically involved in FosB/DFosB accumulation in the brain stress systems after chronic morphine exposure, which might result in lasting changes of gene expression pattern in stress-related areas. The present findings also indicate that FosB/DFosB may contribute to the GC-dependent changes on brain stress system plasticity during opiate dependence. Further studies are necessary to determine the intracellular mechanisms by which chronic opiates induce FosB/ DFosB in selected stress-related regions, as well as the mechanism that is responsible for the suppression of morphine-induced FosB/ DFosB expression by GC.

Adrenalectomy
Rats were bilaterally adrenalectomized and a corticosterone pellet was implanted to ensure low but stable levels of GC [29]. Rats were bilaterally adrenalectomized (ADX) via a dorsal approach under 90 mg/kg ketamine chlorhydrate and 8 mg/kg xylazin (i.p.) anaesthesia, and implanted subcutaneously (s.c.) with slow-release corticosterone pellets at surgery. The composition of steroid pellets (25 mg corticosterone plus 75 mg cholesterol) was chosen to provide stable corticosterone concentration corresponding to circadian nadir up to 20 d after implantation [56]. ADX rats with corticosterone replacement (ADX plus corticosterone) do not mount a challenge-induced increase of plasma corticosterone [56]. After surgery, ADX plus corticosterone rats had free choice to drink isotonic saline (0.9% NaCl) to replace depleted sodium secondary to the loss of aldosterone because of adrenalectomy. Control rats were subjected to the same surgical procedure (sham) without adrenal extirpation. Sham and ADX plus corticosterone rats were allowed to recover from surgery for 5 d before the morphinedependence procedure. Successful bilateral adrenalectomy was confirmed by plasma concentration of corticosterone and ACTH and by post-mortem examination of the ADX animals.

Drug Treatment and Experimental Procedure
Five days after surgery, rats were implanted s.c. with two 75 mg morphine pellets under light ether anaesthesia. Control rats received placebo pellets containing lactose. This procedure has been shown to produce consistent plasma morphine concentrations beginning a few hours after the implantation of the pellets and a full withdrawal syndrome after acute injection of opioid antagonists [57]. Dependence on morphine is achieved 24 h after implantation of pellets and remained constant for 15 days [58]. Ten days after the implantation of morphine or (iv) ADX-morphine. The weight gain of the rats was checked during treatment to ensure that the morphine was liberated correctly from the pellets, because it is known that long-term morphine treatment induces a decrease in body weight gain caused by lower caloric intake [29].

Brain Perfusion and Sectioning
Rats were deeply anaesthetised with an overdose of pentobarbital (100 mg/kg i.p.) and perfused transcardially with saline following by fixative containing paraformaldehyde (4% paraformaldehyde in 0.1 M borate buffer, pH 9.5). After removal of the perfused brains, they were post fixed in the same fixative for 3 h and stored at 4uC in PBS containing 10% sucrose until coronal sections (30-mm thickness) were cut rostrocaudally on a freezing microtome (Leica, Nussloch, Germany). The atlas of Paxinos and Watson (2007) [59] was used to identify different brain regions of rats: NAc(shell), BNST, PVN, CeA and NTS-A 2 ( Figure 1). The sections were cryoprotected and stored at 220uC until use.

Double-labelling Immunohistochemistry of FosB/DFosBimmunoreactive Nuclei and CRH-, TH-and DYN-positive Neurons
For FosB/DFosB-TH double-labelling, tissue sections from each rat in each treatment group were processed for FosB/DFosB immunoreactivity using DAB nickel intensification and then TH was revealed using DAB chromogen only. FosB/DFosB immunostaining was performed as described previously. Following the FosB/DFosB staining, sections were rinsed in PBS, treated with 2% normal goat serum and then incubated overnight with the rabbit polyclonal anti-TH antibody (AB152, Chemicon, USA; 1:6000) at room temperature. The same immunohistochemistry procedures described above were followed. The TH antibodyperoxidase complex was developed in DAB. The sections were mounted onto chrome-alum gelatine coated slides and coverslipped.
For the FosB/DFosB-CRH and FosB/DFosB-pro-DYN double labelling, the process was the same as described before for FosB/DFosB-TH. For detecting DYN-expressing neurons in the NAc, an antigen retrieval procedure was applied by incubating brain sections in citrate buffer (10 mM Citric Acid, 0.05% Tween 20, pH 6.0) at 60uC for 20 min before the blocking procedure. The primary anti-CRH rabbit antiserum was kindly provided by Wylie W. Vale (The Salk Institute, La Jolla, CA, USA), and was used at 1:500 dilution for 72 h at 4uC. The pro-DYN primary antibody was purchased from Neuromics (# GP10110, Neuromics, Edina, MN, USA) and diluted 1:2000 (72 h, 4uC).

Image Analysis
Images were captured by means of Leica microscope (DM 4000B; Leica) connected to a video camera (DFC290, Leica). FosB/DFosB-positive cell nuclei were counted using a computerassisted image analysis system (QWIN, Leica). The boundaries of the NTS-A 2 , BNST, CeA, NAc(shell) and the parvocellular subdivision of the PVN were outlined and the number of positive profiles was recorded. The number of FosB/DFosB nuclear profiles within the confines of cell groups of interest was counted bilaterally in four to five sections from each rat and averaged to obtain a single value for each rat. To avoid observer bias, all sections were quantified by a blinded investigator. Total counts for different brain regions are expressed as mean 6 SEM. Positive nuclei for FosB/DFosB immunoreactivity were detected using the same conventional light microscopy described above, and counted at x40 magnification. FosB/DFosB-positive CRH, TH or DYN cells were identified as cells with brown cytosolic deposits for CRH-positive, TH-positive and DYN-positive staining and blue/dark nuclear staining for FosB/DFosB. A square field (195 mm) was superimposed upon captured image to use as reference area. The number of double-labelled neurons observed bilaterally was counted in four to five sections from each animal in the PVN, BNST and CeA for CRH neurons, NTS for TH neurons and NAc for DYN neurons. The CRH, TH and DYN positive cells without a visible nucleus (FosB/DFosB-negative CRH, TH or DYN cells) were also included in the analysis.

Radioimmunoassay
Blood was collected into ice-cooled tubes containing 5% EDTA and was then centrifuged (500 g; 4uC; 15 min). Plasma was separated, divided into two aliquots and stored at 280uC until assayed for corticosterone or ACTH. Plasma concentration of corticosterone and ACTH were quantified using specific corticosterone and ACTH antibodies for rats ([ 125 I]-CORT and [ 125 I]-ACTH RIA; MP Biomedicals, Orangeburg, NY, USA). The sensitivity of the assay was 7.7 ng/mL for corticosterone and 5.7 pg/mL for ACTH.

Statistical Analysis
Data are presented as mean 6 SEM and were analyzed using the GraphPad Prism statistical package (San Diego, CA, USA). Data were analysed by two-way analysis of variance (ANOVA) with treatment (placebo, morphine) and surgery (sham, ADX) as independent variables. The Newman-Keuls post hoc test was used for individual group comparisons. Differences with a p,0.05 were considered significant. Figure S1 Effects of adrenalectomy (ADX) on plasma ACTH (A) and corticosterone (B) concentrations in controls and in morphine-treated animals. Surgical ADX increased ACTH levels both in placebo-and in morphine-treated rats, whereas decreased plasma corticosterone concentration in morphine-treated rats. Data represent the mean 6 SEM of plasma ACTH and corticosterone levels in rats pretreated with placebo or morphine for 10 days. ***p,0.001 versus sham-placebo; ++ p,0.01, +++ p,0.001 versus sham-morphine. (TIF)