B R A I N R E S E A R C H 1 1 2 2 ( 2 0 0 6 ) 5 6 – 6 4

ava i l ab l e a t www.sc i enced i r ec t . com

www.e l sev i e r. com/ loca te /b ra in res

Research Report

Curcumin reverses the effects of chronic stress on behavior,the HPA axis, BDNF expression and phosphorylation of CREB

Ying Xu, Baoshan Ku⁎, Lu Tie, Haiyan Yao, Wengao Jiang, Xing Ma, Xuejun LiDepartment of Pharmacology, School of Basic Medical Science, Peking University, 38 Xueyuan Road, Beijing, 100083, PR China

A R T I C L E I N F O

⁎ Corresponding author. Fax: +86 10 82801833E-mail address: kubaoshan2002@yahoo.co

0006-8993/$ – see front matter © 2006 Elsevidoi:10.1016/j.brainres.2006.09.009

A B S T R A C T

Article history:Accepted 5 September 2006Available online 3 October 2006

Curcuma longa is a major constituent of the traditional Chinese medicine Xiaoyao-san,which has been used to effectively manage stress and depression-related disorders inChina. Curcumin is the active component of curcuma longa, and its antidepressant effectswere described in our prior studies inmousemodels of behavioral despair.We hypothesizedthat curcumin may also alleviate stress-induced depressive-like behaviors andhypothalamic–pituitary–adrenal (HPA) axis dysfunction. Thus in present study weassessed whether curcumin treatment (2.5, 5 and 10 mg/kg, p.o.) affects behavior in achronic unpredictable stress model of depression in rats and examined what its moleculartargets may be. We found that subjecting animals to the chronic stress protocol for 20daysresulted in performance deficits in the shuttle-box task and several physiological effects,such as an abnormal adrenal gland weight to body weight (AG/B) ratio and increasedthickness of the adrenal cortex as well as elevated serum corticosterone levels and reducedglucocorticoid receptor (GR) mRNA expression. These changes were reversed by chroniccurcumin administration (5 or 10 mg/kg, p.o.). In addition, we also found that the chronicstress procedure induced a down-regulation of brain-derived neurotrophic factor (BDNF)protein levels and reduced the ratio of phosphorylated cAMP response element-bindingprotein (pCREB) to CREB levels (pCREB/CREB) in the hippocampus and frontal cortex ofstressed rats. Furthermore, these stress-induced decreases in BDNF and pCREB/CREB werealso blocked by chronic curcumin administration (5 or 10 mg/kg, p.o.). These results providecompelling evidence that the behavioral effects of curcumin in chronically stressed animals,and by extension humans, may be related to their modulating effects on the HPA axis andneurotrophin factor expressions.

© 2006 Elsevier B.V. All rights reserved.

Keywords:CurcuminDepressionChronic unpredictable stressHPA axisShuttle boxGRBDNFpCREB/CREBRat

1. Introduction

At present, there are three main kinds of classical antidepres-sants in clinical practice, including tricyclic antidepressants,selectiveserotonin reuptake inhibitors (SSRIs) andmonoamineoxidase inhibitors (MAOIs). Most of these drugs, however, haveundesirable side effects and their mechanisms of action have

.m.cn (B. Ku).

er B.V. All rights reserved

not been satisfactorily resolved. Therefore, seeking safe andeffective antidepressant agents from traditional herbs mayenable scientists to uncover novel treatments for depressivedisorders andmay further reveal as yet unknownmechanismsby which depressive symptoms can be alleviated.

Curcuma longa is commonly found in traditional Chineseherbal medicines, such as Jieyu-wan and Xiaoyao-san, which

.

Fig. 1 – The effects of curcumin on shuttle-box behavior instressed rats. Rats were administered vehicle, curcumin (2.5,5 and 10 mg/kg, p.o.) or imipramine (10 mg/kg, i.p.) and themeannumber of escape failures over 30 trialswas quantified.Mean values±SEM are shown (n=8–9 rats per group).##p<0.01 vs. Non-stressed control group. *p<0.05 vs.vehicle-treated, stressed group.

57B R A I N R E S E A R C H 1 1 2 2 ( 2 0 0 6 ) 5 6 – 6 4

are used to treat the symptoms of mental stress, hypochon-driac pain and mania. As the major active component ofcurcuma longa, curcumin has been reported to have antiox-idant, anti-inflammatory, immuno-modulatory and neuro-protective activities (Motterlini et al., 2000; Thiyagarajan andSharma, 2004). Previous studies in our laboratory showed thatacute curcumin administration significantly decreased im-mobility time in the tail suspension and forced swim tests inmice (Xu et al., 2005b). We also demonstrated that chroniccurcumin administration has an antidepressant effect in theolfactory bulbectomy model of depression in rats, whichsuggests that the central monoaminergic neurotransmittersystems may be involved in the anti-depressive effects ofcurcumin (Xu et al., 2005c). However, studies investigating themolecular targets of curcumin relevant to its antidepressanteffects have not been done.

A large number of clinical observations have suggested thatstress can act as a precipitating factor in the onset of affectiveillnesses, especially major depression (Bidzinska, 1984). Thepathophysiology of depression and the neurobiology of stressare linked by their shared association with the hypothalamic–pituitary–adrenal (HPA) axis and in particular by their sharedassociation with serotonin and norepinephrine containingneuronal systems (Breslow et al., 1989). In animals, severalstudies have demonstrated that the behavioral deficits and theabnormalities in the neuroendocrine system that are inducedby exposure to uncontrollable and unpredictable chronicstress, such as hypoactivity in the open field and aberrationsin the HPA system, can be reversed by antidepressanttreatments (Kennett et al., 1986; Murua et al., 1991; Reul et al.,1993; Mizoguchi et al., 2002; Soblosky, 1986). These studies alsorevealed that rats exposed to a chronic unpredictable stressparadigm exhibited attenuated HPA axis feedback and abnor-mal gene expression. Indeed stress-associated behavioralchanges were observed together with elevated corticosteronelevels and decreased glucocorticoid receptor (GR) levels in thehippocampus and some other brain regions (Holsboer, 1999).

To our knowledge, studies examining the relationshipbetween the neuroendocrine effects of chronic stress andthose of curcumin treatment are lacking. Therefore, in thepresent study, we used an unpredictable chronic stressparadigm to determine whether chronic curcumin treatmentcan alleviate stress-induced behavioral abnormalities andcorresponding gene changes in the HPA axis. In addition, tofurther investigate the possible molecular mechanisms thatmay be mediating the therapeutic effects of curcumin, weassessed the expression levels of brain-derived neurotrophicfactor (BDNF) protein and the portion of its upstream targetcAMP response element-binding protein (CREB) that is activat-ed (phosphorylated CREB (pCREB)) in the hippocampus andfrontal cortex.

2. Results

2.1. The effects of curcumin on the number of escapefailures in the shuttle box task

As shown in Fig. 1, chronically stressed rats exhibited escaperesponse deficits. Over the testing period, which included 30

footshock trials, the control rats on average failed to escapeonly twice. Meanwhile, the stressed rats subjected to the sameparadigm failed to escape an average of ten times (F(5,37)=3.13, p<0.01 vs. non-stressed controls). Chronic curcumin(10 mg/kg, p.o.) administration significantly reduced thenumber of escape failures in stressed rats (F(5,37)=3.13,p<0.05 vs. vehicle-treated, stressed rats). A similar reductionof failures was found in stressed rats treated with the tricyclicantidepressant imipramine (10 mg/kg, i.p.; F(5,37)=3.13,p<0.05 vs. vehicle-treated, stressed rats). No differences wereobserved when administered with 10 mg/kg curcumin orimipramine as compared with non-stressed controls.

2.2. The effects of curcumin on body weight, the ratio ofadrenal gland weight to body weight and thickness of adrenalcortex

The effects of chronic stress and the administration ofcurcumin on body weight, the ratio of adrenal gland weightto body weight (AG/B) and the thickness of adrenal cortex aresummarized in Table 1. No difference in the initial bodyweight was observed in any experimental group. Chronicstress significantly decreased body weight (F(5,30)=12.69,p<0.05), and significantly increased both the AG/B (F(5,30)=6.69, p<0.05) and the thickness of adrenal cortex (F(5,30)=23.11, p<0.01) relative to the non-stressed control rats. Thechronic stress-induced decrease in body weight was notsignificantly affected by the 2.5, 5 or 10 mg/kg curcumintreatment; however, the increases in the AG/B and thethickness of adrenal cortex were ameliorated by the drugtreatment (F(5,30)=6.69, p<0.01; F(5,30)=23.11, p<0.001). Thepositive control treatment with imipramine (10 mg/kg, i.p.)had similar effects on the AG/B and the thickness of theadrenal cortex (F(5,30)=6.69, p<0.001; F(5,30)=23.11, p<0.001,).

2.3. The effects of curcumin on serum corticosterone levels

The chronic stress paradigm caused a significant elevation ofbasal serum corticosterone levels relative to the non-stressed

Table 1 – The effects of curcumin on bodyweight (g), the ratio of adrenal glandweight/bodyweight (AG/B) and the thicknessof adrenal cortex (μm) in stressed rats

Group Dose(mg/kg)

Body weight(pre-stress)

Body weight(after stress)

The ratioof AG/B

Thickness ofadrenal cortex

Control 190.5±2.1 360.5±3.9 0.118±0.006 721.2±7.1Stress 190.6±1.8 314.1±6.4# 0.152±0.004# 981.1±5.8##

Curcumin 2.5 190.5±1.2 315.4±6.6 0.136±0.123 824.0±5.6⁎⁎⁎5 190.2±2.1 325.2±5.4 0.128±0.008 751.0±12.9⁎⁎⁎

10 193.5±1.3 334.8±2.9 0.100±0.006⁎⁎ 723.9±14.7⁎⁎⁎Imipramine 10 191.8±1.9 315.8±3.2 0.108±0.005⁎⁎ 725.4±13.8⁎⁎⁎

Values are the mean±SEM with 6 rats in each group. Data analysis was performed using S-N-K test. #p<0.05, ##p<0.01, compared with controlgroup. ⁎⁎p<0.01, ⁎⁎⁎p<0.001, compared with stressed group.

58 B R A I N R E S E A R C H 1 1 2 2 ( 2 0 0 6 ) 5 6 – 6 4

control group (F(5,37)=73.43, p<0.001, Fig. 2). The stress-induced increases in serum corticosterone levels were signif-icantly reduced in rats that were treated with curcumin (5 and10 mg/kg, p.o.) (F(5,37)=73.43, p<0.01; p<0.001) or imipramine(10 mg/kg, i.p.) (F(5,37)=73.43, p<0.001). Although the reduc-tions were robust they were not sufficient to bring hormonelevels down all the way to non-stressed baseline levels.

2.4. The effects of curcumin on GR mRNA levels inhippocampus

As shown in Fig. 3, GR mRNA levels in the hippocampuswere reduced following exposure of rats to chronic stress for20days (F(5,28)=6.72, p<0.05 vs. non-stressed controls).Administering curcumin (5 and 10 mg/kg, p.o.) or imipra-mine (10 mg/kg, i.p.) prior to the stress protocol preventedthese changes (F(5,28)=6.72, p<0.001 vs. vehicle-treated,stressed rats).

2.5. The effects of curcumin on BDNF expression

Western blot analysis demonstrated that BDNF levels weredecreased in the hippocampus and frontal cortex of stressedrats (F(5,28)=10.47, p<0.05 vs. non-stressed controls, Figs. 4and 5). The reduction was ameliorated by the 5 and 10 mg/kgcurcumin treatments (F(5,28)=10.47, p's<0.05 vs. vehicle-

Fig. 2 – The effects of curcumin on serum corticosteronelevels in stressed rats. Each column represents themean±SEM of 8–9 rats. ###p<0.001 vs. non-stressed controlgroup. **p<0.01, ***p<0.001, vs. the stressed group.

treated, stressed rats). The effect of imipramine treatment(10 mg/kg) on BDNF levels was similar to that of curcumin(F(5,28)=10.47, p<0.05 vs. vehicle-treated, stressed rats).

Fig. 3 – The effects of curcumin on hippocampal GR mRNAexpression in stressed rats. (A) Electrophoretic analysis of GRmRNA and β-actin mRNA expression in the hippocampus ofstressed rats. Above: GR mRNA; below: β-actin mRNA. Lane1: non-stressed control group; Lane 2: stress+vehicle group;Lane 3: 2.5 mg/kg curcumin; Lane 4: 5 mg/kg curcumin; Lane5: 10 mg/kg curcumin; Lane 6: 10 mg/kg imipramine. (B) PCRproducts were quantified by densitometric scanning and GRexpression was normalized relative to the steady-stateexpression ofβ-actin used as internal control (intensity ratio:GR toβ-actin). Each column represents the groupmean±SEM(n=5–6 rats per group). #p<0.05, vs. non-stressed controlgroup. *p<0.05, vs. vehicle-treated, stressed group.

Fig. 4 – Western blot analysis of BDNF expression in thehippocampus. The intensity of the protein bands wasquantified with a densitometric scanner. Each columnrepresents the group mean±SEM of 5–6 rats. Band 1:non-stressed control group; Band 2: stress+vehicle group;Band 3: 2.5mg/kg curcumin; Band 4: 5mg/kg curcumin; Band5: 10mg/kg curcumin; Band 6: 10mg/kg imipramine. #p<0.05vs. non-stressed control group. *p<0.05, vs. vehicle-treatedstressed group.

Fig. 5 – Western blot analysis of BDNF expression in thefrontal cortex. The intensity of the protein bands wasquantified with a densitometric scanner. Each columnrepresents the group mean±SEM of 5–6 rats. Band 1:non-stressed control group; Band 2: stress+vehicle group;Band 3: 2.5mg/kg curcumin; Band 4: 5mg/kg curcumin; Band5: 10mg/kg curcumin; Band 6: 10mg/kg imipramine. #p<0.05vs. non-stressed control group. *p<0.05 vs. vehicle-treatedstressed group.

Fig. 6 – Western blot analysis of pCREB/CREB in thehippocampus. The intensity of Western bands wasquantified with a densitometric scanner. Each columnrepresented the group means±SEM of 5–6 rats. Band 1:non-stressed control group; Band 2: stress+vehicle group;Band 3: 2.5mg/kg curcumin; Band 4: 5mg/kg curcumin; Band5: 10 mg/kg curcumin; Band 6: 10 mg/kg imipramine.##p<0.01 vs. control group. *p<0.05 and **p<0.01 vs.vehicle-treated, stressed group.

59B R A I N R E S E A R C H 1 1 2 2 ( 2 0 0 6 ) 5 6 – 6 4

2.6. The effects of curcumin on pCREB/CREB ratio

As shown in Figs. 6 and 7, CREB expression in the hippocam-pus and frontal cortex of non-stressed control rats andstressed rats did not differ (F(5,28)=19.11, p>0.05; F(5,28)=14.73, p>0.05). Neither curcumin nor imipramine inducedchanges in CREB expression (p's>0.05 vs. vehicle-treated,stressed rats). The amount of pCREB in these two brainregions however did differ between the stressed and non-stressed control groups (F(5,28)=19.11, p<0.05; F(5,28)=14.73,p<0.05). Chronic administration of curcumin increasedpCREB/CREB ratios in the hippocampus (5 or 10 mg/kg, p.o.,F(5,28)=19.11, p's<0.05 vs. vehicle-treated, stressed rats) andin the frontal cortex (10 mg/kg, F(5,28)=14.73, p<0.01). Similarresults were obtained with the 10 mg/kg imipramine admin-istration (p's<0.01).

3. Discussion

There is a growing body of evidence showing that the chronicadministration of various uncontrollable stresses, a procedureknown as “chronic unpredictable stress”, is an appropriatemodel for the experimental investigation of depression (Katzand Schmaltz, 1980; Willner, 1991; Willner et al., 1992). Bothvariability and unpredictability during the stress regime arecritical triggers in the induction of depressive-like behaviors,such as escape deficits in the shuttle box task (Murua et al.,

Fig. 7 – Western blot analysis of pCREB/CREB in the frontalcortex. The intensity ofWestern bandswas quantifiedwith adensitometric scanner. Each column represented the groupmean±SEM of 5–6 rats. Band 1: non-stressed control group;Band 2: stress+vehicle group; Band 3: 2.5 mg/kg curcumin;Band 4: 5 mg/kg curcumin; Band 5: 10 mg/kg curcumin; Band6: 10 mg/kg imipramine. ##p<0.01, vs. non-stressed controlgroup. **p<0.01 vs. vehicle-treated, stressed group.

60 B R A I N R E S E A R C H 1 1 2 2 ( 2 0 0 6 ) 5 6 – 6 4

1991; Soblosky, 1986). The theoretical premise behind thismethod is that depression is the outcome of an eventualinability to cope with a stream of dissimilar unpleasantstimuli imposed by the environment (Ferretti et al., 1995).Indeed, most animal models of depression are based on thebehavioral deficits induced by stressors and the reversal ofsuch effects following antidepressant treatments (Katz andHersh, 1981; Maier, 1984; Kennett et al., 1986). In the modelused in the present study, rats exposed to a regime of chronicstress exhibited altered escape performance to an aversivestimulus, namely increased failures to escape from an electricfootshock. The present results showed that animals subjectedto a chronic unpredictable stress regime for 20 consecutivedays failed to acquire a normal escape response when testedfor shuttle-box responding. However, escape deficits could bereversed if animals were chronically administered curcumin(10 mg/kg, p.o.) or imipramine (10 mg/kg, i.p.) during thestressor period. No differences were observed when adminis-tered with curcumin (10 mg/kg) or imipramine (10 mg/kg) ascompared with non-stressed controls. These results showedthat curcumin treatment completely reversed chronic stressinduced the escape deficits in shuttle box.

Stress and depression have been linked in a variety ofways. Some stress-provoked disturbances seem to be associ-ated with the pathophysiology of depression (Kioukia-Fougiaet al., 2002). Many of the neurobiological abnormalities foundin chronically stressed rats parallel those found in humandepressed patients. For instance, a long-lasting impairment of

HPA axis feedback inhibition has been reported both inchronically stressed rats and clinically depressed patients(Morley-Fletcher et al., 2004; Tafet and Bernardini, 2003). Thisdisruption of the HPA axis negative feedback system ischaracterized by a failure to suppress the secretion of serumcorticosterone and adrenal hypertrophy (Mizoguchi et al.,2002).

In our study, we found rats that were chronically exposedto chronic stress showed a depression-like dysfunction of HPAaxis feedback, associated with an increase in the AG/B and thethickness of adrenal cortex, as well as spontaneous cortico-sterone secretion. Previous researchers have shown that theadrenal hypertrophy and elevated corticosterone secretioninduced by chronic stress can be reversed by administrationwith the antidepressants amitriptyline and imipramine (Reulet al., 1993; McEwen, 2005). The present findings complementthese prior studies and show that chronic curcumin admin-istration can also alleviate stress-associated physiologicalchanges in the AG/B, in the thickness of adrenal cortex andin serum corticosterone levels. The effects of curcumin weresimilar to those of imipramine, which was used as a positivecontrol. The effects of chronic stress on body weight were notameliorated by curcumin and imipramine administration.Thus it may be that the chronic stress-induced decrease in thebody weight may occur through a mechanism that isindependent from the variables that could be reversed bythe drug treatment (Mizoguchi et al., 2002).

Elevated corticosterone (in rodents) or cortisol (in primates)levels is a hallmark of HPA axis feedback inhibition (Centenoand Volosin, 1997). This feedback is mediated by two types ofcorticosteroid receptors in the brain, the mineralocorticoidreceptor (MR) and the GR (McEwen, 2000). When normalsecretion of glucocorticoids is altered, leading to increasedlevels of corticosterone (or cortisol), this may result in a down-regulation of hippocampal glucocorticoid receptors (GRs)(Tafet and Bernardini, 2003). The hippocampus is known tocontain an exceptionally high level of corticosteroid receptors,which are involved in negative feedback inhibition of the HPAaxis in learning and memory processes (Budziszewska, 2002).Studies examining the involvement of corticosteroid receptorsin the control of HPA axis activity have demonstrated thatcorticosterone via activation of GRs inhibits stress-inducedbiosynthesis and release of proopiomelanocortin (POMC) andcorticotrophin releasing hormone (CRH), which in turndecreases corticosterone secretion (Pariante et al., 2004).Long-term administration of some antidepressant drugs,such as MAOIs and SSRIs, increases GR mRNA and GR proteinlevels, especially in the hippocampus (Pariante and Miller,2001). Consistent with these previous findings, we found thatchronic stress-induced down-regulation of GR mRNA expres-sion in the rat hippocampus could be blocked by chroniccurcumin treatment. Curcuminat 5 and10mg/kgper day for 21days resulted in a 36.7% and a 43.8% reduction in corticoste-rone levels respectively relative to stressed rats not given thedrug. Likewise, stress-induced reduction of GRmRNA levels inthe hippocampuswas significantly ameliorated in rats admin-istered 5 or 10mg/kg curcumin. These effects are in agreementwith previouswork examining the effects of chronic treatmentwith tricyclic antidepressants (imipramine or desipramine) ormonoamine oxidase inhibitors (moclobemide) in stressed rats

61B R A I N R E S E A R C H 1 1 2 2 ( 2 0 0 6 ) 5 6 – 6 4

(Centeno and Volosin, 1997; Pepin et al., 1989; Holsboer, 1999).Previous studies in our laboratory suggested that chroniccurcumin administration (14days) increased monoamine (5-HT, noradrenaline and dopamine) levels in olfactory bulbecto-mized rats (Xu et al., 2005c). Although further researchinvestigating the putative effects of curcumin on GR mRNAup-regulation and increases inmonoamine levels is indicated,our findings provide strong evidence that the antidepressanteffects of curcumin may be the result of normalization of theHPA axis.

Chronic stress-induced elevation of glucocorticoids anddecreased GR mRNA levels are accompanied by structuralchanges and neuronal damage in certain brain areas, such asthe hippocampus and frontal cortex (Vollmayr et al., 2001). Inanimal models of chronic stress, neurons in the hippocampusand the frontal cortex respond to repeated stress by showingatrophy and a down-regulation BDNF expression that isassociated with memory impairment (McEwen, 2005;Vollmayr et al., 2001). The BDNF gene contains a cAMPresponse element (CRE), to which phosphorylated cAMPresponse element-binding protein (CREB) binds and therebyenhances transcription. Clinical observations suggest thatglucocorticoids may interfere indirectly this process when thecortisol-GR complex binds CREB, preventing its phosphoryla-tion, and therefore blocking the expression of CRE-regulatedgenes such as BDNF. Chronic administration of variousantidepressants increases the expression, phosphorylationand function of CREB and its downstream target gene BDNF inrat hippocampus and other limbic brain regions thought to beinvolved in depression (Duman, 1996; Thome et al., 2000). Inaccordance with this view, we found in the present study thatchronic (21days) curcumin treatment before stress ameliorat-ed the reduction of BDNF and pCREB/CREB levels that wasproduced by the stress protocol. These results are in confor-mity with previous studies which have shown that someantidepressants, such as fluoxetine, can increase BDNF levelsand the pCREB/CREB ratio in rat hippocampus and frontalcortex (Tiraboschi et al., 2004; Odagaki et al., 2001). Moreover,these findings raise the possibility that curcumin treatment,via up-regulation of pCREB and BDNF, may reverse or protecthippocampal and frontal cortical neurons from furtherdamage in response to chronic stress or other environmentalinsults. These findings shed light on themechanism by whichcurcumin acts at the psycho-neuroendocrine levels, particu-larly as a modulator of the HPA axis.

Rats subjected to an uncontrollable aversive stress subse-quently showed a severe blunting of the behavioral escaperesponse. The behavioral deficit in the footshock avoidance(shuttle-box) taskwas associatedwith dysfunction of HPA axisfeedback, characterized by an exaggerated corticosteroneresponse (Murua et al., 1991; Centeno and Volosin, 1997;Eede and Claes, 2004). It was subsequently shown that thesechanges may be related to alterations in brain function andmay be reversed by chronic concominant treatment withsome traditional antidepressants, such as tricyclic antidepres-sants and monoamine oxidase inhibitors, through activationof CREB and neurotrophins such as BDNF (Soblosky, 1986;Henn et al., 2004; Butterweck et al., 2001). Our results showedthat curcumin reversed stress-induced changes in behavior,physiology and expression of genes involved in HPA axis

control. Chronic curcumin treatment also altered transcrip-tion factor and neurotrophic factor levels in the brain in amanner that was similar to the changes elicited by the tricyclicantidepressants imipramine; this finding has not beenpreviously reported in an in vivo model. The effects ofcurcumin treatment upon gene and protein expression weremanifest at lower doses than the effects on escape behavior.This distinctionmay be due to behavioral complications, suchas individual differences in rats and limitations of thesensitivity of the task to reveal effects on HPA axis function.

Curcumin is highly lipophilic and should readily cross theblood–brain barrier. Studies on the toxicity of curcumin haveincluded in vitro animal and human studies. Oral administra-tion of curcumin as high as 5 g/kg bodyweight in rats producesno apparent toxicity (Wahlstrom and Blennow, 1978; Deodharet al., 1980). Human trials using up to 8000mg of curcumin perday for 3 months have also found it to be safe (Chainani-Wu,2003). Curcumin has a long history of use as a traditionalmedicine which is consistent with prospect that it not onlysafe but alsomay have therapeutic and protective applicationsin the treatment of depression. However, the efficacy andsafety of curcumin at relevant doses need to be established byfurther study.

In conclusion, the main finding from this study is that theeffects of curcumin on the behavioral deficits induced bychronic stress may be related to their modulating effects onHPA axis dysfunction. Moreover, the data presented hereinsuggest that selective increases in BDNF and pCREB proteins inspecific brain regions, namely the hippocampus and frontalcortex, may be involved in themechanism by which curcuminis effective as an antidepressive therapeutic agent. Furtherstudies are currently being conducted in our laboratory toconfirm whether and to examine how the antidepressanteffects of curcumin are related to central monoaminergicneurotransmitters, HPA axis function and neurotrophic factorexpression.

4. Experimental procedures

4.1. Animals

Male Sprague–Dawley (SD) rats, weighing between 190 and200 g at the start of the experiment, were obtained from theAnimal Center of Peking University Health Science Center. Therats were housed six per cage under standard colony condi-tions, with a 12-h light/12-h dark cycle and provided food andwater ad libitum. They were allowed to acclimatize to thecolony for at least 5days prior to any experimentation. Theexperimental procedures were in compliance with theNational Institutes of Health Guide for Care and Use ofLaboratory Animals and with the European CommunitiesCouncil Directive of 24 November 1986 (86/609/EEC).

4.2. Drugs and drug administration

Curcumin and the clinically effective tricyclic antidepressantimipramine hydrochloride were purchased from Sigma Che-mical Co., (USA). The anti-CREB, anti-phospho-CREB, anti-BDNF antibodies and the respective secondary antibody were

62 B R A I N R E S E A R C H 1 1 2 2 ( 2 0 0 6 ) 5 6 – 6 4

purchased from Zhongshan Technology Co., Ltd. (Beijing,China). For oral (p.o.) administration, fresh curcumin wasdissolved in peanut oil and diluted to the desired concentra-tion on the day of experiment. For intra-peritoneal (i.p.)injection, imipramine was dissolved in double-distilledwater. Curcumin (2.5, 5 and 10 mg/kg, p.o.) or imipramine(10 mg/kg, i.p.) was administered daily for 21 days. Behav-ioral testing commenced 60min after the last drug treat-ment. In a preliminary experiment, peanut oil (p.o., 1 ml/kg)and redistilled water (i.p., 1 ml/kg) were used as controltreatments and the behavioral data did not differ betweenthe rats that received the two vehicle solutions (Table 2).Therefore, we chose to present only the peanut oil controlgroup data for comparison.

4.3. Chronic stress procedure

The stressed rats were subjected to the following conditionsused by Molina et al. (1990) and Murua et al. (1991) with minormodifications. Stress was administered once per day over aperiod of 20 days between 8:00 am to 12: 00 am. The order ofstressors used was as follows: shaker stress (high speed,45 min), cold swim (10 °C, 5 min), restraint (1.5 h), tail pinch(1 min), water deprivation (24 h), foot shock for 30 min(1 mA, 1 s duration, average 1 shock/min), cold swim (10 °C,5 min), food deprivation (24 h), restraint (2 h), shaker stress(high speed, 1 h), tail pinch (1 min), water deprivation(24 h), 24-h social isolation (the rats were placed individuallyin 30×15×10 cm acrylic cages in another housing room andwere returned to their home cage after 24 h), foot shock for45 min (1 mA, 1 s duration, average 1 shock/min for 45 min),cold swim (8 °C, 5 min), shaker stress (high speed, 1.5h),restraint (2.5h), tail pinch (2min), food deprivation (24 h), 24-h social isolation. On day 21, behavior of the stressed rats wasobserved and the rats were sacrificed to assess any neuroen-docrine changes (including body weight, the ratio of adrenalgland weight to body weight, thickness of adrenal cortex andthe serum corticosterone concentrations) that may haveoccurred and to measure gene expression levels in thehippocampus and frontal cortex.

4.4. Shuttle-box testing

On the 21st day following the final drug or vehicle treatment,rats were subjected to shuttle box testing. The escape–avoidance test was carried out in a two-way shuttle box(60×20×20 cm) with Plexiglas walls fitted with a floor consist-ing of stainless steel rods separated by 1.0 cm. The floor wasdivided into two equal size chambers by a wood partition

Table 2 – The different effects of peanut oil (p.o., 1 ml/kg) and rweight, the ratio of adrenal gland weight/body weight (AG/B) a

Group Number ofescape failure

Body weight(pre-stress)

Peanut oil 2.5±0.6 195.7±4.6Redistilled water 2.2±0.9 197.8±7.3

Values are the mean±SEM with 6 rats in each group.

(1.5 cm above the grid floor). Subjects were placed in theshuttle-box, allowed to habituate to the test environment for3 min, and then were submitted to 30 avoidance trials (30-sinter-trial interval). During the first 4 s of each trial, a whitenoise signal was presented. If the rat did not escape to theother compartment, a 4-s 0.8 mA shock was delivered via thegrid floor. The number of escape failures, defined as theabsence of a crossing response before or during shock delivery,was recorded. The shuttle-box training was always initiated24 h after the last stressful event (Murua et al., 1991).

4.5. Measurement of serum corticosterone

Following behavioral testing, rats were sacrificed by decap-itation and serum samples were collected to measurecorticosterone concentrations. Radioimmunoassay (RIA) ofcorticosterone was performed using [125I]-labeled corticoste-rone, antiserum and a standard solution in a kit obtainedfrom the China Institute of Atomic Energy (Beijing, China).The RIA was performed according to manufacturer's instruc-tion. The inter- and intra-assay coefficients of variance were6.5% and 4.5%, with a detection limit of 25 ng/ml.

4.6. Semiquantitative reverse transcription-polymerasechain reaction (RT-PCR)

RT-PCR was performed according to a previously describedprotocol (Xu et al., 2005a). Total RNA from rat hippocampuswas isolated with Trizol® reagent (Dingguo biotechnologyCompany, Beijing, China). The hippocampus was homoge-nized (PF 1200E, Switzerland) in 1 ml Trizol®, and then 200μlchloroform was added and gently but thoroughly mixed. Thehomogenate was centrifuged at 12,000 rpm for 10 min at 4 °C.The colorless supernatants were collected carefully andmixedwith an equal volume of isopropanol, leave them at 20 °C for30 min. The mixture was centrifuged at 12,000 rpm for 10 minat 4 °C. The supernatant was discarded and the pellet wasresuspended in 1 ml of 75% ethanol, vortexed well and thencentrifuged at 12,000 rpm for 5 min at 4 °C. The supernatantwas again discarded, and the pellet was dried and resus-pended with ribonuclease (RNase)-free water. The amount oftotal RNA was determined by a spectrophotometer (LambdaBio 4.0, Perkin-Elmer, USA) at 260 nm.

The reverse transcription was carried out with 2 μg of totalRNA by M-MLV reverse transcriptase (Promega, USA) using anOligo (dT) primer in a 25-μl final solution. The reaction solutionincluded 5 μl of a 5× buffer stock solution (250mMTris–HCl, pH8.3, 375 mM KCl, 15 mM MgCl2, 50 mM DTT), 0.8 mM of amixture of all four dNTPs and 1U/μl ribonucleasin. The

edistilled water (i.p., 1 ml/kg) on shuttle-box behavior, bodynd the thickness of adrenal cortex (μm) in stressed rats

Body weight(after stress)

The ratioof AG/B

Thickness ofadrenal cortex

360.3±5.1 0.117±0.007 724.2±7.1363.0±6.9 0.119±0.008 721.7±5.9

63B R A I N R E S E A R C H 1 1 2 2 ( 2 0 0 6 ) 5 6 – 6 4

reverse transcription was carried out at 37 °C for 1 h,heated for 5 min at 95 °C and then cooled to 4 °C tosynthesize the first cDNA strand. The synthesized templatecDNA (2 μl) was amplified by PCR in a reaction mixture(25μl) containing 200 IU Taq DNA polymerase (Sangon,Shanghai, China), 2.5 μl of a 10× buffer (20 mM MgSO4, 100mM KCl, 80 mM (NH4)2SO4, 100 mM Tris–HCl, pH 9.0, 0.5%NP400), 0.2 mM of each of dNTP and 0.25 mM of thefollowing primers: β-actin: forward (5′) primer: 5′-GTC ACCCAC ACT GTG CCC ATC T-3′; reverse (3′) primer: 5′-ACA GAGTAC TTG CGC TCA GGA G-3′ ( β-actin PCR production:542 bp); GR: forward (5′) primer: 5′-ATCCCACAGACCAAAGCACCT T-3′; reverse (3′) primer: 5′-TCC AGT TTT CAG AAC CAACAG G-3′ (GR PCR production: 540 bp): All primers weresynthesized by Sangon (Shanghai, China). Reaction solutionswere brought to their final volumes by the addition ofnuclease-free nanopure water.

In general, PCR was performed with a preheating cycle at94 °C for 5 min, denaturation, annealing and elongation werecarried out at 94 °C for 30 s, at 58 (GR) or 55 °C (β-actin) for30 s, and at 72 °C for 40 s, respectively. The reactions wererepeated for 31 (GR) or 32 cycles (β-actin). The amplifiedproducts were separated on 5% agarose gel and visualizedby ethidium bromide staining. The RT-PCR reaction pro-ducts were photographed and analyzed by Image MasterVDS Software (Hoefer Pharmacia Biotech Inc.). The negativecontrol contained all reagents, except that 4 μl of nanopurewater substituted for the RT reaction product.

4.7. Western blot analysis

Rats were decapitated and their brains were rapidly removedand stored at −70 °C. The frontal cortex andhippocampusweredissected out on a cold plate (−16 °C) (Franklin and Paxinos,1997; Paxinos and Watson, 1986). The tissue samples weresuspended in a solution containing 2% sodium dodecyl sulfate(SDS), 10% glycerol, 100 mM dithiothreitol, 0.01% (w/v) bromo-phenol blue and 60mMTris–HCl (pH 6.8) and sonicated for 10 s.The homogenates were centrifuged at 12,000×g for 15 min at4 °C. Protein concentrations of each sample were measuredby the Bradford method (Bradford, 1976). Aliquots werestored at −70 °C prior to use. Separating (10%) and stacking(5%) polyacrylamide gels containing 0.1% SDS were used.Tissue proteins were suspended in a sample buffer contain-ing 2% SDS, 50mM Tris–HCl buffer (pH 6.8), 10% glycerol (v/v)and 100 mM dithiothreitol and heated at 100 °C for 5 minin a water bath. A sample consisting of 50 μg of proteinwas loaded into each lane of the polyacrylamide gels, whichwere electrophoresed at 150V constant power for 1 h. Proteinwas transferred electrophoretically at 50 mA to a polyvinyli-dene difluoride (PVDF) immobilion-P membrane (0.45 μmporesize; Millipore Corp, Bedford, MA) in a transfer buffer (pH 8.3)composed of 25 mM Tris–HCl, 192 mM glycine and 20%methanol at 4 °C overnight. The membranes were incubatedin Tris-buffered saline (TBS, 100 mM Tris–HCl and 0.9%NaCl, pH 7.5) containing 5% nonfat milk for 1h at 20–22 °Cand subsequently incubated with anti-CREB, anti-phospho-CREB at Ser133 or anti-BDNF primary antibody at a 1:1000dilution and shaken on a rotator at 20–22 °C for 1.5 h. Afterthree washes in TBS containing 0.1% Tween-20 (TBS-T) for

5 min each, the membranes were incubated with 1:3000(CREB, pCREB) or 1:5000 (BDNF) of anti-HRP-conjugatedsecondary antibody with 5% nonfat milk for 1 h at 20–22 °C.Following the post-secondary washes (3×5min), the resultingantigen–antibody–peroxidase complexes were detected byenhanced chemiluminescent autoradiography (ECL kit;Amersham Pharmacia, Denver, CO, USA) and visualized byexposures of various lengths (approximately 25 s to 2 min) toKodak film. Densitometer readings were used to quantitatethe amount of protein in each treatment situation. Themean optical density is the ratio, based upon densitometricdata, of pCREB/CREB.

4.8. Statistical analysis of data

All data are presented as means±SEM. A one-way analysis ofvariance (ANOVA) followed by a Student–Newman–Keuls(S-N-K) test was used for statistical evaluation. The p valuesless than 0.05 were considered statistically significant.

Acknowledgments

This work was supported by the National Nature ScienceFoundationofChina (No.30270528), 973Programof theMinistryof Science and Technology (No.2004CB518902), research fundfrom Ministry of Education of China No.20020001082 and 985Program from Ministry of Education of China. The authorsthank Dr. J.H. Liang for the suggestion that improve the qualityof the manuscript.

R E F E R E N C E S

Bidzinska, E.J., 1984. Stress factors in affective diseases. Br. J.Psychiatry 144, 161–166.

Bradford, M.M., 1976. A rapid and sensitive method for thequantitation of microgram quantities of protein utilizing theprinciple of protein dye binding. Anal. Biochem. 72, 248–254.

Breslow, M.F., Fankhauser, M.P., Potter, R.L., Meredith, K.E.,Misiaszek, J., Hope, D.G., 1989. Role of gamma-aminobutyricacid in antipanic drug efficacy. Am. J. Psychiatry 146,353–356.

Budziszewska, B., 2002. Effect of antidepressant drugs on thehypothalamic pituitary adrenal axis activity and glucocorticoidreceptor function. Pol. J. Pharmacol. 54, 343–349.

Butterweck, V., Winterhoff, H., Herkenham, M., 2001. St Jonw'swort, hypericin, and imipramine: a comparative analysis ofmRNA levels in brain areas involved in HPA axis controlfollowing short-term and long-term administration in normaland stressed rats. Mol. Psychiatr. 6, 547–564.

Centeno, V.A., Volosin, M., 1997. Chronic treatment withdesipramine: effect on endocrine and behavioralresponses induced by inescapable stress. Psysiol. Behav.62, 939–944.

Chainani-Wu, N., 2003. Safety and anti-inflammatory activity ofcurcumin: a component of tumeric (Curcuma longa). J Altern.Complement. Med. 9, 161–168.

Deodhar, S.D., Sethi, R., Srinal, R.C., 1980. Preliminary study onantirheumatic activity of curcumin. Indian J. Med. Res. 71,632–634.

Duman, R.S., 1996. A role for CREB and BDNF in the actions ofantidepressants. Biol. Psychiatry 39, 558–559.

64 B R A I N R E S E A R C H 1 1 2 2 ( 2 0 0 6 ) 5 6 – 6 4

Eede, F.V.D., Claes, S., 2004. Mechanisms of depression: role of theHPA axis. Drug Discov. Today 1, 413–418.

Ferretti, C., Marinella, B., Gamalero, S.R., Ghi, P., 1995. Biochemicaland behaviour changes induced by acute stress in a chronicvariate stress model of depression: the effect of amitriptyline.Eur. J. Pharmacol. 280, 19–26.

Franklin, K.B.J., Paxinos, G., 1997. The Mouse Brain in StereotaxicCoordinates. Academic Press, San Diego.

Henn, F., Vollmayr, B., Sartorius, A., 2004. Mechanisms ofdepression: the role of neurogenesis. Drug Discov. Today 1,407–411.

Holsboer, F., 1999. The rationale for corticotrophin-releasinghormone receptor (CRH-R) antagonists to treat depression andanxiety. J. Psychiatr. Res. 33, 181–214.

Katz, R.J., Hersh, S., 1981. Amitriptyline and scopolamine in ananimal model of depression. Neurosci. Biobehav. Rev. 5,265–271.

Katz, R.J., Schmaltz, K., 1980. Dopaminergic involvement inattention, a novel animal model. Prog.Neuropsychopharmacol. 4, 585–590.

Kennett, G.A., Chaouloff, F., Marcou, M., Curzon, G., 1986. Femalerats are more vulnerable than males in an animal model ofdepression: the possible role of serotonin. Brain Res. 382,416–421.

Kioukia-Fougia, N., Antoniou, K., Bekris, S., Liapi, C., Christofidis, I.,Papadopoulou-Daifoti, Z., 2002. The effects of stress exposureon the hypothalamic–pituitary–adrenal axis, thymus, thyroidhormones and glucose levels. Prog. Neuro-Psychopharmacol.26, 823–830.

Maier, S.F., 1984. Learned helplessness and animal models ofdepression. Prog. Neuro-Psychopharmacol. Biol. Psychiatry 8,435–446.

McEwen, B.S., 2000. The neurobiology of stress: from serendipity toclinical relevance. Brain Res. 886, 172–189.

McEwen, B.S., 2005. Glucocorticoids, depression, and mooddisorders: structural remodeling in the brain. Metabolism 54,20–23.

Mizoguchi, K., Yuzurihara, M., Ishige, A., Sasaki, H., Tabira, T.,2002. Saiko-ka ryukotsu-borei-to, an herbal medicine, preventschronic stress-induced disruption of glucocorticoid negativefeedback in rats. Life Sci. 72, 67–77.

Molina, V.A., Volosin, M., Cancela, L., Keller, E., 1990. Effect ofchronic variable stress on monoamine receptors: influence ofimipramine administration. Pharmcol. Biochem. Behav. 35,335–340.

Morley-Fletcher, S., Darnaudery, M., Mocaer, E., Froger, N.,Lanfumey, L., Laviola, G., Casolini, P., Zuena, A.R., Marzano, L.,Hamon, M., Maccari, S., 2004. Chronic treatment withimipramine reverses immobility behaviour, hippocampalcorticosteroid receptors and cortical 5-HT1A receptor mRNA inprenatally stressed rats. Neuropharmacology 47, 841–847.

Motterlini, R., Foresti, R., Bassi, R., Green, C.J., 2000. Curcumin, anantioxidant and anti-inflammatory agent, induces hemeoxygenase-1 and protects endothelial cells against oxidativestress. Free Radic. Biol. Med. 28, 1303–1312.

Murua, V.S., Gomez, R.A., Andrea, M.E., Molina, V.A., 1991.Shuttle-box deficits induced by chronic variable stress: reversalby imipramine administration. Pharmacol. Biochem. Behav.38, 125–130.

Odagaki, Y., Garcia-Sevilla, J.A., Huguelet, P., Harpe, R.L., Koyama,T., Guimon, J., 2001. Cyclic AMP-mediated signalingcomponents are upregulated in the prefrontal cortex ofdepressed suicide victims. Brain Res. 898, 224–231.

Pariante, C.M., Miller, A.H., 2001. Glucocorticoid receptors in majordepression: relevance to pathophysiology and treatment. Biol.Psychiatry 49, 391–404.

Pariante, C.M., Thomas, S.A., Lovestone, S., Makoff, A., Kerwin,R.W., 2004. Do antidepressants regulate how cortisol affectsthe brain? Psychoneuroendocrinology 29, 423–447.

Paxinos, G., Watson, C., 1986. The Rat Brain in StereotaxicCoordinates. Academic Press, Orlando.

Pepin, M.C., Beaulieu, S., Barden, N., 1989. Antidepressantsregulate glucocorticoid receptor messenger RNAconcentrations in primary neuronal cultures. Mol. Brain Res. 6,77–83.

Reul, J.M.H.M., Stec, I., Soder, M., Holsbore, F., 1993. Chronictreatment of rats with the antidepressantamitriptyline attenuates the activity of thehypothalamic-pituitary-adrenocortical system.Endocrinology 133, 312–320.

Soblosky, J.S., 1986. Biochemical and behavioral correlates ofchronic stress: effects of tricyclic antidepressants. Pharm.Biochem. Behav. 24, 1361–1368.

Tafet, G.E., Bernardini, R., 2003. Psychoneuroendocrinologicallinks between chronic stress and depression. Prog.Neuro-Psychopharmacol. 27, 893–903.

Thiyagarajan, M., Sharma, S.S., 2004. Neuroprotective effect ofcurcumin in middle cerebral artery occlusion induced focalcerebral ischemia in rats. Life Sci. 74, 969–985.

Thome, J., Sakai, N., Shin, K.H., Steffen, C., Zahang, Y.J., Impey, S.,Storm, D., Duman, R.S., 2000. cAMP responseelement-mediated gene transcription is upregulated bychronic antidepressant treatment. J. Neurosci. 20, 4030–4036.

Tiraboschi, E., Moraschi, S., Tardito, D., Kasahara, J., Gennarelli, M.,Racagni, G., Popoli, M., 2004. Selective effects ofantidepressants on CREB phosphorylation and related kinasecascade. Eur. Neuropsychopharmacol. 14, S10–S11.

Vollmayr, B., Faust, H., Lewicka, S., Henn, F.A., 2001. Brain derivedneurotrophic factor (BDNF) stress response in rats bred forlearned helplessness. Mol. Psychiatry 6, 471–474.

Willner, P., 1991. Animal models as simulations of depression.Trends Pharmacol. Sci. 12, 131.

Willner, P., Muscat, R., Papp, M., 1992. Chronic mild stress inducedanhedonia: a realistic animal model of depression. Neurosci.Biobehav. Rev. 16, 525.

Wahlstrom, B., Blennow, G., 1978. A study on the fate of curcuminin the rat. Acta Pharm. Toxicol. 43, 86–92.

Xu, J.F., Chen, X.Q., Du, J.Z., Wang, T.Y., 2005a. CRF receptor type 1mediates continual hypoxia-induced CRF peptide and CRFmRNA expression increase in hypothalamic PVN of rats.Peptides 26, 639–646.

Xu, Y., Ku, B.S., Yao, H.Y., Lin, Y.H., Ma, X., Zhang, Y.H., Li, X.J.,2005b. The effects of curcumin on depressive-like behaviors inmice. Eur. J. Pharmacol. 518, 40–46.

Xu, Y., Ku, B.S., Yao, H.Y., Lin, Y.H., Ma, X., Zhang, Y.H., Li, X.J.,2005c. Antidepressant effects of curcumin in the forced swimtest and olfactory bulbectomy models of depression in rats.Pharmacol. Biochem. Behav. 82, 200–206.