Dynamic regulation of mitochondrial functions by glucocorticoids and stress

Dynamic regulation of mitochondrial functionby glucocorticoidsJing Dua, Yun Wanga, Richard Hunterb, Yanling Weia, Rayah Blumenthala, Cynthia Falkea, Rushaniya Khairovaa,Rulun Zhoua, Peixiong Yuana, Rodrigo Machado-Vieiraa, Bruce S. McEwenb,1, and Husseini K. Manjia,1

aLaboratory of Molecular Pathophysiology, Mood and Anxiety Disorders Program, National Institute of Mental Health, National Institutes of Health,Bethesda, MD 20892; and bLaboratory of Neuroendocrinology, The Rockefeller University, New York, NY 10065

Contributed by Bruce S. McEwen, December 22, 2008 (sent for review April 7, 2008)

Glucocorticoids play an important biphasic role in modulating neuralplasticity; low doses enhance neural plasticity and spatial memorybehavior, whereas chronic, higher doses produce inhibition. Wefound that 3 independent measures of mitochondrial function—mitochondrial oxidation, membrane potential, and mitochondrialcalcium holding capacity—were regulated by long-term corticoste-rone (CORT) treatment in an inverted ‘‘U’’-shape. This regulation ofmitochondrial function by CORT correlated with neuroprotection;that is, treatment with low doses of CORT had a neuroprotectiveeffect, whereas treatment with high doses of CORT enhanced kainicacid (KA)-induced toxicity of cortical neurons. We then undertookexperiments to elucidate the mechanisms underlying these biphasiceffects and found that glucocorticoid receptors (GRs) formed a com-plex with the anti-apoptotic protein Bcl-2 in response to CORTtreatment and translocated with Bcl-2 into mitochondria after acutetreatment with low or high doses of CORT in primary cortical neurons.However, after 3 days of treatment, high, but not low, doses of CORTresulted in decreased GR and Bcl-2 levels in mitochondria. As with thein vitro studies, Bcl-2 levels in the mitochondria of the prefrontalcortex were significantly decreased, along with GR levels, afterlong-term treatment with high-dose CORT in vivo. These findingshave the potential to contribute to a more complete understandingof the mechanisms by which glucocorticoids and chronic stress reg-ulate cellular plasticity and resilience and to inform the future devel-opment of improved therapeutics.

allostasis � Bcl-2 � mitochondria � mood disorders � glucocorticoid receptor

Glucocorticoids, the adrenal steroid hormones secreted duringstress, are necessary for survival of the organism. However,

sustained elevations of glucocorticoids (such as those seen duringchronic stress) are associated with numerous deleterious effects,including on neuronal survival (1). Accumulating data have shownan inverted ‘‘U’’-shaped relationship between corticosterone(CORT) levels and performance on behavioral tasks such as spatiallearning and memory (2) and passive avoidance tasks (3). Inaddition, long-term potentiation and primed burst potentiation alsodemonstrate an inverted ‘‘U’’-shaped relationship with CORTlevels (4). Furthermore, considerable data have shown that lowdoses of glucocorticoids have trophic actions on neuronal branchingand survival (5), whereas higher doses are detrimental to neuronalsurvival (6). The role of these hormones in mediating both theprotective, as well as the potentially damaging, effects of stress hasled to the introduction of the terms ‘‘allostasis’’ (the process ofmaintaining homeostasis by active means) and ‘‘allostatic load’’ (thedeleterious consequences of sustained allostasis) (7).

However, the precise cellular mechanisms underlying glucocor-ticoids’ intriguing biphasic effects have not been fully elucidated.Some data suggest that—at least in the hippocampus—these effectsare mediated by distinct intracellular receptor subtypes: the glu-cocorticoid receptor (GR) and the mineralocorticoid receptor(MR). The GR has a lower affinity for endogenous glucocorticoidsand is therefore believed to be more important in the response to,and regulation of, the stress response when endogenous levels ofglucocorticoids are high; in contrast, MRs are activated at lower

glucocorticoid concentrations (8). However, it is unclear if thispostulated mechanism fully explains the biphasic effects of glu-cocorticoids, particularly in areas outside the hippocampus, such asthe frontal cortex.

It is notable that in addition to their well-characterized nucleartranslocation in many cell lines, recent reports indicate that GRsalso translocate to the mitochondria (9). Mitochondria have thewell-known function of energy production through the Krebstricarboxylic-acid cycle and oxidative phosphorylation (10). How-ever, it is now clear that mitochondria play additional importantroles in regulating intracellular calcium, cytoprotection, and neuralplasticity (11). Thus, mitochondrial function plays an important rolein both ‘‘here and now’’ synaptic function, as well as long-termstructural plasticity—both of which are known to be regulated byglucocorticoids.

In the present study, we therefore sought to determine, in asystematic manner, the effects of both physiologic and pathophys-iologic levels of glucocorticoids on various aspects of mitochondrialfunction. We found that glucocorticoids exert biphasic effects onneuronal mitochondrial function, with low levels potentiating var-ious aspects of mitochondrial function, and chronically high levelsattenuating function. These findings have the potential to contrib-ute to a more complete understanding of the mechanisms by whichglucocorticoids and chronic stress regulate cellular plasticity andresilience and to inform the future development of improvedtherapeutics.

ResultsCORT Regulates Mitochondrial Oxidation in a Dose- and Time-Depen-dent Manner. Mitochondrial oxidation is one of the importantmitochondrial functions involved in ATP synthesis. We investigatedmitochondrial oxidation after physiological (low, 100 nM, 500 nM)and pathological (high, 1 �M) doses of CORT using MitoTrackerRed (MTR) and MitoTracker Green (MTG) staining. MTR dyewill become a red fluorescent color only when it is oxidized. Wemeasured both the fluorescent intensity of MTR and the ratio ofMTR/MTG; this ratio was able to filter the bias caused by quench-ing of the fluorescence signals. After 1 day of treatment with 100nM, 500 nM, and 1 �M of CORT, mitochondrial oxidation wassignificantly higher than control, by 143%, 129%, and 137%,respectively (Fig. 1 A, C, and E). For the low-dose (100 nM) CORTtreatment, the enhancement of mitochondrial oxidation was sus-tained for 3 days (Fig. 1 A-C, and E). However, after long-term (3days) treatment with high-dose CORT, cortical neurons showed a

Author contributions: J.D., B.S.M., and H.K.M. designed research; J.D., Y. Wang, R.H., Y.Wei, R.B., C.F., R.K., R.Z., P.Y., and R.M.-V. performed research; R.H. and B.S.M. contributednew reagents/analytic tools; J.D., Y. Wang, R.H., Y. Wei, R.B., and R.M.-V. analyzed data; J.D.and H.K.M wrote the paper.

The authors declare no conflict of interest.

Freely available online through the PNAS open access option.

1To whom correspondence may be addressed. E-mail: [email protected] [email protected].

This article contains supporting information online at www.pnas.org/cgi/content/full/0812671106/DCSupplemental.

www.pnas.org�cgi�doi�10.1073�pnas.0812671106 PNAS � March 3, 2009 � vol. 106 � no. 9 � 3543–3548

NEU

ROSC

IEN

CE

http://www.pnas.org/cgi/content/full/0812671106/DCSupplemental


significant reduction in mitochondrial oxidation (from 137% to77%) (Fig. 1 B, C, and E). When the ratio of MTR to MTG was usedto represent mitochondrial oxidation, the data showed a similarpattern (Fig. 1 D and F).

Mitochondrial Membrane Potential Was Modulated by CORT in a Dose-and Time-Dependent Manner. We also measured mitochondrialmembrane potential by JC-1 staining in cultured cortical neurons[10 days in vitro (DIV)] after CORT treatment. We found thatmitochondrial membrane potential was significantly increased after1.5 hr treatment with 100 nM, 500 nM, or 1 �M of CORT, and thisenhancement was sustained for 1 day (Fig. 2 A, C, and D).Moreover, 100 nM and 500 nM of CORT continued to increasemitochondrial membrane potential after 3 days of treatment (Fig.2 B-D). However, high-dose CORT (1 �M) decreased mitochon-drial membrane potential after 3 days (Fig. 2 C and D). Mitochon-dria play a key function in buffering intracellular calcium (12).Therefore, we explored calcium holding capacity after treatmentwith CORT (100 nM, 1 �M) for 1.5 h or 3 days. The dihydrorhod-2colocalized with the MTG signal showed by confocal image,suggesting that the calcium indicator dyhydrorhod-2 (Rhod-2) waslocalized in the mitochondria (Fig. 3A). We calculated Fi/Fo(before and after thapsigargin stimulation), and found that calciumholding capacity was enhanced after 1.5 h of both low- andhigh-dose CORT treatment. However, after 72 h of treatment withCORT, the low dose showed significantly enhanced mitochondrialholding capacity. High doses of CORT (1 �M) reduced the calciumholding capacity of the neurons (Fig. 3 B-D). Indeed, the regulationof calcium holding capacity correlated with the pattern of Bcl-2levels in the mitochondria.

The Biphasic Effects of Glucocorticoids on Mitochondrial FunctionWere Accompanied by Similar Effects on Neuronal Viability. Becauseour findings suggested that CORT regulates mitochondrial functionin a biphasic manner, we further investigated the impact of thisalteration of mitochondrial function on neuronal viability. Cortical

neurons (10–12 DIV) were treated with 100 nM or 1 �M of CORTfor 1 or 3 days. After treatment with various concentrations ofCORT, the cortical neurons were challenged with kainic acid (KA)

Fig. 1. Mitochondrial oxidation has an inverted ‘‘U’’-shaped relationship with CORT dose. Cortical neurons (10–12DIV) were treated with 100 nM, 500 nM, or 1 �M CORT for1 h, 1 day, or 3 days and stained with MTR and MTG. Datawere combined from 2–3 independent experiments andpresented as mean � SE (*one-way ANOVA ***P � 0.001,**P � 0.05,*P � 0.01 compared to control; # P � 0.05 com-pared to group of 1 �M and 72-hr treatment, n � 32–70). (A)The sample image of MTR and MTG staining after CORTtreatment for 24 h. (B) The sample image of MTR and MTGstaining after CORT treatment for 72 h. (C) Time course ofmitochondrial oxidation by MTR. (D) Time course of mito-chondrial oxidation by MTR/MTG ratio. (E) Dose-dependentcurve for mitochondrial oxidation by MTR. (F) Dose-dependent curve for mitochondrial oxidation by MTR/MTGratio.

Fig. 2. CORT modulates membrane potential in a dose- and time-dependentmanner. Data were combined from 2–3 independent experiments and presentedas mean � SE (*one-way ANOVA, *P � 0.05, **P � 0.01, ***P � 0.001 comparedto control, #P � 0.01 compared to 72 h and 1 �M CORT treatment, n � 34–90). (A)JC-1 staining image after 24 h treatment with 100 nM and 1 �M of CORT. (B) JC-1staining image after 72 h treatment with 100 nM and 1 �M of CORT. (C) Timecourse of JC-1 staining after CORT treatment. (D) Dose-dependent curve for JC-1staining after CORT treatment.

3544 � www.pnas.org�cgi�doi�10.1073�pnas.0812671106 Du et al.

(50 �M) for 12 h and TUNEL assay was performed to determinecell death. As has been previously demonstrated, KA treatmentenhanced apoptosis of cortical neurons (13). As predicted, wefound biphasic effects of CORT on KA-mediated toxicity; 1 day or3 days of treatment with 100 nM of CORT had significant neuro-protective effects against KA-induced apoptosis. However, 1 day or3 days of treatment with the higher dose of CORT (1 �M) enhancedthe apoptotic effects of KA (Fig. 4).

GRs Translocated to Mitochondria in a Dose- and Time-DependentManner. We sought to determine the mechanism whereby glucocor-ticoids regulate mitochondrial functions. We postulated that GRstranslocate to mitochondria and modulate mitochondrial functionin cortical neurons. First, we set out to confirm the purity of themitochondrial fraction, and found a significant 9-fold enrichment ofthe 57 kDa mitochondrial protein cytochrome oxidase subunit 1(COX-1) in the mitochondrial fraction (Fig. S1A). When investi-gating whether the mitochondrial fractions were contaminated bynuclear or cytosol fractions, we found that the levels of both 70 kDamarker nuclear pore protein, and the cytosolic marker glyceralde-hyde 3-phosphate dehydrogenase (GADPH), were low in mito-chondrial fractions (Fig. S1 B and C), suggesting that the mito-chondrial fraction was not contaminated by nuclear or cytosolfractions. We chose 100 nM, 500 nM, and 1 �M of CORT to covera physiologically and pathologically relevant range in our in vitroexperiments. We confirmed that in our experiments, GRs translo-cated into nucleus in response to CORT (Fig. 5A). We thensystematically characterized the dose- and time-dependence of GRmitochondrial translocation after CORT treatment. Cortical neu-rons (10 DIV) were treated with a physiological dose (100 nM) anda supraphysiological/pathological dose (1 �M) for 1.5, 24, and 72 h.After 1.5 h of treatment, both low- and high-dose CORT enhancedmitochondrial localization of GRs by 130% and 136%, respectively(Fig. 5 B and C). However, after long-term treatment (3 days), GRcontent in mitochondria was reduced to 57% only in neurons

treated with high doses of CORT (Fig. 5 B and C). In all of the lanes,the molecular weight of the protein band was 97 kDa, correspond-ing to the reported molecular size of the GR protein. We usedsubunit specific antibodies—anti-GR� and anti-GR�—to deter-mine the subtype and found that the GR band we recognized wasat the same molecular weight as anti-GR� antibody (data notshown). The mitochondrial translocation of GRs was further con-firmed by double immunostaining and confocal microscopy (Fig. S2A and B). Because MRs are also a target for glucocorticoids, wefurther examined potential mitochondrial translocation of MRsafter CORT stimulation. In contrast to the effects observed withGRs, we found that MRs did not translocate into mitochondria(Fig. S3).

GRs Formed a Complex with Bcl-2, and Mitochondrial Bcl-2 Levels WereRegulated by CORT. We thought it possible that GRs bind to andrecruit Bcl-2 into mitochondria to regulate mitochondrial function.Indeed, immunoprecipitation with anti-GR antibody—but not thecontrol anti-IGF Ab—was able to pull down Bcl-2 after low-dose(100 nM) and high-dose (1 �M) CORT treatment (Fig. 6A).Coimmunoprecipitation of GRs and Bcl-2 showed that, afterCORT treatment for 30 mins, formation of the GR/Bcl-2 complexwas significantly enhanced (Fig. 6B). Previous studies have shownthat GRs form protein complexes with heat shock protein 90(HSP90) after 5 mins of treatment with dexomethasone (14). Wetherefore investigated the dynamics of the formation of GR/Bcl-2complex after 5 mins of treatment. After the neurons were treatedwith CORT (1 ��) for 5 mins, the GR/Bcl-2 complex wassignificantly decreased in cytosol fraction to 40.3 � 13.9% of thecontrol (paired Student’s t test, n � 3, P � 0.05) and significantlyincreased in mitochondrial fraction up to 253.7 � 13.7% of thecontrol (paired Student’s t test, n � 3, P � 0.05).

Similar to the GR increase seen in mitochondrial fractions, wefound that Bcl-2 levels in the mitochondria were enhanced after1.5 h with the low (100 nM) and high (1 �M) dose of CORT by146% and 132%, respectively (Fig. 6C). In the low-dose treatmentgroup, this increase of mitochondrial Bcl-2 was sustained for 72 hby 168%. However, in the high-dose treatment group, Bcl-2 levelsin mitochondria were significantly reduced by 68% in corticalneurons after 3 days of treatment (Fig. 6C). The whole cellexpression levels of Bcl-2 did not change after short or long-termtreatment, suggesting that the Bcl-2 increase in mitochondria wasdue to a Bcl-2 translocation into mitochondria in response to CORTtreatment (data not shown). We further performed Bcl-2/GRdouble immunostaining to confirm the interaction of Bcl-2 and GRin response to CORT treatment. We found that after 1 h oftreatment with CORT (1 �M), the colocalization of GR/Bcl-2 wassignificantly increased (Fig. S4). There were more Bcl-2 (green)puncta in the control group, and these did not colocalize with theGRs (red puncta). After CORT treatment, the number of GR/Bcl-2colocalized puncta increased (Fig. S4).

RU486, a GR Partial Agonist, Inhibited CORT-Induced GR Trafficking toMitochondria and Changes in Mitochondrial Function. RU486 is aligand that acts both as a partial GR agonist and antagonist (15).

Fig. 3. Modulation of mitochondrial calcium holdingcapacitybyCORTinatime-anddose-dependentmanner.Data are presented as mean � SE (one-way ANOVA,n�4,n�223.*P�0.05,**P�0.01comparedtocontrol,# P � 0.001 compared to the group of CORT 1 �M for 3days). (A) Rhod-2 colocalizes with MTG in cortical neu-rons. (B) Mitochondrial calcium levels (Fi/Fo) in responseto thapsigargin stimulation in cortical neurons after1.5-hr treatment with CORT. (C) Mitochondrial calciumlevels in response to thapsigargin stimulation in corticalneurons after 3-day treatment with CORT. (D) Compari-son of calcium holding capacity 10 mins after thapsigar-gin stimulation in CORT-treated cortical neurons.

Fig. 4. The effect of low and high concentrations of CORT in KA-inducedapoptosis. Cortical neurons (10–12DIV) were treated with CORT (100 nM, 1 �M)for 1 day in Neurobasal and 3 days in N2 plus Neurobasal. KA (50 �M) was thenappliedfor12htochallengetheneurons.NeuronalapoptosiswasdeterminedbyTUNEL assay (n � 2, n � 17–29 for each group. One-way ANOVA, *P � 0.05,**P � 0.01). Data are presented as mean � SE.

Du et al. PNAS � March 3, 2009 � vol. 106 � no. 9 � 3545

NEU

ROSC

IEN

CE

To determine whether CORT’s effects were mediated throughGRs, we investigated whether RU486 blocked CORT-inducedmitochondrial translocation of GRs or enhancement of mitochon-drial potential. Cortical neurons were treated with CORT 1 �M,RU486 (10 �M), or CORT plus RU486 for 1.5 h. As expected dueto its partial agonistic effects, treatment of the neurons with 10 �MRU486 induced a translocation of GRs from cytosol into mito-chondria similar to that observed with CORT alone (Fig. 7A).However, RU486 blocked CORT-induced GR mitochondrialtranslocation; together, these data show that the effect is mediatedthrough CORT binding to GRs (Fig. 7A). Next, we determinedwhether CORT-induced changes in mitochondrial function werealtered by the GR inhibitor RU486. As before, RU486 treatmentalone significantly increased membrane potential compared tocontrol (Fig. 7B). However, cotreatment of the cells with 10-foldexcess concentrations of RU486 (10 �M) and CORT (1 �M)resulted in a significant inhibitory effect on the mitochondrialmembrane potential compared with CORT 1 �M alone (Fig. 7B).Thus, the antagonistic effects of RU486 were observed for theregulation of membrane potential, as well as for GR translocationto mitochondria, suggesting that these effects were mediatedthrough GRs.

Chronic in Vivo CORT Administration Also Regulated Mitochondrial GRand Bcl-2 Levels in Rodent Prefrontal Cortex. Finally, we sought todetermine whether these in vitro biochemical changes also occurredin vivo. We treated the animals chronically for 21 days with 2 dosesof CORT (50 �g/ml or 400 �g/ml in drinking water) or vehicle. Wemeasured body weights and thymus weights to ensure adequacy of

the chronic CORT treatment. The body weights of the animal andthe weights of the thymus were indeed significantly lower in thegroup treated with 400 �g/ml of CORT. We found that in theprefrontal cortex, GR levels in mitochondria were significantlydecreased by about 30–40% after chronic treatment with high- orlow-dose CORT (50 �g/ml or 400 �g/ml); total cellular levels ofGRs remained unchanged (Fig. 8). In addition, mitochondrial Bcl-2levels were also significantly reduced by approximately 35% only inthe group treated with high-dose CORT (400 �g/ml) (Fig. 8).

DiscussionIn this study, we demonstrated for the first time that glucocorticoidsexert biphasic effects on neuronal mitochondrial dynamics, with lowlevels potentiating and chronic high levels attenuating variousaspects of mitochondrial function. Overall, the study had 5 keyfindings: (i) 3 independent mitochondrial measures—mitochon-drial oxidation, membrane potential, and calcium holding capaci-ty—demonstrated a biphasic response to glucocorticoids; (ii) theenhancement of mitochondrial function by low-dose CORT wasassociated with a neuroprotective effect against KA-induced neu-rotoxicity; in contrast, high-dose CORT enhanced KA neurotox-icity; (iii) GRs translocated into mitochondria in primary corticalneurons in a biphasic relationship with concentrations of CORT aswell as time of treatment; (iv) GRs formed a complex with theantiapoptotic protein Bcl-2 and facilitated Bcl-2 translocation intomitochondria. The expression of Bcl-2 in mitochondria also showeda biphasic response to CORT dose and time; and (v) in rodentschronically treated with CORT, GR levels in mitochondria weresignificantly reduced, with high doses of CORT also reducingmitochondrial Bcl-2 levels.

Glucocorticoids Exert Biphasic Effects on Various Aspects of NeuronalMitochondrial Function. We undertook a series of studies to deter-mine if—as hypothesized—the biphasic effects of glucocorticoidson long-term potentiation (LTP) and behavior are accompanied by

Fig. 5. GRs translocate into mitochondria in responseto low and high concentrations of CORT. Data werecombined from 2–3 independent experiments and pre-sented as mean � SE. (A) GR translocation into nucleusafter CORT treatment after 1.5 h treatment (one-wayANOVA, P � 0.05, n � 18). (B) GR translocation intomitochondria after 100 nM CORT treatment (*Student’st test P � 0.05, n � 8–10). (C) GR translocation intomitochondria after 1 �M CORT treatment (*Student’st test P � 0.05, n � 8–10)

Fig. 6. Bcl-2 coimmunoprecipitated with GRs and translocated into mitochon-dria after CORT treatment in a time- and dose-dependent manner. (A) Bcl-2coimmunoprecipitates with GRs in cultured cortical neurons after 30-min treat-ment (lane1–3: IP with anti-GR antibody; lane 4: total protein from corticalneurons; lane 5: IP with anti-IGF antibody; lane 6: no antibody). (B) The formationof GR/Bcl-2 complexes was enhanced in cortical neurons after 30-min treatmentwith CORT (1 �M and 100 nM) (n � 6, n � 5–8, Student’s t test P � 0.05). (C) Bcl-2translocatedintomitochondriaafterCORTtreatment inculturedcorticalneurons(n � 2–3, n � 8–14, Student’s t test, **P � 0.01, *P � 0.05). Data are presented asmean � SE.

Fig. 7. RU486 has an antagonist effect on CORT-induced GR translocation andmitochondrial potential in cortical neurons. (A) RU486 exerted antagonist effectson CORT-induced GR translocation into mitochondria (N � 2–3, n � 56. One-wayANOVA, *P � 0.05). (B) RU486 effect on CORT-induced increase in mitochondrialpotential revealedbyJC-1staining(N�3–4,n�346.One-wayANOVA,*P�0.01,**P � 0.05). Data are presented as mean � SE.


similar effects on various aspects of critical mitochondrial functions.We found that short-term treatment with low or high doses ofCORT enhanced mitochondrial oxidation, mitochondrial potential,and calcium holding capacity; these effects were sustained duringlonger-term treatment with low-dose CORT. In contrast, reduc-tions in mitochondrial oxidation, mitochondrial potential, andcalcium holding capacity were observed after longer-term treat-ment with high-dose CORT (Figs. 1, 2, and 3), which may providea novel molecular mechanism for the biphasic effects of glucocor-ticoids on animal behavior during chronic stress.

Biphasic Regulation of Mitochondrial Function by GlucocorticoidsProduces Similar Effects on Neuronal Viability. As noted above,glucocorticoids exert biphasic effects on neuronal viability/resilience with low doses exerting neuroprotective effects but higherdoses enhancing neurotoxicity (16, 17). In view of the critical roleof mitochondrial function in regulating neuronal viability/resilience,we investigated the effects of glucocorticoids on KA-induced celldeath. Strikingly similar to the effects seen on mitochondrialfunction, we found that treatment with low-dose CORT had aneuroprotective effect, whereas long-term and high-dose concen-trations of CORT enhanced KA-induced neurotoxicity. In additionto triggering calcium increases in these cells, CORT also inducesgene expression and other nongenomic effects (e.g., changing ERKand mitochondrial functions) (18), which may help to potentiate orprevent the apoptosis induced by other insults (i.e., KA). Together,the data suggest that regulation of mitochondrial function may playan important role in the biphasic effects of glucocorticoids onneuronal viability.

Regulation of Mitochondrial Function by CORT Is Mediated ThroughGRs. To confirm that the effects of glucocorticoids were indeedmediated by GRs, we used the well-known GR partial agonist/

antagonist, RU486 (19). In many tissues, RU486 exerts partialagonistic effects, while blocking the effects of glucocorticoids.Indeed, RU486-induced nuclear GR translocation has also beenpreviously demonstrated (20). In this study, RU486 clearly hadpartial agonist and antagonist effects on CORT’s effects on mito-chondrial GR translocation and mitochondrial membrane potential(Fig. 7). Therefore, the effects of GR translocation on mitochondriaand regulation of mitochondrial potential do indeed appear to bemediated through GRs.

CORT Regulates Mitochondrial GR and Bcl-2 Levels in a BiphasicManner. As noted above, glucocorticoids exert biphasic concentra-tion/time-dependent effects on various aspects of synaptic andneural function. However, the mechanism(s) underlying thesecomplex effects have not been fully elucidated. Cortisol concen-trations in normal human serum vary from 137 nM to 283 nM(7.5 � 2.6 �g/dl) (21); pathophysiologic states have been associatedwith elevated levels in the 420–779 nM (21.4�/�6.4 �g/dl) range(22). Therefore, we chose 100 nM, 500 nM, and 1 �M of CORT torepresent both physiologic and pathophysiologic levels of cortisol.For the CORT treatment of cortical neurons, we used neurobasalmedium, which does not contain cortisol binding globulin (CBG).Therefore, free CORT may be as high as 100 nM, 500 nM, and 1�M. Because of CBG in serum, the free concentrations in vivo inthe brain might be in the range of the lowest concentration (100nM) used here. Our study intentionally investigated the effects ofvery high concentrations and, indeed, we found the biphasic effectthat leads to cytotoxicity. Whether or not this happens in vivo understress conditions remains unclear. However, the in vivo data showedthat both low and high doses (50 �g/ml or 400 �g/ml) of CORT indrinking water reduced GRs in mitochondria with a biphasic timecourse, whereas only the highest, clearly supraphysiological, dosesreduced Bcl-2 levels in the mitochondria.

The expression of the major cytoprotective protein, Bcl-2, isknown to reduce reactive oxygen species (ROS) production, thuspreventing permeability transition pore opening, and mitochon-drial depolarization (23). Studies of isolated mitochondria haveshown that Bcl-2 overexpression increases mitochondrial calciumuptake capacity (24), making the cells resistant to the deleteriousinfluence of elevated intracellular calcium. Thus, Bcl-2 exertssignificant effects on cellular calcium buffering, not only under theextreme conditions of calcium-induced apoptosis, but likely evenduring normal synaptic activity. This report is the first to show thatGRs interact with Bcl-2 and translocate into mitochondria inresponse to CORT, thus regulating mitochondrial functions (videinfra; Figs. 5 and 6).

GR and Bcl-2 Levels in Mitochondria Were Decreased after ChronicTreatment with CORT in Vivo. Our in vitro studies demonstrated thatlong-term and high-dose treatment with CORT down-regulatedGR and Bcl-2 levels in mitochondria (Figs. 5 and 6), as well asmitochondrial function (Figs. 1, 2, and 3). As expected, we foundthat chronic treatment with high-dose and low-dose glucocorticoidsattenuated prefrontal cortical mitochondrial GR levels. Most im-portantly, anti-apoptotic mitochondrial Bcl-2 levels were reducedafter high-dose and chronic treatment with glucocorticoids in themitochondrial fractions, suggesting an alteration of mitochondrialfunctions (Figs. 4 and 8). This reduction in anti-apoptotic mito-chondrial Bcl-2 levels by glucocorticoids is consistent with previousin vivo findings that glucocorticoids potentiated the KA- andischemia-induced injury to neurons (25, 26).

ConclusionIn conclusion, this study demonstrates for the first time thatglucocorticoids exert major, biphasic effects on various facets ofcortical mitochondrial dynamics. In addition to regulating mi-tochondrial activities, there may well be other mechanismsoperative in the biphasic effects that glucocorticoids exert on

Fig. 8. GR and Bcl-2 levels in mitochondria are decreased in mitochondrialfractions in prefrontal cortex after chronic stress and after CORT treatment.Western blot analysis of GR and Bcl-2 levels was performed in total homogenates(T)andmitochondrial fractions (M) fromprefrontal cortexofCORT(50 �g/ml,400�g/ml)-treated animals. Western blot analysis of GR and Bcl-2 content was per-formedintotalhomogenates (T)andmitochondrial (M)fractionsfromprefrontalcortex. Data are presented as mean � SE. (A) Thymus weights in CORT-treatedanimals (n � 8 animals per group; **one-way ANOVA, P � 0.01). (B) Body weightsof CORT-treated animals (n � 8 animals per group, one-way ANOVA, ** P � 0.01)(C) Chronic CORT treatment significantly reduced mitochondrial GR and Bcl-2levels (Control group n � 12–15; CORT 50 �g/ml, n � 8; CORT 400 �g/ml, n �12–15, *one-way ANOVA, P � 0.05)

Du et al. PNAS � March 3, 2009 � vol. 106 � no. 9 � 3547

NEU

ROSC

IEN

CE

neuronal function. For instance, concentration-dependent ef-fects of glucocorticoids on MRs and GRs may explain thebiphasic effects of glucocorticoids on hippocampal function (27).Consistent with the findings presented here, recent work hasshown that glucocorticoid-mediated cell death was counteractedby the MR agonist aldosterone (27). It is possible that under lowconcentrations of CORT, both MRs and GRs mediate neuro-protective effects via different mechanisms. While the differenteffects of glucocorticoids on MRs and GRs are undoubtedlyimportant for various neural functions, the dynamic regulation ofmitochondrial function by glucocorticoids provides new evi-dence to explain the biphasic effects of glucocorticoids onvarious neural functions. Indeed, the regulation of mitochondrialdynamics that we observed during chronic high-dose glucocor-ticoid administration appears to represent an important com-ponent of the allostatic overload observed as a pathologicalconsequence of chronic stress (1, 7). The possibility that agentsthat enhance mitochondrial function may be useful in counteringthe deleterious effects of excessive glucocorticoid secretionobserved in diverse conditions is an exciting prospect for futureinvestigation.

Materials and MethodsPreparation of Cortical Neuronal Cultures. Cultures of cortical neurons wereprepared according to a previously published procedure with minor modifica-tions (28). Detailed methods are provided in the SI.

MTR and MTG Staining of Cortical Neurons. The staining of MTR CM-H2XROS(Molecular Probes), and MTG (Molecular Probes) was performed according to themanufacturer’s instructions (Molecular Probes). Detailed methods are providedin the SI.

JC-1 Staining of Cortical Neurons. JC-1 staining was performed according to themanufacturer’s instructions (Molecular Probes). Detailed methods are providedin the SI.

Determination of Mitochondrial Calcium by Rhod 2 Staining. The procedure forRhod 2 staining was performed according to the manufacturer’s instructions(Invitrogen). Detailed methods are provided in the SI.

In Situ Cell Death Detection Assay. We used the kit from Roche to detect andquantify apoptosis at the single cell level based on labeling of DNA strandbreaks (TUNEL assay) according to the manufacturer’s instructions. Detailedmethods are provided in the SI.

Preparation of Mitochondrial Fractions from Cortical Neurons. Preparation ofmitochondrial fractions from cortical neurons followed a previously publishedprocedure with minor modifications (29). Detailed methods are provided inthe SI.

Western Blot Analysis. Protein concentrations were determined by using the BCAprotein assay kit (Pierce, Rockford, IL). Western blot analysis was similar to apreviouslypublishedprocedurewithminormodifications (29,30).Detailedmeth-ods are provided in the SI.

CORT Treatment. All animal treatments, procedures, and care were approved bythe NIMH Animal Care and Use Committee and followed the Guide for the Careand Use of Laboratory Animals (31). Animal treatment was similar to a previouslypublished procedure with minor modifications (29, 32). Detailed methods areprovided in the SI.

Preparation of Mitochondrial Fraction from Brain Tissue. Preparation of mito-chondrial fractions from rat brain tissue was similar to a previously publishedprocedure with minor modifications (33). Detailed methods are provided inthe SI.

ACKNOWLEDGMENTS. We thank Ioline Henter for outstanding editorial assis-tance. We acknowledge the support of the Intramural Research Program of theNIMH and the Stanley Medical Research Institute. B.S.M. acknowledges grantsupport from NIH MH41256.

1. McEwen BS (2008) Central effects of stress hormones in health and disease: Under-standing the protective and damaging effects of stress and stress mediators. EurJ Pharmacol 583:174–185.

2. Quirarte GL, Roozendaal B, McGaugh JL (1997) Glucocorticoid enhancement of mem-ory storage involves noradrenergic activation in the basolateral amygdala. Proc NatlAcad Sci USA 94:14048–14053.

3. Gold PE, McGaugh JL (1977) in Neuropeptide Influences on the Brain and Behavior,eds. Miller LH, Sandman CA, Kastin AJ (Raven Press, New York), pp. 127–143.

4. Joels M (2005) Corticosteroid effects in the brain: U-shape it. Trends Pharmacol Sci27:244–249.

5. Gould E, Woolley CS, McEwen BS (1990) Short-term glucocorticoid manipulationsaffect neuronal morphology and survival in the adult dentate gyrus. Neuroscience37:367–375.

6. Sapolsky R (1992) Stress, the Aging Brain and the Mechanisms of Neuron Death. (MITPress Cambridge).

7. McEwen BS (1998) Stress, adaptation, and disease. Allostasis and allostatic load Ann NYAcad Sci 840:33–44.

8. Joels M, De Kloet ER (1992) Coordinative mineralocorticoid and glucocorticoid recep-tor-mediated control of responses to serotonin in rat hippocampus. Neuroendocrinol-ogy 55:344–350.

9. Koufali MM, Moutsatsou P, Sekeris CE, Breen KC (2003) The dynamic localization of theglucocorticoid receptor in rat C6 glioma cell mitochondria. Mol Cell Endocrinol 209:51–60.

10. Huttemann M, Lee I, Samavati L, Yu H, Doan JW (2007) Regulation of mitochondrialoxidative phosphorylation through cell signaling. Biochim Biophys Acta 1773:1701–1720.

11. Toescu EC, Verkhratsky A (2004) Ca2� and mitochondria as substrates for deficits insynaptic plasticity in normal brain ageing. J Cell Mol Med 8:181–190.

12. Parekh AB (2003) Mitochondrial regulation of intracellular Ca2� signaling: More thanjust simple Ca2� buffers. News Physiol Sci 18:252–256.

13. Stein-Behrens BA, et al. (1992) Glucocorticoids exacerbate kainic acid-induced extra-cellular accumulation of excitatory amino acids in the rat hippocampus. J Neurochem58:1730–1735.

14. Bartis D, et al. (2007) Intermolecular relations between the glucocorticoid receptor,ZAP-70 kinase, and Hsp-90 Biochem Biophys Res Commun 354:253–258.

15. Groyer A, Schweizer-Groyer G, Cadepond F, Mariller M, Baulieu EE (1987) Antiglu-cocorticosteroid effects suggest why steroid hormone is required for receptors to bindDNA in vivo but not in vitro. Nature 328:624–626.

16. Wenzel A, et al. (2001) Prevention of photoreceptor apoptosis by activation of theglucocorticoid receptor. Invest Ophthalmol Vis Sci 42:1653–1659.

17. Sapolsky RM (2000) The possibility of neurotoxicity in the hippocampus in majordepression: a primer on neuron death. Biol Psychiatry 48:755–765.

18. Haller J, Mikics E, Makara GB (2008) The effects of non-genomic glucocorticoid mech-anisms on bodily functions and the central neural system. A critical evaluation offindings. Front Neuroendocrinol 29:273–291.

19. Bourgeois S, Pfahl M, Baulieu EE (1984) DNA binding properties of glucocorticosteroidreceptors bound to the steroid antagonist RU-486. EMBO J 3:751–755.

20. Jewell CM, et al. (1995) Immunocytochemical analysis of hormone mediated nucleartranslocation of wild type and mutant glucocorticoid receptors. J Steroid Biochem MolBiol 55:135–146.

21. Davis HA, Gass GC, Bassett JR (1981) Serum cortisol response to incremental work inexperienced and naive subjects. Psychosom Med 43:127–132.

22. Pirich K, Vierhapper H (1988) 24-hour serum concentration profile of cortisol inpatients with Cushing’s disease. Exp Clin Endocrinol 92:275–279.

23. Zamzami N, et al. (1998) The thiol crosslinking agent diamide overcomes the apopto-sis-inhibitory effect of Bcl-2 by enforcing mitochondrial permeability transition. On-cogene 26:1055–1063.

24. Murphy AN, Bredesen DE, Cortopassi G, Wang E, Fiskum G (1996) Bcl-2 potentiates themaximal calcium uptake capacity of neural cell mitochondria. Proc Natl Acad Sci USA93:9893–9898.

25. Sapolsky RM, Pulsinelli WA (1985) Glucocorticoids potentiate ischemic injury to neu-rons: therapeutic implications. Science 229:1397–1400.

26. Roozendaal B, et al. (2001) Memory retrieval impairment induced by hippocampal CA3lesions is blocked by adrenocortical suppression. Nat Neurosci 4:1169–1171.

27. Crochemore C, et al. (2005) Direct targeting of hippocampal neurons for apoptosis byglucocorticoids is reversible by mineralocorticoid receptor activation. Mol Psychiatry10:790–798.

28. Du J, et al. (2003) Regulation of TrkB receptor tyrosine kinase and its internalization byneuronal activity and Ca2� influx. J Cell Biol 163:385–395.

29. Almeida A, Medina JM (1998) A rapid method for the isolation of metabolically activemitochondria from rat neurons and astrocytes in primary culture. Brain Res Brain ResProtoc 2:209–214.

30. Du J, et al. (2008) The role of hippocampal GluR1 and GluR2 receptors in manic-likebehavior. J Neurosci 28:68–79.

31. Institute of Laboratory Animal Research, Commission on Life Sciences, National Re-search Council (1996) Guide for the Care and Use of Laboratory Animals (Superinten-dent of Documents, U.S. Government Printing Office, Washington, DC), p 140.

32. Nacher J, Pham K, Gil-Fernandez V, McEwen BS (2004) Chronic restraint stress andchronic corticosterone treatment modulate differentially the expression of moleculesrelated to structural plasticity in the adult rat piriform cortex. Neuroscience 126:503–509.

33. Zini R, Simon N, Morin C, Thiault L, Tillement JP (1998) Tacrolimus decreases in vitrooxidative phosphorylation of mitochondria from rat forebrain. Life Sci 63:357–368.











Supporting InformationDu et al. 10.1073/pnas.0812671106Materials and MethodsPreparation of Cortical Neuronal Cultures. Whole cortices weredissected from embryonic day 18 (E18) rats, dissociated incalcium and magnesium-free HBSS containing 0.125% trypsinfor 15 min. After trituration, the neurons were plated at 0.4million cells per well in 6-well plates or 2–4 million in each 150mm plate in Dulbecco’s modified Eagle’s medium (DMEM,Invitrogen)/10% FBS and cultured at 37 °C, 5% CO2, and 95%humidity overnight, and then changed to serum-free medium,Neurobasal plus B27 (Invitrogen). The medium was changedevery 3 days. These cultures yielded virtually all neurons (datanot shown).

MTR and MTG Staining of Cortical Neurons. Cortical neurons in4-well chamber glass slides were incubated with MTR CM-H2XROS (Molecular Probes, final concentration 300 nM) andMTG (Molecular Probes, final concentration 60 nM) in 1�neurobasal without phenol red and 10 mM Hepes buffer for 15min and washed twice with phenol red-free Neurobasal � 10mM Hepes. Two to five Z-stack images for each slice wereacquired randomly by using a Zeiss 510-LSM confocal micro-scope under exactly the same conditions. The longest dendritefrom each neuron treated in each group was quantified by 510metamorph software. The average dendrite lengths from quan-tified dendrites were not significantly different among the con-trol and treated groups.

JC-1 Staining of Cortical Neurons. Cortical neurons in 4-well cham-ber glass slide were incubated with JC-1 (Molecular Probes,T3168, 10 �g/ml) in warm neurobasal without phenol red � 10mM Hepes for 20 min at 37 °C and then washed twice. Two tofive Z-stack images were randomly acquired by using a Zeiss510-LSM confocal microscope under exactly the same condi-tions. The longest dendrite from each neuron treated in eachgroup was quantified. The average dendrite lengths were notsignificantly different among the control and treated groups. Thered and green fluorescent signals were analyzed by 510 meta-morph software.

Determination of Mitochondrial Calcium by Rhod 2 Staining. Neuronswere placed in MIC buffer (NaCl 130 mM, KCl 5.3 mM, MgSO40.8 mM, Na2HPO4 1 mM, Glucose 2 mM, Hepes 20 mM,Na-Pyruvate 1 mM, NaHCO3 2.5 mM, Ascorbic acid 1.0 mM,CaCl2 1.5 mM, BSA 1.5 mg/ml) for 10 min at room temperatureand loaded with 4.0 mM dihydrorhod-2 in MIC buffer for 30 minat 37 °C. After washing twice, the fluorescent signals weremeasured by Zeiss LSM 510 microscopy (excitation at 543 nmand a 560 nm long pass filter was used for emission). Theimage-fields were randomly determined and images were cap-tured every 20 sec. After 1 min and 40 sec of baseline measure-ments, 2 mM thapsigargin was added and images were taken foran additional 10 min. The regions of interest for quantificationwere the cell body and the longest dendrite. The average redfluorescent intensity of the cell in each time interval wasdetermined by 510 Metamorph software. We calculated Fi/Fobefore and after thapsigargin treatment.

Preparation of Mitochondrial Fractions from Cortical Neurons. Neu-rons were harvested in MitoBuffer (250 mM sucrose, 20 mMHepes, 10 mM KCl, 1 mM EDTA, 1 mM EGTA, 5 mM MgCl2,1 mM DTT and 1x protease inhibitors) and homogenized by tightglass homogenizers 20 times. One hundred microliters of the

homogenized solution was saved as total homogenates. The restof the solution was centrifuged at 800 � g for 10 min. Theobtained pellet was saved as nuclear fraction. The supernatantwas centrifuged again at 15,000 � g for 10 min. The resultingsupernatant was transferred to another tube as cytosol fraction.The remaining second pellet (mitochondria) was dissolved withlysis buffer (10 mM Tris-HCL, 15 mM NaCl, 1 mM EDTA, 1 mMEGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM�-Glycerophosphate) with phosphatase inhibitors mixture I andII (1:100) and protease inhibitor mixture (1:100) (Sigma). Thefinal samples were kept on ice for 30 min, centrifuged at140,000 � g for 10 min, and then the supernatants were storedat �80 °C as mitochondrial fractions.

in Situ Cell Death Detection Assay. We used the kit from Roche fordetection and quantification of apoptosis at the single cell levelbased on labeling of DNA strand breaks (TUNEL assay) fol-lowing the manufacturer’s instructions. Briefly, cells were fixedin 4% paraformaldehyde in PBS for one hour at 15–25 °C andwashed with PBS, then incubated in permeabilisation solutionfor 2 min on ice. Fifty microliter TUNEL reaction mixture wasadded to the sample and the slide was incubated at 37 °C for 60min in the dark. The slide was rinsed 3 times and mounted withmounting media with DAPI. The pictures were randomly takenunder the microscope and apoptotic and total cells werecounted.

Western Blot Analysis. Protein concentrations were determined byusing the BCA protein assay kit (Pierce). Equal amounts ofproteins were subjected to 4–20% SDS/PAGE gels and sepa-rated by electrophoresis (Invitrogen), transferred to 0.45 �mpore-size PVDF membranes (Millipore) and immunoblottedwith anti-GR (Abcam, 1:200), anti-Bcl-2 (Santa Cruz Biotech-nology, sc-783, 1:200), anti-GADPH (Abcam, 1:500), anti-COX-1 (Molecular Probes, 1:200), anti-MR antibody (AffinityBioreagents, 1:200), and anti-nucleus pore protein antibodies(Abcam, 1:300). HRP-conjugated anti-rabbit antibody andHRP-anti-mouse antibody (Amersham, and Vector Laborato-ries) were used as secondary antibodies. Immunoreactive bandswere visualized by enhanced chemoluminescence (ECL�) (Am-ersham) and exposed to Kodak Biomax or Biolight film. TheECL signal intensities were quantified using a Kodak Imagesystem. All data were analyzed by ANOVA or Student’s t test.

Double Immunocytochemistry Staining of Cortical Neurons. Rat cor-tical neurons were fixed in 4% paraformaldehyde in PBS for onehour on ice. The cells were then blocked with 10% normal goatserum, 1% BSA, and 0.4% triton in PBS for one hour, followedby incubation with rabbit anti-GR (Abcam, 1:40) and mouseanti-COXIV (Molecular Probes, 1:100) at 4 °C overnight. Thesecondary antibodies were FITC-conjugated anti-mouse (Jack-son Labs, for multiple staining, 1:50) and Cy3-conjugated anti-rabbit (Jackson Labs, for multiple staining, 1:150) antibodies.After washing, the cells were mounted onto slides with anti-fademounting media (Molecular Probes). Z-stack imaging was ac-quired by using a Zeiss 510-Meta confocal microscope underexactly the same conditions. Neurons were randomly photo-graphed with 5 images obtained for each slice to ensure thatconditions for each slice for each group were the same. Thelongest dendrite from each neuron treated in each group wasquantified. Colocalization efficiency of GR (red) and COXIV(green) at each longest dendrite was determined using 510-

Du et al. www.pnas.org/cgi/content/short/0812671106 1 of 6

http://www.pnas.org/cgi/content/short/0812671106

Metamorph software, using exactly the same conditions for bothcontrol and experimental groups. The colocalization coefficientrepresents the percentage of red fluorescence (GR) on the greenfluorescence (COXIV), using total red signal as 100% in theregion of interest. Data were analyzed by Student’s t test.

For double staining of GR and Bcl-2, rat cortical neurons werefixed in 4% paraformaldehyde in PBS for 20 min at roomtemperature. The cells were then blocked with 10% normal goatserum and 0.4% triton in PBS for one hour, followed byincubation with rabbit anti-GR (ABR Affinity Bioreagents,1:60) and mouse anti-Bcl-2 (Abcam, 1:60) at 4 °C overnight. Thesecondary antibodies were Dylight-488-conjugated Donkey anti-mouse (Jackson Labs, for multiple staining,1:200) and Cy3-conjugated Donkey anti-rabbit (Jackson Labs, for multiple stain-ing, 1:200) antibodies. After washing, the cells were mountedonto slides with anti-fade mounting media (Molecular Probes).Confocal images (0.5–0.8 �m in thickness) were acquired byusing a Zeiss 510-LSM confocal microscope under exactly thesame conditions. Neurons were randomly photographed with 5to 10 images obtained for each slice to ensure that conditions foreach slice for each group were the same. The cell bodies from allneurons in the image were extracted from the original image andthe colocalization efficiency was quantified by 510-LSM soft-ware, using exactly the same conditions for both control andexperimental groups. The colocalization efficiency representsthe percentage of green fluorescence (Bcl-2) on the red fluo-rescence (GR). Data were analyzed by Student’s t test.

Corticosterone Treatment. All animal treatments, procedures, andcare were approved by the NIMH Animal Care and UseCommittee and followed the Guide for the Care and Use ofLaboratory Animals (1). Animals received corticosterone in theirdrinking water in a final concentration of 50 �g or 400 �gcorticosterone/ml of water for 3 weeks with a final ethanolconcentration of 0.3% or 2.4%, respectively. Vehicle-treatedanimals received 0.3 or 2.4% ethanol in their drinking water for3 weeks. Animals were weighed after the treatment period. Forbiochemical studies, the animals were decapitated and braintissue was immediately processed for mitochondrial preparation.To assess the adequacy of the chronic corticosterone adminis-tration, thymus weight was determined.

Preparation of Mitochondrial Fraction from Brain Tissue. Prefrontalcortices were homogenized in 1 ml of homogenizing buffer (10mM Hepes, pH 7.5, 200 mM mannitol, 70 mM sucrose, 1 mMEGTA) in a 1 ml tissue grinder 4 times with loose and 8 timeswith tight grinder. One hundred microliters of tissue homoge-nates were saved as total homogenates. The rest of the homog-enate was centrifuged at 2,000 � g for 4 min at 4 °C and thepellets were saved as nuclear fractions. The supernatants werecentrifuged again at 12, 000 � g for 8 min at 4 °C. The secondsupernatants were saved as cytosol fractions. The pellets wereresuspended as mitochondrial fractions and dissolved into 1 mlof homogenization buffer and centrifuged one more time at 12,000 � g for 8 min. The pellets were resuspended into the lysisbuffer and stored at �80 °C.

1. Institute of Laboratory Animal Research, Commission on Life Sciences, National Re-search Council (1996) Guide for the Care and Use of Laboratory Animals (Superinten-dent of Documents, U.S. Government Printing Office, Washington, DC), p 140.



Fig. S1. Purity of mitochondrial fraction from cortical neurons. (A) Mitochondrial marker COX1 was enhanced 9-fold in the mitochondrial fraction. (B)Mitochondrial fraction showed low content of cytosol marker GADPH. (C) Absence of nuclear pore protein in the mitochondrial fraction (*Student’s t test P �0.05, # one-way ANOVA, P � 0.01, n � 4 for each group). This experiment was repeated.



Fig. S2. Double immunostaining of cortical neurons with anti-GR and anti-COXIV antibodies after corticosterone treatment. (A) Cortical neurons (10–12DIV)were treated with 100 nM corticosterone for 1.5 h. Double immunostaining was performed with anti-GR and anti-COXIV antibodies. Green arrows indicatemitochondria without GRs and yellow arrows indicate mitochondria with GRs. (B) Quantification of GRs and COXIV colocalization. Colocalization efficiency ofGRs (red) and COXIV (green) at each individual longest dendrite was determined using 510-Meta software. This experiment was repeated (*Student’s t test, P �0.05, n � 10).



Fig. S3. Lack of translocation of MRs to mitochondria in response to corticosterone treatment. Cortical neurons (10–12DIV) were treated with 100 nM or 1 �Mcorticosterone for 1.5 h. Mitochondrial and nucleus fractions were isolated from the cortical neurons. Western blot analyses were performed with anti-MRantibody (n � 7–14 for each group). In contrast to what was observed with GRs, no mitochondrial translocation was observed for the MRs.



Fig. S4. Colocalization of GR and Bcl-2 after corticosterone treatment. Cortical neurons (12 DIV) were cultured in corticosterone-depleted medium (neurobasalwithout B27) for one day. The cells were then treated with corticosterone (1 �M) for one hour and fixed with 4% paraformaldehyde in PBS. Doubleimmunostaining was performed with rabbit-anti-GR antibody and mouse-anti-Bcl-2 antibody. Images were taken by confocal microscope. Colocalizationefficiency of GRs and Bcl-2 was analyzed by 510 LSM software. Light blue arrows indicate the Bcl-2 puncta (green) without GR (red). Yellow arrows indicate thepuncta with both green (Bcl-2) and red (GRs). Note that after treatment with corticosterone, yellow puncta were increased (Student’s t test, **P � 0.01, n � 15).This experiment was repeated.

