Original Contribution

AMPK activation prevents prenatal stress-induced cognitiveimpairment: Modulation of mitochondrial content and oxidative stress

Ke Cao a, Adi Zheng a, Jie Xu a, Hao Li a, Jing Liu a, Yunhua Peng a, Jiangang Long a,Xuan Zou b, Yuan Li a, Cong Chen a, Jiankang Liu a, Zhihui Feng a,n

a Center for Mitochondrial Biology and Medicine, The Key Laboratory of Biomedical Information Engineering of Ministry of Education, School of Life Scienceand Technology and Frontier Institute of Science and Technology, Xi’an Jiaotong University, Xi’an, Chinab Center for Translational Medicine, The Key Laboratory of Biomedical Information Engineering of Ministry of Education,School of Life Science and Technology and Frontier Institute of Science and Technology, FIST, Xi’an Jiaotong University, Xi’an, China

a r t i c l e i n f o

Article history:Received 12 June 2014Received in revised form10 July 2014Accepted 23 July 2014Available online 1 August 2014

Keywords:AMPKResveratrolMitochondrial biogenesisOxidative stressPrenatal stressCognitive function

a b s t r a c t

Prenatal stress induces cognitive functional impairment in offspring, an eventuality in which mitochon-drial dysfunction and oxidative stress are believed to be closely involved. In this study, the involvementof the AMP-activated protein kinase (AMPK) pathway was investigated. A well-known activator,resveratrol (Res), was used to induce AMPK activation in SH-SY-5Y cells. Significant mitochondrialbiogenesis and phase II enzyme activation, accompanied by decreased protein oxidation and GSSGcontent, were observed after Res treatment, and inhibition of AMPK with Compound c abolished theinduction effects of Res. Further study utilizing a prenatal restraint stress (PRS) animal model indicatedthat maternal supplementation of Res may activate AMPK in the hippocampi of both male and femaleoffspring, and that PRS-induced mitochondrial loss in the offspring hippocampus was inhibited by Resmaternal supplementation. In addition, Res activated Nrf2-mediated phase II enzymes and reducedPRS-induced oxidative damage in both male and female offspring. Moreover, PRS markedly decreasedmRNA levels of various neuron markers, as well as resultant offspring cognitive function, based onspontaneous alternation performance and Morris water maze tests, the results of which weresignificantly improved by maternal Res supplementation. Our results provide evidence indicating thatAMPK may modulate mitochondrial content and phase II enzymes in neuronal cells, a process whichmay play an essential role in preventing PRS-induced cognitive impairment. Through the coupling ofmitochondrial biogenesis and the Nrf2 pathway, AMPK may modulate oxidative stress and be apromising target against neurological disorders.

& 2014 Elsevier Inc. All rights reserved.

Introduction

Prenatal stress during gestation has many deleterious effects onboth the development and the behavior of offspring [1]. Clinicalstudies demonstrate that exposure of pregnant mothers to stressfulconditions increases the susceptibility of their offspring to mental

disorders, such as depression, schizophrenia, and cognitive deficits[2]. Animal studies show that prenatal stress causes alterations ofboth the hypothalamic–pituitary adrenocortical axis and the brainneurotransmitter systems and also impairs hippocampal-dependentspatial learning and memory abilities in offspring [3,4]. Pregnantrodents that have suffered restraint stress represent a valid model ofstress with neurobiological and behavioral consequences [5,6].Although the specific mechanisms of prenatal restraint stress (PRS)remain unclear, evidence suggests that both oxidative stress andmitochondrial dysfunction may be involved in PRS-induced neuro-logical damage and cognitive impairment [7,8]. However, thedetailed mechanisms regulating each of these processes have notbeen elucidated in the rodent model.

The involvement of cellular energy metabolism in different condi-tions has become an area of intense interest [9]. As a sensor of cellularenergy status, AMP-activated protein kinase (AMPK) is an attractivetarget for a range of diseases, such as cancer [10], diabetes [11], andcardiovascular disease [12]. Studies also implicate a neuroprotective

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/freeradbiomed

Free Radical Biology and Medicine

http://dx.doi.org/10.1016/j.freeradbiomed.2014.07.0290891-5849/& 2014 Elsevier Inc. All rights reserved.

Abbreviations: AMPK, AMP-activated protein kinase; Arc, activity-regulatedcytoskeleton-associated protein; BDNF, brain-derived neurontrophic factor;EMX2, empty spiracles homeobox 2; GAP43, growth-associated protein-43; GCL,glutamate-cysteine ligase; HO-1, heme oxygenase 1; NMDAR, N-methyl-D-asparticacid receptor; Nrf2, NF-E2-related factor; NQO-1, NAD(P)H dehydrogenase(quinone 1); PGC-1α, peroxisome proliferator-activated receptor gamma, coacti-vator 1 alpha; PRS, prenatal restraint stress; Res, resveratrol; SCG10, stathmin-like2.

n Correspondence to: Center for Mitochondrial Biology and Medicine, School ofLife Science and Technology and Frontier Institute of Science and Technology, Xi’anJiaotong University, 28 W, Xian-ning Road, Xi’an 710049, China.Fax: 86 029 82665849.

E-mail address: zhfeng@mail.xjtu.edu.cn (Z. Feng).

Free Radical Biology and Medicine 75 (2014) 156–166

effect of AMPK both in vitro [13] and in vivo [14]. However, whetherAMPK plays a role in PRS-induced neurological impairment isstill unknown. It has been demonstrated that AMPK activity correlatesstrongly with mitochondrial function by promoting a mitochondrialbiogenesis pathway. Activation of AMPK is dependent on theupregulation of peroxisome proliferator-activated receptor gammacoactivator-1alpha (PGC-1α) and nuclear respiratory factor 1 (NRF1)expression in rat visual cortical neurons [15]. In addition to mitochon-drial dysfunction, oxidative stress was also involved in PRS-inducedneurological damage and cognitive impairment [7,8]. Induction ofphase II detoxifying enzymes is one of the most important pathwaysfor cells to fight against oxidative stress. Nuclear factor erythroid-2-related factor-2 (Nrf2) is an antioxidant transcription factor mediatingthe expression of antioxidant enzymes, such as NADPH quinineoxidoreductase-1 (NQO1) and heme oxygenase-1 (HO-1). It has a widerange of activities in regulating redox state and energy metabolism incells [16]. The response of AMPK to oxidative stress has been recentlyreported, but the downstream signals of this response are largelyunknown. The potential for cross talk between the AMPK and the Nrf2cascades has been reported in Caenorhabditis elegans [17], in humanendothelial cells [18], and in mammalian inflammatory systems [19].

However, no information about the potential for convergence betweenthe AMPK and the Nrf2 pathways or the subsequent exertion of aneuroprotective effect exists.

Resveratrol (3,40,5 trihydroxystilbene, Res), a naturally occur-ring phytoalexin compound present in almost 70 plant species(such as grape, peanut, and soya beans), is considered one of themost effective known antioxidants. Similar to other chemicalactivators of AMPK, such as 5-aminoimidazole-4-carboxamideribonucleoside (AICAR) and metformin, Res has been widelyaccepted as a natural AMPK activator [20,21]. The impact of Reson the prevention of prenatal stress-induced cognitive impairmenthas been recently reported, but the exact mechanisms involved inits neuroprotective effects are still poorly characterized [22–24].

In the present study, to investigate the potential regulatoryeffect of AMPK on mitochondrial biogenesis and phase II enzymeinduction, and their subsequent involvement in prenatal stress-induced cognitive dysfunction, an AMPK activator was employedboth in SH-SY-5Y cells and in a PRS animal model. We propose thatthe upregulation of mitochondrial biogenesis and Nrf2 pathwaysby AMPK activation may play an important role in promotingneuron survival and related improvement in cognitive function.

Fig. 1. Effects of Res on mitochondria and phase II enzymes in SH-SY-5Y cells. Cells were treated with Res at doses of 10, 50, and 100 μM for 24 h. Mitochondrial biogenesis-related proteins were analyzed by Western blot (A, Western blot image; B, statistical analysis of PGC-1α; C, statistical analysis of Complex I; D, statistical analysis of ComplexIII). Phase II enzyme-related proteins were determined by Western blot (E, Western blot image; F, statistical analysis of Nrf2; G, statistical analysis of NQO1; H, statisticalanalysis of HO-1). Protein oxidation was determined by measuring carbonyl protein content (I, Western blot image; J, statistical analysis). GSSG was evaluated (K). GSH/GSSGratio was calculated (L). Values are means 7 SEM from at least three independent experiments. nP o 0.05, nnP o 0.01.

K. Cao et al. / Free Radical Biology and Medicine 75 (2014) 156–166 157

Material and methods

Chemicals

Antibodies against β-actin were purchased from Sigma (St.Louis, MO, USA). Antibodies against complexes I (30 kDa), II(30 kDa), III (51 kDa), IV (40 kDa), and V (55 kDa) were purchasedfrom Invitrogen (Carlsbad, CA, USA). Antibodies against PGC-1α,Nrf2, NQO1, and HO-1 were purchased from Santa Cruz Biotech-nology (Santa Cruz, CA, USA). Antibodies against AMPK andp-AMPK (Thr172) were purchased from Cell Signaling Technology(Danvers, MA, USA). PCR primers were synthesized by BaiaokeBiotech (Beijing, China). TRIzol and other reagents were purchasedfrom Invitrogen (Carlsbad, CA). Resveratrol was purchased fromAPP-Chem Bio (Xi’an, China).

Animals and treatments

Sprague-Dawley female and male rats were purchased from theSLAC laboratory Animal Co. Ltd. (Shanghai, China). Female ratsweighing 230–250 g and male rats weighing 280–350 g wereused. All animals were housed in a temperature (24–27 1C) andhumidity (60%) controlled animal room and maintained on a 12 hlight/12 h dark cycle (light from 8:00 to 20:00), with food andwater provided during the experiments. All of the procedureswere performed in accordance with the United States PublicHealth Services Guide for the Care and Use of Laboratory Animals,and all efforts were made to minimize both the suffering andnumber of animals used in this study. After 1 week of acclimatiza-tion, female rats were randomly divided into the following threegroups: Control, Stress, and Res treatment (100 mg/kg/day).

Fig. 2. Effects of AMPK on mitochondrial biogenesis and Phase II enzyme induction by Res in SH-SY-5Y cells. Cells were treated with Res at doses of 10, 50, and 100 μM for 24h, and AMPK activation was determined by measuring the expression of p-AMPK and AMPK (A, Western blot image; B, statistical analysis of p-AMPK; C, statistical analysis ofp-AMPK/AMPK ratio). Cells were treated with 100 μM Res, with or without Compound c (10 μmol/L) for 24 h, and the protein expression of p-AMPK, AMPK, PGC-1α, ComplexI, Nrf2, and HO-1 was determined by Western blotting (D, Western blot image; E, statistical analysis of PGC-1α; F, statistical analysis of Complex I; G, statistical analysis ofNrf2; H, statistical analysis of HO-1). Values are means 7 SEM from at least three independent experiments. nP o 0.05, nnP o 0.01.

K. Cao et al. / Free Radical Biology and Medicine 75 (2014) 156–166158

For each group, 14 female and 3 male rats were used. Virgin femalerats were placed with adult male rats (3:1) overnight for mating.Vaginal smears were examined on the following morning. The dayon which a smear was determined to be sperm-positive was set asembryonic day 0. Each pregnant rat was then housed separately,and Res gavages were implemented on a daily basis until baby ratswere born.

Stress procedure

Each group of pregnant rats, except the Control group, wasexposed to restraint stress on days 14–20 of pregnancy, threetimes each day and for 2 h each time. To prevent the habituation ofthe animals as a result of being subjected to daily procedures,restraint periods were randomly shifted to different time periods(08:00 AM to 11:00 AM, 11:00 AM to 02:00 PM, and 04:00 PM to07:00 PM). The restraint device was a transparent plastic tube(6.8 cm in diameter) with closed ends and air holes for breathing.The length could be adjusted to accommodate the size of theanimal. After birth, the offspring of all groups were housed in thesame animal room and kept together with their biologicalmothers. The pregnant rats of the Control group were leftundisturbed. On day 21, after all offspring were weaned, maleand female pups were housed separately until testing at 1 monthof age.

Morris water maze

The Morris water maze (MWM) was performed as previouslyreported [8]. The day before beginning the MWM test, each rat wasallowed to swim freely for 120 s in order to become familiar with thenovel environment of the maze and to locate and climb onto theescape platform. For four and half consecutive days, each animalperformed eight swimming trials per day, with two trials beginning ateach of the four start positions (N, E, S, W). The average result of thefour start positions was defined as the result for that session. The timelapse between 2 sessions was 6 h. The order of start positions wasrandomized each day, and across days, for all animals. Latency andswim patterns were digitally tracked by HVS Image Water 2020. Foreach trial, rats were allowed to swim until they reached the platformand climbed onto it, subject to a 120 s cutoff. Any animal that failedwas guided to find the platform and allowed to stay on it for 3 s. Thevariable recorded was escape latency, a measure of the time requiredfor an animal to reach the platform and remain there for more than3 s. For cases in which rats did not reach the platform, the latencyvalue was set at 120 s. Data were recorded and then analyzed byparametric ANOVA (repeated measures and multivariate measures).

T maze for spontaneous alternation performance

Spontaneous alternation is the natural tendency of rats andmice to alternate their choice of goal arm (left or right arm) during

Fig. 3. Effects of PRS and Res on the AMPK pathway in both male and female offspring hippocampus. Total proteins were prepared from offspring hippocampus, and p-AMPKand AMPK levels were detected by Western blot. In male offspring as follows: (A) Western blot image, (B) statistical analysis of p-AMPK, and (C) ratio of p-AMPK/AMPK; infemale offspring as follows: (D) Western blot image, (E) statistical analysis of p-AMPK, and (F) ratio of p-AMPK/AMPK. Values are means 7 SEM; n Z 8. nP o 0.05,nnP o 0.01.

K. Cao et al. / Free Radical Biology and Medicine 75 (2014) 156–166 159

exploration of a T maze. The size of the maze was based onparameters used for rats in a previous publication [25]. The ratswere confined to the start area for 30 s before testing. The slidingdoor was then removed, and the rats were allowed to freelyexplore the rest of the maze for an 8-min trial [26,27]. Numbers ofvisits to each goal arm and spontaneous alternation performancewere recorded by the observer. The apparatus was cleaned aftereach rat’s use to remove any residual odors and debris.

Cell culture and treatment

The human SH-SY-5Y cell line was obtained from ATCC (Man-assas, VA, USA). Cells were maintained in Dulbecco’s modifiedEagle’s medium (DMEM) supplemented with 10% (v/v) fetalbovine serum and maintained at 37 1C in a humidified atmosphereof 95% air and 5% CO2 by incubator. The culture medium waschanged every other day. Resveratrol was dissolved in dimethylsulfoxide (DMSO) at 100 mM and was stored at �20 1C. For

cellular treatment, resveratrol was dissolved in DMEM (with 10%fetal bovine serum) to reach a final concentration of 100 μM.

Biochemical analysis

Small portions of hippocampal tissue were collected andhomogenized in ice cold phosphate buffered saline (PBS). Aftercentrifugation (1000g, 10 min), the supernatant was collected foranalysis. GSH and GSSG content was analyzed using commercialclinical diagnosis kits according to the manufacturer’s standardsand protocols (Jiancheng, Nanjing, China). Briefly, total glutathione(oxidized þ reduced) assay is based on the reaction with the thiol-specific reagent dithionitrobenzoic acid after glutathione reduc-tase treatment to change GSSG to GSH. The adduct was measuredspectrophotometrically at 412 nm. GSSG assay is based on thesame reaction with a pretreatment for clearance of GSH. Thus, theGSH content is equal to total glutathione content minus GSSGcontent.

Fig. 4. Effects of PRS and Res on mitochondrial content in both male and female offspring hippocampus. Total proteins and DNAwere prepared from offspring hippocampus,and mitochondria subunit expression and mtDNA contents were detected by Western blot or real-time PCR. In male offspring as follows: (A) Western blot image,(B) statistical analysis of PGC-1α, Complex I, III, IV, and V subunits, and (C) mtDNA content; in female offspring as follows: (D) Western blot image, (E) statistical analysis ofPGC-1α, Complex I, and IV, and (F) mtDNA content. Values are means 7 SEM; n Z 8. nP o 0.05, nnP o 0.01.

K. Cao et al. / Free Radical Biology and Medicine 75 (2014) 156–166160

Protein carbonylation assay

Protein carbonyls in soluble proteins were assayed using theOxyblot protein oxidation detection kit (Cell Biolabs, San Diego,CA, USA). Protein carbonyls were labeled with 2,4-dinitrophenyl-hydrazine and detected by Western blot. As a negative loadingcontrol, equal amounts of samples were subjected to 10% SDS-PAGE and stained with Coomassie brilliant blue.

Real-time PCR

Total RNA was isolated from hippocampal samples using TRIzolReagent (Invitrogen, Carlsbad, CA, USA), and 1 μg of RNA wasreverse-transcribed into cDNA using a RT-PCR kit (TaKaRa, DaLian,China) according to the manufacturer’s protocol, followed bysemiquantitative real-time PCR with specific primers. For the

mtDNA copy analysis, total DNA was extracted using the QIAampDNA Mini Kit (Qiagen, Hilden, Germany), followed by quantitativereal-time PCR with mitochondrial D-loop primers. Data werenormalized to the mRNA of actin as a housekeeping gene andwere analyzed by the 2-△△Ct method. Final results were presentedas a percentage of the control. The specific primers used in thestudy are presented in Supplemental Table 1.

Western blotting

Hippocampal samples and SH-SY-5Y cellular samples werelysed with Western and IP lysis buffer (Beyotime, Jiangsu, China).The lysates were homogenized, and the homogenates were cen-trifuged at 13,000g for 15 min at 4 1C. The supernatants werecollected, and the protein concentrations were determined with aBCA protein assay kit. Equal aliquots (20 μg) of the protein samples

Fig. 5. Effects of PRS and Res on phase II enzymes in both male and female offspring hippocampus. Total RNA and proteins were prepared from offspring hippocampus, andphase II enzyme regulator Nrf2 and its target genes were analyzed. In male offspring as follows: (A) mRNA expression of Nrf2, NQO1, HO-1, GCLc, and GCLm, (C) Western blotimage, (D) statistical analysis of Nrf2, (E) statistical analysis of NQO-1, and (F) statistical analysis of HO-1; in female offspring as follows: (B) mRNA expression of Nrf2, NQO1,HO-1, GCLc, and GCLm, (G) Western blot image, (H) statistical analysis of Nrf2, (I) statistical analysis of NQO-1, and (J) statistical analysis of HO-1. Values are means 7 SEM;n Z 8. nP o 0.05, nnP o 0.01.

K. Cao et al. / Free Radical Biology and Medicine 75 (2014) 156–166 161

were loaded for Western blotting. Chemiluminescent detectionwas performed using an ECL Western blotting detection kit andquantified by scanning densitometry.

Statistical analysis

Data are presented as the means 7 SEM. Statistical analyseswere conducted using one-way ANOVA followed by least signifi-cant difference post hoc analysis. For water maze data, statisticalanalysis was conducted using two-way ANOVA repeated mea-sures. For all analyses, values of Po0.05 were considered statis-tically significant.

Results

Res promotes mitochondrial biogenesis and induces phase II enzymesin SH-SY-5Y cells

As an activator of AMPK, Res was used to treat SH-SY-5Y cellsfor 24 h. As illustrated in Fig. 1, Res may significantly increaseexpression of PGC-1α and Complex I and III subunits (Fig. 1A–D) in

a dose-dependent manner. Similar induction was observed withNrf2, the key regulator of phase II enzymes, as well as its targetgenes, NQO1 and HO-1 (Fig. 1E–H), indicating the simultaneousactivation of mitochondrial biogenesis and Nrf2 pathways byRes. In addition, Res may effectively ameliorate oxidative stressin SH-SY-5Y cells by decreasing both carbonyl protein (Fig. 1I and J)and GSSG content (Fig. 1K) in a dose-dependent manner. AlthoughGSH content was not affected after Res treatment (data notshown), GSH/GSSG ratio exhibited a dose-dependent increaseafter Res treatment, indicating an enhancement in antioxidantsystem (Fig. 1L).

AMPK inhibition diminishes the induction effect of Res onmitochondria and phase II enzymes

To further confirm that the induction effect of Res on mito-chondria and phase II enzymes was a result of AMPK activation,we first confirmed that Res may significantly induce AMPKactivation by increasing p-AMPK content and the p-AMPK/AMPKratio in a dose-dependent manner in SH-SY-5Y cells (Fig. 2A–C).The application of Compound C, a common AMPK inhibitor,sufficiently blocked the induction effect of Res, as evidenced by

Fig. 6. Effects of PRS and Res on oxidative stress in both male and female offspring hippocampus. Total proteins were prepared from offspring hippocampus, and carbonylprotein as marker of protein oxidation was measured by Western blot: (A) Western blot image and (B) statistical analysis in male offspring; (D) Western blot image and(E) statistical analysis in female offspring. Tissue homogenates were prepared from offspring: GSH and GSSG were measured, and the ratio of GSH/GSSG was determined inmale (C) and female (F) offspring. Values are means 7 SEM; n Z 8. nP o 0.05, nnP o 0.01.

K. Cao et al. / Free Radical Biology and Medicine 75 (2014) 156–166162

the resulting expression of PGC-1α, Complex I, Nrf2, and HO-1(Fig. 2D–H).

PRS suppresses AMPK activation in offspring hippocampus

To study the involvement of AMPK in prenatal restraint stress-induced offspring cognitive dysfunction, Res was supplementedduring the pregnancy at a dose of 100 mg/kg/day. As shown inFig. 3A, PRS induced significant declines in both p-AMPK contentand p-AMPK/AMPK ratio in the hippocampus of male offspring(Fig. 3A–C). In female offspring, PRS did not have a significanteffect on p-AMPK content compared with controls, whereas thep-AMPK/AMPK ratio was decreased (Fig. 3D–F). Supplementationof Res effectively increased both p-AMPK content and thep-AMPK/AMPK ratio in the hippocampus of both male and femaleoffspring (Fig. 3), suggesting significant activation of AMPK by Res.

Maternal Res supplementation inhibits PRS-induced mitochondrialloss in offspring hippocampus

To determine whether mitochondrial biogenesis was affected inthe PRS-induced cognitive impairment model, expression of PGC-1α,mitochondrial Complex subunits, and mtDNA content were exam-ined. As shown in Fig. 4A-C, PRS decreased protein expression ofPGC-1α, Complex I, III, IV, and V subunits in male offspring, as well asmtDNA content, indicating decreased mitochondrial content. Asexpected, Res significantly increased the expression of those mito-chondrial biogenesis-related proteins, as well as mtDNA content(Fig. 4A–C). In female offspring, Res demonstrated the ability tonormalize the decline of PGC-1α, Complex IV, and mtDNA contentinduced by PRS (Fig. 4D–F). Although PRS had no significant effect onComplex I, Res nonetheless markedly increased its expression(Fig. 4D and E). Moreover, PRS induced a decrease in the mRNAcontent of PGC-1α, PGC-1β, NRF1, NRF2α, NRF2β, TFAM, and TFB1Min both male and female offspring, an eventuality that could also benormalized by Res treatment (Fig. S1). Taken together, these dataindicated that PRS-induced mitochondrial loss may be prevented byRes supplementation.

Maternal Res supplementation activates Nrf2-mediated phase IIenzymes in offspring

In vitro results indicated that AMPK may regulate phase IIenzyme expression, and decreased AMPK activation was observedin PRS-treated offspring (Fig. 3). Therefore, we further evaluatephase II enzymes expression in offspring hippocampus. As shownin Figs. 5A and B, PRS significantly decreased the mRNA content ofNQO1, HO-1, GCLc, and GCLm in male offspring, as well as GCLm infemale offspring, whereas Res supplementation effectivelyincreased mRNA levels of Nrf2 and its target genes, NQO1, HO-1,GCLc, and GCLm, in both male and female offspring comparedwith a PRS control (Fig. 5A and B).Western blot results confirmedthat the protein levels of Nrf2, NQO-1, and HO-1 were all markedlyincreased in the Res-treated group compared to the PRS control(Fig. 5C–J), although the decline of these proteins induced by PRSwas not significant in female offspring (Fig. 5G–J). Taken together,these data suggested that Res may activate the Nrf2 pathway inthe offspring hippocampus.

Effects of PRS and Res on oxidative stress

Oxidative stress was suggested to be involved in PRS-inducedneurological damage and cognitive impairment [7,8], and mito-chondrial dysfunction and inhibition of phase II enzymes may eachcontribute to increased oxidative damage. As shown in Fig. 6, PRSinduced a significant increase in carbonyl protein (Fig. 6A and B)

and a decrease in the GSH/GSSG ratio (Fig. 6C) in male offspring,whereas no significant effects were observed in female offspring(Fig. 6D–F). Despite the difference between male and femaleoffspring, Res supplementation sufficiently decreased oxidativedamage, as evidenced by the decrease in carbonyl protein and theincreased GSH/GSSG ratio in both male and female offspringcompared to the PRS control group (Fig. 6).

Effects of PRS and Res on mRNA levels of neurotrophic factors

Oxidative damage is known to contribute to neuron loss andassociated disease progression. We therefore examined neuro-trophic factors in offspring, as their presence may partially reflectneuron condition. As shown in Fig. 7A and B, PRS significantlydecreased the mRNA content of several factors, including Arc, BDNF,EMX2, GAP43, NMDAR, and SCG10 in male offspring, as well as Arc,EMX2, and NMDAR in female offspring. Res treatment significantlyincreased the mRNA levels of each of those factors in both male andfemale offspring compared with PRS control groups.

Effects of PRS and Res on cognitive function

The T maze and Morris water maze tests were used to assessspatial learning and memory abilities in rat offspring. As shown inFig. 8A, B, D, and E, PRS caused decreased learning and memoryabilities compared to the T maze controls in both male and femaleoffspring, although the total amount of exploration in femaleoffspring was not significant compared with the control group

Fig. 7. Effects of PRS and Res on mRNA levels of neurotrophic factors in both male andfemale offspring hippocampus. Total RNA were prepared from offspring hippocampus:mRNA expressions of Arc, BDNF, EMX2, GAP43, NMDAR, and SCG10were determined byreal-time PCR in both male (A) and female (B) offspring. Values are means 7 SEM; n Z8. nP o 0.05, nnP o 0.01.

K. Cao et al. / Free Radical Biology and Medicine 75 (2014) 156–166 163

(Fig. 8D). Res sufficiently increased the total amount of and successrate of exploration of the T maze, indicating improved spatiallearning and memory abilities (Fig. 8A, B, and E). Similar to the Tmaze test, the results of the Morris water maze test showed thatPRS significantly increased escape latency time in both maleand female offspring (Fig. 8C and F). Res significantly decreasedescape latency time compared with the control group in both maleand female offspring (Fig. 8C and F). Taken together, these dataillustrated that the cognitive impairment induced by PRS wasefficiently ameliorated through maternal Res supplementation.

Discussion

Prenatal stress reportedly has many deleterious effects on thedevelopment and behavior of offspring [1,5,6]. Although oxidativestress and mitochondrial dysfunction were believed to be involvedin PRS-induced neurological damage and cognitive impairment[7,8], the details regarding specific mechanisms remain largelyunknown. Recent studies reveal AMPK, a major regulator ofcellular and whole-body energy homeostasis, as an attractivetherapeutic target for a range of diseases, including neurologicaldisorders [28,29]. Therefore, in the current study, we investigated

the involvement of AMPK in PRS-induced neurological impair-ment, and we proposed that AMPK may affect neuron survival andcognitive function through the modulation of mitochondrialbiogenesis and phase II enzyme expression.

As a well-known ATP producer, mitochondria are also a majorsource of ROS. Studies suggest that oxidative stress resulting frommitochondrial dysfunction may play an important role in neuro-logical impairment and cognitive decline [30–32]. Previously, wefound that PRS-induced cognitive decline was closely associatedwith a functional disorder in hippocampal mitochondria [8,33].Recent data indicate that the capacity to undergo mitochondrialbiogenesis, which is dysregulated in disease states, may play a vitalrole in determining neuron cell survival [34]. As a key regulator ofmitochondrial biogenesis, PGC-1α was believed activated byAMPK in rat visual cortical neurons, as previously reported [15].We consistently observed the dose-dependent induction ofPGC-1α through activation of AMPK by Res, a known naturalAMPK activator, accompanied by the increased presence of mito-chondrial complex subunits in SH-SY-5Y cells. Mitochondrialbiogenesis induction was efficiently abolished with an AMPKinhibitor, Compound c. In further animal analysis, we founddecreased p-AMPK/AMPK levels in the hippocampus of both maleand female offspring, suggesting the PRS-induced inactivation of

Fig. 8. Effects of PRS and Res on cognitive impairment in both male and female offspring. After baby rats were born and grown to 1 month of age, both male and femaleoffspring were tested for cognitive function. In male offspring as follows: (A) Total amount of exploration and (B) success rate of exploration in T maze, and (C) escape latencyin Morris water maze; in female offspring as follows: Total amount of exploration and (D) success rate of exploration in T maze, and (E) escape latency in Morris water maze(F). All groups of animals were able to learn the task through consecutive trials. The latency per testing session represents the average of four trials of all animals in eachgroup. Values are means 7 SEM. For T maze (n Z10) and for Morris water maze (n Z15). nP o 0.05, nnP o 0.01.

K. Cao et al. / Free Radical Biology and Medicine 75 (2014) 156–166164

AMPK. As a result, mitochondrial content was decreased, asevidenced by the decreased amount of mitochondrial complexsubunits and mtDNA copy number. However, via maternal sup-plementation of Res, increased AMPK activation was observed inoffspring, accompanied by increased mitochondrial biogenesis.Therefore, we proposed that modulation of AMPK activity mayaffect mitochondrial biogenesis both in vivo and in vitro.

Nrf2 is an antioxidant transcription factor mediating theexpression of phase II antioxidative enzymes, and has a widerange of activities in regulating redox state and energy metabolismin cells [16]. Upregulation of Nrf2 and its downstream genes (suchas NQO1, HO-1, and GCL) is usually considered evidence for Nrf2pathway activation. Recent studies indicate the existence of crosstalk between the AMPK and the Nrf2 pathways in Caenorhabditiselegans [17], human endothelial cells [18], and mammalian inflam-matory systems [19]. In the current study, we found that activationof AMPK may increase the protein content of Nrf2 and its targetgenes, NQO-1 and HO-1, in SH-SY-5Y cells in a dose-dependentmanner. Through maternal Res supplementation, significantincreases in both mRNA and protein content were observed inboth male and female offspring of the Res-treated group comparedwith the PRS-only group, suggesting the obvious activation ofNrf2-mediated phase II enzymes. Although Res is known as anactivator of the AMPK pathway, and previous studies have indi-cated the potential neuroprotective effects of Res both in vitro[35,36] and in vivo [37], the relationship between AMPK and Nrf2was not elucidated. Therefore, combining cellular and animal data,we proposed for the first time that AMPK may regulate neuronalphase II antioxidative enzyme expression, which may contribute toreduced oxidative damage.

Oxidative stress is usually defined as a state of imbalance favoringthe factors that generate reactive oxygen species (e.g., superoxide orhydroxyl radicals) and working against the factors that protectcellular macromolecules from these reactants, including antioxidants.An increasing number of studies have revealed the involvement ofoxidative stress in the pathogenesis of various diseases [38,39]. Ourprevious studies have shown that cognitive dysfunction was relatedto neuronal oxidative damage in the hippocampus, which mayaccount for impaired spatial learning and memory in stressedoffspring [8,32,33]. In the current study, we found that AMPKactivation induced both mitochondrial biogenesis and phase IIenzyme activation, accompanied by decreased protein oxidationand GSSG content in SH-SY-5Y cells. Although GSH content wasunchanged by Res treatment, GSH/GSSG ratio exhibited a dose-dependent increase by Res treatment. Consequently, maternal sup-plementation of Res may activate AMPK, induce mitochondrialbiogenesis, and trigger antixodative enzymes in offspring. As a result,significant reduction in carbonyl protein levels, as well an increasedGSH/GSSG ratio, was observed in Res-supplemented offspring. Even-tually, Res efficiently inhibited the decreased expression of neuron-related factors, indicating the promotion of neuron survival, andresults from both the T maze and the Morris water maze showedsignificant improvement in cognitive function facilitated by supple-mentation of Res. Collective data from the current study havesuggested a functional association between AMPK and oxidativestress through the modulation of mitochondrial content and the Nrf2pathway. However, the details regarding the mechanism of AMPK’smodulation of Nrf2 were not elucidated in this study, and whetherthere is potential cross talk between PGC-1α-mediated mitochon-drial biogenesis and Nrf2-mediated phase II enzyme activationshould be investigated further.

Our results showed that Res could activate AMPK pathways,and then may activate mitochondrial biogenesis and Nrf2 path-ways in both male and female offspring, although in femaleoffspring the difference between the Stress group and the Controlgroup was not so significant compared with the male offspring.

The similar activation of AMPK, mitochondrial biogenesis, andNrf2 pathways in both male and female offspring confirmed thecross talk among AMPK, mitochondrial biogenesis, and Nrf2 path-ways. As for the difference, it has been reported that estrogen isinvolved in protecting neural function and regulating learning andmemory mechanisms, and therefore may reduce the damagecaused by prenatal stress [40,41]. Thus, female offspring mayexhibit less sensitivity to prenatal stress.

In conclusion, our study demonstrated for the first time thatRes-mediated AMPK activation is closely associated with mito-chondrial biogenesis and Nrf2 pathways in both SH-SY-5Y cellsand PRS-induced rat offspring and that AMPK may work upstreamof these two pathways simultaneously. This finding provides theevidence necessary to further explore the mechanism underlyingthe neuroprotective effects of Res, as well as the subsequentprevention of PRS-induced cognitive impairment, and the crosstalk among energy homeostasis, mitochondrial dysfunction, andoxidative clearance. Understanding and rationally utilizing thisfunctional relationship may be helpful in the establishment ofeffective therapies for neurological diseases.

Acknowledgments

The authors are supported by the National Natural ScienceFoundation of China (81201023, 31370844, and 31070740), the 973program for young scientists (No. 2014CB548200), National"Twelfth Five-Year" Plan for Science & Technology Support(2012BAH30F03), the Fundamental Research Funds for the CentralUniversities, and the 985 and 211 projects of Xi’an JiaotongUniversity.

Appendix A. Supplementary Information

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.freeradbiomed.2014.07.029.

References

[1] Weinstock, M. Alterations induced by gestational stress in brain morphologyand behaviour of the offspring. Prog. Neurobiol. 65:427–451; 2001.

[2] O'Connor, T. G.; Heron, J.; Golding, J.; Glover, V. Maternal antenatal anxietyand behavioural/emotional problems in children: a test of a programminghypothesis. J. Child Psychol. Psychiatry 44:1025–1036; 2003.

[3] Beydoun, H.; Saftlas, A. F. Physical and mental health outcomes of prenatalmaternal stress in human and animal studies: a review of recent evidence.Paediatr. Perinat. Epidemiol. 22:438–466; 2008.

[4] Bustamante, C.; Bilbao, P.; Contreras, W.; Martinez, M.; Mendoza, A.; Reyes, A.;Pascual, R. Effects of prenatal stress and exercise on dentate granule cellsmaturation and spatial memory in adolescent mice. Int. J. Dev. Neurosci.28:605–609; 2010.

[5] Lemaire, V.; Koehl, M.; Le Moal, M.; Abrous, D. N. Prenatal stress produceslearning deficits associated with an inhibition of neurogenesis in the hippo-campus. Proc. Natl. Acad. Sci. USA 97:11032–11037; 2000.

[6] Zuena, A. R.; Mairesse, J.; Casolini, P.; Cinque, C.; Alema, G. S.; Morley-Fletcher, S.;Chiodi, V.; Spagnoli, L. G.; Gradini, R.; Catalani, A.; Nicoletti, F.; Maccari, S. Prenatalrestraint stress generates two distinct behavioral and neurochemical profiles inmale and female rats. PLoS One 3:e2170; 2008.

[7] Zhu, Z.; Li, X.; Chen, W.; Zhao, Y.; Li, H.; Qing, C.; Jia, N.; Bai, Z.; Liu, J. Prenatalstress causes gender-dependent neuronal loss and oxidative stress in rathippocampus. J. Neurosci. Res. 78:837–844; 2004.

[8] Feng, Z.; Zou, X.; Jia, H.; Li, X.; Zhu, Z.; Liu, X.; Bucheli, P.; Ballevre, O.; Hou, Y.;Zhang, W.; Wang, J.; Chen, Y.; Liu, J. Maternal docosahexaenoic acid feedingprotects against impairment of learning and memory and oxidative stress inprenatally stressed rats: possible role of neuronal mitochondria metabolism.Antioxid. Redox Signal. 16:275–289; 2012.

[9] Ye, J.; Keller, J. N. Regulation of energy metabolism by inflammation: afeedback response in obesity and calorie restriction. Aging (Albany NY)2:361–368; 2010.

[10] Kim, J. H.; Lee, J. O.; Lee, S. K.; Kim, N.; You, G. Y.; Moon, J. W.; Sha, J.; Kim, S. J.;Park, S. H.; Kim, H. S. Celastrol suppresses breast cancer MCF-7 cell viability via

K. Cao et al. / Free Radical Biology and Medicine 75 (2014) 156–166 165

the AMP-activated protein kinase (AMPK)-induced p53-polo like kinase 2(PLK-2) pathway. Cell Signal. 25:805–813; 2013.

[11] Li, Y. Y.; Yu, L. F.; Zhang, L. N.; Qiu, B. Y.; Su, M. B.; Wu, F.; Chen, D. K.; Pang, T.;Gu, M.; Zhang, W.; Ma, W. P.; Jiang, H. W.; Li, J. Y.; Nan, F. J.; Li, J. Novel small-molecule AMPK activator orally exerts beneficial effects on diabetic db/dbmice. Toxicol. Appl. Pharmacol. 273:325–334; 2013.

[12] Sasaki, H.; Asanuma, H.; Fujita, M.; Takahama, H.; Wakeno, M.; Ito, S.; Ogai, A.;Asakura, M.; Kim, J.; Minamino, T.; Takashima, S.; Sanada, S.; Sugimachi, M.;Komamura, K.; Mochizuki, N.; Kitakaze, M. Metformin prevents progression ofheart failure in dogs: role of AMP-activated protein kinase. Circulation119:2568–2577; 2009.

[13] Park, Y. J.; Ko, J. W.; Jang, Y.; Kwon, Y. H. Activation of AMP-activated proteinkinase alleviates homocysteine-mediated neurotoxicity in SH-SY5Y cells.Neurochem. Res. 38:1561–1571; 2013.

[14] Liu, F.; Benashski, S. E.; Persky, R.; Xu, Y.; Li, J.; McCullough, L. D. Age-relatedchanges in AMP-activated protein kinase after stroke. Age (Dordr.) 34:157–168;2012.

[15] Yu, L.; Yang, S. J. AMP-activated protein kinase mediates activity-dependentregulation of peroxisome proliferator-activated receptor gamma coactivator-1alpha and nuclear respiratory factor 1 expression in rat visual corticalneurons. Neuroscience 169:23–38; 2010.

[16] Vomhof-Dekrey, E. E.; Picklo Sr M. J. The Nrf2-antioxidant response elementpathway: a target for regulating energy metabolism. J. Nutr. Biochem.23:1201–1206; 2012.

[17] Onken, B.; Driscoll, M. Metformin induces a dietary restriction-like state andthe oxidative stress response to extend C. elegans health span via AMPK, LKB1,and SKN-1. PLoS One 5:e8758; 2010.

[18] Liu, X. M.; Peyton, K. J.; Shebib, A. R.; Wang, H.; Korthuis, R. J.; Durante, W.Activation of AMPK stimulates heme oxygenase-1 gene expression and humanendothelial cell survival. Am. J. Physiol. Heart Circ. Physiol 300:H84–H93; 2011.

[19] Mo, C.; Wang, L.; Zhang, J.; Numazawa, S.; Tang, H.; Tang, X.; Han, X.; Li, J.;Yang, M.; Wang, Z.; Wei, D.; Xiao, H. The crosstalk between Nrf2 and AMPKsignal pathways is important for the anti-inflammatory effect of berberine inLPS-stimulated macrophages and endotoxin-shocked mice. Antioxid. RedoxSignal. 20:574–588; 2014.

[20] Kim, M. Y.; Lim, J. H.; Youn, H. H.; Hong, Y. A.; Yang, K. S.; Park, H. S.; Chung, S.;Ko, S. H.; Shin, S. J.; Choi, B. S.; Kim, H. W.; Kim, Y. S.; Lee, J. H.; Chang, Y. S.;Park, C. W. Resveratrol prevents renal lipotoxicity and inhibits mesangial cellglucotoxicity in a manner dependent on the AMPK-SIRT1-PGC1alpha axis indb/db mice. Diabetologia 56:204–217; 2013.

[21] Tillu, D. V.; Melemedjian, O. K.; Asiedu, M. N.; Qu, N.; De Felice, M.; Dussor, G.;Price, T. J. Resveratrol engages AMPK to attenuate ERK and mTOR signaling insensory neurons and inhibits incision-induced acute and chronic pain. Mol.Pain 8:5; 2012.

[22] Sahu, S. S.; Madhyastha, S.; Rao, G. M. Neuroprotective effect of resveratrolagainst prenatal stress induced cognitive impairment and possible involve-ment of Na(þ), K(þ)-ATPase activity. Pharmacol. Biochem. Behav. 103:520–-525; 2013.

[23] Madhyastha, S.; Sekhar, S.; Rao, G. Resveratrol improves postnatal hippocam-pal neurogenesis and brain derived neurotrophic factor in prenatally stressedrats. Int. J. Dev. Neurosci. 31:580–585; 2013.

[24] Madhyastha, S.; Sahu, S. S.; Rao, G. Resveratrol for prenatal-stress-inducedoxidative damage in growing brain and its consequences on survival ofneurons. J. Basic Clin. Physiol. Pharmacol. :1–10; 2013.

[25] Deacon, R. M.; Rawlins, J. N. P. T-maze alternation in the rodent. Nat. Protoc.1:7–12; 2006.

[26] Fernandez, F.; Morishita, W.; Zuniga, E.; Nguyen, J.; Blank, M.; Malenka, R. C.;Garner, C. C. Pharmacotherapy for cognitive impairment in a mouse model ofDown syndrome. Nat. Neurosci. 10:411–413; 2007.

[27] Gué, M.; Bravard, A.; Meunier, J.; Veyrier, R.; Gaillet, S.; Recasens, M.;Maurice, T. Sex differences in learning deficits induced by prenatal stress injuvenile rats. Behav. Brain Res. 150:149–157; 2004.

[28] Paintlia, A. S.; Paintlia, M. K.; Mohan, S.; Singh, A. K.; Singh, I. AMP-activatedprotein kinase signaling protects oligodendrocytes that restore central ner-vous system functions in an experimental autoimmune encephalomyelitismodel. Am. J. Pathol. 183:526–541; 2013.

[29] Frosig, C.; Jensen, T. E.; Jeppesen, J.; Pehmoller, C.; Treebak, J. T.;Maarbjerg, S. J.; Kristensen, J. M.; Sylow, L.; Alsted, T. J.; Schjerling, P.;Kiens, B.; Wojtaszewski, J. F.; Richter, E. A. AMPK and insulin action—responses to ageing and high fat diet. PLoS One 8:e62338; 2013.

[30] Bishop, N. A.; Lu, T.; Yankner, B. A. Neural mechanisms of ageing and cognitivedecline. Nature 464:529–535; 2010.

[31] Gibson, G. E.; Starkov, A.; Blass, J. P.; Ratan, R. R.; Beal, M. F. Cause andconsequence: mitochondrial dysfunction initiates and propagates neuronaldysfunction, neuronal death and behavioral abnormalities in age-associatedneurodegenerative diseases. Biochim. Biophys. Acta 1802:122–134; 2010.

[32] Liu, J.; Head, E.; Gharib, A. M.; Yuan, W.; Ingersoll, R. T.; Hagen, T. M.;Cotman, C. W.; Ames, B. N. Memory loss in old rats is associated with brainmitochondrial decay and RNA/DNA oxidation: partial reversal by feedingacetyl-L-carnitine and/or R-alpha -lipoic acid. Proc. Natl. Acad. Sci. USA99:2356–2361; 2002.

[33] Feng, Z.; Jia, H.; Li, X.; Bai, Z.; Liu, Z.; Sun, L.; Zhu, Z.; Bucheli, P.; Ballevre, O.;Wang, J.; Liu, J. A milk-based wolfberry preparation prevents prenatal stress-induced cognitive impairment of offspring rats, and inhibits oxidative damageand mitochondrial dysfunction in vitro. Neurochem. Res. 35:702–711; 2010.

[34] Zhu, J.; Wang, K. Z.; Chu, C. T. After the banquet: mitochondrial biogenesis,mitophagy, and cell survival. Autophagy 9:1663–1676; 2013.

[35] Kim, D. W.; Kim, Y. M.; Kang, S. D.; Han, Y. M.; Pae, H. O. Effects of resveratroland trans-3,5,40-trimethoxystilbene on glutamate-induced cytotoxicity, hemeoxygenase-1, and sirtuin 1 in HT22 neuronal cells. Biomol. Ther. (Seoul) 20:306–312; 2012.

[36] Son, Y.; Byun, S. J.; Pae, H. O. Involvement of heme oxygenase-1 expression inneuroprotection by piceatannol, a natural analog and a metabolite of resver-atrol, against glutamate-mediated oxidative injury in HT22 neuronal cells.Amino Acids 45:393–401; 2013.

[37] Ren, J.; Fan, C.; Chen, N.; Huang, J.; Yang, Q. Resveratrol pretreatmentattenuates cerebral ischemic injury by upregulating expression of transcrip-tion factor Nrf2 and HO-1 in rats. Neurochem. Res. 36:2352–2362; 2011.

[38] Sims-Robinson, C.; Hur, J.; Hayes, J. M.; Dauch, J. R.; Keller, P. J.; Brooks, S. V.;Feldman, E. L. The role of oxidative stress in nervous system aging. PLoS One 8:e68011; 2013.

[39] Yavari, A.; Azizova, G. I. Effect of oxidative stress on immunological parametersin type 2 diabetes mellitus in the Azerbaijan Republic. Diabetes Metab. Syndr6:195–198; 2012.

[40] Suzuki, S.; Brown, C. M.; Dela Cruz, C. D.; Yang, E.; Bridwell, D. A.; Wise, P. M.Timing of estrogen therapy after ovariectomy dictates the efficacy of itsneuroprotective and antiinflammatory actions. Proc. Natl. Acad. Sci. USA104:6013–6018; 2007.

[41] Gulinello, M.; Lebesgue, D.; Jover-Mengual, T.; Zukin, R. S.; Etgen, A. M. Acuteand chronic estradiol treatments reduce memory deficits induced by transientglobal ischemia in female rats. Horm. Behav. 49:246–260; 2006.

K. Cao et al. / Free Radical Biology and Medicine 75 (2014) 156–166166