Experimental Neurology 261 (2014) 386–395

Contents lists available at ScienceDirect

Experimental Neurology

j ourna l homepage: www.e lsev ie r .com/ locate /yexnr

Regular Article

Secretions from placenta, after hypoxia/reoxygenation, can damagedeveloping neurones of brain under experimental conditions☆

Daniel J. Curtis a, Aman Sood a, Tom J. Phillips a, Veronica H.L. Leinster a, Akihiro Nishiguchi b,Christopher Coyle a, Lizeth Lacharme-Lora a, Oliver Beaumont a, HelenaKemp c, Roberta Goodall c, Leila Cornes c,Michele Giugliano d, Rocco A. Barone d, Michiya Matsusaki b, Mitsuru Akashi b, Hiroyoshi Y. Tanaka e,Mitsunobu Kano e, Jennifer McGarvey f, Nagaraj D. Halemani f, Katja Simon g, Robert Keehan a, William Ind a,Tracey Masters h, Simon Grant h, Sharan Athwal h, Gavin Collett i, Dionne Tannetta i, Ian L. Sargent i,Emma Scull-Brown j, Xun Liu j, Kristian Aquilina k, Nicki Cohen l, JonD. Lane f, Marianne Thoresen j, Jon Hanley f,Andrew Randall m, C. Patrick Case a,⁎a Musculoskeletal Research Unit, School of Clinical Sciences, Bristol University, Southmead Hospital, Bristol BS10 5NB, UKb Department of Applied Chemistry, Osaka University, Japanc Newborn Screening, Southmead Hospital, Bristol, UKd Biomedical Sciences, University of Antwerp, Belgiume Department of Molecular Pathology, University of Tokyo, Japanf School of Biochemistry, University of Bristol, UKg Weatherall Institute of Molecular Medicine, University of Oxford, Oxford, UKh Dept. Obstetrics and Gynaecology, Southmead Hospital, Bristol, UKi Nuffield Department of Obstetrics and Gynaecology, University of Oxford, Oxford, UKj Department of Neonatology, University of Bristol, Bristol, UKk Department of Neurosurgery, Frenchay Hospital, Bristol, UKl Department of Neuropathology, University of Bristol, Bristol, UKm School of Physiology and Pharmacology, University of Bristol, Bristol, UK

☆ The authors declare no competing financial interests.⁎ Corresponding author. Fax: +44 117 3238476.

E-mail address: c.p.case@bristol.ac.uk (C.P. Case).

http://dx.doi.org/10.1016/j.expneurol.2014.05.0030014-4886/© 2014 Elsevier Inc. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 6 November 2013Revised 25 April 2014Accepted 1 May 2014Available online 10 May 2014

Keywords:PlacentaNeuroneDendriteParvalbuminHypoxiaReoxygenationCerebral cortexDevelopmentNeurodevelopmental disorderSchizophrenia

Some psychiatric diseases in children and young adults are thought to originate from adverse exposures duringfoetal life, including hypoxia and hypoxia/reoxygenation. The mechanism is not understood. Several authorshave emphasised that the placenta is likely to play an important role as the key interface between mother andfoetus. Here we have explored whether a first trimester human placenta or model barrier of primary humancytotrophoblasts might secrete factors, in response to hypoxia or hypoxia/reoxygenation, that could damageneurones. We find that the secretions in conditioned media caused an increase of [Ca2+]i and mitochondrialfree radicals and a decrease of dendritic lengths, branching complexity, spine density and synaptic activity in dis-sociated neurones from embryonic rat cerebral cortex. There was altered staining of glutamate and GABA recep-tors.We identify glutamate as an active factorwithin the conditionedmedia and demonstrate a specific release ofglutamate from the placenta/cytotrophoblast barriers in vitro after hypoxia or hypoxia/reoxygenation. Injectionof conditionedmedia into developing brains of P4 rats reduced the numerical density of parvalbumin-containingneurones in cortex, hippocampus and reticular nucleus, reduced immunostaining of glutamate receptors andaltered cellular turnover. These results show that the placenta is able to release factors, in response to alteredoxygen, that can damage developing neurones under experimental conditions.

© 2014 Elsevier Inc. All rights reserved.

Introduction

Some human diseases originate from events during foetal life(Barker, 2004). A variety of prenatal insults, including hypoxia, hypoxia/

reoxygenation, and infection, as early as latefirst trimester, are associatedwith an increased risk of neurodevelopmental disorders including schizo-phrenia, attention deficit/hyperactivity disorder and autism (Fatemi andFolsom, 2009). In animal models even relatively brief periods of foetalhypoxia may lead to the death of susceptible neuronal populations(Rees et al., 2011). The principal pathways for this are initiated by energydepletion followed by an increased neuronal release and a reducedglial uptake of glutamate, an accumulation of cytosolic calcium and a

387D.J. Curtis et al. / Experimental Neurology 261 (2014) 386–395

generation of reactive oxygen species (Rees et al., 2011). The death ofneurones is thought to occur in three stages. There is an initial periodof cell dysfunction and oxidative stress. 8 to 48 h later there is a second-ary phase of injury, which results in a neuroinflammatory response,mitochondrial permeabilisation, reperfusion and a loss of cerebral auto-regulation.Weeks ormonths later theremay be tertiary injuries as a re-sult of a sensitization to inflammation, persistent gliosis and epigeneticchanges (Baburamani et al., 2012; Chicha et al., 2014).

However the precise mechanisms of how transient gestationalchallenges can lead to neurodevelopmental diseases in later life arelargely unknown, and it is thought that the placenta is likely to play akey role (Hsiao and Patterson, 2012; Rapoport et al., 2012). As Hsiaoand Patterson emphasise, disruptions to the maternal or intrauterineenvironment are necessarily conveyed to the developing embryovia the placenta. Where gestational challenges are confined to theuteroplacental compartment, inmodels of intrauterine infection and in-trauterine growth restriction, primary insults to placenta can manifestin perinatal brain damage in the offspring (Hsiao and Patterson, 2012;Mikaelsson et al., 2013).

The past 5 years have seen particular interest in the contribution ofplacental pathology to neurodevelopmental disorders (Rapoport et al.,2012). Furthermore the placenta actively secretes molecules that areimportant for infant brain development and which might be affectedby gestational challenges (Bonnin et al., 2011; McKay, 2011).

Previously we showed that the placenta or amodel placental barrierresponds to toxins or altered oxygen by secreting factors that cause ge-netic damage in fibroblasts or human embryonic stem cells (Bhabraet al., 2009; Sood et al., 2011). Therefore we explored here whether itwould also respond to altered oxygen by secreting factors that coulddamage developing neurones under experimental conditions.

We show that the placenta responds in vitro to hypoxia or hypoxia/reoxygenation by secreting factors that increase calcium andmitochon-drial free radicals in embryonic cortical neurones in vitro and reducesynaptic activity, dendritic length, branching complexity and spine den-sity. Exposure of a developing brain to media conditioned by placentaunder hypoxia and hypoxia reoxygenation results in decreased densityof parvalbumin containing neurones in cortex hippocampus and reticu-lar nucleus. We identify glutamate as an active factor mediating thesechanges.

Methods

Preparation of barriers/explants

Bilayered BeWo cell barriers were prepared on transwell inserts at2% or 21% oxygen according to our previous protocols (Sood et al.,2011). Human placenta from 1st and 3rd trimester was obtained withethical approval and patient consent from patients with normal preg-nancies at voluntary termination of pregnancy or elective caesareansection.

Primary villous cytotrophoblasts, were extracted from 3rd trimesterplacenta according to our previous protocols (Tannetta et al., 2008).They were used to create barriers using a modification of the cell-accumulation technique (Nishiguchi et al., 2011). This allows cell bar-riers to be made in 1 day by coating the cells with extracellular matrixnanofilms approximately 5 nm in thickness.

The primary cytotrophoblasts were alternately incubated nine timeswith 0.04 mg/ml type IV-collagen (Sigma) and laminin (Invitrogen) in50mMTris–HCl (pH=7.4) for 1min each. The collagen laminin coatedcells were seeded into transwell inserts and cultured under 2% and 21%oxygen conditions. After 1 day of incubation, the barriers were obtainedand subsequently cultured for 1 daywith or without a change of the ox-ygen conditions from 2% to 8% and 21%. Themedia beneath the barrierswere collected and exposed to neurones for 6 days. All barriers wereassessed with confocal imaging of E cadherin (Cell Signaling) immuno-cytochemistry or vital dye staining.

1st trimester placenta (mean 9 weeks 4 days gestation n = 20)wastransferred to the laboratory under atmospheric conditions and dis-sected into 2.5 × 2.5 mm pieces to cut across and expose placental villi.

Preparation of conditioned media

Neurobasal media or HEPES buffered saline were conditioned byplacing placental explants within them or prepared barriers abovethem at 37 °C in a Ruskin 5ive hypoxic chamber under 4 different oxygenconditions; 21% or 2% for 48 h, or transfer from 2% (24 h) to either 8% or21% (24 h) (Hung et al., 2004) (Fig. 1).

Neurone exposures

Neuronal cultureswere prepared from the cerebral cortex of embry-onic day 18 Wistar rat embryos as previously described (Gray et al.,2011). Following enzymatic and mechanical dissociation, cells werecounted and plated onto poly L-lysine-coated 13-mm coverslips at50,000–100,000 cells/coverslip, and cultured in DMEM supplementedwith 2% B27 (Gibco, Paisley, UK) and 1% penicillin/streptomycin.Neurones were left to settle out overnight in plating medium beforebeing transferred to neurobasal feeding medium. They were grown foreither 8 or 12 days, in case of critical periods of sensitivity (Metzger,2010), and exposed to filtered conditioned media for up to 6 days.

Ca2+-imaging

Changes in [Ca2+]i after immediate exposure to conditioned mediawere recorded with Fura 2 imaging according to our previous protocols(Sood et al., 2011).

Flow cytometry

Mitochondrial stainswere performed after 24 h of exposure by incu-bating at 37 °C for 30minwith either 100nMNaOor 5 μMMitoSOXRed(all from Molecular Probes, Invitrogen) and directly analysed withoutfixing by flow cytometry using a FACScalibur (BD). Annexin V-PE(BD Pharmingen) staining was performed according to themanufacturer's instructions followed by analysis on FACScalibur (BD).

Primary neuronal cultures on MEAs

Neurones dissociated from P0 Wistar rat cortices were plated (at3000 cells/mm2) and cultured on glass-substrate arrays of microelec-trodes (MEAs; Multichannel Systems, Germany) (Potter and DeMarse,2001). Prior to seeding, 12 MEAs were treated by polyethylenimine-laminin (10–0.02 mg/ml, respectively) and sister cultures were pre-pared and maintained on each of them at 37 °C, 5% CO2. During thefirst 8–12 days in vitro (DIVs), the culturemedium contained neurobasal,2% B-27 supplement, 1% L-glutamine, 1% penicillin–streptomycin, and10% horse serum (Gibco, Invitrogen Corporation), and was replacedthree times perweek. The 12 cultureswere thendivided into twogroups:one exposed to the medium conditioned at 2–8%, the other to a controlconditioned at 21%. Each group was further divided in two: one whereexposure occurred at 8 DIVs, the other at 12 DIVs.

Electrophysiological recordings

Spontaneous electrophysiological (multiunit) activity was moni-tored immediately after the media change, as well as one, four, andeight days after. Extracellular electric fields were monitored from upto 59 independent electrodes in eachMEAs, sampled at 25 kHz/channel,1200× amplified, bandpass-filtered (200–3000 Hz), and digitallyrecorded for 20 min per session. Simple spike-sorting, based on spike-unit polarity, was performed. Epochs of spontaneous synchronizedfiring across the MEA electrodes were identified over 25 ms bins by

Fig. 1. Schematic diagram showing the method used to prepare the conditioned media. For media conditioned by BeWo barriers (upper diagram), the BeWo cells were seeded ontotranswell inserts and grown under 21% oxygen or 2% oxygen for 7 days to form bilayered barriers (according to our previous protocols and as confirmed by Sood et al.(2011). On day8 new media were placed below the barrier and the oxygen was maintained at its previous concentration or changed to 8% or 21% for 24 h to give four types of conditioned media.This new media was collected on day 9 and snap frozen. For media conditioned by the placenta (lower diagram) the placenta was dissected into small explants and these explantswere placed in culture media at either 21% or 2% oxygen for 24 h. The next day the culture media was replaced and the oxygen level was either maintained or changed to 8% or 21%for 24 h to give four types of conditioned media. This new media was collected on day 3 and snap frozen.

388 D.J. Curtis et al. / Experimental Neurology 261 (2014) 386–395

standard procedures (Van Pelt et al., 2004), and their number, intervaldistribution, and duration were extracted from each recording session.Data analysis (MATLAB, The MathWorks, Inc., Natick, USA) involvedthe extraction and manipulation of spike times, deriving a variety ofindicators (Van Pelt et al., 2004) (e.g. the mean spiking rate, and thedegree of synchronization across MEA electrodes). Reported data areexpressed as the mean ± SD. The Kolmogorov–Smirnov test wasemployed to assess the significance of differences between cumulativedistributions of inter-burst intervals (*** p b 0.001). Statistics werenot performed on spiking activity as patterned spike trains are non-stationary and unbiased statistical indicators are much more relevantfor the synchronous elevation of the neuronal spiking rate. Note thatprimary neuronal cultures in vitro do not display the intrinsic electricalphenotype of bursting-neurones: therefore the synchronous episodicevents observed both in control and in condition result from recurrentglutamatergic interaction and from activity-dependent intrinsic or syn-aptic refractory processes.

Dendritic length

After exposure to conditioned media the neurones on coverslipswere fixed with 4% paraformaldehyde (PFA) for 20 min, washed inPBS and then stained with MAP2 1:2000 primary antibody (SynapticSystems) followed by goat anti-mouse 488 (1:500 Invitrogen) second-ary antibody and mounted with Vectashield/DAPI (Vector).

Using Image J software dendritic length and cell number wereassessed after staining of the neurones with DAPI or MAP 2 antibody.Cultured cortical neurones were transfected 2 days prior to treatmentswith plasmid encoding mCherry using Lipofectamine 2000 as previously

described (Nakamura et al., 2011). Dendritic complexity and spine num-ber were assessed using Sholl analysis after confocal imaging of theseneurones.

Receptors

Neurones on coverslips were stained for NMDAR1 (1:500Millipore),NMDAR3A (1:500 Millipore), GABA Aα1 (1:500 Abcam) and GABA B1(1:500 Abcam). Surface staining was achieved by fixating neuroneswith 4% PFA/5% sucrose for 3 min whereas total receptor staining wasachieved by fixing with supercold methanol before blocking andincubation in primary antibody overnight at 4 °C. Antibodies werevisualised using either goat anti rabbit 488 (1:500 Invitrogen) or goatanti mouse 488 (1:500 Invitrogen) before mounting in Vectashield.

Images were taken at ×64 (oil) after excitation with a fluorescentmicroscope at 488. Using Image J images were converted to RGB filesthen measurements were taken of the mean grey value of each imageproviding an average of the relative intensity of the staining. IBM SPSSwas then used to convert each condition set into a mean averagevalue that was presented in a histogram. Statistics were run using oneway ANOVA comparing each set with its control.

Cortical neurones were grown on 3 cm glass bottom dishes andimaged live with phase contrast using an Olympus IX-71 microscopedriven by MetaMorph software.

Cytotoxicity

Cytotoxicity and caspase activation were measured with ApoTox-Glo™ assay according to the manufacturer's instructions (Promega).

389D.J. Curtis et al. / Experimental Neurology 261 (2014) 386–395

Biochemical analysis

Levels of sodium, potassium, creatinine, urea, bicarbonate, glucose,lactate dehydrogenase, ammonia, total protein, albumin, amylase, li-pase, total cholesterol, triglycerides and ammonia in the conditionedmedia were analysed by standard laboratory methods on Cobas(r)6000 and 8000 analysers (Roche Diagnostics Ltd, Sussex, UK).Qualitative organic acid analysis was performed using the GCMSQP2010 Plusinstrument (Shimadzu UK Ltd., Milton Keynes, UK).Amino acids were analysed with the Waters Acquity UPLC systemusing the MassTrak Amino Acid Analysis Solution (WatersLtd.,Elstree, Herts UK). Butyrylcholinesterase activity was determined by astandard assay which uses butyrylthiocholine as the enzyme substrateand dithiobis(nitrobenzoate) (DTNB) as chromogenwith reagents sup-plied byRandox Laboratories Limited on anOlympusAU400 instrument(Beckman Coulter UK Ltd., High Wycombe, UK).

In vivo

All animal procedures were approved by ethic committee at Univer-sity of Bristol and conducted underHomeOffice animal research licence.Forty five Wistar rat pups of either sex, 4 days old (P4, day of birth P1),

Fig. 2. Changes in neurones (12 DIVs) after exposure to the conditionedmedia. A)Mean emissioneurones (×20 objective n=50) after exposure tomediawithout (control open bars) or with (field of view (n = 15) 6 days after exposure to conditioned medium using MAP 2 staining (3 pE–G) Representative images of mCherry transfected neurones. H–J) Neurones transfected withm298198+ MPEP) to conditionedmedia. H)Mean dendrite length per neurone (n=40). I) Sholl an = 3. *p b 0.05, ** p b 0.01, *** p b 0.001.

were randomised to 11 intervention groups. Rats were anaesthetizedwith isoflurane (3% induction and 1% maintenance) and N2O (1:1 N2Oto O2 ratio). A slow injection of either 40 μl or 80 μl of filtered condi-tioned mediumwas injected through a 27 G needle into the left ventri-cle (Aquilina et al., 2011) (frontal region). After injection, rats wereallowed to recover and sent back to the dam. Their brainswere harvest-ed at P10 after a trans-cardiac perfusionwith 4% PFA under anaesthesia.6 or 25 days later the rats were perfused under anaesthesia with 3%paraformaldehyde in 0.1 M phosphate buffer.

The brains were removed and processed for paraffin wax histology.Sections were cut at 3 μm and stained with haematoxylin and eosin oravitin biotin immunocytochemistry for parvalbumin (1/2000 Sigma)or GFAP (Abcam) using a full automated Bond 3 immunostainingmachine with bond polymer refined detection (Leica) to ensure thesame immunostainingmethodwas applied for all sections. Paraffin em-bedded sections were rehydrated and stained for NMDAR1 (1:500Millipore) and using citrate buffer antigen retrieval methods. Sectionswere incubated in primary antibody overnight at 4 °C. Antibodieswere visualised using goat anti rabbit 488 (1:500 Invitrogen) beforemounting in Vectashield.

Staining intensity was assessed using image J.All experiments were performed in triplicate.

n ratio for 340 and380 nmexcitation of Fura-2 (proportional to [Ca2+]i) forwholefields ofblack bars) placenta under different O2 conditions (n=3). B) Total length of dendrites perlacentae/oxygen condition). C–D) Representative images of neurones stained with MAP2.Cherry with or without adding a cocktail of glutamate antagonists (L689560+NBQX+ YMnalysis of neurite branching complexity. J)Mean number of spines per unit length of dendrite

390 D.J. Curtis et al. / Experimental Neurology 261 (2014) 386–395

Statistics

Statistical tests unless otherwise stated were performed using oneway ANOVA.

Results

Ca2+, ROS

The media conditioned by placenta after hypoxia/reoxygenationcaused a large rapid increase in intracellular [Ca2+]i in embryonic corti-cal neurones grown for 12 days before exposure (Fig. 2A) in comparisonto media conditioned by placenta at 21% oxygen without change. Therewere increased mitochondrial reactive oxygen species (SupplementaryFigs. 1A–E)without oxidative damage of the inner mitochondrial mem-brane or apoptosis (Supplementary Figs. 1F–I). Media conditioned by aBeWo cell barrier, amodel of the placenta (Bhabra et al., 2009), caused asimilar change (Supplementary Fig. 2).

Dendritic length

The media conditioned by placenta under hypoxia or hypoxia/reoxygenation caused a substantial reduction of dendritic length(Figs. 2B–H, Movies 1–3), dendritic branching complexity (Fig. 2I) anddendritic spine density (Fig. 2J). The dendrite length did not recoverby 6 days after replacement of conditionedmediawith normalmedium(Supplementary Fig. 3). There was little evidence of neuronal loss or al-teration in cell body diameter up to 6 days after exposure (Movies 1–2,

Fig. 3. Electrophysiology: Reduction of spike activity and its rate of transient network-wide syn(A) microelectrodes (MEAs), allowing non-invasive chronic monitoring of neuronal electr(spikes) across multiple electrodes within the array. The average rate of spiking activity (rate (D) (i.e., referred for brevity herein as bursts) were monitored following the acute resthereafter. Mean ± SD, *** p b 0.001, Kolmogorov–Smirnov test over the inter-burst cum

Supplementary Figs. 1J–O). No changewas seen after exposure tomediaconditioned by placenta at 21% oxygen.

Electrophysiology

There was a clear electrophysiological correspondence withthe changes in morphology. Media conditioned by placenta afterhypoxia/reoxygenation caused a reduction of spiking after exposure(Figs. 3A–C) compared tomedia conditioned by placenta at 21% oxygen.The irregularly recurring transient network-wide synchronous eleva-tion of the neuronal spiking rate of neuronal spiking activity was re-duced (Fig. 3D), suggesting a specific down-regulation of effectivesynaptic connectivity.

Composition of conditioned medium

In order to identify the active factor, we analysed the media condi-tioned by placenta with clinical assays relevant to brain damage, anoxiaand reoxygenation (Fanos et al., 2012). This revealed a strikingly specif-ic, large increase in glutamate in the conditionedmedia that had causeddendritic shortening compared to media conditioned at 21% oxygenalone (Table 1, Supplementary Fig. 4). There was a 2.7 fold increase inmedia conditioned under hypoxia (2%) and a 3.7 fold increase in mediaconditioned under 2–8% oxygen. There were small relative changes inother amino acids in media conditioned at 2%. There was no increase inlactate, LDH, electrolytes or glucose to suggest non-specific damage tothe placenta after altered oxygen (Table 1, Supplementary Fig. 4). Organicacidswere detected inmedia conditioned by placenta but not specifically

chronous elevation. Cortical neurones were cultured on substrate-integrated arrays ofical activity by means of (B) extracellular recordings of the firing of action potentialsC) and of the transient network-wide synchronous elevation of the neuronal spikingponse of the networks to the conditioned media application, as well as 1, 4, and 8 daysulative distributions.

Table 1Mean ± SDmeasurements of amino acids (μM/l) by ultrahigh performance liquid chromatography, organic acids (μM/l) by gas chromatography mass spectrometry, cholinesterase by aspectrometric method (units/l enzyme), electrolytes (μM/l) and other metabolites(μM/l) in the conditioned media surrounding human placenta exposed to different oxygen conditions.ND = not detected. *p b 0.05, **p b 0.01, ***p b 0.001 (shaded).

Oxygenexposure

Lactate

µmol/l

His Asn Taurine Ser Glutamine Arg Gly Asp Glutamate Thr

µmol/l µmol/l µmol/l µmol/l µmol/l µmol/l µmol/l µmol/l µmol/l µmol/l

Oxygenexposure

Ala

µmol/l

Pro Orn Cys Lys Tyr Met Val Ile Leu Phe

µmol/l µmol/l µmol/l umol/l µmol/l µmol/l µmol/l µmol/l µmol/l µmol/l

Oxygenexposure

Trp

µmol/l

Homocystine Citrulline Cystathionin Sarcosine Carnosine Ammonia Sodium Potassium Bicarbonate Chloride

µmol/l µmol/l µmol/l µmol/l µmol/l Arbitrary units mmol/l mmol/l mmol/l mmol/l

Oxygenexposure

Urea Creatinine Glucose LDH Protein Albumin Amylase Lipase Cholesterol Triglyceride

µmol/l mmol/l IU/l g/l g/l mmol/l

2%

21%

2-8%

2-21%

Control

2%

21%

2-8%

2-21%

Control

2%

21%

2-8%

2-21%

Control

2%

21%

2-8%

2-21%

Control

ND

ND

ND

ND ND ND ND

ND ND ND ND ND

ND ND ND ND ND

ND ND ND ND ND

ND ND ND

ND ND ND

ND ND ND

ND ND ND

ND ND ND

ND ND ND

ND ND

ND

ND

ND

ND ND

16.9 ± 5.1

14.7 ± 3.9

13.5 ± 0.6

12.7 ± 1.1

226 ± 35.5

202 ± 6.5

168 ± 28.7

185 ± 4.4

139 ± 27.4

275 ± 85.8

194 ± 21.0

219 ± 12.7

250 ± 43.2

22 ± 3.8

121 ± 16.2

104 ± 7.3

101 ± 12.6

119 ± 3.1

52 ± 9.8

54 ± 0.6

30 ± 4.9

44 ± 23.8

30 ± 8.5

<5 ± 0.0

32 ± 7.1

35 ± 4.6

24 ± 9.1

27 ± 7.6

<5 ± 0.0

44 ± 8.4

29 ± 9.4

31 ± 2.9

37 ± 3.0

6 ± 1.3

86.2 ± 12.5***

74.6 ± 2.9*

58.3 ± 11.1

52.7 ± 0.4

56.1 ± 11.1

1.662 ± 0.2

1.712 ± 0.2

1.601 ± 0.1

1.831 ± 0.1

0.785 ± 0.1

6 ± 0.6

6 ± 0.0

6 ± 0.0

5 ± 0.6

5 ± 0.0

17.7 ± 2.0

18.0 ± 2.3

16.7 ± 1.0

15.1 ± 0.4

24.4 ± 0.5

2 ± 0.6

2 ± 0.0

2 ± 0.6

1 ± 0.6

2 ± 0.0

4 ± 0.0

4 ± 0.7

4 ± 0.6

3 ± 0.6

3 ± 0.6

2 ± 1.5

2 ± 1.4

1 ± 1.0

0 ± 0.0

1 ± 0.6

0.3 ± 0.0

0.2 ± 0.1

0.2 ± 0.1

0.2 ± 0.1

0.0 ± 0.0

3 ± 1.5

6 ± 3.5

4 ± 0.6

5 ± 0.6

15 ± 1.5

80 ± 9.9

77 ± 0.7

72 ± 1.2

69 ± 3.2

70 ± 2.1

27 ± 7.2*

13 ± 2.4

19 ± 2.7

16 ± 6.8

<5 ± 0.0

453 ± 57.9**

382 ± 15.7*

318 ± 64.3

298 ± 4.5

302 ± 58.7

2263 ± 413.2

2223 ± 201.8

1648 ± 291.8

1911 ± 201.8

1771 ± 343.0

391 ± 62.8*

374 ± 10.6

277 ± 31.0

316 ± 16.2

296 ± 58.6

893 ± 75.5*

823 ± 10.9

715 ± 105.3

759 ± 16.9

679 ± 114.4

468 ± 72.8*

429 ± 5.8

358 ± 57.1

377 ± 13.2

307 ± 61.5

543 ± 76.6

491 ± 14.8

419 ± 62.0

470 ± 3.9

326 ± 63.5

211 ± 30.3

204 ± 19.3

157 ± 26.2

173 ± 5.8

135 ± 26.0

369 ± 76.5

142

562 ± 56.8

501 ± 226.5

104 ± 12.2

101 ± 2.1

95 ± 1.2

90 ± 3.8

92 ± 5.7

905 ± 115.4*

822 ± 10.6

678 ± 114.9

716 ± 8.0

636 ± 122.0

863 ± 110.0*

784 ± 10.8

640 ± 113.9

663 ± 4.9

630 ± 121.0

909 ± 116.9*

813 ± 20.1

675 ± 116.8

706 ± 7.0

633 ± 121.7

467 ± 70.3*

423 ± 7.9

343 ± 57.9

370 ± 7.8

309 ± 60.1

9 ± 3.3

10 ± 1.9

11 ± 3.0

13 ± 0.0

<5 ± 0.0

120 ± 15.4*

163 ± 43.1**

100 ± 24.8

45 ± 37.2

7 ± 0.8

917 ± 133.1*

837 ± 8.8

675 ± 115.9

726 ± 14.4

622 ± 122.2

391D.J. Curtis et al. / Experimental Neurology 261 (2014) 386–395

relating to any oxygen condition. Butyrylcholinesterase was not detected(Supplementary Table 1). Adding antagonists of glutamate NMDA,AMPA/kainate, and mGluR5 receptors (listed in order of potency) tomedia conditioned by placenta partly blocked intracellular Ca2+

transients (Fig. 4A) and dendritic retraction (Figs. 4B–C). A morecomplete block was seen with a cocktail of all antagonists or a gluta-mate scavenger. Tetrodotoxin was ineffective indicating no require-ment for synaptic activity. Glutamate application in normal mediumimitated the effects of medium conditioned by placenta. The mediaconditioned by placenta caused a reduction of GABA Aα1 (p b 0.05)and NMDAR1 (GluN1) (p b 0.05), an overall increase of NMDAR3A(GluN3) (p b 0.05), and no overall change in GABA B1 receptor immu-nostaining intensity (Figs. 4D–K). A similar patternwas seen after gluta-mate treatment (data not shown). Together, the experiments show thatglutamate is a major active component of the medium conditioned byplacenta.

Origin of signal

In order to determine whether the signal originated from thecytotrophoblast (i.e. facing the foetus), we prepared, for the first time,a confluent barrier of primary villous cytotrophoblast cells extractedfromhumanplacenta (Fig. 5A). Themedia conditioned by these barriersafter hypoxia/reoxygenation, but not hypoxia, caused shortening ofdendrites (Fig. 5B). Glutamate levels were higher in the conditionedmedia (2–21% (75 μM), 2–8% (80 μM) and 2% (68 μM)oxygen comparedto 21% (47 μM) oxygen).

The dendrite shortening and release of glutamate were reduced bytreating the cytotrophoblast barriers with antagonists of pannexin andconnexin 43 channels and glutamate transporters (not EAAT2, locatedon syncytiotrophoblast (i.e. facing the mother) (Noorlander et al.,2004)) (Figs. 5C–D). We compared the effects of direct hypoxia/reoxygenation (with normal glucose) of cortical neurones in culture

with the indirect effects exposing them tomedia conditioned byplacen-ta after hypoxia/reoxygenation. Dendrite retraction was only caused byexposure to conditioned media (Supplementary Fig. 5).

Effects in vivo

During infection/hypoxia the placentamay release reactive productswhich accumulate in the foetal circulation and cause foetal brain dam-age (Rees et al., 2011). Furthermore, under conditions of reoxygenationfollowing an episode of hypoxia, the foetal blood brain barrier becomesleaky potentially exposing the foetal brain to molecules within the foe-tal blood (Chen et al., 2012). It was therefore of interest to test whetherthe factors that were secreted by the placenta under hypoxia or hypoxiareoxygenation would have the potential to alter brain developmentin vivo, should the brain be exposed to them. To test this we carriedout intracerebroventricular injections of filtered conditioned mediainto neonatal rat brain. We used P4 rats as the earliest practical model,whose EEG pattern corresponds to a human foetus between 23 and30 weeks of gestation (Tucker et al., 2009). Although the brain archi-tecture was normal 6 days after injection, the media conditioned byplacenta under hypoxia caused increased proliferation and apoptosisin dentate gyrus and cerebral cortex (Fig. 6A). There was reducedparvalbumin staining in the ventrobasal complex of thalamus afterexposure to media conditioned by placenta or by model primarycytotrophoblast barriers under hypoxia/reoxygenation (Figs. 6B–J).There was decreased NMDAR1 staining in cortex (Figs. 6N–P), similarto the effects in vitro (Fig. 3E).

Twenty five days after injection there was a reduced numerical den-sity of parvalbumin-containing neurones. This wasmore pronounced inneurones that had not yet appeared at the time of the injection, inhippocampus and cortex (Figs. 6L–M), than in those that were alreadyestablished, in the reticular nucleus (Fig. 6K) (Solbach and Celio,1991). These data indicate that the placental trophoblast secretions

Fig. 4.Glutamate as amediator of increased [Ca2+]i and reduced dendrite lengths. A)Mean emission ratio for 340 and 380 nmexcitation of Fura-2 (proportional to [Ca2+]i). and B–C) totaldendrite lengths/field of view (MAP2 staining). The effect of glutamate antagonists or agonists is shown on dissociated cortical neurones (12 day growth (DIV)) exposed to mediaconditioned by placenta. B) 2–21%, C) 2–8% oxygen. GPT (glutamate transaminase) (10 units/ml), L689560 (NMDA antagonist) (10 μM) NBQX (AMPA/kainate antagonist) (10 μM) YM298198 (mGluR1 antagonist) (10 μM), MPEP (mGluR5 antagonist) (3 μM) cocktail (L689560+ NBQX+ YM 298198 + MPEP), or tetrodotoxin (TTX) (500 nM) was added to the condi-tioned media. Glutamate (150 μM) was added in normal medium) (n= 3mean ± sem). D–K) Images of immunostaining of NMDAR1, NMDAR3A, GABA Aα1 and GABA B1 receptors indissociated cells after exposure to media conditioned at 21% or 2–8% oxygen. *p b 0.05, ** p b 0.01, *** p b 0.001.

Fig. 5. Primary cytotrophoblast barriers. A) Primary cytotrophoblast barrier stained with E cadherin. B) Dendrite lengths (MAP2 staining) per field of view after exposure to media con-ditioned by primary human cytotrophoblast barrier. C–D) The effect of channel blockers on secretion from primary trophoblast barriers with altered oxygen (2–8%). C) Dendrite lengths,and D) glutamate levels. Gap26 (connexin43mimetic peptide) (150 μM), Panx1 (pannexin channel blocker) (0.5 mg/ml), DIDS (anion exchange inhibitor) (400 μM), DLTBoA (antagonistglutamate transporters EAAT1-5) (40 μM), dihydrokainic acid (EAAT2 antagonist) (200 μM). *p b 0.05, ** p b 0.01, *** p b 0.001 (data compared with 2–8% oxygen).

392 D.J. Curtis et al. / Experimental Neurology 261 (2014) 386–395

Fig. 6. Conditioned media alter brain development. (A) Number of apoptosis and mitoses in sections of left and right dentate gyri, hippocampus and cortex of brains 6 days afterbeing injected with (filtered) media conditioned by placenta compared to control media (open bars), n= 3. B) Optical density of parvalbumin staining in ventrobasal complex ofthalamus after injection of 40 μl (filtered) media conditioned by primary trophoblast barriers into neonatal ventricle (stats: compared with control saline injection). C–J)Parvalbumin labelling of (C–F) reticular nucleus and (C–J) adjacent ventrobasal complex, C–F) 6 days after injection of media conditioned with placenta, or G–J) 6 days afterinjection of media conditioned below trophoblast barriers (see B). K–M) Numerical density of parvalbumin containing neurones in (K) reticular nucleus, (L) hippocampusand (M) cortex (primary auditory, somatosensory and retrosplenal averaged) 25 days after injection and N–P) NMDAR1 staining in cortex 6 days after injection of media con-ditioned by placenta. *p b 0.05, **p b 0.01, ***p b 0.001.

393D.J. Curtis et al. / Experimental Neurology 261 (2014) 386–395

394 D.J. Curtis et al. / Experimental Neurology 261 (2014) 386–395

would have the potential to be able to alter neuronal developmentin vivo.

Discussion

The results show that the placenta responds in vitro to hypoxia orhypoxia/reoxygenation by secreting factorswhich cause dendrite short-ening, increased mitochondrial free radicals, increased calcium andaltered immunostaining of both glutamate andGABA receptors in disso-ciated neurones from embryonic cortex in vitro. We identify glutamateas a key active factor that is secreted from the placenta and show thatthis can be released from human cytotrophoblasts, which are knownto face the foetal blood vessels in a placenta. We show that the secre-tions from the placenta would have the potential to alter the devel-opment of a brain should it be directly exposed to the secretionsin vivo. Exposure to the conditioned media in vitro caused mildexcitotoxicity and only had effects on dendritic morphology and fir-ing properties of the neurones without affecting neuronal survival.Gorman and Docherty suggest that a similar process of dendrite re-traction and spine loss in vivo may protect neurones against apopto-sis but results in mood disorders (Gorman and Docherty, 2010).Exposure to conditioned media in vivo did lead to a long term lossof parvalbumin containing neurones. However this was more prom-inent in brain regions where such neurones had not yet appeared atthe time of exposure. This raises the possibility that this may havebeen more of a developmental than a cytotoxic effect. The increasedmitosis and apoptosis after exposure to media conditioned underhypoxia may imply that additional factors are present in this condi-tioned media.

Increased glutamate is present in the foetal and not maternalblood in several conditions in which there is an increased risk ofneurodevelopmental disorders. A 1.4 fold increase is seen aftermaternalexposure to alcohol (ethanol) or stress, which increase the risk of au-tism or attention deficit/hyperactivity disorder (Karl et al., 1995; Fryeret al., 2007; Class et al., 2013; Zhang et al., 2013). Ethanol induces oxida-tive stress in placental villi, reduces uteroplacental perfusion and causesplacental hypoxia (Bosco and Diaz, 2012). Maternal stress, particularlytraumatic stress, also reduces uteroplacental perfusion by activatingthe hypothalamus–pituitary–adrenal axis, causing high cortisol and cy-tokine release and diverting blood to maternal muscle (Sandman et al.,1994;Wallace et al., 2011). Evenmore striking than the increase of glu-tamate after maternal exposure to ethanol or stress, is the specific and 3fold increase of glutamate with a smaller 1.4 fold increase of ornithineand taurine in humanvenous and arterial foetal blood (but notmaternalblood) in gestational diabetes. Gestational diabetes increases the risk ofschizophrenia 7 fold (Van Lieshout and Voruganti, 2008) and causeschronic hypoxic stress in placenta (Cetin et al., 2005; Li et al., 2013). Itis interesting that this is almost exactly the same type and magnitudeof change of all amino acids as in our media conditioned by human pla-centa or cytotrophoblast barriers, after exposure to hypoxia or hypoxia/reoxygenation.

These neurodevelopmental disorders are characterised by similarchanges in neurones in the brain to what we have seen in our in exper-imental in vitro or in vivo systems.

A loss of parvalbumin-containing neurones in hippocampus andcortex and of NMDA receptors and a reduction of dendrite length,spine density and dendrite complexity of cortical neurones, are alsolisted as key changes in the brains of patients with schizophrenia(Fatemi and Folsom, 2009; Coyle, 2012) as well as in experimentalmodels of neurodevelopmental disorders using maternal hypoxia, alco-hol and/or stress (Gerstein et al., 2005; Jia et al., 2010; Samudio-Ruizet al., 2010).

The questions then arise as to whether any secretion of glutamatefrom the placenta would enter the foetal blood and would reach thefoetal brain directly or whether such glutamate would have the poten-tial for an indirect effect on foetal brain development. It is known for

example that hypoxia can damage the placenta without causingfoetalhypoxia (Stanek, 2013). Levels of foetal glutamate are known toincrease peripherally and in the brain after short term uterine hypoxiaand reoxygenation (Engidawork et al., 1997). The foetal brain is alsoknown to be particularly sensitive to glutamate so that even a short ex-posure will cause long term changes (Morrison et al., 2012). Thesequestions have been considered by other authors in other studies asfollows.

Under normal conditions glutamine passes from the maternal tofoetal circulation through the placenta and is converted to gluta-mate in the foetal liver. Glutamate in the foetal blood is efficientlytaken up by the placenta, limited only by the foetal supply, and ismetabolised to CO2 (Moores et al., 1994). This uptake is thoughtto protect against foetal brain injury (Danbolt, 2001; Noorlanderet al., 2004). The raised levels of glutamate in the foetal blood inthe conditions, which predispose to neurodevelopmental disor-ders, suggest a dysregulation at least of the placenta's protectivefunction. Astrocytes perform a similar protective role by taking upglutamate at a synapse. However under conditions of hypoxia, as-trocytes, like cardiac myocytes, secrete glutamate through connexinand pannexin channels and a reversal of glutamate transporters(Malarkey and Parpura, 2008). In our experiments applying blockersof these channels to the placenta prevented the increase of glutamatein the conditioned medium after hypoxia/reoxygenation and neuronaldamage.

Increasing the levels of systemic glutamate in the mother (outsidethe placenta) is known to cause a change in foetal brain receptors, neu-ronal survival and behaviour. According to some reports the glutamatepasses across the placenta to reach the foetal brain (Takasaki, 1978;Yu et al., 1997, 2006; López-Zapata et al., 2011). Injections of glutamateanalogues or NMDA antagonists or agonists into the maternal circu-lation will lead to neuronal degeneration and altered behaviour andparvalbumin staining in the brains of the offspring (Takai et al., 2003;Tanemura et al., 2009; Abekawa et al., 2011). Even in a neonate,whoseblood brain barrier may be more protective than a foetus (Baburamaniet al., 2012), an increase of glutamate in the peripheral blood will causealterations in cholinergic andNMDA receptors in the cerebral cortex, hip-pocampus and striatum, loss of dendritic arborisation and spine densityin the hippocampus and associated changes in behaviour (Beas-Zárateet al., 2002; Olvera-Cortés et al., 2005).

The effects of placental secretions could potentially be exacerbatedunder conditions of altered oxygen or infection/inflammation. Reoxy-genation after hypoxia causes a leak in the foetal blood brain barrierand a potentially increased exposure of foetal brain to molecules inthe blood (Chen et al., 2012). Such a leak could allow or alternativelyincrease the exposure of foetal brain to raised glutamate in gestationaldiabetes/maternal stress or alcohol. Furthermore as Rees and colleaguespoint out “fetal brain damage might arise not only because oxygendelivery to the brain is impaired but because of the accumulation ofreactive products in the fetal circulation released from the placentathat affect fetal cerebral vascular resistance and cerebral metabolism”

(Rees et al., 2011).The importance of the placenta in the aetiology of neuro-

developmental disorders has been emphasised by several authors(Rees et al., 2011; Hsiao and Patterson, 2012; Rapoport et al., 2012;Stolp et al., 2012; Mikaelsson et al., 2013).

It would therefore be of interest to investigate whether one ofthe mechanisms for this might be a lack of uptake (or secretion) of glu-tamate as a result of placental pathology, disease, hypoxia or hypoxia/reoxygenation (in preeclampsia). An additional and complimentarymechanism to our finding is suggested by the work of Bonnin andcolleagues,who show that disruption of placentalmetabolismof trypto-phan could prejudice the foetal forebrain's supply of 5-HT (Bonnin et al.,2011).

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.expneurol.2014.05.003.

395D.J. Curtis et al. / Experimental Neurology 261 (2014) 386–395

Acknowledgments

The financial support of the Perivoli Trust is gratefully acknowl-edged. MG acknowledges financial support from the EC-FP7 (MarieCurie Network “NAMASEN” grant n. 264872), the Research FoundationFlanders (grant n. G.0888.12N) and the Interuniversity Attraction PolesProgram (IUAP), initiated by the Belgian Science Policy Office.

References

Abekawa, T., Ito, K., Nakato, Y., Koyama, T., 2011. Developmental GABAergic deficit en-hances methamphetamine-induced apoptosis. Psychopharmacology (Berl) 215,413–427.

Aquilina, K., Chakkarapani, E., Love, S., Thoresen, M., 2011. Neonatal rat model of intraven-tricular haemorrhage and post-haemorrhagic ventricular dilatation with long-termsurvival into adulthood. Neuropathol. Appl. Neurobiol. 37, 156–165.

Baburamani, A.A., Ek, C.J., Walker, D.W., Castillo-Melendez, M., 2012. Vulnerability of thedeveloping brain to hypoxic–ischemic damage: contribution of the cerebral vascula-ture to injury and repair? Front. Physiol. 3, 424.

Barker, D.J., 2004. The developmental origins of adult disease. J. Am. Coll. Nutr. 23,588S–595S.

Beas-Zárate, C., Perez-Vega, M.I., Gonzalez-Burgos, I., 2002. Neonatal exposure tomonosodium L-glutamate induces cell death and cytoarchitectural impairments inhippocampal CA1 pyramidal neurons of adult rats. Brain Res. 952, 275–281.

Bhabra, G., Sood, A., Fisher, B., Cartwright, L., Saunders, M., Evans, W.H., Surprenant, A.,Lopez-Castejon, G., Mann, S., Davis, S.A., Hails, L.A., Ingham, E., Verkade, P., Lane, J.,Heesom, K., Newson, R., Case, C.P., 2009. Nanoparticles can cause DNA damage acrossa cellular barrier. Nat. Nanotechnol. 4, 876–883.

Bonnin, A., Goeden, N., Chen, K., Wilson, M.L., King, J., Shih, J.C., Blakely, R.D., Deneris, E.S.,Levitt, P., 2011. A transient placental source of serotonin for the fetal forebrain.Nature 472, 347–350.

Bosco, C., Diaz, E., 2012. Placental hypoxia and foetal development versus alcohol expo-sure in pregnancy. Alcohol Alcohol. 47, 109–117.

Cetin, I., de Santis, M.S., Taricco, E., Radaelli, T., Teng, C., Ronzoni, S., Spada, E., Milani, S.,Pardi, G., 2005. Maternal and fetal amino acid concentrations in normal pregnanciesand in pregnancies with gestational diabetes mellitus. Am. J. Obstet. Gynecol. 192,610–617.

Chen, X., Threlkeld, S.W., Cummings, E.E., Juan, I., Makeyev, O., Besio, W.G., Gaitanis, J.,Banks, W.A., Sadowska, G.B., Stonestreet, B.S., 2012. Ischemia–reperfusion impairsblood–brain barrier function and alters tight junction protein expression in theovine fetus. Neuroscience 226, 89–100.

Chicha, L., Smith, T., Guzman, R., 2014. Stem cells for brain repair in neonatal hypoxia–ischemia. Childs Nerv. Syst. 30, 37–46.

Class, Q.A., Abel, K.M., Khashan, A.S., Rickert, M.E., Dalman, C., Larsson, H., Hultman, C.M.,Långström, N., Lichtenstein, P., D'Onofrio, B.M., 2013. Offspring psychopathology fol-lowing preconception, prenatal and postnatal maternal bereavement stress. Psychol.Med. 17, 1–14.

Coyle, J.T., 2012. NMDA receptor and schizophrenia: a brief history. Schizophr. Bull. 38,920–926.

Danbolt, N.C., 2001. Glutamate uptake. Prog. Neurobiol. 65, 1–105.Engidawork, E., Chen, Y., Dell'Anna, E., Goiny, M., Lubec, G., Ungerstedt, U., Andersson, K.,

Herrera-Marschitz, M., 1997. Effect of perinatal asphyxia on systemic and intracere-bral pH and glycolysis metabolism in the rat. Exp. Neurol. 145, 390–396.

Fanos, V., Antonucci, R., Barberini, L., Noto, A., Atzori, L., 2012. Clinical application of meta-bolomics in neonatology. J. Matern. Fetal Neonatal Med. 25 (Suppl. 1), 104–109.

Fatemi, S.H., Folsom, T.D., 2009. The neurodevelopmental hypothesis of schizophrenia,revisited. Schizophr. Bull. 35, 528–548.

Fryer, S.L., McGee, C.L., Matt, G.E., Riley, E.P., Mattson, S.N., 2007. Evaluation of psycho-pathological conditions in children with heavy prenatal alcohol exposure. Pediatrics119, e733–e741.

Gerstein, M., Huleihel, M., Mane, R., Stilman, M., Kashtuzki, I., Hallak, M., Golan, H., 2005.Remodeling of hippocampal GABAergic system in adult offspring after maternal hyp-oxia and magnesium sulfate load: immunohistochemical study. Exp. Neurol. 196,18–29.

Gorman, J.M., Docherty, J.P., 2010. A hypothesized role for dendritic remodeling in the eti-ology of mood and anxiety disorders. J. Neuropsychiatry Clin. Neurosci. 22, 256–264.

Gray, E., Ginty, M., Kemp, K., Scolding, N., Wilkins, A., 2011. Peroxisome proliferator-activatedreceptor-α agonists protect cortical neurons from inflammatory mediatorsand improve peroxisomal function. Eur. J. Neurosci. 33, 1421–1432.

Hsiao, E.Y., Patterson, P.H., 2012. Placental regulation of maternal–fetal interactions andbrain development. Dev. Neurobiol. 72, 1317–1326.

Hung, T.H., Charnock-Jones, D.S., Skepper, J.N., Burton, G.J., 2004. Secretion of tumor ne-crosis factor-alpha from human placental tissues induced by hypoxia–reoxygenationcauses endothelial cell activation in vitro: a potential mediator of the inflammatoryresponse in preeclampsia. Am. J. Pathol. 164, 1049–1061.

Jia, N., Yang, K., Sun, Q., Cai, Q., Li, H., Cheng, D., Fan, X., Zhu, Z., 2010. Prenatal stress causesdendritic atrophy of pyramidal neurons in hippocampal CA3 region by glutamate inoffspring rats. Dev. Neurobiol. 70, 114–125.

Karl, P.I., Kwun, R., Slonim, A., Fisher, S.E., 1995. Ethanol elevates fetal serum glutamatelevels in the rat. Alcohol. Clin. Exp. Res. 19, 177–181.

Li, H.P., Chen, X., Li, M.Q., 2013. Gestational diabetes induces chronic hypoxia stress andexcessive inflammatory response in murine placenta. Int. J. Clin. Exp. Pathol. 6,650–659.

López-Zapata, A., León, D., Castillo, C.A., Albasanz, J.L., Martín, M., 2011. Maternal gluta-mate intake during gestation and lactation regulates adenosine A and A(2A) recep-tors in rat brain from mothers and neonates. Neuroscience 199, 133–142.

Malarkey, E.B., Parpura, V., 2008. Mechanisms of glutamate release from astrocytes.Neurochem. Int. 52, 142–154.

McKay, R., 2011. Developmental biology: remarkable role for the placenta. Nature 472,298–299.

Metzger, F., 2010. Molecular and cellular control of dendrite maturation during brain de-velopment. Curr. Mol. Pharmacol. 3, 1–11.

Mikaelsson, M.A., Constância, M., Dent, C.L., Wilkinson, L.S., Humby, T., 2013. Placentalprogramming of anxiety in adulthood revealed by Igf2-null models. Nat. Commun.4, 2311.

Moores Jr., R.R., Vaughn, P.R., Battaglia, F.C., Fennessey, P.V., Wilkening, R.B., Meschia, G.,1994. Glutamate metabolism in fetus and placenta of late-gestation sheep. Am. J.Physiol. 267, R89–R96.

Morrison, G., Fraser, D.D., Cepinskas, G., 2012. Mechanisms and consequences of acquiredbrain injury during development. Pathophysiology 2013 (20), 49–57.

Nakamura, Y., Wood, C.L., Patton, A.P., Jaafari, N., Henley, J.M., Mellor, J.R., Hanley, J.G.,2011. PICK1 inhibition of the Arp2/3 complex controls dendritic spine size and synap-tic plasticity. EMBO J. 30, 719–730.

Nishiguchi, A., Yoshida, H., Matsusaki, M., Akashi, M., 2011. Rapid construction of three-dimensional multilayered tissues with endothelial tube networks by the cell-accumulation technique. Adv. Mater. 23, 3506–3510.

Noorlander, C.W., de Graan, P.N., Nikkels, P.G., Schrama, L.H., Visser, G.H., 2004. Distribu-tion of glutamate transporters in the human placenta. Placenta 25, 489–495.

Olvera-Cortés, E., López-Vázquez, M.A., Beas-Zárate, C., González-Burgos, I., 2005. Neona-tal exposure to monosodium glutamate disrupts place learning ability in adult rats.Pharmacol. Biochem. Behav. 82, 247–251.

Potter, S., DeMarse, T., 2001. A new approach to neural cell culture for long-term studies.J. Neurosci. Methods 110, 17–24.

Rapoport, J.L., Giedd, J.N., Gogtay, N., 2012. Neurodevelopmental model of schizophrenia:update 2012. Mol. Psychiatry 17, 1228–1238.

Rees, S., Harding, R.,Walker, D., 2011. The biological basis of injury and neuroprotection inthe fetal and neonatal brain. Int. J. Dev. Neurosci. 29, 551–563.

Samudio-Ruiz, S.L., Allan, A.M., Sheema, S., Caldwell, K.K., 2010. Hippocampal N-methyl-D-aspartate receptor subunit expression profiles in a mouse model of prenatal alco-hol exposure. Alcohol. Clin. Exp. Res. 34, 342–353.

Sandman, C.A., Wadhwa, P.D., Dunkel-Schetter, C., Chicz-DeMet, A., Belman, J., Porto, M.,Murata, Y., Garite, T.J., Crinella, F.M., 1994. Psychobiological influences of stress andHPA regulation on the human fetus and infant birth outcomes. Ann. N. Y. Acad. Sci.739, 198–210.

Solbach, S., Celio, M.R., 1991. Ontogeny of the calcium binding protein parvalbumin in therat nervous system. Anat. Embryol. 184, 103–124.

Sood, A., et al., 2011. Signalling of DNA damage and cytokines across cell barriers exposedto nanoparticles depends on barrier thickness. Nat. Nanotechnol. 6, 824–833.

Stanek, J., 2013. Hypoxic patterns of placental injury: a review. Arch. Pathol. Lab. Med.137, 706–720.

Stolp, H., Neuhaus, A., Sundramoorthi, R., Molnár, Z., 2012. The long and the short of it:gene and environment interactions during early cortical development and conse-quences for long-term neurological disease. Front. Psychiatry 3, 50.

Takai, H., Katayama, K., Yasoshima, A., Uetsuka, K., Nakayama, H., Doi, K., 2003. NMDA-induced apoptosis in the developing rat brain. Exp. Toxicol. Pathol. 55, 33–37.

Takasaki, Y., 1978. Studies on brain lesions after administration ofmonosodium L-glutamateto mice. II. Absence of brain damage following administration of monosodiumL-glutamate in the diet. Toxicology 9, 307–318.

Tanemura, K., Igarashi, K., Matsugami, T.R., Aisaki, K., Kitajima, S., Kanno, J., 2009. J.Toxicol. Sci. 34 (Suppl. 2), SP279–SP286.

Tannetta, D.S., Sargent, I.L., Linton, E.A., Redman, C.W., 2008. Vitamins C and E inhibit ap-optosis of cultured human term placenta trophoblast. Placenta 29, 680–690.

Tucker, A.M., Aquilina, K., Chakkarapani, E., Hobbs, C.E., Thoresen, M., 2009. Developmentof amplitude-integrated electroencephalography and interburst interval in the rat.Pediatr. Res. 65, 62–66.

Van Lieshout, R.J., Voruganti, L.P., 2008. Diabetes mellitus during pregnancy and increasedrisk of schizophrenia in offspring: a review of the evidence and putative mechanisms.J. Psychiatry Neurosci. 33, 395–404.

Van Pelt, J., Wolters, P., Corner, M., Rutten, W., Ramarkers, G., 2004. Long-term character-ization of firing dynamics of spontaneous bursts in cultured neural networks. IEEETrans. Biomed. Eng. 51, 2051–2062.

Wallace, G.H., Arellano, J.M., Gruner, T.M., 2011. Non-syndromic cleft lip and palate: couldstress be a causal factor? Women Birth 24, 40–46.

Yu, T.X., Zhao, Y., Shi, W.C., Ma, R.D., Yu, L.J., 1997. Effects of maternal oral administrationof monosodium glutamate at a late stage of pregnancy on developing mouse fetalbrain. Brain Res. 747, 195–206.

Yu, L., Zhang, Y., Ma, R., Bao, L., Fang, J., Yu, T., 2006. Potent protection of ferulic acidagainst excitotoxic effects of maternal intragastric administration of monosodiumglutamate at a late stage of pregnancy on developing mouse fetal brain. Eur.Neuropsychopharmacol. 16, 170–177.

Zhang, Y., Zou, S., Zhang, X., Zhang, Y., 2013. Correlation of domestic violence during preg-nancy with plasma amino-acid neurotransmitter, cortisol levels and catechol-o-methyltransferase Val(158)Met polymorphism in neonates. Asia Pac Psychiatry 5,2–10.