NMDA Receptor Hypofunction in the Dentate Gyrus and ImpairedContext Discrimination in Adult Fmr1 Knockout Mice

Brennan D. Eadie,1,2,3 Jesse Cushman,4 Timal S. Kannangara,2,3

Michael S. Fanselow,4 and Brian R. Christie2,3,5,6*

ABSTRACT: Fragile X syndrome (FXS) is the most common form ofinherited intellectual disability in humans. This X-linked disorder is causedby the transcriptional repression of a single gene, Fmr1. The loss of Fmr1transcription prevents the production of Fragile X mental retardation pro-tein (FMRP) which in turn disrupts the expression of a variety of key synap-tic proteins that appear to be important for intellectual ability. A clear linkbetween synaptic dysfunction and behavioral impairment has been elusive,despite the fact that several animal models of FXS have been generated.Here we report that Fmr1 knockout mice exhibit impaired bidirectionalsynaptic plasticity in the dentate gyrus (DG) of the hippocampus. Thesedeficits are associated with a novel decrease in functional NMDARs (N-methyl-D-aspartate receptors). In addition, mice lacking the Fmr1 geneshow impaired performance in a context discrimination task that normallyrequires functional NMDARs in the DG. These data indicate that Fmr1 de-letion results in significant NMDAR-dependent electrophysiological and be-havioral impairments specific to the DG. VVC 2010 Wiley-Liss, Inc.

KEY WORDS: dentate gyrus; fragile X syndrome; synapse; long-termpotentiation; long-term depression

INTRODUCTION

Fragile-X syndrome (FXS) is a form of mental retardation that is ini-tiated by cytosine-guanine-guanine (CGG) trinucleotide repeat expan-sion (>200) and is associated with autism in males. The trinucleotideexpansion causes hypermethylation of the human Fmr1 gene (Turneret al., 1996), resulting in complete transcriptional repression, and loss ofproduction of the fragile X mental retardation protein (FMRP; Pierettiet al., 1991; Verkerk et al., 1991; Ashley et al., 1993). This protein isexpressed throughout the body, but is highly expressed in dentate gyrus(DG) granule neurons (Hinds et al., 1993). FMRP itself can shuttle sev-eral mRNAs from the nucleus to the cytoplasm and then to distal post-synaptic sites (Zalfa et al., 2007; Bassell and Warren, 2008). Among themRNA that FMRP normally binds to are several that are important for

synaptic plasticity, including those that code forNMDAR (N-methyl-D-aspartate receptor) subunitsNR1 and NR2B, and components of the postsynapticdensity, such as PSD-95 and CaMKIIa (Brown et al.,2001; Darnell et al., 2004; Zalfa et al., 2007; Schuttet al., 2009). NMDAR subunits, in particular, arethought to play a vital role in learning and memoryprocesses (Bliss and Collingridge, 1993). Thus it seemslogical that a loss of FMRP should alter the expressionof synaptic proteins integral to synaptic plasticity andlead to cognitive deficits.

Surprisingly, it has proven difficult to elucidate a ro-bust impairment in learning and memory in mice thatlack FMRP (1994; D’Hooge et al., 1997; Yan et al.,2004; Eadie et al., 2009). Impairments in synaptic plas-ticity have been equally elusive (Godfraind et al., 1996;Huber et al., 2002), with the exception of an increase inmGluR (metabotropic glutamate receptor)-mediatedLTD in the CA1 subfield of the hippocampus (Huberet al., 2002; Bear et al., 2004). While this does indicatethat some learning and memory mechanisms could bealtered, a link between mGluR-mediated LTD and anyintellectual impairment in these mice remains unclear(Dolen et al., 2007). Indeed, adult Fmr1 KO mice shownormal learning in the Morris water maze, a task de-pendent on the CA1 subfield of the hippocampus(D’Hooge et al., 1997; Eadie et al., 2009).

Recently, a number of studies have indicated thatdeveloping neurons in the brain may be dispropor-tionately affected by a loss of FMRP expression. Sig-nificant structural deficits are selectively apparent inthe dendrites of young neurons in the neocortex ofFmr1 KO mice, and the transient nature of thesespine abnormalities suggests that this could be a defi-cit that might impact learning capacity during earlycritical periods (Nimchinsky et al., 2001). Similarly,young cultured hippocampal neurons exhibit markeddecreases in morphological complexity (Braun andSegal, 2000; Castren et al., 2005), and recently it hasbeen reported that very young Fmr1 KO mice showrobust impairments in CA1 synaptic plasticity (Huet al., 2008; Pilpel et al., 2009). While deficits in theCA1 subfield of the adult hippocampus have been dif-ficult to document, we have recently shown that theprocess of neurogenesis in the adult DG is abnormalin mice that lack FMRP (Eadie et al., 2009). Thesenewly generated neurons contribute disproportionately

1MD/PhD Program, University of British Columbia, Vancouver, BC,Canada; 2Graduate Program in Neuroscience, University of British Co-lumbia, Vancouver, BC, Canada; 3Department of Cellular and Physio-logical Sciences, University of British Columbia, Vancouver, BC, Can-ada; 4Department of Biology, University of Victoria, Victoria, BC, Can-ada; 5Department of Psychology and Brain Research Institute,University of California, Los Angeles; 6Division of Medical Sciences,University of Victoria, Victoria, BC, CanadaGrant sponsors: SRCFC, FXRFC, CIHR.*Correspondence to: Brian R. Christie, Ph.D., Division of Medical Scien-ces, University of Victoria, 3800 Finnerty Rd, Victoria, BC, Canada V8P5C2. E-mail: brain64@uvic.caAccepted for publication 19 August 2010DOI 10.1002/hipo.20890Published online in Wiley Online Library (wileyonlinelibrary.com).

HIPPOCAMPUS 00:000–000 (2010)

VVC 2010 WILEY-LISS, INC.

to NMDAR-dependent synaptic plasticity in the DG (van Praaget al., 1999; Farmer et al., 2004; Ge et al., 2008). Despite this,investigations into synaptic plasticity in the DG of Fmr1 KOmice have been lacking, as have an assessment of the putative be-havioral ramifications of impaired structural and functional plastic-ity in the DG of Fmr1 KO mice. The present work demonstratesthat a loss of FMRP in the mammalian brain alters NMDAR-de-pendent synaptic plasticity in the DG. In addition, we show thatcontext discrimination, a behavior reliant on intact NMDARfunction in the DG (McHugh et al., 2007; pp 1640), is alteredin Fmr1 KO mice. Therefore, this study introduces NMDARs inthe DG as a novel target in the pathology of FXS.

MATERIALS AND METHODS

Animals

Eighty-six C57BL/6, male Fmr1 knockout (KO) (n 5 42)and Fmr1 wild-type (WT) (n 5 44) littermate mice were pro-duced by breeding WT male C57BL/6 (Jackson Laboratories)mice with female C57BL/6 mice heterozygous for the Fmr1gene. All animals were sexed, weaned, ear-punched, and tail-snipped at postnatal day 24 and group-housed with minimalenrichment (tubes and/or nesting materials). Adult male mice(2–4 months) were randomly assigned to electrophysiology orbehavioral experiments (Fig. 1). Experiments were carried outin accordance with international standards on animal welfareand guidelines set out by the Canadian Council on AnimalCare and the Universities of British Columbia and Victoria.

Genotyping

DNA extraction and purification

For DNA extraction and purification, tissue from each ani-mal was placed in 180 ll digestion buffer and 20 ll Proteinase

K in a DNAse/RNase-free 1.5 ml Eppendorf tube and incu-bated overnight in a thermomixer at 558C while being agitatedat 300 RPM. The sample was centrifuged at 21,000 RCF for 3min and the supernatant was transferred into a new tube.About 20 ll RNase A was then added, vortexed, and incubatedat room temperature for 2 min. First 200 ll lysis buffer andthen 200 ll 100% EtOH was added to each tube and vor-texed. The lysate was transferred to a clean spin column andcentrifuged at 10,000 RCF for 1 min at room temperature.The spin column was then placed in a fresh tube and washedby adding 500 ll Wash Buffer I and centrifuged at 10,000RCF for 1 min at room temperature. This process wasrepeated with 500 ll Wash Buffer II and centrifuged at21,000 RCF for 3 min at room temperature. About 100 llElution Buffer was added to the spin column and incubatedfor 1 min followed by centrifugation at 21,000 RCF for 1min at room temperature.

PCR analysis

The reaction was performed by mixing 11 ll nuclease-freeH2O, 2.5 ll 103 PCR reaction buffer, 2.5 ll (50 mM)MgCl2, 2.0 ll (2.5 mM) dNTP, 1.25 ll of each forward andreverse primer, 2 ll DNA, and 0.5 ll Taq DNA polymerase(Invitrogen Canada; Burlington, Ontario, Canada). The cyclingparameters employed were: first cycle of 5 min at 948C, then35 cycles of 60 s at 948C, 90 s at 658C, and 150 s at 728C.Primers M2 5 50 ATCTAGTCATGCTATGGATATCAGC 30

and N2 5 50 GTGGGCTCTATGGCTTCTGAGG 30 wereused to test for KO allele (amplified fragments of 800 basepairs). Primers S1 5 50 GTGGTTAGCTAAAGTGAGGAT-GAT 30 and S2 5 50 CAGGTTTGTTGGGATTAACAGATC30 were used to test for the WT mouse allele amplifying a frag-ment of 465 base pairs. PCR products were run on a 1.5%agarose gel with 10,0003 SYBR-safe (1:13,333 in 13 TAE)and visualized under a BioRad trans-illuminator.

FIGURE 1. Outline of experimental procedures. Schematicshowing the progression of the experiments and group assign-ments. Male Fmr1 knockout (KO) and wild-type (WT) littermatemice were weaned and ear-punched at postnatal Day 24. Mice

were pseudorandomly assigned to either electrophysiology orbehavior experiments commencing at postnatal Day 60. All experi-ments were conducted on young, adult male mice between 2 and 4months of age.

2 EADIE ET AL.

Hippocampus

Electrophysiology

Slice preparation

Fmr1 KO (n 5 19) and WT littermate (n 5 21) mice wereused for the electrophysiology experiments. Mice were anesthe-tized with isoflurane, rapidly decapitated, and their brainsremoved in oxygenated (95% O2/5% CO2), ice-cold normalACSF (nACSF) [(in mM) 125 NaCl, 2.5 KCl, 1.25 NaHPO4,25 NaHCO3, 2 CaCl2, 1.3 MgCl2, and 10 dextrose, pH 7.3].Transverse hippocampal slices (350 lm) were sectioned using aVibratome 1500 (Ted Pella, Redding, CA). Sections were keptin order using a modified 24-well plate and incubated in con-tinuously oxygenated nACSF at 308C. Sections were allowed torest for a minimum of 1 h before recordings commenced.

Field electrophysiology recordings

Field recordings were collected in nACSF using an AxonMultiClamp 700B amplifier connected to a PC running Clam-pex 10.2 software (Molecular Devices, CA). Electrodes wereplaced under visual guidance using an Olympus BX51 micro-scope and motorized micromanipulators (Siskyou Design, OR).Field excitatory postsynaptic potentials (fEPSPs) were elicitedby delivering a 120 ls (10–40 lA) current pulse to the medialperforant path (MPP) of the DG using a digital stimulus ampli-fier (Getting Instruments, CA) and a single, concentric bipolarstimulating electrode (FHC, Bowdoin, ME). fEPSPs wererecorded using a single glass recording electrode (0.5–1.5 MX)filled with nACSF, and placed in the MPP about 200 lm fromthe stimulating electrode. Stimulation magnitude was set to elicita response of �50% and an input–output experiment was con-ducted with increasing stimulation magnitude (30- to 300-lspulse width; 30-ls intervals). Paired-pulse recordings were subse-quently performed using an interpulse interval of 50 ls (53; 20s between pairings). The GABAA receptor antagonist bicucullinemethiodide (5 lM; Sigma-Aldrich, Oakville, ON, Canada) wasincluded to isolate the excitatory component of synaptic transmis-sion for all recordings. A stable baseline (minimum 20 min) ofthe slope of fEPSPs elicited every 15 s was then achieved before aconditioning stimulus (CS) was delivered to induce synaptic plas-ticity. Baseline stimulation parameters were returned to followingthe CS (minimum 60 min). All analyses were conducted withAxon ClampFit 10.2 software (Molecular Devices, CA).

Conditioning stimulation protocols

Long-term potentiation (LTP) of fEPSPs was induced using aCS consisting of four trains of 50 pulses at 100 Hz, 30 s apart(high-frequency stimulation; HFS); whereas, long-term depres-sion (LTD) of fEPSPs was induced using a CS of 900 pulsesdelivered at 1 Hz over 15 min (low-frequency stimulation; LFS).

Whole cell electrophysiology

Whole-cell recordings were obtained using an Axon Axo-patch 200B amplifier connected to a PC running Clampex10.2 software (Molecular Devices, CA). The intracellular re-

cording solution consisted of (in mM) 20 KCl, 120 KGluco-nate, 4 NaCl, 0.10 EGTA, 4 ATP, 0.3 TrisGTP, 14 Phospho-creatine. Biocytin (0.2%) was also included for post hoc stain-ing of the recorded neuron. The resistance of recordingelectrodes was 5–7 MX. Only recordings with a series resist-ance of less than 30 MX were used in analyses.

The resting membrane potential (RMP) was obtained imme-diately following break-in in voltage clamp mode. Series resist-ance (Rs) was calculated from back-fitting recordings from a15 mV (290 to 285 mV) 300 ls step (average of 100recordings) using Axon ClampFit 10.2 software. AMPAR (a-amino-3-hydroxyl-5-methyl-isoazole-propionate receptor)-medi-ated currents were evoked using extracellular stimulation, whilegranule cell was held at 270 mV. Following an estimation ofthe maximum EPSC amplitude, an input–output function wasobtained. NMDAR-mediated EPSCs were obtained by washingover a low Mg21 (0.1 mM) ACSF containing the AMPAR an-tagonist NBQX (5 lM; Sigma-Aldrich, MO). NMDAR-medi-ated EPSCs were blocked using APV (50 lM).

Behavior

Modified SHIRPA

A separate cohort of mice (WT: n 5 10, KO: n 5 10) wasassessed using a modified version of the SHIRPA test battery(Rogers et al., 1997). The measures were: weight, body length,body position, spontaneous activity, respiratory rate, tremor,urination, defecation, transfer arousal, piloerection, startle, gait,pelvic elevation, tail elevation, touch escape, positional passivity,trunk curl, limb grasp, abnormal behavior, grip strength, bodytone, visual placing, pinna reflex, corneal reflex, wire maneuver,skin color, heart rate, limb tone, abdominal tone, salivation,provoked biting, righting reflex, contact righting reflex, negativegeotaxis, fearfulness, irritability, aggression, and vocalization.

Contextual fear acquisition and generalization

As part of an assay to assess contextual fear discrimination,the ability of Fmr1 KO (n 5 14) and WT (n 5 14) mice toacquire contextual fear was first evaluated (as described previ-ously by McHugh et al., 2007). On Days 1 through 4 of test-ing, a single animal was exposed to a single type of context fora 4-min, 2-s session. Three minutes into the session, a 2 s,650-lA shock was delivered through stainless steel rods com-posing the floor of the apparati to induce conditioned fear.The mice were left in the context for another minute. The de-pendent measure was freezing behavior within the first 3 min(i.e., preshock) of each session. Freezing was measured usingthe Video Freeze software with the freezing threshold set at 19and the minimum freeze duration set at 1 s. Two types of con-texts were used. Context A (used as the S1 context) consistedof four identical conditioning chambers (30 3 25 3 25 cm3;Med-Associates) placed in a sound-attenuating cubicle with a60-dB back ground noise provided by a fan and scented withWindex original glass cleaner. An overlay of two white plasticpanels created a continuous curve to the side and back walls.

NMDAR HYPOFUNCTION IN THE DG OF FMR1 KNOCKOUT MICE 3

Hippocampus

The floor of each chamber was made up of 16 stainless steelrods of alternating diameter (0.4 and 1.0 cm) spaced 1.5-cmapart (center to center) wired to a shock generator and scram-bler (Med-Associates) to deliver foot shock. Each chamber waswiped down with 70% ethanol before conditioning andbetween animals. Context B (used as the S2 context) consistedof four identical conditioning chambers (30 3 25 3 25 cm3;Med-Associates) placed in a sound-attenuating cubicle in a dif-ferent room from Context A with a fan providing a back-ground noise of 60 dB and scented with Simple Green. Thegrid floors were the same as those used in A to increase thesimilarity of the contexts. The side walls consisted of two blackplastic panels joined at the top and sloping down to the side ofthe conditioning chamber to form an A-frame.

Assessment of generalization of contextual fear occurred onDays 5 and 6 of testing and involved two 3-min sessions sepa-rated by 2 h. Mice were exposed to both the context that theywere previously shocked (S1) and a novel context (S2); how-ever, the mice did not receive any shocks during these sessions.The sessions on Day 6 were conducted in reverse order. Freez-ing behavior was assessed as described for assessment of acquisi-tion of contextual fear.

Contextual fear discrimination

Days 7 through 14 of testing consisted of discriminationtraining where animals were placed in both the S1 and S2contexts each day. Again, in the S1 context, mice received a 2s, 650-lA shock after 3 min and were left in the chamber for1 min following shock. In the S2 context, mice were simplyplaced in the chamber for an equivalent 4 min and 2 s. Freez-ing was measured during the 3-min preceding shock on alldays in which shock was administered and the equivalent pe-riod of time in the S2 context. The order of training followeda double alternation schedule: Day 7 S2?S1, Day 8S1?S2, Day 9 S1?S2, Day 10 S2?S1, etc. For statisti-cal analysis and graphical presentation the data was collapsedinto consecutive 2-day blocks so that each block consisted of 1day of S1 ? S2 and 1 day of S2 ? S1.

Extinction of contextual fear

During the first 3 min of Days 5, 6, and 7 of testing, allmice were exposed to the S1 context (i.e., the context that waspaired with a shock during acquisition training), but did notreceive a shock. Note that on Day 7, the mice do receive ashock in the S1 that affects freezing behavior in subsequentsessions, but not the current session. Extinction can thus bemeasured as decreased freezing in the S1 context over Days 5,6, and 7 of testing.

Statistical Analyses

Group data are presented as mean 6 standard error of themean (SEM). Differences between mean values of experimentalgroups were compared using Student’s t test, with Bonferronicorrection, or analysis of variance (ANOVA) followed by

Tukey’s post hoc tests as appropriate using Statistical 7.0 soft-ware (StatSoft, Tulsa, OK). Statistical significance: P < 0.05.

RESULTS

Normal Basal Synaptic Transmission and Short-Term Synaptic Plasticity in the Dentate Gyrus ofFmr1 KO Mice

Stimulating and recording electrodes were placed in the mo-lecular layer of the DG to elicit excitatory postsynaptic poten-tials (fEPSPs) in transverse hippocampal sections obtained fromyoung adult Fmr1 knockout (KO) and wild-type (WT) litter-mate mice. Input/Output (IO) functions and paired-pulsemeasures of presynaptic transmitter release were recorded forWT and Fmr1 KO mice (Fig. 2). The slope of the fEPSP sig-nificantly increased with increasing stimulation (repeated meas-ures ANOVA; WT n 5 10, Fmr1 KO n 5 12; F(8,160) 5362.68, P 5 0.000). A significant interaction between stimula-tion strength and genotype (F(8, 160) 5 1.560, P 5 0.141) or asignificant main effect of genotype (F(1,20) 5 1.795, P 50.195) were not apparent. These data suggest that basal synap-tic transmission is not significantly altered in the dentate gyrus(DG) by loss of FMRP. Short-term synaptic plasticity was thenassayed using a paired-pulse protocol with a 50-ls interpulseinterval. Paired-pulse depression in the medial perforant path(MPP) of the DG was clear in both genotypes (WT: 29.23 63.63, n 5 6; Fmr1 KO: 27.49 6 4.66, n 5 5) but not signif-icantly different between genotypes (t test; t(9) 5 0.102, P 50.921). These data suggest that there is no obvious differencein basal synaptic transmission or short-term synaptic plasticityin the DG of adult Fmr1 KO and WTmice.

Decreased NMDAR-Dependent LTP in theDentate Gyrus of Fmr1 KO Mice

The application of a conditioning stimulus, consisting offour trains of high-frequency stimuli (HFS; 50 pulses at 100Hz), produced a robust LTP in WT animals (Fig. 3; 50–60min Post-HFS: 73.01% 6 8.64% (n 5 15); t test, P <0.001). The magnitude of LTP was significantly decreased inFmr1 KO mice (44.27% 6 7.51%, n 5 15; t test, P 50.034). The inclusion of the NMDAR antagonist APV (50lM) during HFS severely attenuates LTP in both WT andFmr1 KO slices (WT: 8.17% 6 3.75%, n 5 14; Fmr1 KO:3.79% 6 6.98%, n 5 7), indicating that the LTP observed inboth groups was primarily NMDAR-dependent.

Decreased NMDAR-Dependent LTD in theDentate Gyrus of Fmr1 KO Mice

The application of low frequency stimulation (LFS) induceda significant LTD in the DG of sections obtained from WTmice (Fig. 4; 50- to 60-min Post-LFS: 221.63% 6 4.03% (n5 14); t test, P < 0.000). In contrast, LFS failed to induce

4 EADIE ET AL.

Hippocampus

LTD in the DG of sections obtained from Fmr1 KO mice(21.05% 6 3.57%, n 5 9; t test, P > 0.05). The applicationof the NMDAR antagonist APV (50 lM) during administrationof the LFS completely blocked LTD in WT [22.59 6 4.84 (n 57); t test, P 5 0.489] and Fmr1 KO mice [4.61 6 3.71 (n 510), t test, P 5 0.390]. These results suggest that, in addition tothe impaired NMDAR-dependent LTP, NMDAR-dependentLTD is abolished in the DG of Fmr1 KO mice.

Decreased NMDAR-Mediated SynapticTransmission in the DG of Fmr1 KO Mice

The data thus far indicate that a loss of FMRP substantiallylimits bidirectional synaptic plasticity in the DG. TheNMDAR makes a significant contribution to both forms ofsynaptic plasticity in the DG (Vasuta et al., 2007), thus it maybe that NMDAR function is downregulated in these animals.To determine if a loss of FMRP alters NMDAR function, weperformed whole-cell voltage clamp experiments to examineboth AMPAR- and NMDAR-mediated components of theexcitatory postsynaptic currents (EPSCs). All experiments wereperformed with inhibition blocked, and each neuron was filledwith biocytin (0.2%) to allow for morphological confirmationof cell identity at the end of the experiment. AMPAR-mediatedEPSCs were obtained by holding the neurons at 270 mV inartificial cerebrospinal fluid (nACSF). AMPAR-mediatedEPSCs were evoked with increasing afferent stimulation toobtain reliable maximum responses. Both genotypes showedabsolute maximal AMPAR-mediated currents of comparablemagnitudes [Fig. 5; WT: 328.6 6 39.6 pA (n 5 5), Fmr1KO: 399.8 6 94.1 pA (n 5 5); t test, P 5 0.4577]. This isconsistent with our observation of comparable IO functions forfEPSPs between genotypes. Following collection of AMPAR-

mediated IOs, NMDAR-mediated currents were obtained byapplication of the AMPAR antagonist NBQX (5 lM) in low-Mg21 ACSF. In all cells, the EPSC was challenged with APV(50 lm) at the end of the experiment to confirm that thesewere indeed NMDAR-mediated responses. Following acquisi-tion of NMDAR-mediated EPSCs, an IO function was thenobtained in a manner similar to that described for AMPAR-mediated EPSCs. The absolute maximum NMDAR-mediatedEPSC was significantly decreased in Fmr1 KO mice relative toWT littermates [WT: 281.3 6 18.2 pA (n 5 5), Fmr1 KO:168.5 6 48.0 pA (n 5 5); t test, P 5 0.0197]. A comparisonof AMPAR/NMDAR current ratios between the genotypesrevealed a significant increase in the ratio in DG neurons fromFmr1 KO mice [WT 5 1.190 6 0.198 (n 5 5), Fmr1 KO 52.714 6 0.617 (n 5 5); t test, P 5 0.0302]. Loss of FMRPappears to decrease NMDAR-mediated currents in DG granuleneurons, and it is likely that this is responsible for the impairedNMDAR-dependent bidirectional synaptic plasticity.

Intact Passive and Active Membrane Propertiesof Granule Neurons in the Dentate Gyrus ofFmr1 KO Mice

The resting membrane potential (RMP), input resistanceand whole cell capacitance were not significantly differentbetween genotypes (Table 1). In current clamp mode, the rheo-base and action potential frequency were determined and werealso not significantly different between genotypes. In voltageclamp mode, a depolarizing voltage step protocol (270 to 170mV; 10-mV intervals) was employed to assess both a clear fastinward current (i.e., Na1) and a slower outward current (i.e.,K1) and a hyperpolarizing voltage step protocol (270 to2120 mV; 5 mV intervals) was employed to assess an inward

FIGURE 2. Normal basal synaptic transmission and short-termplasticity in Fmr1 KO mice. Field electrophysiology recordings wereconducted by placing a stimulating and recording electrode in themedial perforant path (MPP) of the molecular layer of the dentategyrus. Field excitatory postsynaptic potentials (fEPSP) approximately50% of maximum were elicited using a 120-ls pulse by varying thestimulation intensity (10–40 lA). (a) fEPSPs were evoked usingincreasing pulse widths from 30 to 300 ls to obtain an input-output

(IO) curve. No differences between Fmr1 knockout (KO; black) andwild-type (WT; white) littermates were observed. (b) Separate sec-tions were used to assess paired-pulse synaptic plasticity by deliveringtwo stimuli 50 ls apart to induce consecutive fEPSPs. No significantdifferences in paired-pulse plasticity between the genotypes wereobserved. Only sections without population spikes were used. Sampletraces are shown above for both WT (black) and Fmr1 KO (gray)mice. Vertical scale bar 5 0.5 mV. Horizontal scale bar 5 10 ls.

NMDAR HYPOFUNCTION IN THE DG OF FMR1 KNOCKOUT MICE 5

Hippocampus

rectifying current (i.e., K1 IR; Fig. 6). The peak inward cur-rent to the depolarizing step protocol changed significantlyacross holding potentials [WT (n 5 7) and KO (n 5 5);repeated measures factorial ANOVA, F(14,140) 5 20.217, P 50.000]; however, there was no significant interaction (F(14,140)5 0.975, P 5 0.482) or main effect of genotype (F(1,10) 50.001, P 5 0.975). These data suggest that there is no obviousdifference in Na1 conductance in dentate granule neurons lack-ing FMRP. Analyses of the outward currents revealedsimilar results. The peak outward current to the depolarizingstep protocol changed significantly across holding potentials

[WT (n 5 7) and KO (n 5 4); repeated measures factorialANOVA, F(14,126) 5 68.383, P 5 0.000]; however, there wasno significant interaction (F(14,126) 5 0.850, P 5 0.614) ormain effect of genotype (F(1,9) 5 0.1.685, P 5 0.227). Thesedata suggest that there is no obvious difference in outward K1

conductance in dentate granule neurons lacking FMRP. Thepeak inward current resulting from the hyperpolarizing stepprotocol changed significantly across holding potentials [WT (n5 7) and KO (n 5 5); repeated measures factorial ANOVA,

FIGURE 3. NMDAR-dependent long-term potentiation (LTP)is significantly attenuated in the dentate gyrus of Fmr1 KO mice.(a) Field EPSPs were elicited every 15 s for 20 min in WT andFmr1 KO mice prior to HFS. Field EPSPs were obtained for 60min following HFS. Traces are averages from 10 min immediatelyprior to HFS (‘‘PRE’’) or 50–60 min after HFS (‘‘POST’’). Verticalscale bar 5 0.5 mV. Horizontal scale bar 5 5 ls. (b) The absoluteslope of the fEPSP was significantly increased in both genotype;however, the magnitude of LTP was significantly less in the Fmr1KO mice. The NMDAR antagonist APV (50 lM) blocked LTP inboth WT and Fmr1 KO mice. (c) Bar graphs showing the changein slope at 50–60 min post conditioning for each group. (* denotesP < 0.05 between genotypes).

FIGURE 4. NMDAR-dependent long-term depression (LTD) issignificantly attenuated in the dentate gyrus of Fmr1 KO mice. (a)Field EPSPs were obtained every 15 s for 20 min prior to LFS andfor 60 min following. Traces are averages from 10 min immedi-ately prior to LFS (‘‘PRE’’) or 50–60 min after LFS (‘‘POST’’).Vertical scale bar 5 0.5 mV. Horizontal scale bar 5 5 ls. (b) Sig-nificant LTD was only observed in the DG of WT animals. Inclu-sion of the NMDAR antagonist APV (50 lM) blocked LTD induc-tion in WT animals. (c) Bar graphs showing the change in fEPSPslope at 50- to 60-min post conditioning for each group. (* denotesP < 0.05 between genotypes).

6 EADIE ET AL.

Hippocampus

F(16,160) 5 2.157, P 5 0.008]; however, there was no signifi-cant interaction (F(16,160) 5 1.117, P 5 0.343) or main effectof genotype (F(1,10) 5 1.191, P 5 0.301). These data suggestthat there is no obvious difference in inward-rectifying K1 con-ductance in dentate granule neurons lacking FMRP.

Modified SHIRPA Indicates That Fmr1 KO Miceare Behaviorally Similar to Wild-Type Animals

No significant differences were observed between WT andFmr1 KO mice in any of the measures in the SHIRPA test bat-tery (t tests, P > 0.05 for all measures; Table 2). This indicatesthat there are no significant behavioral abnormalities that couldinterfere with the performance of these animals in the contextfear discrimination procedure used here.

Normal Acquisition and Generalization ofContextual Fear in Fmr1 KO Mice

NMDAR function in the hippocampus is critical to contex-tual fear learning and NMDARs in the DG contribute to theability to learn to discriminate between contexts (Young et al.,1994). Prior to the assessment of context discrimination, bothgroups were assessed on acquisition and generalization of con-textual fear (Fig. 7). The acquisition of the context-shock asso-ciation in Fmr1 KO and WT mice was measured over fourtraining days. Repeated measures ANOVA across training daysshowed no overall effect of genotype (F(1,26) 5 2.87, P 50.102) and the effect of training day (F(3,78) 5 16.38, P 50.000) and the genotype by training day interaction were notsignificant (F(3,78) 5 2.09, P 5 0.108). Overall, Fmr1 KO

FIGURE 5. Decreased NMDAR-mediated synaptic transmis-sion in granule neurons in the DG of Fmr1 KO mice. (a) The am-plitude of the AMPAR-mediated EPSC increased equivalently inboth genotypes with increasing stimulation (increasing pulse widthfrom 0 to 300 lS). Sample AMPAR-mediated EPSCs from bothFmr1 KO and WT mice at �50% of maximum amplitude areshown. (b) The input–output (IO) function of NMDAR-mediatedEPSCs is shown. The EPSC amplitude increased equivalently inboth genotypes with increasing afferent stimulation (increasing

pulse width from 0 to 300 lS). Sample NMDAR-mediated EPSCsfrom both Fmr1 KO and WT mice at �50% of maximum ampli-tude are shown. Vertical scale bar 5 50 pA. Horizontal scale bar5 50 ls. (c) Bar graph showing decreased NMDAR-mediatedEPSCs in Fmr1 KO mice compared to controls, and no differencesin AMPAR-mediated EPSCs. (d) The average AMPAR/NMDAR ra-tio from neurons recorded from Fmr1 KO mice is significantlyhigher than that obtained in WT mice. (* denotes P < 0.05between genotypes).

NMDAR HYPOFUNCTION IN THE DG OF FMR1 KNOCKOUT MICE 7

Hippocampus

mice appear to show normal acquisition of contextual fear.Conditional fear responses will occur not only in response tothe original conditional stimuli but also to stimuli which showsimilar or overlapping features, a process referred to as general-ization. The extent to which Fmr1 KO and WTmice generalizeto a context that was similar, but distinct from, the originaltraining context was tested over a 2-day period. Freezing datafrom S1 and S2 over the two generalization testing days wereaveraged. Two-way ANOVA indicated a significant effect ofcontext (F(1,26) 5 182.4, P < 0.001) but the effect of genotype(F(1,26) 5 2.866, P 5 0.102) and the genotype by contextinteraction (F(1,26) 5 1.298, P 5 0.265) did not reach statisti-cal significance. This indicates that Fmr1 KO mice show nor-mal levels of freezing in the S1 context and equivalent general-ization of contextual fear to the S2 context.

Impaired Context Discrimination inFmr1 KO Mice

The ability of Fmr1 KO and WT mice to discriminatebetween the S1 context (paired with shock) and the S2 con-text (never paired with shock) was assessed over 8 days of dis-crimination training (Fig. 8). Learning this discrimination isthought to require a process of ‘‘pattern separation’’ where thepattern of multi-modal stimuli that defines the S1 context isseparated from the pattern of stimuli that define the S2 con-text. This information is utilized to determine the ‘‘safe’’ andthe ‘‘dangerous’’ context. Repeated measures ANOVA for freez-ing in the S1 context showed a significant effect of TrainingBlock (F(3,78) 5 16.38, P 5 0.000) but did not show a Train-ing Block by Genotype interaction (F(3,78) 5 2.09, P 5 0.108)or an overall effect of Genotype (F(1,26) 5 2.87, P 5 0.102).

This indicates that Fmr1 KO and WT mice freeze to a similarextent in the S1 context. Repeated measures ANOVA forfreezing in the S2 context showed a significant effect of Train-ing Block (F(3,78) 5 28.42, P 5 0.000) and an overall effect ofgenotype (F(1,26) 5 10.52, P 5 0.003), but did not show atraining block by genotype interaction (F(3,78) 5 0.444, P 50.722). This analysis indicates that Fmr1 KO mice show ele-vated freezing in the S2 context.

A discrimination ratio was calculated for each individual ani-mal for each training block using the following formula:(Freezing in S1)/((Freezing in S1) 1 (Freezing in S2)). Dis-crimination ratios above 0.5 are indicative of significant dis-crimination. Repeated measures ANOVA on the discriminationratios indicated a significant effect of training block (F(3,78) 5

TABLE 1.

Passive Membrane Properties and Intrinsic Excitability of Dentate

Granule Neurons From Wild-Type and Fmr1 Knockout Mice

Wild type Fmr1 knockout P

RMP (mV)a 288.64 (63.73) 289.87 (63.19) 0.710

Input resistance (MX)b 173.03 (634.11) 195.60 (630.88) 0.478

Capacitance (pF)c 20.04 (62.11) 18.41 (62.58) 0.872

Rheobase (pA)d 102.86 (68.33) 103.50 (66.21) 0.955

AP frequency (Hz)e 6.70 (60.44) 6.88 (60.48) 0.708

Passive membrane properties and intrinsic excitability of dentate granule neu-rons from adult, male wild-type (n 5 9 cells; three mice) and Fmr1 knockout(n 5 8 cells; three mice) mice.aResting membrane potential (RMP) was obtained immediately following break-inin voltage clamp mode and subsequently adjusted for liquid junction potential.bInput resistance was calculated from a 15 mV step from 290 to 285 mV.cCapacitance was calculated from fitting a double exponential curve to theuncorrected waveform.dIn current clamp, the minimum current required to induce a single AP (rheo-base) was obtained by injecting increasing amounts of current (1 pA intervalsstarting at 90 pA; 300 ms).eIn current clamp, a 500 ms depolarizing current injections (20) equal to rheo-base 120% was used to obtain the action potential (AP) frequency. Statisticalanalyses employed two-tailed t tests.

FIGURE 6. Normal active membrane properties of dentategranule neurons from wild-type and Fmr1 knockout mice. Activemembrane properties of dentate granule neurons from adult, malewild-type (n 5 8 sections, three mice) and Fmr1 knockout (n 5 5sections, three mice) mice. (a) I–V curve for voltage-gated Na1

currents and K1 currents, displaying similar amplitudes for bothgenotypes. (b) Inward rectifying currents evoked by current stepsfrom 270 to 2120 mV (step 5 mV, 300 ls) recorded from bothgenotypes, displaying similar amplitudes for both genotypes.

8 EADIE ET AL.

Hippocampus

28.42, P 5 0.000) and a significant effect of genotype (F(1,26)5 5.02, P 5 0.034) but no training block by genotype interac-tion (F(3,78) 5 0.71, P 5 0.549). This analysis indicates thatFmr1 KO mice show a significantly reduced discrimination ra-tio relative to WTmice.

Impaired Extinction of Contextual Fearin Fmr1 KO Mice

WT mice showed clear evidence of extinction across Days 5,6, and 7 in the above behavioral paradigm (Fig. 9). In contrast,Fmr1 KO mice did not exhibit extinction. Repeated measuresANOVA across Days 6 and 7 showed significant main effectsof genotype (F(1,26) 5 5.79, P 5 0.0,235) and day (F(1,26) 59.60, P 5 0.0,046) as well as a genotype by day interaction(F(1,26) 5 5.90, P 5 0.0,224). Post hoc comparisons indicatedthat Fmr1 KO mice show greater freezing on Day 7 comparedto WT littermates (t test, t(28) 5 3.205, P < 0.01), suggestingthat Fmr1 KO mice show impaired fear extinction. Overall, thebehavioral analyses indicate that Fmr1 KO mice are able to ac-

quire contextual fear, however, they have difficulty in learningto suppress responding to a similar, but safe, context (impairedextinction and discrimination).

DISCUSSION

The present work indicates that a loss of FMRP can pro-foundly impact the capacity for DG to exhibit synaptic plastic-ity and facilitate discrimination learning. We show for the firsttime that deletion of the Fmr1 gene impairs long-term poten-tiation (LTP) and depression (LTD) in the DG subfield of thehippocampus. Both forms of DG synaptic plasticity areNMDAR-dependent, and appear to be associated withNMDAR hypofunction. A significant decrease in NMDAR-mediated currents was apparent in DG granule cells, whileAMPAR-mediated currents did not differ significantly betweenFmr1 KO and WT mice. Assessment of the ability of Fmr1KO mice to discriminate between two similar contexts, a task

TABLE 2.

Modified SHIRPA Assessment of Fmr1 KO and WT Mice

Weight Squares Body length Body posit Spont act Resp rate Tremor Urination Defecation Trans arous Piloerect

WT Mean 33.6 18.66667 9.1 3 1.1 2 0 0 0 5 0

SE 0.67 0.89 0.12 0.00 0J0 0.00 0.00 0.00 0.00 0.00 0.00

KO Mean 31.68 18 9.03 3 1 2 0 0 0 4.9 0

SE 0.85 1.22 0.08 0.00 0.00 0.00 0.00 0.00 0.00 0.10 0.00

Startle Gait Pelvic elev Tail elevat Touch esc Posit pass Trunk curl Limb grasp Abn beh Vis placing Grip

WT Mean 0.9 0 2 1 1 3 1 1 3.1 3

SE 0.10 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.10 0.00

KO Mean 1 0.1 2 1.1 1.1 3 1 1 1 3 3

SE 0.00 0.10 0.00 0.10 0.10 0.00 0.00 0.00 0.00 0.00 0.00

Body tone Pinna ref Corneal ref Toe pinch Wire man Skin color Heart rate Limb tone Ab tone Saliv Prov bite

WT Mean 1 0.9 1 3 0.6 1 1 1 1 1 0.8

SE 0.00 0.10 0.00 0.00 0.16 0.00 0.15 0.15 0.00 0.00 0.13

KO Mean 1 0.9 1 2.9 0.2 1 1.1 1.2 1 1 0.6

SE 0.00 0.10 0.00 0.10 0.13 0.00 0.10 0.13 0.00 0.00 0.16

Right ref Ctct right Neg geotax Fear Irritability Aggression Vocalization Total Score

WT Mean 0 0.8 0.2 0 0.4 0.7 0 43.5

SE 0.00 0.13 0.20 0.00 0.16 0.15 0.00 0.40

KO Mean 0 1 0 0 0.8 0.6 0.1 43.6

SE 0.00 0.00 0.00 0.00 0.13 0.16 0.10 0.40

The basic behavioral phenotype of WT and FMR1 KO mice was assessed using the SHIRPA test battery (Reproduced with permission from Rogers et al., Mamma-lian Genome, 1997, 8, 711–713, 1997). No significant differences were observed after Bonferonni correction (P < 0.001 required for significance). Abbreviationsare Body Position, Spontaneous Activity, Respiratory Rate, Transfer Arousal, Piloerection, Pelvic Elevation, Tail Elevation, Touch Escape, Positional Passivity, Abnor-mal Behavior, Visual Placing, Pinna Reflex, Corneal Reflex, Wire Maneuver, Abdominal Tone, Salivation, Provoked Biting, Righting Reflex, Contact RightingReflex, Negative Geotaxis.

NMDAR HYPOFUNCTION IN THE DG OF FMR1 KNOCKOUT MICE 9

Hippocampus

dependent on functional NMDARs and NMDAR-dependentsynaptic plasticity specifically in the DG, revealed a robustlearning impairment (McHugh et al., 2007). In addition, animpairment in extinction of contextual fear was apparent inFmr1 KO mice. These data illuminate robust impairments inmajor forms of synaptic plasticity associated with learningimpairment in the mouse model of FXS.

The most influential theory of the pathophysiology of FXShas been the mGluR theory (Bear et al., 2004). The theory isbased on the finding that application of the group 1 mGluRagonist DHPG induces LTD of �23% in Fmr1 KO mice com-pared to 12% in controls (Huber et al., 2002). Although manyinteresting parallels exist between mGluR activation and symp-toms of FXS, the theory does not account for the key symptomof FXS: intellectual disability. For example, the enhancedmGluR-LTD in Fmr1 KO mice was found in the CA1 subfieldof the hippocampus but Fmr1 KO mice perform as well ascontrols on learning tasks dependent on this brain region such

as acquisition of the classic Morris water maze and one-trial in-hibitory avoidance (D’Hooge et al., 1997; Eadie et al., 2009).

It has recently been suggested that enhanced mGluR-LTD isassociated with impaired inhibitory-avoidance extinction (IAE)in Fmr1 KO mice (Dolen et al., 2007); however, this behav-ioral link is not direct. The association is based on the postu-late that enhanced mGluR-LTD is protein synthesis dependentand IAE is also protein synthesis dependent (Power et al.,2006). This relationship is confounded by the fact that theenhanced mGluR-LTD in Fmr1 KO mice is not protein syn-thesis dependent (Nosyreva and Huber, 2006), despite the factthat mGluR-LTD is protein synthesis in WT mice (Nosyrevaand Huber, 2006). The mGluR theory of FXS has also beencalled into question because the enhanced LTD observed inFmr1 KO mice can also occur in response to metabotropic re-ceptor activation in general (Volk et al., 2007). As attractive asthe theory is, it does not seem to fully explain the deficitsobserved in humans and animals, and thus it may be that

FIGURE 7. Normal acquisition and generalization of contex-tual fear in Fmr1 KO mice. (a) Experimental design for the assess-ment of contextual fear acquisition. (b) Experimental design forthe generalization of contextual fear. (c) Acquisition training ofcontextual fear across 4 days as measured by a significant increase

in the percentage of freezing. Fmr1 KO and WT mice showedequivalent acquisition of contextual fear. (d) Generalization con-textual fear to a novel context occurred to a similar degree in bothFmr1 KO and WT mice. Percentage freezing in the shock-paired(S1) and novel context (S2) was collapsed over 2 days of testing.

10 EADIE ET AL.

Hippocampus

FMRP plays different roles in the cell during development andadulthood (Bear et al., 2004).

NMDA receptors have largely been ignored in the study ofFXS, perhaps due to an early study reporting normalNMDAR-dependent LTP in the CA1 subfield of Fmr1 KOmice (Godfraind et al., 1996). Consistent with this observationwas the report that NMDAR-dependent LTD is normal in theCA1 subfield of the hippocampus (Huber et al., 2002). In con-trast to these results conducted in adult mice, recent studies inves-tigating NMDAR-dependent LTP and LTD in young mice hasrevealed deficits. Hu et al. (2008) show impaired NMDAR-de-pendent LTP in the CA1 subfield of 2-week-old Fmr1 KO mice.In a separate recent study, Pilpel et al. (2009) noted abnormallyenhanced NMDAR-dependent LTP in the CA1 subfield of 2-week-old, but not 6- to 7-week-old, Fmr1 KO mice. In addition,the latter study reported a decrease in the AMPAR/NMDAR ra-tio only in 2-week postnatal Fmr1 KO mice (Pilpel et al., 2009).This appeared to be primarily a function of a decrease inAMPAR-mediated synaptic transmission; however, the relationship

between enhanced NMDAR-dependent LTP in Fmr1 KO miceand learning was not investigated in these studies. In the presentstudy, input-output functions for AMPAR and NMDAR-medi-ated EPSCs from dentate granule neurons revealed decreasedNMDAR-mediated synaptic transmission and normal AMPAR-mediated synaptic transmission. These results are consistent withthe hypothesis that the AMPAR/NMDAR ratio is altered by aloss of the FMRP in developing regions of the brain; even if ourfindings in the adult DG are the opposite of the findings in theCA1 subfield of 2-week postnatal Fmr1 KO mice (Pilpel et al.,2009). That said, another group has also reported results consist-ent with ours (i.e., decreased NMDAR-mediated synaptic trans-mission and normal AMPAR-mediated synaptic transmission inthe DG of Fmr1 KO mice) (Yun et al., 2009 SFN abstract) andit may be that some of these changes are developmentally regu-lated (Harlow et al., 2010). In the present study, decreasedNMDAR-mediated synaptic transmission appears to underlieimpaired bidirectional, NMDAR-mediated synaptic plasticity inthe DG of adult Fmr1 KO mice.

FIGURE 8. Impaired context discrimination in Fmr1 KOmice. (a) Experimental design for discrimination training. Datawas collapsed over each consecutive 2-day period into trial blocks1–4 for the S1 and S2 contexts. (b) Percent freezing in the 3 minprior to shock in the S1 context and in the equivalent 3-min pe-riod in the S2 context across discrimination training. Fmr1 KO

mice show a persistent elevation of freezing in the S2 context rela-tive to WT mice. (c) Discrimination ratios, calculated as (Percentfreezing in S1)/((Percent freezing in S1) 1 (Percent freezing inS2)) across discrimination training. Fmr1 KO mice showedreduced discrimination ratios relative to WT mice. (* denotes P <0.05 between genotypes).

NMDAR HYPOFUNCTION IN THE DG OF FMR1 KNOCKOUT MICE 11

Hippocampus

We hypothesized that impaired NMDAR-mediated synaptictransmission and plasticity in the DG would manifest as learn-ing impairment on a contextual discrimination task. We haverecently shown that genetic deletion of the obligatory NMDARsubunit NR1 specifically in the DG impairs NMDAR-depend-ent LTP and learning in a contextual discrimination task(McHugh et al., 2007). Prior to the contextual discriminationtesting, the acquisition, generalization, and extinction of con-textual fear were assessed. Similar to previous studies, alteredextinction was observed (Dolen et al., 2007). These authorsreported greater extinction following inhibitory-avoidance (IA)training. In contrast, we observed less extinction following con-textual fear conditioning. These differences may be partiallyexplained by the nature of the task. In IA, rodents are placedinto an illuminated chamber of a two-chamber apparatus andadministered a foot shock upon entrance into the dark chamber(Power et al., 2006). The latency of the rodents to enter thedark chamber upon a subsequent trial is measured. Theassumption is that if the rodent has formed a memory of thepairing trial then they will exhibit an increased latency to re-enter the dark chamber. This assumption is confounded by theobservations that rodents seek out a dark environment whenconfronted with an anxiety-invoking situation (even if theyhave been previously shocked in that context) and the consist-ent observation that Fmr1 KO mice have a tendency to spend

more time in the exposed, well-lit regions of behavioral appa-rati such as the open field (Eadie et al., 2009). In addition, aconsistent lack of difference across trials in an IA task mayreflect an inability of mice to maintain the ability to discrimi-nate between two similar contexts. Thus our data provide someinsight into the potential cause of greater extinction of IAreported (Dolen et al., 2007). Further studies into the effectsof loss of FMRP on extinction are warranted.

The physiological basis of fear extinction may be related toabolished NMDAR-dependent LTD in the DG of Fmr1 KOmice. Recently, Dalton et al. (2008) reported that administrationof a peptide that interferes with NMDAR-dependent LTDimpairs the extinction, but not acquisition, of contextual fear(Dalton et al., 2008). Interestingly, these authors reported thatthis effect was dependent on NR2B-containing NMDARs. A sep-arate recent study has shown that FMRP normally binds mRNAfor the NR2B subunit suggesting that loss of FMRP may specifi-cally lead to a decrease in NR2B-containing NMDARs (Schuttet al., 2009). Future studies should elucidate the subtype ofNMDAR that is decreased in the DG of Fmr1 KO mice and therelationship to LTD and impaired fear extinction.

Learning impairments in Fmr1 KO mice parallel learningimpairments in mice lacking functional NMDARs specificallyin the DG (DG-NR1 KO mice; McHugh et al., 2007). Wehave shown that Fmr1 KO mice, similar to DG-NR1 KOmice, are not impaired on acquisition of the Morris water maze(Eadie et al., 2009). DG-NR1 KO mice do show impairmentswhen assessed on a contextual fear discrimination task. This isconsistent with modern theories that the function of the DG ispattern separation (Lisman, 2003; Rolls and Kesner, 2006).Assessment of Fmr1 KO mice using a similar paradigm revealeda novel learning impairment in the ability to discriminatebetween two similar or overlapping contexts. In fact, the impair-ment in context discrimination was strikingly similar betweenDG-NR1 KO and Fmr1 KO mice: both transgenic mice appearto manifest impaired context discrimination because of anincrease in freezing behavior in the context that they were notpreviously shocked. These results extend upon our previous find-ings suggesting that a reduction in functional NMDARs in theDG is sufficient to impair context discrimination and illuminatea novel learning impairment in the mouse model of FXS.

Decreased NMDAR-mediated synaptic transmission in theDG may extend beyond impaired learning ability. Niewoehneret al. (2007) have recently shown that DG-NR1 KO mice ex-hibit decreased anxiety-like behavior (Niewoehner et al., 2007).This is consistent with the observation of anxiolysis resultingfrom intrahippocampal infusion of NMDAR antagonists. Thesedata have led to the suggestion that NMDARs in the DG areat the interface between cognition and emotion (Barkus et al.,2010). Our group and several others have shown that Fmr1KO mice exhibit decreased anxiety-like behavior similar to thatobserved in DG-NR1 KO mice and rodents receiving intrahip-pocampal infusions of NMDAR antagonists (Hackl and Caro-brez, 2006; Niewoehner et al., 2007; Eadie et al., 2009; Barkuset al., 2010). It is possible that decreased NMDA-mediatedsynaptic transmission and plasticity reported here underlies

FIGURE 9. Impaired extinction of contextual fear in Fmr1KO mice. Assessment of extinction of contextual fear occurred fol-lowing acquisition of contextual fear. During the 2 days of gener-alization testing and the first 3 min of context discriminationlearning, mice were not shocked in the context that they previouslyreceived a shock (S1). The WT mice exhibited significant extinc-tion across these 3 days; whereas, the Fmr1 KO mice did not ex-hibit significant extinction of contextual fear. (* denotes P < 0.05between genotypes).

12 EADIE ET AL.

Hippocampus

other key behavioral abnormalities observed in the mousemodel of FXS.

Intellectual disability is generally assumed to result from del-eterious alterations in several genes combined with, often un-identified, environmental factors. The resulting abnormalitiesin the brain that underlie the intellectual disability are oftenassumed to be equally complex. Here we demonstrated that ro-bust learning impairments can result from mutation of a singlegene, and that the effects on the brain can be clear and robust.The illumination of this brain-behavior relationship was de-pendent upon investigation of a brain region previously unin-vestigated in Fmr1 KO mice: the dentate gyrus. We show thatloss of FMRP decreases NMDAR-mediated synaptic transmis-sion and NMDAR-dependent bidirectional synaptic plasticityspecifically in the DG. We show that learning dependent onNMDA receptors specifically in the DG is also impaired. Theseexperiments improve our understanding of the pathophysiologyof intellectual disability in general and identify a novel targetfor therapeutics aimed at ameliorating intellectual disability inFXS and related neurodevelopmental disorders.

Acknowledgments

The authors thank E. Wiebe, N. Jacobs and Drs. J. John-ston, C. Brown, P. Nahirney, A. Titterness, and Y.T. Wang forassistance with this work. BDE holds a Vancouver CoastalHealth-CIHR-UBC MD/PhD Studentship Award. TSK holdsa NSERC-CGS award. BRC is a MSFHR Senior Scholar. Thiswork was initiated with funding from Fragile-X Canada andthe Scottish Rites Charitable Foundation.

REFERENCES

1994. Fmr1 knockout mice: A model to study fragile X mental retar-dation. The Dutch-Belgian fragile X consortium. Cell 78:23–33.

Ashley CT Jr, Wilkinson KD, Reines D, Warren ST. 1993. FMR1protein: Conserved RNP family domains and selective RNA bind-ing. Science 262:563–566.

Barkus C, McHugh SB, Sprengel R, Seeburg PH, Rawlins JN, Ban-nerman DM. 2010. Hippocampal NMDA receptors and anxiety:At the interface between cognition and emotion. Eur J Pharmacol626:49–56.

Bassell GJ, Warren ST. 2008. Fragile X syndrome: Loss of localmRNA regulation alters synaptic development and function. Neu-ron 60:201–214.

Bear MF, Huber KM, Warren ST. 2004. The mGluR theory of fragileX mental retardation. Trends Neurosci 27:370–377.

Bliss TV, Collingridge GL. 1993. A synaptic model of memory: Long-term potentiation in the hippocampus. Nature 361:31–39.

Braun K, Segal M. 2000. FMRP involvement in formation of synapsesamong cultured hippocampal neurons. Cereb Cortex 10:1045–1052.

Brown V, Jin P, Ceman S, Darnell JC, O’Donnell WT, Tenenbaum SA,Jin X, Feng Y, Wilkinson KD, Keene JD, et al. 2001. Microarrayidentification of FMRP-associated brain mRNAs and altered mRNAtranslational profiles in fragile X syndrome. Cell 107:477–487.

Castren M, Tervonen T, Karkkainen V, Heinonen S, Castren E, Lars-son K, Bakker CE, Oostra BA, Akerman K. 2005. Altered differen-tiation of neural stem cells in fragile X syndrome. Proc Natl AcadSci USA 102:17834–17839.

D’Hooge R, Nagels G, Franck F, Bakker CE, Reyniers E, Storm K,Kooy RF, Oostra BA, Willems PJ, De Deyn PP. 1997. Mildlyimpaired water maze performance in male Fmr1 knockout mice.Neuroscience 76:367–376.

Dalton GL, Wang YT, Floresco SB, Phillips AG. 2008. Disruption ofAMPA receptor endocytosis impairs the extinction, but not acquisi-tion of learned fear. Neuropsychopharmacology 33:2416–2426.

Darnell JC, Warren ST, Darnell RB. 2004. The fragile X mental retar-dation protein, FMRP, recognizes G-quartets. Ment Retard DevDisabil Res Rev 10:49–52.

Dolen G, Osterweil E, Rao BS, Smith GB, Auerbach BD, Chattarji S,Bear MF. 2007. Correction of fragile X syndrome in mice. Neuron56:955–962.

Eadie BD, Zhang WN, Boehme F, Gil-Mohapel J, Kainer L, SimpsonJM, Christie BR. 2009. Fmr1 knockout mice show reduced anxietyand alterations in neurogenesis that are specific to the ventral den-tate gyrus. Neurobiol Dis 36:361–373.

Farmer J, Zhao X, van Praag H, Wodtke K, Gage FH, Christie BR.2004. Effects of voluntary exercise on synaptic plasticity and geneexpression in the dentate gyrus of adult male Sprague-Dawley ratsin vivo. Neuroscience 124:71–79.

Ge S, Sailor KA, Ming GL, Song H. 2008. Synaptic integration andplasticity of new neurons in the adult hippocampus. J Physiol586:3759–3765.

Godfraind JM, Reyniers E, De Boulle K, D’Hooge R, De Deyn PP,Bakker CE, Oostra BA, Kooy RF, Willems PJ. 1996. Long-termpotentiation in the hippocampus of fragile X knockout mice. Am JMed Genet 64:246–251.

Hackl L, Carobrez A. 2006. Anxiolytic-like effects of NMDA-receptorantagonist injected into the ventral but not dorsal hippocampus inrats submitted to the elevated plus-maze. Eur Neuropsychopharma-col 16:S255–S256.

Harlow EG, Till SM, Russell TA, Wijetunge LS, Kind P, ContractorA. 2010. Critical period plasticity is disrupted in the barrell cortexof Fmr1 knockout mice. Neuron 65:385–398.

Hinds HL, Ashley CT, Sutcliffe JS, Nelson DL, Warren ST, HousmanDE, Schalling M. 1993. Tissue specific expression of FMR-1 pro-vides evidence for a functional role in fragile X syndrome. NatGenet 3:36–43.

Hu H, Qin Y, Bochorishvili G, Zhu Y, van Aelst L, Zhu JJ. 2008.Ras signaling mechanisms underlying impaired GluR1-dependentplasticity associated with fragile X syndrome. J Neurosci 28:7847–7862.

Huber KM, Gallagher SM, Warren ST, Bear MF. 2002. Altered synap-tic plasticity in a mouse model of fragile X mental retardation.Proc Natl Acad Sci USA 99:7746–7750.

Lisman J. 2003. Long-term potentiation: Outstanding questions andattempted synthesis. Philos Trans R Soc Lond B Biol Sci 358:829–842.

McHugh TJ, Jones MW, Quinn JJ, Balthasar N, Coppari R, ElmquistJK, Lowell BB, Fanselow MS, Wilson MA, Tonegawa S. 2007.Dentate gyrus NMDA receptors mediate rapid pattern separationin the hippocampal network. Science 317:94–99.

Niewoehner B, Single FN, Hvalby O, Jensen V, Meyer zum AltenBorgloh S, Seeburg PH, Rawlins JN, Sprengel R, Bannerman DM.2007. Impaired spatial working memory but spared spatial refer-ence memory following functional loss of NMDA receptors in thedentate gyrus. Eur J Neurosci 25:837–846.

Nimchinsky EA, Oberlander AM, Svoboda K. 2001. Abnormal devel-opment of dendritic spines in FMR1 knock-out mice. J Neurosci21:5139–5146.

Nosyreva ED, Huber KM. 2006. Metabotropic receptor-dependent long-term depression persists in the absence of protein synthesis in themouse model of fragile X syndrome. J Neurophysiol 95:3291–3295.

Pieretti M, Zhang FP, Fu YH, Warren ST, Oostra BA, Caskey CT,Nelson DL. 1991. Absence of expression of the FMR-1 gene infragile X syndrome. Cell 66:817–822.

NMDAR HYPOFUNCTION IN THE DG OF FMR1 KNOCKOUT MICE 13

Hippocampus

Pilpel Y, Kolleker A, Berberich S, Ginger M, Frick A, Mientjes E,Oostra BA, Seeburg PH. 2009. Synaptic ionotropic glutamatereceptors and plasticity are developmentally altered in the CA1field of Fmr1 knockout mice. J Physiol 587 (Part 4):787–804.

Power AE, Berlau DJ, McGaugh JL, Steward O. 2006. Anisomycininfused into the hippocampus fails to block ‘‘reconsolidation’’ butimpairs extinction: The role of re-exposure duration. Learn Mem13:27–34.

Rogers DC, Fisher EM, Brown SD, Peters J, Hunter AJ, Martin JE.1997. Behavioral and functional analysis of mouse phenotype:SHIRPA, a proposed protocol for comprehensive phenotype assess-ment. Mamm Genome 8:711–713.

Rolls ET, Kesner RP. 2006. A computational theory of hippocampalfunction, and empirical tests of the theory. Prog Neurobiol 79:1–48.

Schutt J, Falley K, Richter D, Kreienkamp HJ, Kindler S. 2009. Frag-ile X mental retardation protein regulates the levels of scaffold pro-teins and glutamate receptors in postsynaptic densities. J BiolChem 284:25479–25487.

Turner G, Webb T, Wake S, Robinson H. 1996. Prevalence of fragileX syndrome. Am J Med Genet 64:196–197.

van Praag H, Christie BR, Sejnowski TJ, Gage FH. 1999. Runningenhances neurogenesis, learning, and long-term potentiation inmice. Proc Natl Acad Sci USA 96:13427–13431.

Vasuta C, Caunt C, James R, Samadi S, Schibuk E, Kannangara T,Titterness AK, Christie BR. 2007. Effects of exercise on NMDAreceptor subunit contributions to bidirectional synaptic plasticity inthe mouse dentate gyrus. Hippocampus 17:1201–1208.

Verkerk AJ, Pieretti M, Sutcliffe JS, Fu YH, Kuhl DP, Pizzuti A,Reiner O, Richards S, Victoria MF, Zhang FP, et al. 1991. Identifi-cation of a gene (FMR-1) containing a CGG repeat coincidentwith a breakpoint cluster region exhibiting length variation in frag-ile X syndrome. Cell 65:905–914.

Volk LJ, Pfeiffer BE, Gibson JR, Huber KM. 2007. Multiple Gq-coupled receptors converge on a common protein synthesis-de-pendent long-term depression that is affected in fragile X syndromemental retardation. J Neurosci 27:11624–11634.

Yan QJ, Asafo-Adjei PK, Arnold HM, Brown RE, Bauchwitz RP.2004. A phenotypic and molecular characterization of the fmr1-tm1Cgr fragile X mouse. Genes Brain Behav 3:337–359.

Young SL, Bohenek DL, Fanselow MS. 1994. NMDA processes medi-ate anterograde amnesia of contextual fear conditioning induced byhippocampal damage: Immunization against amnesia by contextpreexposure. Behav Neurosci 108:19–29.

Zalfa F, Eleuteri B, Dickson KS, Mercaldo V, De Rubeis S, di PentaA, Tabolacci E, Chiurazzi P, Neri G, Grant SG, et al. 2007. A newfunction for the fragile X mental retardation protein in regulationof PSD-95 mRNA stability. Nat Neurosci 10:578–587.

14 EADIE ET AL.

Hippocampus