NF-κB/Rel Regulates Inhibitory and Excitatory Neuronal ......10.1128/MCB.00510-06. Mol. Cell. Biol.€2006, 26(19):7283. DOI: LeFevour, Shikha Chakraborty-Sett and Warner C. Greene

10.1128/MCB.00510-06.

2006, 26(19):7283. DOI:Mol. Cell. Biol. LeFevour, Shikha Chakraborty-Sett and Warner C. GreeneFoehr, Victor Han, Shao-ming Lu, Hakju Kwon, Anthony Alison O'Mahony, Jacob Raber, Mauricio Montano, Erik PlasticityExcitatory Neuronal Function and Synaptic

B/Rel Regulates Inhibitory andκNF-

http://mcb.asm.org/content/26/19/7283Updated information and services can be found at:

These include:

REFERENCEShttp://mcb.asm.org/content/26/19/7283#ref-list-1at:

This article cites 78 articles, 27 of which can be accessed free

CONTENT ALERTS more»articles cite this article),

Receive: RSS Feeds, eTOCs, free email alerts (when new

http://journals.asm.org/site/misc/reprints.xhtmlInformation about commercial reprint orders: http://journals.asm.org/site/subscriptions/To subscribe to to another ASM Journal go to:

on February 24, 2013 by P

EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/cgi/alerts

http://mcb.asm.org/

MOLECULAR AND CELLULAR BIOLOGY, Oct. 2006, p. 7283–7298 Vol. 26, No. 190270-7306/06/$08.00�0 doi:10.1128/MCB.00510-06Copyright © 2006, American Society for Microbiology. All Rights Reserved.

NF-�B/Rel Regulates Inhibitory and Excitatory Neuronal Functionand Synaptic Plasticity

Alison O’Mahony,1† Jacob Raber,3,4,5† Mauricio Montano,1† Erik Foehr,1†‡ Victor Han,5 Shao-ming Lu,6Hakju Kwon,1§ Anthony LeFevour,3 Shikha Chakraborty-Sett,7 and Warner C. Greene1,2*

Gladstone Institute of Virology and Immunology1 and Departments of Medicine and Microbiology and Immunology,2 University ofCalifornia, San Francisco, California 94141; Departments of Behavioral Neuroscience3 and Neurology4 and Division ofNeuroscience, Oregon National Primate Research Center,5 Oregon Health and Science University, Portland, Oregon 97239;

and Department of Neurology, Center for Aging and Developmental Biology,6 and Department of Microbiologyand Immunology,7 University of Rochester Medical Center, Rochester, New York 14642

Received 22 March 2006/Returned for modification 20 April 2006/Accepted 13 July 2006

Changes in synaptic plasticity required for memory formation are dynamically regulated through opposingexcitatory and inhibitory neurotransmissions. To explore the potential contribution of NF-�B/Rel to theseprocesses, we generated transgenic mice conditionally expressing a potent NF-�B/Rel inhibitor termed I�B�superrepressor (I�B�-SR). Using the prion promoter-enhancer, I�B�-SR is robustly expressed in inhibitoryGABAergic interneurons and, at lower levels, in excitatory neurons but not in glia. This neuronal pattern ofI�B�-SR expression leads to decreased expression of glutamate decarboxylase 65 (GAD65), the enzymerequired for synthesis of the major inhibitory neurotransmitter, �-aminobutyric acid (GABA) in GABAergicinterneurons. I�B�-SR expression also results in diminished basal GluR1 levels and impaired synapticstrength (input/output function), both of which are fully restored following activity-based task learning.Consistent with diminished GAD65-derived inhibitory tone and enhanced excitatory firing, I�B�-SR� miceexhibit increased late-phase long-term potentiation, hyperactivity, seizures, increased exploratory activity, andenhanced spatial learning and memory. I�B�-SR� neurons also express higher levels of the activity-regulated,cytoskeleton-associated (Arc) protein, consistent with neuronal hyperexcitability. These findings suggest thatNF-�B/Rel transcription factors act as pivotal regulators of activity-dependent inhibitory and excitatoryneuronal function regulating synaptic plasticity and memory.

Stimulus-coupled changes in synaptic plasticity are requiredfor the storage, retrieval, and removal of acquired informationcollectively referred to as memory formation (28, 32, 39). Suchchanges are facilitated by both modifications of existing syn-aptic effectors and the de novo synthesis of new gene productsregulated by various transcriptional regulators. These pro-cesses are tightly controlled by the coordinated action of bothexcitatory and inhibitory neurotransmitters derived from glu-tamatergic neurons and GABAergic (where GABA is �-amino-butyric acid) interneurons, respectively (47, 54). While the vastmajority of studies to date have focused on the cyclic AMP-responsive transcription factor (CREB) regulating excitatoryneuron function (7, 32–34, 62, 72), more recently, other tran-scription factors, including members of the NF-�B/Rel familyof transcription factors, have been implicated in experience-based synaptic adaptations (38, 45, 49, 55). However, our un-derstanding of their precise role in regulating synaptic plasticityremains rudimentary at best.

Although NF-�B/Rel factors were originally implicated ascentral regulators of the immune and inflammatory responses,

both basal expression and stimulus-coupled induction of NF-�B/Rel factors occur in neurons and glial cells (23, 30, 31, 45,48, 55). Activation of NF-�B/Rel proceeds through the site-specific phosphorylation, polyubiquitylation, and proteasome-mediated degradation of the major NF-�B/Rel inhibitor pro-tein, I�B� (41). The newly liberated NF-�B/Rel complexrapidly translocates into the nucleus, where it engages cognate�B enhancer elements in a variety of cellular target genesincluding the I�B� gene eliciting an auto-inhibitory feedbackloop (65). Substitution of the two key serine phospho-acceptorsites in I�B� with alanines (S32A/S36A) generates a potent,nondegradable inhibitor of NF-�B/Rel activation termed theI�B� superrepressor (I�B�-SR) (6, 61, 74), which serves as auseful tool to probe NF-�B action in vivo (40, 71).

As transcriptional regulators, NF-�B/Rel proteins can po-tentially either positively or negatively regulate the expressionof genes governing changes in synaptic plasticity and cognitivefunctions (73). Several reports support a positive link betweenthe activation of NF-�B/Rel factors and the induction of long-term potentiation (LTP) or long-term depression, experimen-tal correlates of learning and memory (1, 17, 20, 48). Obser-vations in the crab also support a positive role for NF-�B/Relaction in emotional learning and fear responses (16, 50). Sim-ilarly, mice lacking the p50 subunit of NF-�B/Rel display im-paired emotional learning and decreased anxiety-related re-sponses but exhibit increased exploratory activity (37, 38). Incontrast, other studies suggest a negative correlation betweenNF-�B action and synaptic function. For example, NF-�B/Relactivation has also been shown to impair the generation of

* Corresponding author. Mailing address: Gladstone Institute ofVirology and Immunology, University of California, San Francisco,1650 Owens St., San Francisco, CA 94158. Phone: (415) 734-4804. Fax:(415) 355-0153. E-mail: [email protected].

† A.O., J.R., M.M., and E.F. contributed equally to this work.‡ Present address: BioMarin Pharmaceutical, 105 Digital Drive, No-

vato, CA 94949.§ Present address: PDL BioPharma, Inc., 34801 Campus Drive, Fre-

mont, CA 94555.

7283


EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

synaptic currents in hippocampal neurons (20). Increased NF-�B/Rel action is also associated with the accelerated onset ofcognitive deficits in an experimental model of Alzheimer’sdisease (2).

These apparently conflicting observations could reflect dis-tinct roles for the various NF-�B/Rel factors in regulatingdifferent cognitive behaviors in select brain regions. Alterna-tively, these different outcomes may involve specific NF-�B/Rel actions in distinct neuronal subtypes and/or in glia. Insupport of the latter, Meffert et al. have reported that loss ofthe RelA/p65 subunit of NF-�B, in the context of concomitanttumor necrosis factor receptor 1 (TNFR1) deficiency, in bothneuronal and glial cells results in impaired spatial learning andmemory (49). Additionally, Fridmacher et al. and Kaltschmidtet al. (19, 29), using the CamKII promoter to direct expressionof the NF-�B/Rel inhibitor exclusively in excitatory neurons ofthe forebrain, have demonstrated a clear positive requirementfor NF-�B/Rel action in regulating synaptic plasticity andmemory. To date the potential function of NF-�B/Rel in in-hibitory GABAergic interneurons remains unexplored (47).

In our present study, we have utilized the prion promoterand enhancer (70) to direct the expression in neurons of apotent, nondegradable inhibitor of NF-�B/Rel activation,termed the I�B� superrepressor. I�B�-SR� mice display ro-bust I�B�-SR expression in inhibitory interneurons, somewhatlower levels of expression in excitatory neurons, and no detect-able expression in glia. These I�B�-SR animals were used toexplore the role of NF-�B/Rel in various electrophysiologicaland biochemical parameters in the brain. Our findings revealthat pan-neuronal inhibition of NF-�B action results in amarked enhancement of activity-dependent synaptic signalingand select cognitive functions including learning and memory.

MATERIALS AND METHODS

Generation and characterization of I�B�-SR bigenic mice. A total of eighttransgenic pTet-O-HA-I�B�-SR founder lines were generated by pronuclearinjection of linearized DNA into DBA inbred zygotes, and the resulting micewere screened by PCR to detect the presence of the transgene. Positive trans-genic mice were crossed with FVBN mice carrying a Prp/TtA transgene (Tet-off).Bigenic mice were screened by PCR for both transgenes. At 1 month of age,FVBN/DBA bigenic litters were divided into two groups and maintained oneither a standard or a doxycycline-supplemented diet. At 4 to 6 months of age,mice were anesthetized with isoflurane and sacrificed by cervical dislocation.Brains from control and I�B�-SR� mice were removed, fixed in 4% parafor-maldehyde, embedded in paraffin, sectioned, and stained with Nissl stain forhistological examination. Strict adherence to institutional and NIH guidelineswas maintained in all procedures relating to the care and treatment of mice. Asdiscussed in the results section, three independently derived transgenic lines,demonstrating readily detectable I�B�-SR transgene in the absence of doxycy-cline but near complete inhibition of I�B�-SR expression in the presence ofdoxycycline, were selected for further study. Use of three independently derivedfounder lines minimized effects related to the site of transgene integration.

Immunoprecipitation, immunoblotting, and immunofluorescence. Mice fromeach group were anesthetized with isoflurane, perfused with saline followed by4% paraformaldehyde, at 4 to 8 months of age. Brains were removed, and lysatesfrom whole tissue or specific regions were prepared with a low-stringency buffercontaining 50 mM Tris-HCl, pH 8, 120 mM NaCl, 5 mM EDTA, 0.5% (wt/vol)NP-40, supplemented with fresh 1� protease inhibitor cocktail (Calbiochem).Membrane-enriched fractions were isolated as described in Pharmingen proto-cols online. Lysates were either resolved by sodium dodecyl sulfate-polyacryl-amide gel electrophoresis directly or were first immunoprecipitated with hem-agglutinin (HA)-specific monoclonal antibodies (262K at a dilution of 1:200; CellSignaling) conjugated to protein G agarose (25 �l of packed beads). Membraneswere immunoblotted with HA-specific polyclonal antibodies (Y11 at a dilution of1:1,00; Santa Cruz Biotechnology) and the antibodies indicated in the legends

overnight at 4°C and visualized by chemiluminescence. For immunohistochem-ical and immunofluorescence analysis, brains were removed from selected micefollowing anesthetization with isoflurane and perfusion with saline and 4% para-formaldehyde and embedded in cryo-matrix mounting medium (22-oxyacalcitriol[OCT]; Tissue-Tek), frozen, and cryosectioned into 10-�m sections. Addition-ally, primary neuronal cultures (hippocampal or cortical) were grown on treatedslides and probed. Frozen sections or slides were fixed in 4% paraformaldehydeand permeabilized in 0.2% Triton X-100. Sections were either stained with Nisslor incubated with primary antibodies (indicated in the legend) overnight at 4°Cand probed with specific antibodies as outlined in the legends, followed withfluorescein isothiocyanate-conjugated or Alexa-conjugated secondary antibodies(1:1,000) for 60 min at room temperature. Sections were counter-stained withDAPI (4�,6�-diamidino-2-phenylindole; 1:500) for 5 min for nuclear staining.Sections were mounted in Gel/Mount (Biomeda) and visualized under UV lighton a Nikon E600 microscope connected to a SPOT advanced software camera.Several fields were compared for intensity of positive immunostaining by atechnician blinded to the genotypes.

Evaluation of NF-�B/Rel signaling by electrophoretic mobility shift assay(EMSA) and RPA. Bigenic I�B�-SR� and I�B�-SR� mice were treated withkainic acid (KA) (22 mg/kg) by intraperitoneal injection at 4 to 6 months of age(44). Mice were anesthetized, and hearts were perfused with saline at 7 to 8 hafter KA injection. Brains were removed and lysed in nuclear extraction buffer(20 mM Tris-HCl, pH 7.8, 125 mM NaCl, 5 mM MgCl2, 0.2 mM EDTA, 12%[wt/vol] glycerol, and 0.1% [wt/vol] NP-40) supplemented with 10 �g/ml aproti-nin, 1 mM phenylmethylsulfonyl difluoride, and 1 mM dithiothreitol. Sampleswith matched protein concentrations were incubated with a �B/Rel enhancerprobe radiolabeled with [32�-P]ATP. Nucleoprotein-�B/Rel complexes were sep-arated on nondenaturing gels and visualized by autoradiography. Control andbigenic mice were treated with KA as described above. After 8 h, brains wereremoved, and total RNA was isolated with an RNA isolation kit (Access RTPCR system; Promega) according to the manufacturer’s directions. An RNaseprotection assay (RPA) was performed using the Riboquant system (Pharmin-gen), and NF-�B-targeted RNA sequences were detected with a specific probetemplate set (mAPO-3).

Establishing primary neuronal cultures. Embryonic day 15 embryos wereremoved from gestating mice and placed in ice-cold 1� Hanks balanced saltsolution. Whole brains from these embryos were rapidly removed, and thehippocampal formation and cortex were isolated and washed in ice-cold 1�Hanks balanced salt solution. The tissue was dissociated with papain using aWorthington papain dissociation system, and isolated cells were plated at aconcentration of 4 � 104 cells/ml on poly-L-lysine-coated six-well plates in neu-robasal medium supplemented with B-27. Cortical cultures enriched inGABAergic neurons (95%) were grown in the presence of valine (25 �g/ml),pyruvate (2 mM), and �-ketoglutarate (5 mM) and in the absence of glutamine.Hippocampal cultures enriched in glutamatergic neurons (87%) were main-tained in medium supplemented with 2 mM glutamine as described previously(76). Cultures were incubated at 37°C in a humidified atmosphere containing 5%CO2. On day 3 the medium was changed and supplemented with conditionedculture medium and mitotic inhibitor solution (5-fluoro-2-deoxyuridine, cyto-sine-D-arabinofuranoside, and uridine). Fresh medium was added every 24 h. Onday 6, cultures were incubated with fresh medium without glutamic acid andmaintained for up to 2 weeks. In experiments using enriched glial cultures,cultures were shifted into Dulbecco’s modified Eagle’s medium/F12 L-valinelacking KCl but supplemented with 10% fetal calf serum, G5 (Gibco), andpenicillin/streptomycin to induce neuronal cell death.

Establishing postmitotic (adult) brain neurons and glial cells from the hip-pocampus. Hippocampal regions were rapidly dissected from the brains of post-natal day 1 (PN1) mice in 2 ml of Hibernate-A, supplemented with B27 and 0.5mM L-glutamine. Meninges and excess white matter were removed. The hip-pocampus was sliced perpendicularly to the long axis and transferred to a tubecontaining fresh Hibernate-A/B27 and 0.5 mM L-glutamine and rocked at 30°Cfor 30 min. The slices were transferred to fresh Hibernate-A with papain (20U/mg) and rocked at 30°C for 45 min. Slices were transferred to fresh Hibernate-A/B27 incubated at room temperature for 5 min. Slices were pipetted 15 times,and the pieces were allowed to settle for 2 min. The supernatant (cell suspension)was removed and carefully applied to the top of a gradient (35%, 25%, 20%, and15%) formed with OptiPrep in Hibernate-A/B27 medium and centrifuged at800 � g for 15 min. The top 6 ml (debris) was discarded. The underlying 2 ml,designated fraction 1, was enriched for oligodendrocytes. The next layer, fraction2, contained neurons with accessory glia cells. Fraction 3 was enriched forneurons, and the pellet, fraction 4, was enriched for microglia. Each fraction waspelleted at 200 � g for 1 min and then resuspended in Neurobasal-A/B27 withL-glutamine and 1� gentamicin. Neurons were plated at 90 to 320 cells/mm2 on

7284 O’MAHONY ET AL. MOL. CELL. BIOL.


EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

poly-D-lysine-coated plates in fresh Neurobasal-A/B27 medium supplementedwith 0.5 mM L-glutamine, 10 �g/ml gentamicin, and 5 ng/ml fibroblast growthfactor-. Glial cells from each representative fraction (oligodendrocytes, micro-glia, and mixed glial fraction), were resuspended at a concentration of 3 � 106

cells in 12 ml of MEM10 containing 1% penicillin/streptomycin, and 2 ml of eachsuspension was added to each well of a six-well plate precoated with poly-D-lysine. All cultures were maintained in a highly humidified atmosphere contain-ing 5% CO2 at 37°C.

Reverse transcription-PCR analysis of GAD65 mRNA levels. Whole hip-pocampus from either untrained or trained (subjected to maze testing) mice ineach genotypic group was collected for RNA quantification. RNA was extractedusing an RNAeasy kit from QIAGEN. The purified RNA was DNase treated(Ambion, Austin, TX) and reverse transcribed from mRNA to cDNA using afirst-strand synthesis kit (Invitrogen, Carlsbad, CA). The amount of cDNA wasquantified using real-time PCR and primers sets designed to amplify GAD65.Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and -actin mRNA lev-els were used as internal controls for normalization. Amplified mRNA tran-scripts were visualized on 1.5% agarose gels.

Characterization of I�B�-SR bigenic mice. At 1 month of age, bigenic litter-mates were screened by PCR to confirm the presence of the transgene, dividedinto two groups, and maintained on a normal or doxycycline-supplemented diet.At 4 to 8 months, male mice from three independently derived founder lineswere selected for further analysis. Progeny from each founder line behavedsimilarly in each of the various assays.

Electrophysiological analysis. (i) Slice preparation. The animals were firstanesthetized with halothane, and brains were rapidly removed and cooled inice-cold artificial cerebrospinal fluid (ACSF) containing the following: 125 mMNaCl, 5 mM KCl, 1.25 mM NaH2PO4, 25 mM NaHCO3, 25 mM D-glucose, 2 mMCaCl2, 1 mM MgSO4, 0.01 mM glycine, and 1 mM kynurenic acid bubbled witha mixture gas of 95% O2 and 5% CO2. For slicing, 1 mM kynurenic acid was alsoadded to ACSF. The solution pH was adjusted to 7.3 to 7.4, and osmolarity wasset at 300 � 10. The hippocampus was dissected, and 350-�m slices were cuttransversely. A cut was routinely made between the CA1 and CA3 area rightafter slicing, and in some recordings, both CaCl2 and MgSO4 concentration wasincreased to 4 mM to reduce seizure-like activity and population spikes. All sliceswere collected in a holding chamber in the same solution maintained at 31 � 1°C.A single slice was transferred to an interface recording chamber that was con-stantly perfused with the gas-saturated ACSF at 1 ml/min at 29 � 1°C. Therecording chamber was also constantly superfused with gas-saturated moist airduring recording. The remaining slices were kept in the holding chamber untiltested.

(ii) Extracellular field recordings. Field recordings were used to assess syn-aptic transmission and plasticity in the Schaffer collateral/commissural CA1pathway. For field excitatory postsynaptic potential (fEPSP) recordings, a glasspipette was filled with the ACSF or 1 mM NaCl (3 to 5 M resistance) and placedwithin the striatum radiatum. For high-frequency stimulation, two bipolar tung-sten electrodes (S1 and S2) were placed on opposite sides of the recordingelectrode along the Schaffer collateral fibers in the striatum radiatum. The teststimulation was delivered alternately through S1 and S2 once per minute. After30 to 60 min of recording for control tetanus stimulation, three or four trains ofsquare pulses (100 pulses at 100 Hz) were delivered with 3- to 4-min intervalsthrough S1, while the S2 was turned off. In some cases, theta burst stimulationwas also used. These two stimulation methods were compared in the previouswhole-cell recordings, and no clear difference was noted. For late-phase long-lasting LTP (L-LTP), field potentials evoked by S1 and S2 were monitored in thesame way as in the control for a minimum of 180 min and, in most cases, up to300 min after the LTP induction protocol. The peak amplitude (or 30 to 70%rising slope) of all fEPSPs recorded from an individual slice were normalized tothe mean peak amplitude (or slope) during the 30 min before the theta burst orhigh-frequency stimulation, and these normalized values were used to comparedLTPs induced in various treatment groups. Field potentials were recorded withan Axon amplifier 2B (Axon Instruments, Union City, CA). Data acquisition andanalysis were done off-line using P-Clamp 9 software and Origin. All data arepresented as the mean � standard error of the mean. Significance was assessedat a P value of �0.05, using Dunnett’s t test. Any recordings lasting less than 3 hwere excluded from the final analysis. An average of three slices/mouse and threemice/group were used for the analysis of L-LTP. A researcher blinded to thegenotypes of the mice performed all recordings.

Behavioral analysis. Control and bigenic mice were assessed at 6 to 8 monthsof age in a blind-controlled series of behavioral tests, including (i) a Morris watermaze test and (ii) a radial arm maze.

(i) Morris water maze test. A pool (diameter 140 cm) was filled with opaquewater (24 � 1°C), and mice were trained to locate first a visible platform (days

1 to 2) and then a submerged hidden platform (days 3 to 5) in two daily sessions3.5 h apart, each consisting of three 60-s trials (10-min intertrial intervals). Miceacquired spatially encoded information with visual cues outside the maze tolocate the platform. Mice that failed to find the hidden platform within 60 s wereplaced on the platform for 15 s. For analysis of data, the pool was divided intofour quadrants. During the visible platform training, the platform was moved toa different quadrant for each session. During the hidden platform training, theplatform location was kept constant for each mouse (in the center of the targetquadrant). The starting point at which the mouse was placed into the water waschanged for each trial. Time to reach the platform (latency), path length, andswim speed were recorded with a Noldus Instruments EthoVision video trackingsystem set to analyze two samples per second. Since there were no significantdifferences in average swim speeds between the different groups of mice duringthe visible platform sessions, the time required to locate the platform (latency)was used as the main measure for analysis. A 60-s probe trial (platform removed)was carried out as described in the legend.

(ii) Radial arm maze. One week before testing, animals were placed on arestricted diet of 80% to 90% of food levels so that their initial body weightdecreased by a maximum of 15%. This diet was maintained throughout thetesting period; however, mice were given free access to water. Mice received foodreward pellets in their home cages 1 day before pretraining to become accus-tomed to the novel food in a familiar environment. For pretraining, each mousewas placed in the central platform of the eight-arm radial maze and allowed toexplore and consume food pellets scattered throughout the entire maze for a15-min period. During training, mice were allowed to take a pellet from eachfood dispenser located at the distal end of each arm. A trial was finished after thesubject mouse consumed the pellet in each of the eight arms. As a mouseconsumed a pellet from each arm, the next arm opened, and when the mouseentered this arm, the first arm closed. Thus, the mouse was sequentially guidedthrough each arm of the maze. Immediately after the training, maze acquisitiontrials were performed with all eight arms baited with food pellets. Mice wereplaced on the central platform and allowed access to all eight arms for 15 min.The session was terminated immediately after all eight food pellets concealed atthe end of the arms were consumed as measured by breaking a sensor beam(“head poke”) or after 15 min. An “arm visit” was defined as entering the armand breaking both sensor light beams. Mice were confined in the center platformfor 2 s after each arm choice, thus reducing behaviors such as clockwise serialsearching strategies. Animals were subjected to one session per day. A MedPCsoftware program was used for both the training and testing phases. For eachtrial, the following were automatically recorded: (a) latency to complete themaze and retrieve all pellets, (b) number of errors, (c) choice of arms, and (d)number of different arms chosen within the first eight choices. The operator alsomanually recorded the number of pellets eaten by each mouse. Each of theseparameters was used to detect abnormalities in the acquisition and retention ofspatially encoded information. In each case, the operator was blinded to thegenotype of the mice being examined. Data are expressed as mean � standarderror of the mean. Differences among means were evaluated by analysis ofvariance (ANOVA) and a Tukey-Kramer test. Learning curves were comparedby a repeated-measures ANOVA using contrasts to assess differences betweenspecific groups of mice. For all analyses, the null hypothesis was rejected at the0.05 level.

RESULTS

Generation and characterization of transgenic mice ex-pressing an I�B�-SR inhibitor of NF-�B action in neurons. Toexplore the relationship between NF-�B/Rel factor activation,synaptic signaling, and higher-order cognitive function, wegenerated multiple independent lines of transgenic mice ex-pressing an HA-tagged version of the NF-�B/Rel inhibitor,I�B�-SR. The I�B�-SR transgene was cloned downstream ofa tetracycline transactivator (tTA)-responsive promoter(Tet-O) facilitating regulatable expression. By breeding thesemice to a second transgenic line of mice where the prionpromoter (Prp) was used to direct tTA expression (a generousgift from Stanley Prusiner, University of California, San Fran-cisco), expression of the Tet-O-I�B�-SR transgene was re-stricted to neurons. In the presence of the blood-brain barrier-permeable tetracycline analogue, doxycycline, transcription of

VOL. 26, 2006 NF-�B REGULATES INHIBITORY INTERNEURON FUNCTION 7285


EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

the tTA-regulated gene is blocked (Fig. 1A) (22). In the pres-ence or absence of doxycycline, bigenic Prp-tTA/Tet-O-HA-I�B�-SR mice were both viable and fertile. Furthermore, thesemice displayed anatomically and structurally normal gross

brain morphology as indicated by Nissl and anti-calbindinimmunofluorescence staining of the hippocampal regions(Fig. 1B).

In each of the eight independently derived lines, expression

FIG. 1. Generation and biochemical characterization of bigenic I�B�-SR mice. (A) Transgenic mice encoding an N-terminal, HA-taggedI�B�-(SS32/36AA) superrepressor under the control of tetracycline response element (Tet-O) were crossed with transgenic mice expressing thetetracycline transactivator (tTA) under the control of the prion promoter-enhancer (Prp/tTA) to generate bigenic mice that express I�B�-SR inthe absence of doxycycline. (B) I�B�-SR� bigenic mice from three independently derived bigenic lines (FVBN/DBA), exhibit grossly normalhippocampal morphology as assessed by Nissl staining and calbindin immunostaining. (C) Doxycycline regulation of I�B�-SR transgene expres-sion. HA-I�B�-SR expression in bigenic lines (lanes 1 to 6) in the absence and presence of doxycycline versus control Prp/tTA (asterisk denotesbigenic lines selected for further analysis) (lane 7) and pTet-O-HA-I�B�-SR (lane 8) singly transgenic lines. Bigenic litters were divided at 1 monthof age and maintained for 28 to 31 days on a normal or doxycycline-supplemented diet (asterisk denotes transgenic lines selected for furtheranalysis). (D) Evaluation of transgene expression in various tissues in I�B�-SR� mice. (E) I�B�-SR transgene expression in distinct regions of thebrain in an I�B�-SR� mouse. (F) Levels of I�B�-SR protein expressed in the hippocampal region were similar to the levels of endogenous I�B�protein in both I�B�-SR� and control mice. (G) Detection of I�B�-SR transgene expression using indirect immunofluorescence with anti-HAantibodies (green fluorescence). Glial and neuronal cells were identified using with anti-GFAP or anti-MAP2 antibodies, respectively, withAlexa568-conjugated secondary antibodies (red fluorescence). Certain sections were counterstained with DAPI (blue fluorescence) to identifynuclear regions. All mice designated I�B�-SR� are on a diet supplemented with doxycycline to suppress transgene expression.



EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

of the I�B�-SR transgene was clearly detectable (designatedI�B�-SR� or SR�), and its expression was suppressed in thepresence of doxycycline (designated I�B�-SR� or SR�) (Fig.1C). Three independently derived lines of FVBN/DBA bigenicmice exhibiting the tightest regulation of transgene expressionby doxycycline and control mice were selected for further anal-ysis (Fig. 1C, designated by asterisks) to ensure that the ob-served effects were attributable to transgene expression ratherthan to a positional effect related to the site of transgeneinsertion. Among several tissues examined, expression of theI�B�-SR transgene appeared confined to brain, and within thebrain, transgene expression was detected in all regions exam-ined, including cerebellum, cortex, thalamus, and mid-brain(Fig. 1D and E). The levels of HA-tagged transgenic I�B�-SRprotein in hippocampal neurons isolated from I�B�-SR� miceproved comparable to the levels endogenous I�B� proteinobserved in either I�B�-SR� or control mice (Fig. 1F). Ofnote, the level of endogenous I�B� protein was lower in I�B�-SR� neurons due to the loss of NF-�B/Rel action required forthe auto-regulatory induction of the I�B� gene (65). Theseequivalent levels of transgene and endogenous protein miti-gate against an artifactual gain-of-function phenotype some-times observed when transgenes are overexpressed in vivo.

Immunofluorescence staining of I�B�-SR� brain sectionsconfirmed I�B�-SR transgenic protein expression in neuronsfrom the hippocampus, cortex, medulla, hypothalamus, andPurkinje cells of the cerebellum (data not shown). The HA-tagged I�B�-SR protein was not detected in matched neuro-anatomical regions from control or I�B�-SR� mice. In vivoexpression of the HA-I�B�-SR gene product was not detectedin macroglia when sections were probed with both anti-HA

and anti-glial fibrillary acidic protein (GFAP) antibodies, butexpression was detectable in MAP2-positive neurons (Fig. 1G).

To confirm functional inhibition of NF-�B in vivo followingI�B�-SR expression, we assessed both DNA binding and in-duction of NF-�B target genes. Using EMSAs, we comparedKA-induced NF-�B/Rel DNA binding activity in I�B�-SR�

mice and I�B�-SR� mice. While robust NF-�B/Rel activationwas detected in various brain regions from I�B�-SR� mice,this response was markedly impaired in the I�B�-SR� mice(Fig. 2A). These KA-induced nucleoprotein complexes con-tained p50, p52, c-Rel, and RelA protein species as determinedby supershifting with specific antibodies (data not shown). Pri-mary neuronal or glial cultures from embryonic brain tissuesfrom I�B�-SR�, I�B�-SR�, and control mice were either un-treated (�) or exposed to TNF-� or high concentrations ofKCl (22 mM) to induce membrane depolarization. In I�B�-SR� neurons, the activation of NF-�B/Rel in response to ei-ther stimuli was markedly inhibited under both basal and stim-ulated conditions. In contrast, NF-�B/Rel DNA binding wasevident in both unstimulated and, to a greater extent, in stim-ulated in neurons from both I�B�-SR� and control mice (Fig.2B). Consistent with expression of I�B�-SR in neurons, butnot glia, NF-�B/Rel activation was detected in glial culturesfrom I�B�-SR� mice as well as from I�B�-SR� and controlmice (Fig. 2C).

RPAs were used to detect KA induction of various endog-enous NF-�B/Rel-regulated target genes, specifically Fas andthe p55 TNF-� receptor (Fig. 2D). RPAs revealed thatI�B�-SR expression in neurons effectively inhibited KA induc-tion of the Fas and p55 TNF-� receptor genes. In contrast,both genes were effectively induced by KA in both I�B�-SR�

FIG. 2. Functional inhibition of NF-�B/Rel DNA binding by I�B�-SR expression. (A) NF-�B/Rel activation induced by in vivo administrationof KA in various brain regions from I�B�-SR� and I�B�-SR� mice. (B and C) NF-�B/Rel activation in neurons or glia under basal, untreatedconditions (�) or induced by either TNF-� or membrane depolarization (22 mM KCl). NF-�B DNA binding was inhibited in primary neuronalcultures but was intact in glial cultures isolated from I�B�-SR� mice. (D) RNase protection assays revealed a KA-induced temporal increase inthe mRNA expression levels for two NF-�B/Rel target genes, Fas and TNFR p55, in whole-brain preparations from control mice. (E) After 12 hstimulation with KA, neither Fas nor TNFR p55 mRNA expression was induced in I�B�-SR� mice. In contrast, robust mRNA transcript levelswere seen in whole-brain RNA preparations from I�B�-SR� and control mice. GAPDH was used as a loading control in the RPA. NS, nonspecificband.



EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

and control mice (Fig. 2E). Together these findings demon-strate that expression of the I�B�-SR transgene is restricted toneurons, is tightly regulated by doxycycline, can functionallyinhibit KA-induced nuclear translocation of NF-�B/Rel factorsin neurons but not in glial cells, and is able to effectively impairthe activation of various endogenous NF-�B/Rel-inducible tar-get genes.

I�B�-SR is expressed in both cortical inhibitory GABAergicinterneurons and hippocampal excitatory glutamatergic neu-rons. To assess potential differences in I�B�-SR expression infunctionally distinct neuronal subtypes, we probed HA-specificimmunoprecipitations of lysates prepared from primary hip-pocampal excitatory neurons or from cortical inhibitoryGABAergic neuronal cultures. HA-I�B�-SR transgene ex-pression was detected only in I�B�-SR� lysates and, takinginto account -tubulin protein input levels, expression levels ofI�B�-SR approximately seven times higher were detected incortical cultures enriched for GABAergic interneurons relativeto levels detected in hippocampal cultures enriched for exci-tatory glutamatergic neurons (Fig. 3A). The relative levels ofinhibitory GABAergic neurons and excitatory glutamatergicneurons present in cortical cultures and hippocampal cultureswere quantitated by immunofluorescence staining using anti-bodies specific for vesicular GABA transporter VGAT(GABAergic) versus vesicular glutamate transporter VGLUT(glutamatergic), respectively. Cortical neuronal cultures wereenriched for GABAergic interneurons (98% VGAT� cells),whereas hippocampal cultures primarily contained glutamater-gic cells (83% VGLUT� cells) (Fig. 3B). Enhanced prion-promoter driven HA-I�B�-SR expression in GABAergic in-terneurons is consistent with the activity of the prion promoterin vivo, where either the endogenous prion protein or a Prp-enhanced green fluorescent protein transgene was robustlyexpressed in these neuronal cells with little or no protein de-tected in glia (5, 51).

Inhibition of NF-�B by I�B�-SR results in deceased GAD65expression in inhibitory GABAergic interneurons. Probinghippocampal lysates from I�B�-SR� mice revealed markedlydecreased levels of glutamate decarboxylase (GAD65), therate-limiting enzyme required for the synthesis of the inhibi-tory neurotransmitter, GABA (Fig. 3C). However, no differ-ence in the levels of the vesicular GABA transporter protein(VGAT) was observed in lysates from mice in all three groups,arguing against a selective loss of these inhibitory GABAergicinterneurons in I�B�-SR� mice (Fig. 3C). Similarly, primaryneuronal cultures isolated from I�B�-SR� mice expressed sig-nificantly lower levels of GAD65 (Fig. 3D).

Consistent with the immunoblotting results, GAD65 mRNAlevels were decreased relative to levels seen in either I�B�-SR� or control mice. Both end-point PCR and real-time PCR(Fig. 3E) analysis confirmed lower GAD65 mRNA transcriptsin GABAergic interneurons in I�B�-SR� mice, suggestingthat the GAD65 gene is regulated either directly or indirectlyby NF-�B/Rel. Together, these findings indicate that NF-�B/Rel is required for expression of GAD65 in GABAergic inter-neurons and suggest that NF-�B/Rel factors may regulate in-hibitory neuronal function.

Inhibition of NF-�B/Rel action by I�B�-SR expression inneurons leads to increased LTP in the hippocampus. We nextinvestigated the potential impact of I�B�-SR expression in

both inhibitory and excitatory neurons on the induction of LTPfollowing tetanus or high-frequency stimulation. The averagesof peak amplitudes of fEPSPs (S1) from all trials collected 15to 30 min before the induction of LTP (100%) were used tonormalize peak amplitude of fEPSPs of each individual trialacquired over 300 min after tetanus stimulation (Fig. 4A-D).Similar recordings were obtained using theta burst high-fre-quency stimulation (data not shown). While tetanus stimula-tion of the Schaffer-collateral pathway induced higher fEPSPamplitude (LTP) in all three groups, we observed significantlygreater potentiation in hippocampal slices from several I�B�-SR� mice (Fig. 4D). In all cases, the enhanced S1 fEPSP lastedfor a minimum of 180 min, and in most cases, responses wererecorded for 300 min, consistent with enhanced L-LTP. Tomonitor basal synaptic signaling, a second stimulating elec-trode was placed at the other side of the recording electrode toevoke control fEPSPs (S2) between S1 trials. The S2 fEPSPsevoked by this unstimulated pathway did not show any signif-icant change after the tetanus stimulation of the S1 pathwayand remained steady for the duration of the recording. TheLTP induced in hippocampal slices from I�B�-SR� mice wasconsistently higher than that observed in either I�B�-SR� orcontrol mice, (P � 0.05, Dunnett’s t test) (Fig. 4D). The ob-served enhancement of synaptic signaling did not appear toreflect an overall increase in synapse numbers in I�B�-SR�mice. Similar levels of synapsin, synaptophysin, and complexinas well as myelin basic protein were observed in sections orhippocampal membrane fractions isolated from either I�B�-SR�, I�B�-SR� or control mice. In addition, overall neuronaland glial cell content was equivalent in mice from all threegroups as evidenced by comparable levels of the neuronal cellmarker MAP2 or the glial cell marker GFAP (data not shown).

The induction of an immediate early gene, arc (for activity-regulated, cytoskeleton-associated), was used as a marker ofincreased excitability of glutamatergic neurons potentially as aresult diminished GABA-mediated inhibitory inputs (64). Asexpected, Arc protein was not expressed under basal condi-tions in hippocampal cultures isolated from either I�B�-SR�,I�B�-SR�, or control mice (Fig. 4E). However, using lowMg2�/high K� supplemented medium to facilitate neuronalexcitation (3), robust Arc expression was detected in culturesfrom I�B�-SR� mice but not from I�B�-SR� or control mice(Fig. 4F). Arc induction was blocked in I�B�-SR� culturesfollowing pretreatment with the NMDA (N-methyl-D-aspar-tate) antagonist, AP5, but not with the AMPAR ((alpha-ami-no-3-hydroxy-5-methyl-4-isoxazole propionic acid receptor)antagonist, CNQX (6-cyano-2,3-dihydroxy-7-nitro-quinoxa-line) (Fig. 4F). Of note, the induction of Arc by CNQX in allthree mice lines is consistent with a recent report in whichinhibiting AMPAR activity strongly potentiates activity-depen-dent Arc expression at the level of transcription by blockingAMPAR-coupled Gi signaling (60). Taken together, thesefindings support the notion that I�B�-SR expression in neu-rons may disturb the homeostatic balance between inhibitoryinterneurons and excitatory neuronal functions, leading to hy-perexcitability, increased LTP, and induction of activity-depen-dent target genes like arc.

Selective inhibition of NF-�B/Rel action in neurons alterscognitive behaviors. To explore in vivo the impact of I�B�-SRexpression in neurons on cognitive function, four groups of



EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

FIG. 3. I�B�-SR expression in both excitatory glutamatergic neurons and inhibitory GABAergic interneurons. (A) Neuronal cultures wereisolated from embryonic day 17 brains from I�B�-SR� and I�B�-SR� mice. Lysates from primary cortical or hippocampal regions enriched inglutamatergic or GABAergic neurons, respectively, were immunoprecipitated and probed by immunoblotting using HA-specific antibodies.(B) Primary cortical neuronal cultures (85% GABAergic interneurons) were maintained in neurobasal medium supplemented with B27containing valine (25 �g/ml) and in the absence of glutamine. Hippocampal neuronal cultures enriched in glutamatergic neurons were maintainedin neurobasal medium supplemented with B27 and glutamine (2 mM). Relative numbers of glutamatergic or GABAergic neurons in each culturewere assessed using immunofluorescence staining to quantify VGLUT versus VGAT expression, respectively. Mean fluorescence intensity wasassessed from a total of 25 fields from three individual cultures. (C) Hippocampal lysates were prepared from three individual controls (CTL) orI�B�-SR� (on doxycycline) or I�B�-SR� mice. Lysates were probed by immunoblotting to assess expression levels of GABAergic interneuron-specific marker proteins, specifically GAD65, VGAT relative to tubulin levels. (D) Hippocampal neuronal cultures from mice in each cohort werealso probed by immunofluorescence with GAD65-specific antibodies. (E) GAD65 mRNA levels isolated from hippocampal lysates were assessedby PCR or real-time PCR. GAPDH and -actin mRNAs were used as control (CTL) for reverse transcription-PCR. �, P � 0.01.

7289


EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

FVBN/DBA genetically matched male mice at 6 to 8 months ofage were evaluated in a range of behavioral tests. Each cohortincluded mice from each of the following groups: (i) nontrans-genic or singly transgenic mice (control; n 57), (ii) bigenicmice from three independently derived lines maintained on

doxycycline from 1 month of age (I�B�-SR�; n 63), (iii)bigenic mice off doxycycline and expressing the I�B� superre-pressor (I�B�-SR�; n 91), and (iv) bigenic mice on doxycy-cline in utero to suppress transgene expression during devel-opment but subsequently switched to a nondoxycycline diet at

FIG. 4. I�B�-SR� transgenic mice exhibit enhanced LTP and increased Arc expression as indicators of neuronal hyperexcitability. Repre-sentative recordings of tetanus stimulation induced LTP along the Schaffer collateral/commissural CA1 pathway in hippocampal slices taken fromcontrol (A), I�B�-SR� (B), and I�B�-SR� (C) mice. (D) Total LTP levels induced in hippocampal slices from mice in each group. The datarepresents the mean � standard error of the mean of three slices/brain and three mice per genotype. (E) Induction of Arc as a marker of neuronalhyperexcitability in primary neuronal cultures incubated in either normal or low Mg 2�/high K� supplemented medium. Under basal conditions,Arc expression is undetected (upper panel). In contrast, in low Mg 2� conditions, Arc induction is only induced in I�B�-SR� neurons, reflectinghyperexcitability in the context of diminished inhibitory tone. (F) Arc induction was blocked by the NMDAR antagonist AP5 but not by theAMPAR antagonist CNQX. �, P � 0.05.



EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

1 month of age to allow transgene expression (I�B�-SR�*; n 29). Of note, all I�B�-SR� mice appeared to be significantlymore active in their home cages and were much more sensitiveto handling-induced seizures.

Using the elevated plus maze to assess anxiety levels, nodifferences were detected in mice from any of the groups. Thisnormal or unimpaired anxiety phenotype was confirmed usingthe zero maze, with mice in all groups spending equivalentamounts of time in the maze (Table 1). Interestingly, in theplus maze, both I�B�-SR� and I�B�-SR�* mice consistentlyexhibited increased exploratory activity as indicated by thetotal distance moved in both the open arms (P � 0.01, Tukey-Kramer test) and in the closed arms of the plus maze (P � 0.01,Tukey-Kramer test) (Table 1). I�B�-SR� and I�B�-SR�*mice also exhibited heightened exploratory activity with morerearing events relative to their doxycycline-treated littermates(I�B�-SR�) or control mice, (P � 0.01) in the open-field test(52). This increased exploratory activity could reflect dimin-ished inhibitory tone leading to increased excitatory neuronalactivity in the I�B�-SR� mice. However, direct effects relatedto I�B�-SR expression in excitatory neurons cannot be ex-cluded.

In evaluating hippocampus- and cortex-dependent nonspa-tial learning and memory by novel object recognition testing(59), we observed that mice in all three groups displayed anormal response. All mice explored both objects equivalentlyon day 1 and spent more time exploring the novel object versusthe familiar object on day 2 of testing (P � 0.01) (data notshown). Of note, I�B�-SR� mice spent significantly more timeoverall exploring both objects on each day, consistent with anenhanced exploratory behavior. Examination of sensorimotorfunction by rotorod testing also revealed no significant differ-ences between these various transgenic animals (data notshown).

I�B�-SR� mice exhibit enhanced performance in spatialmemory tests. In view of a recent report describing impairedspatial learning in mice lacking p65/RelA expression andNF-�B action in all neurons and in glial cells (49), we evalu-ated spatial memory in the I�B�-SR�, I�B�-SR�, and controlmice. Using the Morris water maze, we observed that the

I�B�-SR� mice located both the visible and hidden platformsin significantly less time than their control counterparts (P �0.01) (Fig. 5A). Swim speeds for mice in all groups were similar(control, 17.9 � 0.7 cm/s; I�B�-SR�, 14.8 � 0.3 cm/s; I�B�-SR�, 16.7 �0.6 cm/s; I�B�-SR�*, 16.5 � 0.5 cm/s). These dataindicate an overall enhancement of spatial learning and mem-ory in I�B�-SR� mice. There was a significant genotype bysession interaction in both phases of the test (P 0.0433,repeated-measures ANOVA). Interestingly, I�B�-SR�* miceexpressing the I�B�-SR transgene after 1 month of age exhib-ited even stronger acceleration of spatial learning and memory.Bigenic I�B�-SR� mice learned much faster than I�B�-SR�

or control mice (Fig. 5A, inset), as shown by the change inoverall performance between the first and last sessions of eachtest phase and by the changes in latency (per mouse) betweentwo subsequent sessions of the test (P � 0.05, Tukey-Kramertest). Conversely, bigenic I�B�-SR� mice on doxycycline,which silenced transgene expression, displayed levels of spatiallearning and memory similar to control mice.

The retrieval of spatially encoded information (i.e., mem-ory) was assessed by a probe trial (57), in which the platformwas removed from the water maze and the mice were exam-ined for their ability to remember the specific target quadrantwhere the platform was previously located. After completionof the test, both I�B�-SR� and I�B�-SR� mice spent signif-icantly more time in the target quadrant than in any otherquadrant, indicating that all mice had learned and recalled theplatform location by the end of testing (P � 0.01;, Tukey-Kramer) (data not shown). However, when probe trials wereperformed at earlier times during testing with a second cohortof I�B�-SR� and I�B�-SR� mice (n 5 for each group), weobserved that I�B�-SR� mice spent significantly more time inthe target quadrant than in any other quadrant even after onlyone day of testing (for SR�, P � 0.05; target versus any otherquadrant) (Fig. 5B). In contrast, the I�B�-SR� mice failed toshow any preference for the target quadrant on day 1. On day2, the I�B�-SR� mice showed an even stronger preference forthe target quadrant (for SR�, P 0.0051; P � 0.05 targetversus right quadrant and P � 0.01 target versus left andopposite quadrants). While the I�B�-SR� mice exhibited atrend toward favoring the target quadrant on day 2, this resultdid not reach statistical significance (P 0.198). Even on day3, the I�B�-SR� mice continued to show a preference for thetarget quadrant, (SR�, P � 0.05; target versus any other quad-rant). In contrast to the accelerated retrieval of spatial memoryobserved in I�B�-SR� mice, I�B�-SR� mice display a prefer-ence for the target quadrant only on day 3 of testing (SR�, P 0.0025; P � 0.05 target versus right quadrant and P � 0.01target versus left and opposite quadrants). These findings areconsistent with the overall enhanced performance of I�B�-SR� mice in the Morris water maze.

Since the Morris water maze test may involve a componentof stress relayed to exposure to water, we tested a secondcohort of mice using the radial arm maze, paralleling the studydescribed by Meffert et al. (49). Using time to complete themaze (latency) as a measure of spatial learning and memory,we observed that I�B�-SR� mice again displayed improvedspatial learning and memory relative to either I�B�-SR� orcontrol mice (P � 0.01) (Fig. 5C). After two pretraining ses-sions, faster times were recorded for I�B�-SR� mice on day 1

TABLE 1. Activity and anxiety phenotypes as assessedby plus and zero mazesa

GenotypeDistance moved in plus maze (cm) Time spent

in zeromaze (s)Open arms Closed arms

I�B�-SR� 507 � 54† 2,045 � 128† 61.6 � 8.1I�B�-SR�* 834 � 111† 2,402 � 96† 66.9 � 21.6I�B�-SR� 144 � 17 998 � 97 63.9 � 6.9Control 257 � 49 1,111 � 174 51.6 � 5.2

a Anxiety levels were assessed using the elevated plus maze (consisting of twoopen arms and two perpendicular arms enclosed by high walls, intersecting toform a plus sign shape) and recording the amount of time mice spent exploringthe open arms relative to time in the closed arms. While no differences in anxietylevels (i.e., time) were observed among mice in all groups, both I�B�-SR� miceand I�B�-SR�* mice (i.e., expressing the superrepressor transgene only after 1month of age) consistently exhibited increased exploratory activity, as indicatedby the total distance moved in both the open arms (†, P � 0.01, Tukey-Kramer)and the closed arms of the plus maze (†, P � 0.01, Tukey-Kramer). A lack ofdifference in anxiety was confirmed using the zero maze, a modified circular plusmaze with two enclosed and two open quadrants. All values are the means �standard error of the means.



EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

7292


EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

of the maze test and the I�B�-SR� mice continued to exhibitenhanced performance throughout the remainder of the test.Additionally, using the “number of errors” in the radial armmaze paradigm as reported by Meffert et al. for p65�/�

TNFR�/� mice, the I�B�-SR� mice also performed betterthan the I�B�-SR� or control mice (Fig. 5C, inset).

To determine the relationship between maze performanceand synaptic signaling, we assessed potential correlations be-

tween the magnitude of LTP recorded (percent over baseline)and latency in radial arm maze-trained mice. Striking correla-tions between the increase in LTP and either the time recorded(r �0.95; P 0.0134; n 5) or in errors made in the lasttrial (r �0.98; P 0.0048; n 5) were observed. Correctingthe maze data for potential differences in performance in trial1 [percent improvement measure calculated as the (perfor-mance in the first trial � performance in the last trial)/perfor-

FIG. 5. I�B�-SR� bigenic mice exhibit enhanced spatial learning and memory. (A) I�B�-SR� mice located a visible platform significantlyfaster than control or I�B�-SR� mice (P � 0.01). I�B�-SR� mice also showed enhanced performance in the hidden platform phase of the test(P � 0.01). I�B�-SR�* bigenic mice were exposed to doxycycline in utero and up to 1 month of age to silence transgene expression andconsequently express the superrepressor only after 1 month of age. (Inset) I�B�-SR� bigenic mice had a faster overall rate of learning in bothphases of water maze test. In determining the rate of learning, sessions 4 and 5 were not included as a pair as they represent the border betweenthe visible and hidden sessions. (B) A probe trial was performed after each day of hidden platform testing. At earlier times, I�B�-SR� mice spendsignificantly more time in the target quadrant than their genetically identical I�B�-SR� siblings, suggesting that these mice remember the platformlocation earlier. (C) I�B�-SR mice completed the radial maze in consistently less time and made fewer errors (inset) than either control orI�B�-SR� mice (P � 0.01), indicating enhanced spatial learning and memory. �, P � 0.05, ��, P � 0.01.

FIG. 6. Activity dependent recovery of synaptic transmissions and GluR1 levels in trained I�B�-SR� mice. (A) Examples of fEPSPs evokedby stimulating Schaffer collaterals and recording in CA1 stratum radiatum and recorded from individual hippocampal slices from naive or trainedcontrol, I�B�-SR�, and I�B�-SR� mice over a range of stimulus intensities from 12 to 600 mA. Both naive and trained control and I�B�-SR�

mice exhibited comparable traces. (B) I/O curves generated by averaging the peak amplitude of fEPSPs obtained at each of the same stimulusintensities from slices prepared from either naive or trained groups of mice. Naive I�B�-SR� and control mice had comparable curves. In contrast,I�B�-SR� mice exhibited significantly impaired I/O function in Schaffer-CA1 basal synaptic transmission. This impairment was reversed to controllevels in I�B�-SR� mice after induction of spatial learning and memory. (C) In membrane preparations of postsynaptic densities isolated fromthe hippocampus from various untrained mice, the basal levels of membrane-localized GluR1 from I�B�-SR� mice (lanes 1 to 4) were lowerrelative to levels seen in either I�B�-SR� (lanes 5 and 6) and control mice (lanes 7 and 8). A similar decrease in basal GluR1 levels was observedin whole hippocampal tissue lysates from I�B�-SR� mice (compare panel D, lane 1 with lanes 3 and 5). (D) Conversely, in whole hippocampaltissue lysates isolated from trained I�B�-SR� mice, GluR1 levels were significantly higher than in basal (i.e., untrained) tissue lysates. Activity-dependent increases in GluR1 levels were also observed in I�B�-SR� and control mice. No changes in the levels of other GluR types relative tothe levels of myelin basic protein (MBP) (loading control) were seen in untrained or trained mice. (E) Water maze training induced NF-�B-mediated gel shifts in I�B-SR� and control mice. In contrast, nuclear extracts from trained I�B�-SR� mice did not contain activated NF-�Bcomplexes. CTL, control.



EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

mance in the first trial], the magnitude of the LTP recorded(percentage over baseline) was strongly correlated with per-cent improvement in time to complete the maze (r 0.88; P 0.0239; n 5) and in errors made (r 0.81; P 0.049; n 5).These findings reveal a strong correlation between improvedLTP and enhanced performance in the radial arm maze. Thus,based on these two independent tests, we suggest that theneuronal pattern of expression of I�B�-SR in these mice re-sults in enhanced spatial learning and memory.

Activity-dependent recovery of synaptic signaling and gluR1levels in trained I�B�-SR� mice. In recording synaptic signalingalong the Schaffer collateral/commissural pathway, the intensityof the electrical stimulation (input [I]) and the peak amplitude ofthe fEPSPs (output [O]) evoked were used to establish an I/Ofunction as a measure of the basal synaptic transmission. Strik-ingly, untrained I�B�-SR� mice exhibited significantly impairedbasal synaptic transmission relative to levels in similarly untrainedcontrol mice (Fig. 6A). In contrast, I/O functions in I�B�-SR�

mice were equivalent to levels detected in control mice (data notshown). In sharp contrast to the impaired I/O function observedin naive mice, maze-trained I�B�-SR� mice displayed a com-pletely restored I/O function (Fig. 6B).

One possible explanation for the impaired basal synapticstrength in untrained I�B�-SR� mice relates to markedly de-creased GluR1 levels detected in I�B�-SR� mice. Immunoblotanalysis showed that levels of AMPA-type GluR1 subunits insynaptosomal membrane preparations were significantly de-creased in naive I�B�-SR� mice compared to untrained I�B�-SR� littermates or nontransgenic control mice (Fig. 6C). In con-trast, maze training resulted in a much greater activity-dependentincrease in GluR1 levels in the I�B�-SR� hippocampus than ineither I�B�-SR� littermates or control mice (Fig. 6D, lane 1versus lanes 3 and 5; note the lower levels of basal GluR1 inuntrained SR� lysates relative to SR� or control lysates). A sim-ilar activity-dependent restoration of GAD65 levels was not ob-served in I�B�-SR� mice. This activity-dependent increase inGluR1 may be a consequence of hyperexcitation of glutamatergic

neurons occurring in the context of diminished inhibitory toneand/or altered excitatory circuits in I�B�-SR� mice.

Synaptic signaling induced by behavioral training was eval-uated by EMSA using CREB and NF-�B/Rel-specific radiola-beled probes. NF-�B/Rel DNA binding was increased intrained I�B-SR� and control mice but, consistent with expres-sion of I�B�-SR, was unaffected in I�B�-SR� mice (Fig. 6E).Of note, CREB DNA activity was slightly increased in I�B�-SR� mice under both basal and maze-trained conditions (Fig.6E), suggesting that this factor may be activated in response topan-neuronal NF-�B inhibition. While direct effects ofI�B�-SR expression on excitatory neuronal activity cannot beexcluded, the activity-dependent restoration of synapticstrength (I/O), improved synaptic signaling (LTP and Arc lev-els), and enhanced cognitive functions are consistent withheightened excitatory activity resulting from impaired GAD65-dependent inhibitory neuronal function in I�B�-SR� mice.

DISCUSSION

NF-�B as a regulator of synaptic plasticity and memoryformation. Alterations in synaptic plasticity reflect compositechanges occurring not only in excitatory neurons but alsowithin inhibitory interneurons (32–34, 67). Our findings sug-gest a surprising and previously unrecognized role for the NF-�B/Rel family of transcriptional factors as critical modulatorsof the homeostatic interplay occurring between inhibitory andexcitatory neuronal function. Further, our studies reveal thatNF-�B is an important positive regulator of GAD65, an en-zyme that is critical for establishment of GABAergic interneu-ron-mediated inhibitory tone in vivo.

Using the prion promoter-enhancer, we have generated atransgenic mouse model in which a dominantly acting inhibitorof NF-�B action is exclusively expressed in neurons. This in-hibitor, termed I�B�-SR, is strongly expressed in GABAergicinhibitory interneurons and, to a lesser extent, in excitatoryneurons. As noted, I�B�-SR expression results in decreased

FIG. 7. Our proposed model for altered basal and activity-dependent excitatory synaptic signaling in I�B�-SR� versus I�B�-SR� mice. Underbasal conditions, signaling through AMPA-type GluR1 in I�B�-SR� (and control) mice maintains homeostatic basal synaptic strength (I/Ofunction). In contrast, I�B�-SR expression results in lower basal GluR1 and impaired I/O function. However, after mice are subjected to anexperience- or task-based activity (e.g., maze training), GluR1 levels are increased, facilitating LTP generation. I�B�-SR expression also resultsin decreased GAD65 expression in GABAergic interneurons, leading to impaired inhibitory tone. Consequentially, I�B�-SR� mice exhibitedmarkedly enhanced excitatory firing involving higher GluR1 levels, restored I/O function, and higher LTP. These synaptic changes are reflectedin improved spatial learning and memory and increased exploratory activity observed in I�B�-SR� mice.



EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

expression of GAD65 in GABAergic interneurons. I�B�-SRexpression also leads to impaired basal synaptic signaling,likely due to decreased synaptosomal AMPAR-type glutamatereceptor (GluR1) expression, resulting in sharply impaired I/Ofunction in untrained or naive mice (Fig. 7). However, afterthese I�B�-SR� mice are subjected to an experience- or task-based activity (e.g., maze training), AMPAR-type GluR1 levelsare markedly increased in synaptosomal membranes, and I/Ofunction is completely restored. We suspect that this trainingconverts formerly “silent” dendritic spines into active ones,facilitating increased neuronal excitation (21). Enhanced ac-tivity-dependent synaptic signaling in I�B�-SR� mice couldreflect increased AMPAR-mediated neuronal excitation oc-curring as a consequence of diminished GAD65-derived inhib-itory tone, although direct effects of I�B�-SR expression inexcitatory neurons may also contribute to the observed phe-notype. Consistent with this proposed model, I�B�-SR� miceexhibit increased L-LTP and synaptic activity-dependent geneexpression, enhanced physical and exploratory activity, higherincidence of seizures, and improved performance in varioustests of spatial learning.

Expression of I�B�-SR in GABAergic interneurons resultsin decreased GAD65 expression. Excitatory neurons and inhib-itory interneurons represent the opposing “yin-yang” of syn-aptic function and memory formation. Notably, during anygiven behavioral task, 90% of excitatory neurons remainsilent, whereas almost all of the inhibitory interneurons areactive (18, 27). Recently, a number of studies have implicatedthe NF-�B/Rel family of transcriptional regulators in excita-tory neuronal function and spatial memory (19, 49). However,the role of these factors in inhibitory GABAergic interneurons,which comprise more than 30% of all neurons in the adultmammalian central nervous system, has not been explored.

I�B�-SR� expression resulted in decreased transcription ofGAD65, a rate-limiting enzyme required for GABA synthesisin GABAergic interneurons and generation of the inhibitorytone. GABA is formed from the alpha-decarboxylation of glu-tamate by the GAD isoforms, GAD65 and GAD67 (10). In thedentate gyrus and CA1 region of the rat hippocampus, GAD65is localized primarily in synaptosomes and regulates the vesic-ular pool of GABA, allowing responses to short-term increasesin demand during activity-dependent synaptic signaling (56,68). GAD67 appears to be primarily responsible for the syn-thesis of the metabolic GABA pool and supports tonic levels ofsynaptic transmission (11, 12). Mice in which the GAD67 geneis disrupted die at birth, likely as a result of the dramaticallylower production of GABA synthesis; GAD67�/� mice have90% less GABA levels than normal mice (4, 9). In contrast,GAD65�/� mice survive, and the total GABA content is onlymarginally decreased (26, 35, 36). However, GAD65�/� miceare prone to seizures (35), have diminished GABA releasefollowing K� stimulation of the visual cortex, and exhibit al-tered visual cortical plasticity (26). These findings suggest thatGAD65 plays an important role in GABAergic synaptic trans-mission. Indeed, in view of the large amounts of GABA inneuronal cell bodies and the different intraneuronal distribu-tions of GAD65 and GAD67, it has been suggested thatGAD67 might be involved in the synthesis of GABA for gen-eral metabolic activity through the tricarboxylic acid cycle,

whereas GAD65-derived GABA participates in regulating syn-aptic transmission at active spines (63).

The long-term regulation of GAD is complex, involving bothtranscriptional and posttranscriptional mechanisms. Studies ofgad67 and gad65 gene expression as well as analysis of theirsignificantly different regulatory regions suggest that transcrip-tional regulation involves different intracellular mechanisms(63). Our observation of a specific decrease in GAD65, but notGAD67, mRNA transcripts and protein levels in I�B�-SR�

mice raises the possibility that GAD65 may correspond to anNF-�B target gene in GABAergic interneurons. Alternatively,changes in GAD65 expression may involve more indirectmechanisms resulting from the pan-neuronal expression ofI�B�-SR� in these mice.

Loss of NF-�B action in neurons leads to hyperexcitabilityand enhanced LTP. Experience-based neuronal activity resultsin progressive depolarization of the postsynaptic neuron inresponse to glutamate. In contrast, stimulated inhibitoryGABAergic neurons synthesize and release GABA (47), trig-gering hyperpolarization of the postsynaptic neuron. ThisGABA-mediated inhibitory tone essentially acts as a neuro-chemical brake to inhibit the presynaptic release of other ex-citatory neurotransmitters and attenuate the excitatory signal(18, 27). It has previously been reported that disruption ofinhibitory inputs on neurons results in unopposed excitatoryfiring leading to increased seizure activity (69). Consistent withthis finding is the observation that GABA withdrawal or phar-macological inhibition of GABAergic function triggers neuro-nal hyperexcitability (8). Thus, these I�B�-SR� animals pro-vided a unique opportunity to explore in vivo the impact ofinhibiting NF-�B/Rel action on the interplay between inhibi-tory interneurons and excitatory neurons regulating synapticsignaling.

As an experimental correlate of activity-dependent synapticplasticity, LTP is typically induced in a biphasic manner fol-lowing high-frequency stimulation. The early-phase LTP oc-curs independently of new gene expression and involves theactivation of several protein kinases and the recruitment ofexisting AMPAR into active synapses (42, 53). In contrast,L-LTP requires new gene transcription and protein synthesisand is thus considered to be the most likely mechanism under-lying the long-lasting changes required for long-term memory.Several candidate genes have been identified as molecularanalogs of long-term memory. One such synaptic target is theimmediate-early gene encoding the protein Arc (24). Typicallyinduced following neuronal activation through NMDA recep-tor (NMDAR), synapse-specific Arc expression serves to facil-itate synaptic plasticity and long-term memory consolidationand is a strong indicator of activity-dependent synapse excit-ability (64). Spontaneous induction of Arc expression in lowMg2�/high K� medium (3) confirmed the hyperexcitability ofI�B�-SR� neurons likely due to reduced GAD65 expression inGABAergic interneurons leading to impaired inhibitory tonewhich may, in turn, explain the increased LTP and enhancedArc expression. However, effects due to I�B�-SR expression inexcitatory neurons on synaptic signaling and excitatory firingmay also contribute to this phenotype and cannot be excluded.

Neuronal I�B�-SR expression alters both basal and activ-ity-dependent synaptic signaling. Under normal conditions,experience- or activity-based neuronal activity reflects a bal-



EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

ance achieved primarily through glutamatergic neuronal exci-tation and GABAergic interneuron-mediated inhibitory tone.Our studies suggest that NF-�B may play a dual role in mod-ulating synaptic signaling by regulating select functions in thesedistinct neuronal subtypes. Consistent with prior reports, ourstudies demonstrate that NF-�B is required during the main-tenance phase of the synaptic response for regulating basalAMPAR expression and function (43, 54, 78). Conversely,during the constructive phase of activity-dependent synapticsignaling, activation of transcriptional factors including CREB,CREM (32, 33), or serum response factor (58), in addition toNF-�B, leads to increased AMPAR expression and enhancedexcitatory firing (32, 77). Simultaneously, activity-dependentinduction of NF-�B/Rel action in GABAergic interneuronsincreases GAD65 levels, resulting in enhanced GABA-medi-ated inhibitory tone effectively attenuating excitatory neuronfiring.

Increased LTP and enhanced learning and memory havealso been reported in two other studies involving overexpres-sion of either the NMDA-type glutamate receptor, NR2B, orthe KIF17 kinesin motor protein that are required for traffick-ing of these receptors to active spines in excitatory neurons (66,75). Our findings suggest that I�B�-SR regulation of GABAer-gic neuronal function may also result in increased GluR1 traf-ficking to active spines as a consequence of decreased GAD65-derived inhibitory tone. Additionally, I�B�-SR expression inexcitatory neurons could promote increased GluR1 expression,resulting in enhanced LTP as a consequence of altering theratios of silent synapses in the postsynaptic density. However,this latter possibility seems less likely, given the inhibitoryeffects of the I�B�-SR on neuronal synaptic strength underbasal conditions.

Inhibition of NF-�B action by I�B�-SR expression in neu-rons leads to enhanced cognitive functions. The predominantphenotypes resulting from I�B�-SR expression include in-creased late long-term potentiation, neuronal hyperexcitabil-ity, increased incidence of handling seizures, hyperactivity, andheightened exploratory activity. Of note, consistent with theincreased exploratory phenotype induced by I�B�-SR expres-sion, mice lacking the gene for the p50 subunit of NF-�B/Relalso exhibit increased exploratory activity (38). Altered explor-atory activity has not been reported for either the p65-deficientmice or the CamKII-promoter I�B�-SR mice (19, 49). Fur-ther, transgenic mice overexpressing TNF-�, a potent activatorof NF-�B, displayed decreased exploratory behavior in open-field tests (15).

However, one of the most striking phenotypes resulting fromI�B�-SR expression in vivo is an enhanced performance in twoindependent tests of spatial learning and memory. Indeed, weobserve a strong positive correlation between synaptic signal-ing (LTP) and radial arm maze performance (46). This en-hanced spatial memory in Prp-I�B�-SR� mice is in markedcontrast with prior reports in which mice either lacking expres-sion of RelA subunit of NF-�B (RelA�/� TNFR�/�) in neu-rons and glia (49) or exclusively expressing the I�B�-SR trans-gene in excitatory neurons (19) exhibit impaired spatialmemory. One explanation for these disparate findings relatesto the cell types involved. While neurons are key mediators ofsynaptic signaling, there is mounting evidence that glial cells,by far the most numerous cell type in the brain, are also

essential for learning and memory (14, 25). Glial cells act askey modulators of glutamate-mediated neurotransmission (13,14) and, thus, can potently affect excitatory stimulation withoutimpacting inhibitory tone. Consequently, loss of both TNF-�receptor and RelA/p65 expression in both neurons and glialcells in the study of Meffert et al. (49) could affect synapticplasticity quite differently than inhibiting NF-�B/Rel actiononly in neurons. Another important difference between thesetwo models relates to the fact that Meffert et al. selectivelydeleted only the RelA subunit of NF-�B, while we have em-ployed a more broadly acting inhibitor that potently impairsmany, if not all, members of the NF-�B/Rel family. Potentialcompensatory effects of other Rel family members could con-tribute to the differences observed.

In the other I�B�-SR transgenic model described byKaltschmidt et al. (19), the CamKII promoter was used tospecifically target I�B�-SR transgene expression to neurons inthe forebrain. Such expression leads to diminished LTP andimpaired spatial learning and memory (29). CamKII-SR miceexpress the SR transgene exclusively in excitatory neurons inthe forebrain, while Prp-I�B�-SR mice express the transgenerobustly in inhibitory GABAergic interneurons and, to a lesserextent, in excitatory neurons in multiple brain regions. Not-withstanding the key differences between these two model sys-tems with respect to promoter activity, cell types, and brainregions involved as well as developmental expression, thesemice do serve as interesting contrasts with respect to inhibitingNF-�B action in different neuronal cell types that regulatesynaptic plasticity through fundamentally opposing forces, i.e.,excitatory versus inhibitory neurotransmissions. Thus, it is notsurprising that quite different spatial learning and memoryphenotypes are observed in mice with altered excitatory cir-cuits in the context of intact versus impaired inhibitory neuro-nal function.

Various forms of memory formation and retrieval are likelyto be contingent on different transcriptional mechanisms oc-curring in discrete cell types in select brain regions and storedwith different time constants. While a number of studies haveimplicated the NF-�B/Rel family of transcription factors aspositive regulators of excitatory neuronal function and spatialmemory, our study now also identifies NF-�B/Rel as an im-portant regulator of inhibitory interneuron function throughits effects on GAD65 expression. Overall, we suggest thatNF-�B plays an important role in determining the balancebetween inhibitory neuronal activity and excitatory neuronalfiring that importantly shapes changes in synaptic plasticity,LTP, and memory formation.

ACKNOWLEDGMENTS

We thank Stanley Prusiner (University of California, San Francisco)for supplying the Prp/tTa mice and Paul Worley (Johns Hopkins) forgenerously providing Arc antibodies. We also thank Bobby Benitez,JoDee Fish, Jorge Palop-Esteban, and Vikram Rao (Gladstone Insti-tutes) for their technical expertise and helpful comments. We alsothank members of the Greene laboratory, Lennart Mucke, StevenFinkbeiner, and Michael Cowley, for helpful discussions. We are verygrateful to Tom MacMahon, Kevin Deitchman, and Robert Messing(Ernest Gallo Clinic and Research Center) for their collegiality andinvaluable assistance with the radial arm maze. We thank AngelaRizk-Jackson and Timothy F. Pfankuch for their assistance with thebehavioral testing. We thank John Carroll, Jack Hull, Stephen Gonza-les, and Chris Goodfellow for assistance in the preparation of figures,



EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

Stephen Ordway and Gary Howard for editorial assistance, and RobinGivens and Sue Cammack for administrative support.

This work was supported by the J. David Gladstone Institutes(W.C.G.) and NIH grant AG 20904 (J.R.) and the Extramural Re-search Facilities Improvement Program Project (C06 RR018928).

REFERENCES

1. Albensi, B. C., and M. P. Mattson. 2000. Evidence for the involvement ofTNF and NF-�B in hippocampal synaptic plasticity. Synapse 35:151–159.

2. Arancio, O., H. P. Zhang, X. Chen, C. Lin, F. Trinchese, D. Puzzo, S. Liu, A.Hegde, S. F. Yan, A. Stern, J. S. Luddy, L. F. Lue, D. G. Walker, A. Roher,M. Buttini, L. Mucke, W. Li, A. M. Schmidt, M. Kindy, P. A. Hyslop, D. M.Stern, and S. S. Du Yan. 2004. RAGE potentiates A-induced perturbationof neuronal function in transgenic mice. EMBO J. 23:4096–4105.

3. Arias, R. L., and M. R. Bowlby. 2005. Pharmacological characterization ofantiepileptic drugs and experimental analgesics on low magnesium-inducedhyperexcitability in rat hippocampal slices. Brain Res. 1047:233–244.

4. Asada, H., Y. Kawamura, K. Maruyama, H. Kume, R. Ding, F. Y. Ji, N.Kanbara, H. Kuzume, M. Sanbo, T. Yagi, and K. Obata. 1996. Mice lackingthe 65 kDa isoform of glutamic acid decarboxylase (GAD65) maintain nor-mal levels of GAD67 and GABA in their brains but are susceptible toseizures. Biochem. Biophys. Res. Commun. 229:891–895.

5. Barmada, S., P. Piccardo, K. Yamaguchi, B. Ghetti, and D. A. Harris. 2004.GFP-tagged prion protein is correctly localized and functionally active in thebrains of transgenic mice. Neurobiol. Dis. 16:527–537.

6. Brown, K., S. Gerstberger, L. Carlson, G. Franzoso, and U. Siebenlist. 1995.Control of I�B� proteolysis by site-specific, signal-induced phosphorylation.Science 267:1485–1488.

7. Carlezon, W. A., Jr., R. S. Duman, and E. J. Nestler. 2005. The many facesof CREB. Trends Neurosci. 28:436–445.

8. Casasola, C., T. Montiel, E. Calixto, and S. Brailowsky. 2004. Hyperexcit-ability induced by GABA withdrawal facilitates hippocampal long-term po-tentiation. Neuroscience 126:163–171.

9. Condie, B. G., G. Bain, D. I. Gottlieb, and M. R. Capecchi. 1997. Cleft palatein mice with a targeted mutation in the gamma-aminobutyric acid-producingenzyme glutamic acid decarboxylase 67. Proc. Natl. Acad. Sci. USA 94:11451–11455.

10. Erlander, M. G., N. J. Tillakaratne, S. Feldblum, N. Patel, and A. J. Tobin.1991. Two genes encode distinct glutamate decarboxylases. Neuron 7:91–100.

11. Erlander, M. G., and A. J. Tobin. 1991. The structural and functional het-erogeneity of glutamic acid decarboxylase: a review. Neurochem. Res. 16:215–226.

12. Feldblum, S., M. G. Erlander, and A. J. Tobin. 1993. Different distributionsof GAD65 and GAD67 mRNAs suggest that the two glutamate decarboxyl-ases play distinctive functional roles. J. Neurosci. Res. 34:689–706.

13. Fields, R. D. 2004. The other half of the brain. Sci. Am. 290:54–61.14. Fields, R. D., and B. Stevens-Graham. 2002. New insights into neuron-glia

communication. Science 298:556–562.15. Fiore, M., L. Probert, G. Kollias, K. Akassoglou, E. Alleva, and L. Aloe. 1996.

Neurobehavioral alterations in developing transgenic mice expressingTNF-� in the brain. Brain Behav. Immun. 10:126–138.

16. Freudenthal, R., and A. Romano. 2000. Participation of Rel/NF-�B tran-scription factors in long-term memory in the crab Chasmagnathus. BrainRes. 855:274–281.

17. Freudenthal, R., A. Romano, and A. Routtenberg. 2004. Transcription factorNF-�B activation after in vivo perforant path LTP in mouse hippocampus.Hippocampus 14:677–683.

18. Freund, T. F., and G. Buzsaki. 1996. Interneurons of the hippocampus.Hippocampus 6:347–470.

19. Fridmacher, V., B. Kaltschmidt, B. Goudeau, D. Ndiaye, F. M. Rossi, J.Pfeiffer, C. Kaltschmidt, A. Israel, and S. Memet. 2003. Forebrain-specificneuronal inhibition of nuclear factor-�B activity leads to loss of neuropro-tection. J. Neurosci. 23:9403–9408.

20. Furukawa, K., and M. P. Mattson. 1998. The transcription factor NF-�Bmediates increases in calcium currents and decreases in NMDA- andAMPA/kainate-induced currents induced by tumor necrosis factor-� in hip-pocampal neurons. J. Neurochem. 70:1876–1886.

21. Gazzaley, A., S. Kay, and D. L. Benson. 2002. Dendritic spine plasticity inhippocampus. Neuroscience 111:853–862.

22. Gossen, M., and H. Bujard. 1992. Tight control of gene expression in mam-malian cells by tetracycline-responsive promoters. Proc. Natl. Acad. Sci.USA 89:5547–5551.

23. Guerrini, L., F. Blasi, and S. Denis-Donini. 1995. Synaptic activation ofNF-�B by glutamate in cerebellar granule neurons in vitro. Proc. Natl. Acad.Sci. USA 92:9077–9081.

24. Guzowski, J. F., G. L. Lyford, G. D. Stevenson, F. P. Houston, J. L.McGaugh, P. F. Worley, and C. A. Barnes. 2000. Inhibition of activity-dependent Arc protein expression in the rat hippocampus impairs the main-tenance of long-term potentiation and the consolidation of long-term mem-ory. J. Neurosci. 20:3993–4001.

25. Haydon, P. G. 2001. GLIA: listening and talking to the synapse. Nat. Rev.Neurosci. 2:185–193.

26. Hensch, T. K., M. Fagiolini, N. Mataga, M. P. Stryker, S. Baekkeskov, andS. F. Kash. 1998. Local GABA circuit control of experience-dependentplasticity in developing visual cortex. Science 282:1504–1508.

27. Henze, D. A., Z. Borhegyi, J. Csicsvari, A. Mamiya, K. D. Harris, and G.Buzsaki. 2000. Intracellular features predicted by extracellular recordings inthe hippocampus in vivo. J. Neurophysiol. 84:390–400.

28. Hong, E. J., A. E. West, and M. E. Greenberg. 2005. Transcriptional controlof cognitive development. Curr. Opin. Neurobiol. 15:21–28.

29. Kaltschmidt, B., D. Ndiaye, M. Korte, S. Pothion, L. Arbibe, M. Prullage, J.Pfeiffer, A. Lindecke, V. Staiger, A. Israel, C. Kaltschmidt, and S. Memet.2006. NF-�B regulates spatial memory formation and synaptic plasticitythrough protein kinase A/CREB signaling. Mol. Cell. Biol. 26:2936–2946.

30. Kaltschmidt, C., B. Kaltschmidt, and P. A. Baeuerle. 1993. Brain synapsescontain inducible forms of the transcription factor NF-�B. Mech. Dev. 43:135–147.

31. Kaltschmidt, C., B. Kaltschmidt, and P. A. Baeuerle. 1995. Stimulation ofionotropic glutamate receptors activates transcription factor NF-�B in pri-mary neurons. Proc. Natl. Acad. Sci. USA 92:9618–9622.

32. Kandel, E. R. 2001. The molecular biology of memory storage: a dialogbetween genes and synapses. Biosci. Rep. 21:565–611.

33. Kandel, E. R., and C. Pittenger. 1999. The past, the future and the biologyof memory storage. Philos. Trans. R. Soc. Lond. B 354:2027–2052.

34. Kaplan, M. P., and T. Abel. 2003. Genetic approaches to the study ofsynaptic plasticity and memory storage. CNS Spectr. 8:597–610.

35. Kash, S. F., R. S. Johnson, L. H. Tecott, J. L. Noebels, R. D. Mayfield, D.Hanahan, and S. Baekkeskov. 1997. Epilepsy in mice deficient in the 65-kDaisoform of glutamic acid decarboxylase. Proc. Natl. Acad. Sci. USA 94:14060–14065.

36. Kash, S. F., L. H. Tecott, C. Hodge, and S. Baekkeskov. 1999. Increasedanxiety and altered responses to anxiolytics in mice deficient in the 65-kDaisoform of glutamic acid decarboxylase. Proc. Natl. Acad. Sci. USA 96:1698–1703.

37. Kassed, C. A., and M. Herkenham. 2004. NF-kappaB p50-deficient miceshow reduced anxiety-like behaviors in tests of exploratory drive and anxiety.Behav. Brain Res. 154:577–584.

38. Kassed, C. A., A. E. Willing, S. Garbuzova-Davis, P. R. Sanberg, and K. R.Pennypacker. 2002. Lack of NF-�B p50 exacerbates degeneration of hip-pocampal neurons after chemical exposure and impairs learning. Exp. Neu-rol. 176:277–288.

39. Kelleher, R. J., III, A. Govindarajan, and S. Tonegawa. 2004. Translationalregulatory mechanisms in persistent forms of synaptic plasticity. Neuron44:59–73.

40. Lavon, I., I. Goldberg, S. Amit, L. Landsman, S. Jung, B. Z. Tsuberi, I.Barshack, J. Kopolovic, E. Galun, H. Bujard, and Y. Ben-Neriah. 2000. Highsusceptibility to bacterial infection, but no liver dysfunction, in mice com-promised for hepatocyte NF-�B activation. Nat. Med. 6:573–577.

41. Lilienbaum, A., and A. Israel. 2003. From calcium to NF-�B signaling path-ways in neurons. Mol. Cell. Biol. 23:2680–2698.

42. Lissin, D. V., S. N. Gomperts, R. C. Carroll, C. W. Christine, D. Kalman, M.Kitamura, S. Hardy, R. A. Nicoll, R. C. Malenka, and M. von Zastrow. 1998.Activity differentially regulates the surface expression of synaptic AMPA andNMDA glutamate receptors. Proc. Natl. Acad. Sci. USA 95:7097–7102.

43. Marini, A. M., X. Jiang, X. Wu, F. Tian, D. Zhu, P. Okagaki, and R. H.Lipsky. 2004. Role of brain-derived neurotrophic factor and NF-�B in neu-ronal plasticity and survival: from genes to phenotype. Restor. Neurol. Neu-rosci. 22:121–130.

44. Matsuoka, Y., Y. Kitamura, M. Okazaki, K. Terai, and T. Taniguchi. 1999.Kainic acid-induced activation of nuclear factor-�B in rat hippocampus. Exp.Brain Res. 124:215–222.

45. Mattson, M. P., and S. Camandola. 2001. NF-�B in neuronal plasticity andneurodegenerative disorders. J. Clin. Investig. 107:247–254.

46. Mazzucchelli, C., C. Vantaggiato, A. Ciamei, S. Fasano, P. Pakhotin, W.Krezel, H. Welzl, D. P. Wolfer, G. Pages, O. Valverde, A. Marowsky, A.Porrazzo, P. C. Orban, R. Maldonado, M. U. Ehrengruber, V. Cestari, H. P.Lipp, P. F. Chapman, J. Pouyssegur, and R. Brambilla. 2002. Knockout ofERK1 MAP kinase enhances synaptic plasticity in the striatum and facilitatesstriatal-mediated learning and memory. Neuron 34:807–820.

47. McBain, C. J., and A. Fisahn. 2001. Interneurons unbound. Nat. Rev. Neu-rosci. 2:11–23.

48. Meberg, P. J., W. R. Kinney, E. G. Valcourt, and A. Routtenberg. 1996. Geneexpression of the transcription factor NF-�B in hippocampus: regulation bysynaptic activity. Brain Res. Mol. Brain Res. 38:179–190.

49. Meffert, M. K., J. M. Chang, B. J. Wiltgen, M. S. Fanselow, and D. Balti-more. 2003. NF-�B functions in synaptic signaling and behavior. Nat. Neu-rosci. 6:1072–1078.

50. Merlo, E., R. Freudenthal, and A. Romano. 2002. The I�B kinase inhibitorsulfasalazine impairs long-term memory in the crab Chasmagnathus. Neuro-science 112:161–172.

51. Mironov, A., Jr., D. Latawiec, H. Wille, E. Bouzamondo-Bernstein, G. Legname,



EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

R. A. Williamson, D. Burton, S. J. DeArmond, S. B. Prusiner, and P. J.Peters. 2003. Cytosolic prion protein in neurons. J. Neurosci. 23:7183–7193.

52. Myhrer, T. 1988. Exploratory behavior and reaction to novelty in rats withhippocampal perforant path systems disrupted. Behav. Neurosci. 102:356–362.

53. Nicoll, R. A. 2003. Expression mechanisms underlying long-term potentia-tion: a postsynaptic view. Philos. Trans. R. Soc. Lond. B 358:721–726.

54. Ohba, S., T. Ikeda, Y. Ikegaya, N. Nishiyama, N. Matsuki, and M. K.Yamada. 2005. BDNF locally potentiates GABAergic presynaptic machin-eries: target-selective circuit inhibition. Cereb. Cortex. 15:291–298.

55. O’Neill, L. A., and C. Kaltschmidt. 1997. NF-�B: a crucial transcriptionfactor for glial and neuronal cell function. Trends Neurosci. 20:252–258.

56. Petroff, O. A. 2002. GABA and glutamate in the human brain. Neuroscientist8:562–573.

57. Raber, J., D. Wong, G. Q. Yu, M. Buttini, R. W. Mahley, R. E. Pitas, and L.Mucke. 2000. Apolipoprotein E and cognitive performance. Nature 404:352–354.

58. Ramanan, N., Y. Shen, S. Sarsfield, T. Lemberger, G. Schutz, D. J. Linden,and D. D. Ginty. 2005. SRF mediates activity-induced gene expression andsynaptic plasticity but not neuronal viability. Nat. Neurosci. 8:759–767.

59. Rampon, C., Y. P. Tang, J. Goodhouse, E. Shimizu, M. Kyin, and J. Z. Tsien.2000. Enrichment induces structural changes and recovery from nonspatialmemory deficits in CA1 NMDAR1-knockout mice. Nat. Neurosci. 3:238–244.

60. Rao, V. R., S. A. Pintchovski, J. Chin, C. L. Peebles, S. Mitra, and S.Finkbeiner.2006. AMPA receptors regulate transcription of the plasticity-related immediate-early gene Arc. Nat. Neurosci. 9:887–895.

61. Reuther, J. Y., G. W. Reuther, D. Cortez, A. M. Pendergast, and A. S.Baldwin, Jr. 1998. A requirement for NF-�B activation in Bcr-Abl-mediatedtransformation. Genes Dev. 12:968–981.

62. Scott, R., R. Bourtchuladze, S. Gossweiler, J. Dubnau, and T. Tully. 2002.CREB and the discovery of cognitive enhancers. J. Mol. Neurosci. 19:171–177.

63. Soghomonian, J. J., and D. L. Martin. 1998. Two isoforms of glutamatedecarboxylase: why? Trends Pharmacol. Sci. 19:500–505.

64. Steward, O., and P. Worley. 2002. Local synthesis of proteins at synaptic siteson dendrites: role in synaptic plasticity and memory consolidation? Neuro-biol. Learn. Mem. 78:508–527.

65. Sun, S. C., P. A. Ganchi, D. W. Ballard, and W. C. Greene. 1993. NF-kappaB controls expression of inhibitor I kappa B alpha: evidence for an inducibleautoregulatory pathway. Science 259:1912–1915.

66. Tang, Y. P., E. Shimizu, G. R. Dube, C. Rampon, G. A. Kerchner, M. Zhuo,G. Liu, and J. Z. Tsien. 1999. Genetic enhancement of learning and memoryin mice. Nature 401:63–69.

67. Teyler, T. J., and P. DiScenna. 1987. Long-term potentiation. Annu. Rev.Neurosci. 10:131–161.

68. Tian, N., C. Petersen, S. Kash, S. Baekkeskov, D. Copenhagen, and R. Nicoll.1999. The role of the synthetic enzyme GAD65 in the control of neuronalgamma-aminobutyric acid release. Proc. Natl. Acad. Sci. USA 96:12911–12916.

69. Treiman, D. M. 2001. GABAergic mechanisms in epilepsy. Epilepsia42(Suppl. 3):8–12.

70. Tremblay, P., Z. Meiner, M. Galou, C. Heinrich, C. Petromilli, T. Lisse, J.Cayetano, M. Torchia, W. Mobley, H. Bujard, S. J. DeArmond, and S. B.Prusiner. 1998. Doxycycline control of prion protein transgene expressionmodulates prion disease in mice. Proc. Natl. Acad. Sci. USA 95:12580–12585.

71. van Hogerlinden, M., B. L. Rozell, L. Ahrlund-Richter, and R. Toftgard.1999. Squamous cell carcinomas and increased apoptosis in skin with inhib-ited Rel/nuclear factor-kappaB signaling. Cancer Res. 59:3299–3303.

72. Waltereit, R., and M. Weller. 2003. Signaling from cAMP/PKA to MAPKand synaptic plasticity. Mol. Neurobiol. 27:99–106.

73. West, A. E., E. C. Griffith, and M. E. Greenberg. 2002. Regulation oftranscription factors by neuronal activity. Nat. Rev. Neurosci. 3:921–931.

74. Whiteside, S. T., M. K. Ernst, O. LeBail, C. Laurent-Winter, N. Rice, and A.Israel. 1995. N- and C-terminal sequences control degradation of MAD3/I�B� alpha in response to inducers of NF-�B activity. Mol. Cell. Biol.15:5339–5345.

75. Wong, R. W., M. Setou, J. Teng, Y. Takei, and N. Hirokawa. 2002. Overex-pression of motor protein KIF17 enhances spatial and working memory intransgenic mice. Proc. Natl. Acad. Sci. USA 99:14500–14505.

76. Yavin, Z., and E. Yavin. 1980. Survival and maturation of cerebral neuronson poly(L-lysine) surfaces in the absence of serum. Dev. Biol. 75:454–459.

77. Yin, J. C., and T. Tully. 1996. CREB and the formation of long-term mem-ory. Curr. Opin. Neurobiol. 6:264–268.

78. Yu, Z., G. Cheng, X. Wen, G. D. Wu, W. T. Lee, and D. Pleasure. 2002.Tumor necrosis factor alpha increases neuronal vulnerability to excitotoxicnecrosis by inducing expression of the AMPA-glutamate receptor subunitGluR1 via an acid sphingomyelinase- and NF-�B-dependent mechanism.Neurobiol. Dis. 11:199–213.



EN

N S

TA

TE

UN

IVhttp://m

cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

Documents

NF-κB/Rel Regulates Inhibitory and Excitatory Neuronal ......10.1128/MCB.00510-06. Mol. Cell. Biol.€2006, 26(19):7283. DOI: LeFevour, Shikha Chakraborty-Sett and Warner C. Greene