Excitotoxicity is a process in which glutamate or other excitatoryamino acids induce neuronal cell death. Accumulating evidencesuggests that excitotoxicity may contribute to human neuronalcell loss caused by acute insults and chronic degeneration in thecentral nervous system1−4. The immediate early gene (IEG) c-fosencodes a transcription factor5−6. The c-Fos proteins form het-erodimers with Jun family proteins, and the resulting AP-1 com-plexes regulate transcription by binding to the AP-1 sequencefound in many cellular genes7−9. Emerging evidence suggeststhat c-fos is essential in regulating neuronal cell survival versusdeath10. Although c-fos is induced by neuronal activity, includingkainic acid−induced seizures11−14, whether and how c-fos isinvolved in excitotoxicity is still unknown. To address this issue,we generated a mouse in which c-fos expression is largely elimi-nated in the hippocampus. We found that these mutant micehave more severe kainic acid–induced seizures, increased neu-ronal excitability and neuronal cell death, compared with control

mice. Moreover, c-Fos regulates the expression of the kainic acidreceptor GluR6 and brain-derived neurotrophic factor (BDNF),both in vivo and in vitro. Our results suggest that c-fos is agenetic regulator for cellular mechanisms mediating neuronalexcitability and survival.We made a mouse with the loxP-c-fos-loxP insertion (designatedas f/fc-fos; Fig. 1a,b). We then crossed the f/fc-fos mouse with aT50 CaMKIIα-cre transgenic mouse15. Southern blotting identi-fied mice carrying both the homozygous f/fc-fos gene (Fig. 1b)and the cre transgene (f/fc-fos-cre; Fig. 1c).

We quantified both basal and induced c-fos expression in the hip-pocampus and dentate of wildtype, f/fc-fos and f/fc-fos-cre micebefore and after injections of kainic acid. Kainic acid is a glutamateanalog that elicits seizures by directly stimulating glutamate recep-tors (GluRs) and indirectly increasing the release of excitatoryamino acids from nerve terminals13,14. In situ hybridizationdemonstrates that whereas basal levels of c-fos are very low and

comparable in nontreated wildtype and f/fc-fos mice,equal levels of c-fos induction are seen in pyramidal neu-rons in CA3, CA2 and CA1 regions of the hippocampusand in granule cells in the dentate gyrus, in both geno-types, upon treatment with kainic acid (Fig. 2a−d,g). Incontrast, f/fc-fos-cre mice at ten weeks of age or older havevirtually no basal c-fos expression and at least a 95%reduction in c-fos induction in the CA3, CA2 and CA1regions and a 70% reduction in the dentate, comparedwith wildtype and f/fc-fos mice (Fig. 2a−g) and with cretransgenic mice (data not shown). Notably, there seems tobe equal c-fos induction by kainic acid in other brainregions, including the entorhinal cortex, median emi-nence and cortex in f/fc-fos-cre, f/fc-fos and wildtype mice(see Web Fig. 1a−j on the supplementary informationpage of Nature Genetics online). Immunostaining con-firms the in situ hybridization results (Fig. 2h−m). f/fc-fos-cre mice show no obvious deficiencies in braindevelopment (Web Fig. 2a−i).

To investigate the role of c-fos in seizures induced bykainic acid, we treated f/fc-fos-cre, wildtype, f/fc-fos andcre transgenic mice with kainic acid intraperitoneally(i.p.) and scored the degrees of seizure16. Kainic acid

letter

416 nature genetics • volume 30 • april 2002

c-fos regulates neuronal excitability and survival

Jianhua Zhang1, Dongsheng Zhang1, Jill Slane McQuade1, Michael Behbehani2, Joe Z. Tsien3 & Ming Xu1

1Department of Cell Biology, Neurobiology and Anatomy and 2Department of Molecular and Cellular Physiology, University of Cincinnati College ofMedicine, Cincinnati, Ohio 45267, USA. 3Department of Molecular Biology, Princeton University, Princeton, New Jersey 08544, USA. Correspondenceshould be addressed to M.X. (e-mail: ming.xu@uc.edu).

Published online: 4 March 2002, DOI: 10.1038/ng859

Fig. 1 Generation of f/fc-fos-cre mice. a, The c-fos genomic DNAlocus, the targeting vector, the floxed c-fos gene locus, and the 5′and 3′ hybridization probes. The neor transcription direction is thesame as that of the c-fos. Open boxes represent c-fos exons; blackarrows indicate the loxP sites. Restriction sites shown: Bg, BglII; Cl,ClaI; Sp, SpeI. b, Identification of the f/fc-fos mice. Tail DNA fromtwo litters of heterozygous intercross was digested with BglII, elec-trophoretically separated, transferred onto membranes andhybridized with a 3′ probe for c-fos. c, Identification of the f/fc-fos-cre mice by Southern hybridization of the same litters of mice as inb, using a cre-specific probe.

5' probe

5.2 kb

Sp

10 kb

8 kb

BgSp Sp

14 kb

Bg

Sp

neo

BgCl

3' probe

BgSp

neo r

BgCl

rtargetingvector

wildtype c-fos

loxP-c-fos-loxP

loxP loxP

loxP loxP

cre

wildtype

loxP-c-fos-loxP

a

b

c

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://g

enet

ics.

nat

ure

.co

m

letter

nature genetics • volume 30 • april 2002 417

Fig. 2 Basal expression and induction of c-fos is significantly reduced in the hip-pocampus in f/fc-fos-cre mice. a–f, We hybridized 3–5 coronal brain sectionsfrom each wildtype (a,b), f/fc-fos (c,d), and f/fc-fos-cre mouse (e,f) with a c-fos–specific probe. Panesl a, c and e show brain sections from untreated mice. Panelsb, d and f show brain sections from mice treated with 20 mg kg–1 of kainic acid. g, Quantification of c-fos expression in the hippocampus and dentate of f/fc-fos-cre and two groups of control mice using in situ hybridization. Data represent mean ± s.e.m. NT, no treatment; DG, dentate gyrus. h–m, We stained 3–5 coronalbrain sections from each wildtype (h,i), f/fc-fos (j,k), and f/fc-fos-cre mouse (l,m) with an anti–c-Fos antibody. Panels h, j and l show brain sections from untreatedmice. Panels i, k and m show brain sections from mice treated with 20 mg kg–1 of kainic acid. For these experiments, 2 untreated mice and 3–6 mice treated withkainic acid from each genotype were used. Mice treated with 30 mg kg–1 of kainic acid gave results parallel to those obtained for mice treated with 20 mg kg–1.

induced seizures of progressive severity in all groups of mice(Fig. 3a,b). Whereas there are no differences among the three con-trol groups, f/fc-fos-cre mice show a higher degree of seizure at 20 mgkg−1 of kainic acid than the control groups (Fig. 3a; P<0.001).Although all four groups of mice showed apparently similar degreesof seizure when injected with 30 mg kg–1 of kainic acid (Fig. 3b),f/fc-fos-cre mice died more frequently, after extensive convulsiveseizures, than the three control groups (Fig. 3c, P<0.05). Thus,

kainic acid also induced more severe seizures in f/fc-fos-cre micethan in control mice at the 30 mg kg−1 dose. The increased seizureinduction is specific to kainic acid, as f/fc-fos-cre mice do not differfrom the three control groups of mice in seizures induced by aGABAergic inhibitor pentetrazole (PTZ) at doses of both 30 mgkg−1 (Fig. 3d) and 50 mg kg−1 (Fig. 3e; P > 0.05). These results indi-cate that hippocampal mutation of c-fos leads to an increase in theseverity of kainic acid–induced seizures in f/fc-fos-cre mice.

perc

ent p

ositi

ve c

ells

NT KA NT KA NT KA NT KA

0

20

40

60

80

100

120 CA1+/+

cre

CA2 CA3 DGf/fc-fosf/fc-fos-cre

a b

c d

e f

g

h i

j k

l m

0 5 10 15 20 25 30 35 400

1

2

3

4

5

cre

seiz

ure

scor

e

time (min)

+/+

–/–f/f

Fig. 3 Kainic acid induces exceptionally severeseizures in the f/fc-fos-cre mouse. a–c, Wildtype(+/+), cre transgenic (cre), f/fc-fos (f/f) andf/fc-fos-cre (–/–) mice were injected with kainicacid at 20 mg kg–1 (a, n=4–6 mice each) or 30 mgkg–1 (b, n=15–24 mice each). The percentage ofmice that died after kainic acid administrationat 30 mg kg–1 (c) is shown. d,e, Wildtype (+/+),cre transgenic (cre), f/fc-fos (f/f) and f/fc-fos-cre(–/–) mice were injected with PTZ at 30 mg kg–1

(d, n=4–5 mice each) or 50 mg kg–1 (e, n=4–7mice each). Seizures were recorded and plotted.Data in a,b,d,e represent mean ± s.e.m.

0

1

2

3

4

5

6

seiz

ure

scor

e

cre f/ f

genotype

+/+ –/–0

1

2

3

4

5

6

seiz

ure

scor

e

genotypecre f/f+/+ –/–

0 5 10 15 20 25 30 35 400

1

2

3

4

5

time (min)

seiz

ure

scor

e

0

10

20

30

40

50

60

% d

eath

genotype

cre f/f+/+ –/–

a b c

d e

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://g

enet

ics.

nat

ure

.co

m

letter

418 nature genetics • volume 30 • april 2002

To test whether the increased severity of kainic acid–inducedseizures seen in f/fc-fos-cre mice reflects increased neuronalexcitability, we compared the electroencephalogram (EEG) pat-terns in cortex of wildtype and f/fc-fos-cre mice17. There was nosignificant difference in baseline EEGs between f/fc-fos-cre andwildtype mice in the amplitudes of spike-wave discharges in theabsence of kainic acid (Fig. 4a). When we injected kainic acid i.p.at the 10 mg kg−1 dose, which is a sub-threshold for seizure induc-tion, both f/fc-fos-cre and wildtype mice showed amplitudes ofspike-wave discharges at higher than baseline levels (Fig. 4a,b).However, both the amplitudes of kainic acid–induced spike-wavedischarges (Fig. 4b) and the frequency of spike-wave dischargeswhose amplitudes are higher than baseline levels (Fig. 4c, P<0.05)are significantly higher in f/fc-fos-cre mice than in wildtype mice.Thus, the c-fos mutation leads to increased neuronal excitabilityinduced by kainic acid in f/fc-fos-cre mice.

Rodent CA3 pyramidal neurons are more susceptible to kainicacid–induced death than other neurons in the hippocampus14,18.To test whether c-fos may be neuroprotective against kainicacid–induced excitotoxicity, we assessed the degree of neuronaldamage in the hippocampus in the four groups of mice that sur-vived the 30 mg kg−1 of kainic acid treatment. Nissl staining ofvarious brain sections indicates that f/fc-fos-cre mice have agreater amount of neuronal damage induced by kainic acid in theCA3 region (Fig. 5b,d) than wildtype (Fig. 5a,c), f/fc-fos and cretransgenic mice (data not shown). Immunostaining for the glialfibrillary acidic protein (GFAP) indicates that there is more glio-

sis in the CA3 region in f/fc-fos-cre mice (Fig. 5f) than in wildtype(Fig. 5e), f/fc-fos and cre transgenic mice (data not shown). Theexcitotoxic neuronal cell death in the mutant mice is accompa-nied by more TUNEL staining in the CA3 region (Fig. 5h) than inwildtype mice (Fig. 5g). Quantification of lesion volumes in theCA3 region in all groups of mice shows a markedly greateramount of CA3 cell death induced by kainic acid in f/fc-fos-cremice than in the three control groups (Fig. 5i; P<0.05). Thus,f/fc-fos-cre mice have more cell death in the CA3 region than con-trol mice after 30 mg kg−1 of kainic acid exposure.

It is thought that c-Fos functions through AP-1 transcriptioncomplexes to influence target gene expression. Supershift experi-ments19 indicate that there are changes in both level and compo-sition in AP-1 complexes in the hippocampus and dentate, owingto a lack of c-Fos in f/fc-fos-cre mice, compared with controlmice, before and after exposure to kainic acid (Web Fig. 3a−c).Moreover, western blotting indicates that in the absence of c-fosinduction, absolute FosB levels are reduced; by contrast, Fra-1and Fra-2 levels are higher in the hippocampus and dentate inf/fc-fos-cre mice, compared with wildtype mice, four hours aftertreatment with kainic acid (Fig. 6a). Together, our results suggestthat c-fos regulates the overall composition of AP-1 transcriptioncomplexes both by direct participation and by regulation of otherIEG expression.

The altered dynamic regulation of AP-1 complexes may lead tochanges in target gene expression. To test this possibility, weselected two classes of candidate genes and investigated whether

Fig. 4 The f/fc-fos-cre mouse shows greater kainicacid–induced neuronal excitability than the wild-type mouse. a,b, We recorded EEG patterns inf/fc-fos-cre (–/–) and wildtype (+/+) mice (n=6each) in the absence (a) or presence (b) of 10 mgkg–1 of kainic acid. c, Histogram of the frequencyof spike-wave discharges whose amplitudes arehigher than baseline levels in the two groups ofmice after injections of kainic acid. Data repre-sent mean ± s.e.m. of 2.5 s of sampling. The aster-isk indicates P<0.05.

2 s

26 µ V

0 2 4 6

+/+

–/–

aver

age

num

ber

of s

pike

s

0

10

20

30

40

50

*

+/+ –/–

a b c

genotype+/+ cre f/ f –/–

0

1

2

3

4

5

*

volu

me

of le

sion

(× 1

05

µm3)

Fig. 5 The f/fc-fos-cre mouse has a high rate of kainic acid–induced CA3 celldeath. Five days after seizure induction by 30 mg kg–1 of kainic acid, mice wereperfused and their brains sectioned. a–d, We carried out Nissl staining (a and c,n=7 wildtype mice; b and d, n=5 f/fc-fos-cre mice). e–h, We also carried outimmunostaining for GFAP (e,f) and TUNEL staining (g,h), using the above brainsections for both wildtype (e,g) and f/fc-fos-cre (f,h) mice. The scale bars indicate1 mm for a, b, e, f, and 65 µm for c, d, g, h. Arrows indicate lesioned brainregions. i, We carried out quantification of lesion volumes in the CA3 region ofthe hippocampus for all groups of mice at 30 mg kg–1 of kainic acid (n=5–7 miceeach). Data represent mean ± s.e.m.; +/+, cre, f/f, and –/– represent wildtype, cretransgenic, f/fc-fos and f/fc-fos-cre mice, respectively. The asterisk indicatesP<0.05. There is no major difference in neuronal damage among the fourgroups of mice in amygdala and pyriform cortex after exposure to kainic acid.

a b

c d

e f

g h

i

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://g

enet

ics.

nat

ure

.co

m

letter

nature genetics • volume 30 • april 2002 419

c-fos regulates their expression both in vivo and in vitro. Theionotropic GluRs mediate the fast excitatory neuronal transmis-sion20. Several GluR genes have AP-1 binding sites in their pro-moter regions. We found a 1.8-fold increase in basal GluR6expression in the hippocampus and dentate in naive f/fc-fos-cremice compared with wildtype mice (Fig. 6b; P<0.01), whereasGluR6 is equally expressed in other brain regions, including cor-tex, striatum and cerebellum, in f/fc-fos-cre and wildtype mice(Fig. 6b; P>0.15). Basal expression of the other tested GluRs aresimilar in naive f/fc-fos-cre and wildtype mice (data not shown).Thus, basal levels of c-Fos contribute to regulation of GluR6expression in the hippocampus.

The neutrophin BDNF and its high- and low-affinity receptors,TrkB and p75NTR, are implicated in neuronal protection againstkainic acid–induced seizures21−25. We examined whether c-fos isinvolved in regulating their expression in the hippocampus anddentate. Four hours after exposure to kainic acid, BDNF levels inf/fc-fos-cre mice remain similar to basal levels, whereas they aretwofold higher than basal levels in wildtype mice (Fig. 6c;P<0.01). Basal and kainic acid–induced BDNF expression at the48-hour time point are not overtly affected by the c-fos mutation.Both basal and kainic acid–induced levels of TrkB and p75NTRare similar in f/fc-fos-cre and wildtype mice at all time points mea-sured (data not shown). These results indicate that there is adelayed BDNF induction by kainic acid in the hippocampus inf/fc-fos-cre mice, compared with wildtype mice.

To further test whether c-fos regulates GluR6, BDNF and fosBexpression, we transiently transfected several cell lines with c-fosand measured the expression of the three genes. Similar to theresults observed in vivo, an increasing amount of c-Fos reducesGluR6 expression (Fig. 6d) while increasing BDNF or fosB expres-sion in vitro (Fig. 6e). The parallel changes observed both in vivoand in vitro suggest that the increased basal GluR6 expression andthe delayed BDNF induction by kainic acid in f/fc-fos-cre mice,compared with wildtype mice, are most likely due to a lack of directc-Fos−mediated transcriptional regulation rather than to variousindirect mechanisms or effects of the c-fos mutation.

Together, our results support a genetic model in which c-Fosparticipates in key cellular mechanisms underlying both neu-ronal excitability and protection by selectively regulating geneexpression in the brain. In normal neurons, basal c-Fos or c-Fosinduced by neuronal activity orchestrates the formation of AP-1

Fig. 6 c-fos regulates fosB, GluR6 and BDNF expressionin vivo and in vitro. Nuclear extracts were isolatedfrom hippocampi, dentate and other brain regions ofindividual wildtype (+/+) and f/fc-fos-cre (–/–) mouse(n=4 mice for each time point) either before or after30 mg kg–1 of kainic acid injections. a–c, Western blotsfor the indicated IEG products (a) and BDNF (c) before(0 h) and after 4 h or 5 d of treatment with kainic acid,and for GluR6 (b) before exposure to kainic acid.Equal amounts of protein were loaded in each lane.The status of only the full-length FosB is included, asour pan-Fos antibody and FosB antibody do not con-sistently detect ∆FosB upon administration of kainicacid. Expression of the R1 subunit of NMDA, GluR2,GluR3 and GluR7 receptors does not differ betweenmutant and control brains. Each experiment wasrepeated four times, and identical results wereobtained. Hip, hippocampus; Ctx, cortex; CPu, caudop-utamen; Cb, cerebellum. Increasing amounts of c-fos(1 and 2) were transfected into the indicated cells induplicate, with mock-transfected cells (M) as controls.d,e, (see Methods) Forty-eight hours after transfec-tion, cell extracts were isolated, and equal amounts ofproteins were used for GluR6 (d), BDNF and FosB (e) aswell as c-Fos (d,e) western blotting. Parallel resultswere obtained both within each transfection andamong different transfections.

+/+ –/– –/– –/–+/+ +/+

M 1 2 M 1 2

M 1 2

GluR6

actin

+/+ +/+ –/– –/– –/– –/– +/+ +/+

+/+ +/+ +/+ +/+–/– –/– –/– –/–

c-Fos

FosB

Fra-1

Fra-2

c-Jun

JunB

JunD

BDNF

actin

GluR6

c-Fos

c-Fos

BDNF

FosB

NT2

0 h 4 h 5 d 0 h 4 h 24 h 48 h

Hip Ctx CPu Cb

Neuro2A U251

a

b

c

d

e

transcription complexes, either by direct participation or by reg-ulation of other IEG expression. Basal AP-1 complexes regulatethe expression of GluR6, which mediates neuronal excitability26.Induced AP-1 complexes regulate BDNF to help cell survivalafter excessive stimulation. The dual regulation of the two cellu-lar mechanisms by c-Fos ensures appropriate neuronal excitabil-ity and protects neurons from potential excitotoxicity. Futurestudies may identify additional molecular targets and cellularpathways underlying excitability, survival and plasticity.

MethodsGeneration of f/fc-fos-cre mice. We designed a c-fos targeting construct todelete the coding sequence for the DNA binding domain and the leucinezipper domain that are important for heterodimerization with the Junfamily proteins. We generated homologous recombinants and chimericmice as described27. We bred male chimeric mice with C57BL/6 femalesand identified germline transmission and f/fc-fos mice by genomic South-ern blotting. We then crossed f/fc-fos mice with the T50 mice carrying a cretransgene driven by a CaMKIIα promoter15. The genotypes of the off-spring were identified by genomic Southern blotting with 5′ and 3′ probesfor loxP-c-fos-loxP and with a cre gene probe. We used mutant and theirvarious control littermates at 10−18 wk for all subsequent analyses.

Seizure scoring and statistical analysis. We injected mice i.p. with kainic acidat 20 mg kg−1 or 30 mg kg−1 of body weight. We then scored kainicacid–induced seizures every 5 min for 2 h, according to Yang et al.16. Thisseizure scoring system does not distinguish between generalized tonic-clonicactivity and death, with death clearly representing a higher degree of seizure.We injected PTZ similarly at 30 mg kg−1 or 50 mg kg−1 of body weight andscored seizures based on the highest degree of seizure within 15 min of thePTZ injection. For seizure data analysis, we constructed a 2 × 4 contingencytable with the degree of response to kainic acid injections grouped as less than3 and greater than or equal to 3, followed by Chi-square analysis. The com-parison between the f/fc-fos-cre group and the other three control groups wascarried out with Freeman-Tukey’s multiple comparison, after the percentagedata were transformed with an Arcsine function. The kainic acid–induceddeath rate comparison was carried out by Chi-square analysis. We analyzedthe PTZ seizure results using an unpaired t-test between the f/fc-fos-cre andthe three control groups of mice.

In situ hybridization and quantification. We fixed and processed brain sec-tions27. We then used mouse c-fos cDNA to make a [35S]UTP-labeled anti-sense riboprobe. We used a sense probe as a control. After hybridization, theslides were dehydrated, dipped in NTB2 photographic emulsion (Kodak) and

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://g

enet

ics.

nat

ure

.co

m

letter

420 nature genetics • volume 30 • april 2002

stored. We stained brain sections with cresyl violet and took pictures of eacharea of each section at ×50 magnification, using NIH Image for quantifica-tion. We then counted the silver grains on each cell in defined areas of CA1,CA2, CA3, dentate gyrus and other brain regions with matching sections forall mice, as well as in cell-sized areas of the background. Any cell that con-tained six times or more the number of grains as the background was countedas a positive cell. We derived this number by comparing values from knownpositives and negatives and determining a fair value for what is positive.

Immunohistochemistry. We used 40-µm brain sections with antibodiesfor c-Fos (Oncogene Research Products), c-Jun (Santa Cruz Biotechnolo-gy), pSer73-c-Jun (New England Biolab) and GFAP (Dako). We incubatedthe sections with a biotinylated secondary antibody27 (Vector Laborato-ries) followed by ABC reagent (ABC kit, Vector Laboratories). We visual-ized the immunoreaction by treating the sections in 0.05% diaminobenzi-dine with hydrogen peroxide.

Electroencephalogram analysis. We placed two blunt-tip tungsten elec-trodes on the dura above the sensory motor cortex and a reference electrodeon the nasal bone17. We used a high-input impedance amplifier for recordingand recorded the signals on a digital tape. We first recorded baseline EEGs inthe absence of kainic acid. We next injected kainic acid i.p. at 5 mg kg−1 andrecorded EEG for 90 min. We then injected kainic acid at 10 mg kg−1 andrecorded EEG for another 90 min. Samples of 2.5-s EEG recordings at every5-min interval for each mouse were compared between f/fc-fos-cre and wild-type mice for spike-wave discharges. We carried out Student’s t-tests betweengroups to determine the significance in genotypes.

Evaluating neuronal damage. We stained serial brain sections from thesurviving mice with cresyl violet, for neuronal damage evaluation, asdescribed16. We quantified neuronal loss by measuring the areas of severeneuronal damage on every second section, using the Metamorph Imagingsystem, and integrated sections to calculate volumes of damage of the fourgenotypes in dorsal hippocampus. An unpaired t-test was done on thelesion volumes between the f/fc-fos-cre and each of the three control groupsof mice. We evaluated neuronal damage−induced gliosis by immunostain-ing for GFAP. We also carried out the TUNEL assay with the above brainsections using the In Situ Cell Death Detection Kit in conjunction with theperoxidase-conjugated anti-fluorescein antibody detection system(TUNEL POD) from Boehringer Mannheim.

Protein extract preparation and analysis of DNA-binding activities.We prepared nuclear extracts from hippocampi and dentate19. We ana-lyzed DNA-binding activities of the nuclear proteins using the elec-trophoretic mobility shift assay19. We mixed nuclear extracts with anend-labeled AP-1 oligonucleotide and electrophoresed the mixtures in anondenaturing gel. Gels were dried and exposed. We scanned autoradi-ographs using a Molecular Dynamics scanner and analyzed the results byImageQuant software. We carried out supershifts as described19. We usedpolyclonal antibodies (Santa Cruz Biotechnology) against c-Fos, FosB,Fra-1, Fra-2, c-Jun, JunB and JunD. We repeated each experiment fourtimes and obtained identical results.

Cell transfection and protein extract preparation. Both the mouse neu-roblastoma Neuro2A and the human glioblastoma U251 cells expressendogenous BDNF, and the human teratocarcinoma NT2 cells showendogenous GluR6 expression. We transfected 80% confluent cells with 1−4 µg of pCMV-c-fos containing rat c-fos with a deletion of the 3′ UTR, tostabilize the c-Fos products by lipofection (FuGENE 6, Roche). Forty-eighthours after transfection, we harvested, washed and homogenized the cellsand used the extracts for western blotting. All transfections were done induplicate and each transfection was repeated at least three times.

Western blotting and data analysis. We isolated hippocampi and dentatefrom each mouse individually at different time points and homogenizedthe tissues. These extracts, and those from the transfection experiments,were quantified and separated by SDS−PAGE, transferred to nitrocellulosemembranes, probed with antibodies (Santa Cruz Biotechnology) against

c-Fos, FosB, Fra-1, Fra-2, c-Jun, JunB, JunD, BDNF, TrkB, p75NTR,NMDA R1, GluR2-3, GluR6-7 and actin, and secondary antibodies. Wevisualized the results by enhanced chemiluminenscence (Amersham).Western blotting for each sample was done at least twice. We quantifieddata by densitometer scanning followed by Student’s t-tests.

Note: Supplementary figures are available on the Nature Geneticsweb site.

AcknowledgmentsWe are grateful to J. Duffy and M. Yin for mouse blastocyst injections. Wethank L. Chen, R. Hennigan, H. Jansen, A. Kuan, H. Lee, D. Lou, M.Privitera, L. Sherman and R. Walsh for various advice, help and discussions.J.Z. and M.X. are supported by grants from the National Institutes of Health,the National Alliance for Research on Schizophrenia and Depression and theEpilepsy Foundation of America. J.S.M. is partially supported by an NIHpredoctoral training grant.

Competing interests statementThe authors declare that they have no competing financial interests.

Received 6 December 2001; accepted 23 January 2002.

1. Choi, D.W. Glutamate neurotoxicity and diseases of the nervous system. Neuron1, 623–634 (1988).

2. Coyle, J.T. & Puttfarcken, P. Oxidative stress, glutamate, and neurodegenerativedisorders. Science 262, 689–695 (1993).

3. Lee, J.-M., Zipfel, G.J. & Choi, D.W. The changing landscape of ischaemic braininjury mechanisms. Nature 399, A7–A14 (1999).

4. McNamara, J.O. Emerging insights into the genesis of epilepsy. Nature 399,A15–A22 (1999).

5. Muller, R., Tremblay, J.M., Adamson, E.D. & Verma, I.M. Tissue and cell type-specific expression of two human c-onc genes. Nature 304, 454–456 (1983).

6. Greenberg, M.E. & Ziff, E.M. Stimulation of 3T3 cells induces transcription of thec-fos proto-oncogene. Nature 311, 433–438 (1984).

7. Chiu, R. et al. The c-Fos protein interacts with c-Jun/AP-1 to stimulatetranscription of AP-1 responsive genes. Cell 54, 541–542 (1988).

8. Halazonetis, T.D., Georgopoulos, K., Greenberg, M.E. & Leder, P. c-Jun dimerizeswith itself and with c-Fos, forming complexes of different DNA binding affinities.Cell 55, 917–924 (1988).

9. Kouzarides, T. & Ziff, E. The role of the leucine zipper in the Fos-Jun interaction.Nature 336, 646–651 (1988).

10. Smeyne, R.J. et al. Continuous c-fos expression precedes programmed cell deathin vivo. Nature 363, 166–169 (1993).

11. Popovici, T. et al. Effects of kainic acid-induced seizures and ischemia on c-fos-likeproteins in rat brain. Brain Res. 536, 183–194 (1990).

12. Smeyne, R.J. et al. Fos-lacZ transgenic mice: mapping sites of gene induction inthe central nervous system. Neuron 8, 13–23 (1992).

13. Ferkany, J.W., Zaczek, R. & Coyle, J.T. The mechanism of kainic acid neurotoxicity.Nature 308, 561–562 (1984).

14. Ben-Ari, Y. Limbic seizure and brain damage produced by kainic acid: mechanismsand relevance to human temporal epilepsy. Neuroscience 14, 375–403 (1985).

15. Tsien, J.Z. et al. Subregion- and cell type-restricted gene knockout in mouse brain.Cell 87, 1317–1326 (1996).

16. Yang, D.D. et al. Absence of excitotoxicity-induced apoptosis in the hippocampusof mice lacking the Jnk3 gene. Nature 389, 865–870 (1997).

17. Medvedev, A., Mackenzie, L., Hiscock, J.J. & Willoughby, J.O. Kainic acid inducesdistinct types of epileptiform discharge with differential involvement ofhippocampus and neocortex. Brain Res. Bull. 52, 89–98 (2000).

18. Nadler, J.V., Perry, B.W. & Cotman, C.W. Intraventricular kainic acid preferentiallydestroys hippocampal pyramidal cells. Nature 271, 676–677 (1978).

19. Mandelzys, A., Gruda, M.A., Bravo, R. & Morgan, J.I. Absence of a persistentlyelevated 37 kDa fos-related antigen and AP-1-like DNA-binding activity in thebrains of kainic acid-treated fosB null mice. J. Neurosci. 17, 5407–5415 (1997).

20. Hollmann, M. & Heinemann, S. Cloned glutamate receptors. Ann. Rev. Neurosci.17, 31–108 (1994).

21. Koh, J.-Y., Gwag, B.J., Lobner, D. & Choi, D.W. Potentiated necrosis of culturedcortical neurons by neurotrophins. Science 268, 573–575 (1995).

22. Tandon, P., Yang, Y., Das, K., Holmes, G.L. & Stafstrom, C.E. Neuroprotectiveeffects of brain-derived neurotrophic factor in seizures during development.Neuroscience 91, 293–303 (1999).

23. Gall, C.M. Seizure-induced changes in neurotrophin expression: implications forepilepsy. Exp. Neurol. 124, 150–166 (1993).

24. Merlio, J.-P. et al. Increased production of the TrkB protein tyrosine kinasereceptor after brain insults. Neuron 10, 151–164 (1993).

25. Roux, P.P., Colicos, M.A., Barker, P.A. & Kennedy, T.E. p75 neurotrophin receptorexpression is induced in apoptotic neurons after seizure. J. Neurosci. 19,6887–6896 (1999).

26. Mulle, C. et al. Altered synaptic physiology and reduced susceptibility to kainate-induced seizures in GluR6-deficient mice. Nature 392, 601–605 (1998).

27. Xu, M. et al. Dopamine D1 receptor mutant mice are deficient in striatalexpression of dynorphin and in dopamine-mediated behavioral responses. Cell79, 729–742 (1994).

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://g

enet

ics.

nat

ure

.co

m