B R A I N R E S E A R C H 1 0 7 8 ( 2 0 0 6 ) 2 2 7 – 2 3 4

ava i l ab l e a t www.sc i enced i rec t . com

www.e l sev i e r. com/ l oca te /b ra in res

Research Report

Facilitation of granule cell epileptiform activity by mossyfiber-released zinc in the pilocarpine model of temporallobe epilepsy

Olga Timofeeva, J. Victor Nadler⁎

Department of Pharmacology and Cancer Biology, Box 3813, Duke University Medical Center, Durham, NC 27710, USADepartment of Neurobiology, Box 3209, Duke University Medical Center, Durham, NC 27710, USA

A R T I C L E I N F O

⁎ Corresponding author. Department of PharmUSA. Fax: +1 919 681 8609.

E-mail address: nadle002@acpub.duke.ed

0006-8993/$ – see front matter © 2006 Elsevidoi:10.1016/j.brainres.2006.01.051

A B S T R A C T

Article history:Accepted 16 January 2006Available online 21 February 2006

Recurrent mossy fiber synapses in the dentate gyrus of epileptic brain facilitate thesynchronous firing of granule cells and may promote seizure propagation. Mossy fiberterminals contain and release zinc. Released zinc inhibits the activation of NMDA receptorsand may therefore oppose the development of granule cell epileptiform activity.Hippocampal slices from rats that had experienced pilocarpine-induced status epilepticusand developed a recurrent mossy fiber pathway were used to investigate this possibility.Actions of released zinc were inferred from the effects of chelation with 1 mM calciumdisodium EDTA (CaEDTA). When granule cell population bursts were evoked by mossy fiberstimulation in the presence of 6 mM K+ and 30 μM bicuculline, CaEDTA slowed the rate atwhich evoked bursting developed, but did not change themagnitude of the bursts once theyhad developed fully. The effects of CaEDTA were then studied on the pharmacologicallyisolated NMDA receptor- and AMPA/kainate receptor-mediated components of the fullydeveloped bursts. CaEDTA increased themagnitude of NMDA receptor-mediated bursts andreduced the magnitude of AMPA/kainate receptor-mediated bursts. CaEDTA did not affectthe granule cell bursts evoked in slices fromuntreated rats by stimulating the perforant pathin the presence of bicuculline and 6mMK+. These results suggest that zinc released from therecurrent mossy fibers serves mainly to facilitate the recruitment of dentate granule cellsinto population bursts.

© 2006 Elsevier B.V. All rights reserved.

Keywords:HippocampusEpilepsyMossy fiberDentate gyrusZinc

Abbreviations:aCSF, artificial cerebrospinal fluidD-AP5, 2-amino-5-phosphonopentanoateCaEDTA, calcium disodium EDTANBQX, 2,3-dihydroxy-6-nitro-7-sulfamyl-benzo(F)quinoxaline

1. Introduction

The formation or expansion of recurrent excitatory circuitsmay play an important role in epileptogenesis. In epileptic, butnot in normal, brain, recurrent mossy fibers interconnectgranule cells of the dentate gyrus, synchronizing their firingand presumably facilitating participation of the dentate gyrusin seizures (Nadler, 2003). Dentate granule cells normally

acology and Cancer Biolo

u (J.V. Nadler).

er B.V. All rights reserved

resist the propagation of seizures from the entorhinal cortexto the hippocampus (Collins et al., 1983; Stringer et al., 1989;Lothman et al., 1992). Thus, a reduced threshold for epilepti-form discharge by this neuronal population may facilitate thedevelopment of limbic seizures.

The hippocampal mossy fibers probably contain more zincthan any other pathway in the brain (Frederickson andDanscher, 1990). Zinc is sequestered in the synaptic vesicles

gy, Box 3813, Duke University Medical Center, Durham, NC 27710,

.

Fig. 1 – CaEDTA slowed the development of mossyfiber-evoked granule cell bursts. (A, B) Representative tracesrecorded 40 (A) and 70 (B) min after the change to superfusionmedium that contained 6 mM KCl and 30 μM bicuculline(Bic) ± 1mMCaEDTA. A smaller response was recorded in thepresence of CaEDTA 40 min after the change of medium, butno consistent difference remained at 70 min. *Antidromic

228 B R A I N R E S E A R C H 1 0 7 8 ( 2 0 0 6 ) 2 2 7 – 2 3 4

of mossy fiber boutons (Perez-Clausell and Danscher, 1985)and can be released in a Ca2+-dependent manner by K+ (Assafand Chung, 1984; Aniksztejn et al., 1987) or by electricalstimulation of the pathway (Howell et al., 1984; Budde et al.,1997; Li et al., 2001a,b; Ueno et al., 2002). Studies of mutantmice that lack vesicular transport of zinc (Lee et al., 2000;Lopantsev et al., 2003) suggest that zinc may also be releasedby a non-vesicular mechanism. Exogenous zinc can eitherpromote (Reece et al., 1994) or inhibit (Xu and Mitchell, 1993)epileptiform activity in the hippocampus. It has been sug-gested that zinc released from the recurrent mossy fiberpathway contributes to granule cell epileptiform activity byoverflowing the synapse and reducing the activation of nearbyGABAA receptors (Buhl et al., 1996; Gibbs et al., 1997; Brooks-Kayal et al., 1998; Cohen et al., 2003). However, tests of thishypothesis in hippocampal slices yielded no supportingevidence (Molnár and Nadler, 2001a). The zinc chelatorcalcium disodium EDTA (CaEDTA; 10 mM) reportedly preventsthe development of long-term potentiation (LTP) at mossyfiber synapses on CA3 pyramidal cells (Lu et al., 2000; Li et al.,2001a). It has been postulated that LTP depends on thetranslocation of zinc into the postsynaptic cell and/or thesynaptic bouton (Li et al., 2001a,b). Thus, zinc may contributeto granule cell excitability by mechanisms other than block ofsynaptic inhibition. Conversely, considerable evidence sug-gests that endogenous zinc reduces excitability in the brain.Mice that lack vesicular zinc (Cole et al., 2000) and rats fed azinc-deficient diet (Takeda et al., 2003) are more susceptiblethan normal animals to limbic seizures evoked by kainic acid.The antiseizure activity of zinc may be explained by its non-competitive antagonismof NMDA receptors (Peters et al., 1987;Westbrook and Mayer, 1987; Chen et al., 1997; Paoletti et al.,1997). Blockade of NMDA receptors attenuates granule cellepileptiform activity supported by the recurrent mossy fiberpathway (Patrylo and Dudek, 1998; Okazaki and Nadler, 2001).Release of zinc at mossy fiber synapses on both pyramidal (Luet al., 2000; Vogt et al., 2000) and granule (Molnár and Nadler,2001b) cells inhibits the ability of glutamate to activatepostsynaptic NMDA receptors. Zinc can also inhibit glutamaterelease from mossy fiber terminals (Bancila et al., 2004;Quinta-Ferreira and Matias, 2004; Takeda et al., 2004; Quinta-Ferreira and Matias, 2005), probably by activating presynapticATP-dependent K+ channels (Bancila et al., 2004). Theseresults suggest that zinc from recurrent mossy fibers shouldserve as a brake on epileptiform activity in the dentate gyrus.We tested this possibility in studies on hippocampal slicesprepared from pilocarpine-treated rats that had developedstatus epilepticus and showed histochemical evidence of arecurrent mossy fiber pathway. Actions of zinc were inferredfrom the effects of CaEDTA on granule cell bursts evoked bymossy fiber stimulation.

population spike. (C) CaEDTA delayed the appearance ofevoked epileptiform activity, but did not alter its ultimatemagnitude (n = 8 slices for each group). Arrow indicates thechange to bicuculline/6 K+medium. (D) CaEDTA increased thehalf-time for development of amaximal response. The size ofthe maximal response was computed by averaging thecoastline indices of the 10 responses recorded from 65.5 to70min after the change to bicuculline/6 K+ medium. *P b 0.05,Student's t test.

2. Results

2.1. CaEDTA slowed the development of mossyfiber-evoked granule cell bursts

Experiments were performed with hippocampal slices pre-pared from rats that had developed status epilepticus after the

administration of pilocarpine. As reported previously (Molnárand Nadler, 1999; Okazaki et al., 1999; Tu et al., 2005), Timmhistochemistry performed at the end of the experimentrevealed evidence of a recurrent mossy fiber pathway in allslices.When themossy fibers were stimulated in the presenceof 30 μM bicuculline and 6 mM KCl, a granule cell populationburst appeared in 10–15min (Fig. 1C). The response reached itshalf-maximal size in 23 ± 5 min and stabilized within 60–70 min. A response of this type can be evoked only when arecurrent mossy fiber pathway is present (Tauck and Nadler,1985; Hardison et al., 2000; Okazaki and Nadler, 2001).

Fig. 3 – CaEDTA (1 mM) altered the magnitude of theAMPA/kainate and NMDA receptor-mediated components ofmossy fiber-evoked granule cell bursts, but not themagnitude of perforant path-evoked bursts. CaEDTA wasadded to the superfusion medium 70 min after the change tomedium that contained 6 mM KCl and 30 μM bicuculline. Itseffect was assessed 15–19.5 min later, when it had reachedits maximum. Grouped data from 5–8 hippocampal slices areshown. *P b 0.05 (paired t test with respect to recordingsmade

229B R A I N R E S E A R C H 1 0 7 8 ( 2 0 0 6 ) 2 2 7 – 2 3 4

CaEDTA (1 mM) slowed the rate at which granule cellbursts developed (Fig. 1). The evoked burst first appeared∼25 min later than it did in the absence of CaEDTA and thehalf-time for its achieving maximal size increased by anaverage of 80%. However, mossy fiber-evoked bursts ulti-mately reached the same magnitude as control. Addition of1 mM CaEDTA to the superfusion medium after the evokedburst had reached its maximal size did not alter the responsesignificantly (Figs. 2A, B, 3).

2.2. CaEDTA differentially affected the NMDAreceptor- and AMPA/kainate receptor-mediated components ofthe mossy fiber-evoked bursts

The failure of 1 mM CaEDTA to alter the magnitude of themossy fiber-evoked burst significantly appeared to conflictwith the involvement of NMDA receptors in the evoked burst(Patrylo and Dudek, 1998; Okazaki and Nadler, 2001) and theability of CaEDTA to enhance the activation of postsynapticNMDA receptors by released glutamate (Molnár and Nadler,

Fig. 2 – When added to the superfusionmedium after mossyfiber-evoked bursts had reached their maximal size, 1 mMCaEDTA did not change the magnitude of bursts to whichboth AMPA/kainate and NMDA receptors contributed (A, B),but increased the magnitude of NMDA receptor-mediatedbursts (C, D) and reduced the magnitude of AMPA/kainatereceptor-mediated bursts (E, F). All superfusion mediacontained 6 mM KCl and 30 μM bicuculline. NMDAreceptor-mediated bursts were recorded during superfusionwith a nominally Mg2+-free medium that contained 10 μMNBQX. AMPA/kainate receptor-mediated bursts wererecorded during superfusion with a medium that contained50 μM D-AP5. Traces shown were obtained just before (left)and 15–19.5 min after (right) CaEDTA was added to thesuperfusion medium. *Antidromic population spike.

just before exposure to CaEDTA). A/K + NMDA, mossyfiber-evoked bursts mediated by both AMPA/kainate andNMDA receptors; A/K, mossy fiber-evoked bursts mediatedby AMPA/kainate receptors only; NMDA, mossy fiber-evokedbursts mediated by NMDA receptors only; PP, bursts evokedby stimulating the perforant path.

2001b). To determine whether CaEDTA could increase theNMDA receptor-mediated component of mossy fiber-evokedbursts, we isolated this component and the AMPA/kainatereceptor-mediated component pharmacologically.

CaEDTA (1 mM) had opposite effects on the two compo-nents of mossy fiber-evoked granule cell bursts. It more thandoubled the average size of the burstsmediated only by NMDAreceptors (Figs. 2C, D, 3). The number and amplitude of spikeswithin the burst increased, as well as the amplitude of theunderlying slowwave (169 ± 23% of control; P b 0.05, Student's ttest). In contrast, CaEDTA reduced by about half the averagesize of the evoked burst mediated only by AMPA/kainatereceptors (Figs. 2E, F, 3). The number and amplitude ofpopulation spikes within the burst declined, as well as theamplitude of the underlying slow wave (63 ± 27% of control;P b 0.001; Student's t test).

2.3. CaEDTA did not affect perforant path-evoked granulecell bursts

The effect of CaEDTAwas tested on granule cell bursts evokedby stimulating the perforant path, a type of epileptiformactivity that does not normally involve the mossy fibers.Untreated control rats were used for these experiments,because hippocampal slices from these animals lack aphysiologically significant recurrent mossy fiber pathway(Tauck and Nadler, 1985; Hardison et al., 2000; Okazaki andNadler, 2001). Thus, perforant path stimulation did not evoke adisynaptic mossy fiber response in dentate granule cells. Inaddition, 1 mM CaEDTA does not inhibit the ability of releasedglutamate to activate NMDA receptors at perforant pathsynapses to any detectable degree (Molnár and Nadler,

Fig. 4 – CaEDTAdid not affect the development of granule cellbursts evoked by perforant path stimulation. (Top)Representative traces recorded 70 min after the switch tomedium that contained 6 mM KCl and 30 μM bicuculline(Bic) ± 1mMCaEDTA. *Stimulus artifact. (Bottom) CaEDTA didnot affect the development of the response when it wasincluded in the bicuculline/6 K+ medium (n = 10) or when itwas added to the medium 70 min later (n = 8). The shorterarrow indicates the change to bicuculline/6 K+

medium ± CaEDTA. The longer arrow indicates the additionof CaEDTA to the previously CaEDTA-free bicuculline/6 K+

medium.

230 B R A I N R E S E A R C H 1 0 7 8 ( 2 0 0 6 ) 2 2 7 – 2 3 4

2001b), even though the lateral division of the perforant pathcontains a small amount of zinc.

Stimulation of perforant path fibers evoked in the granulecell body layer a positively directed field EPSP on which asmall negatively directed population spike was sometimessuperimposed. When stimulation continued in the presenceof 6 mM KCl and 30 μM bicuculline, a population spikeappeared or the amplitude of the preexisting spike increasedwithin 5 min (Fig. 4). The number of population spikes (to amaximum of 7) and their amplitude increased steadily. Thecoastline index did not reach a stable value within 70 min.CaEDTA (1 mM) had no effect on the development of evokedgranule cell bursting under these conditions. Addition ofCaEDTA after the bursts had already developed for 70 minfailed to alter the rate of their further development and hadno statistically significant effect on their average magnitudewithin 19.5 min (Fig. 3). This result demonstrates that theaction of CaEDTA is pathway-specific and that endogenouszinc does not regulate all forms of granule cell epileptiformactivity.

3. Discussion

This study utilized CaEDTA, a high affinity membrane-impermeant chelating agent, to identify actions of endoge-nous zinc. An action of a zinc chelator when applied by itselfimplies that endogenous zinc has the opposite effect. Thus,our finding that CaEDTA slows the development of granule

cell bursts evoked by stimulating the recurrent mossy fiberpathway suggests that zinc released by this pathway accel-erates the development of such bursts. The effects of CaEDTAon fully developed bursts suggest that released zinc hasmultiple actions that can either enhance or depress burstmagnitude. In particular, it increases the AMPA/kainatereceptor-mediated component of the response and reducesthe NMDA receptor-mediated component. Under our standardexperimental conditions, these actions canceled each other,leading to the erroneous impression that released zinc has noeffect on granule cell epileptiform activity. It should be notedthat these conclusions do not depend on the mechanism bywhich zinc is released from the recurrent mossy fiberpathway, whether exocytotic, non-exocytotic or some combi-nation of the two.

Previous studies support the use of CaEDTA as a specificheavy metal chelator in studies of the mossy fiber pathway,at least at a concentration of 1 mM. At this concentration,CaEDTA alters transmission in the dentate gyrus at the zinc-containing recurrent mossy fiber synapses selectively. Itenhances the NMDA receptor-mediated component of therecurrent mossy fiber EPSC in a voltage-dependent manner,without affecting either the AMPA/kainate receptor-mediat-ed component of the same response or the NMDA receptor-mediated component of the perforant path EPSC (Molnárand Nadler, 2001b). CaEDTA modifies granule cell burstingevoked by mossy fiber stimulation, but not bursting drivenby the perforant path. CaEDTA also does not affect theresponse of granule cells to applied NMDA (P. Molnár and J.V.Nadler, unpublished observations; see also Chen et al., 2000).CaEDTA exhibits similarly selective actions at mossy fibersynapses in area CA3 (Vogt et al., 2000; Ruiz et al., 2004).Finally, CaEDTA does not alter the extracellular concentra-tions of Ca2+ or Mg2+ appreciably when applied at aconcentration of 1 mM. Based on these findings, we concludethat CaEDTA modified granule cell epileptiform activitythrough mechanisms related specifically to the recurrentmossy fiber pathway and that it did so by chelatingextracellular zinc. Although CaEDTA can bind a number ofheavy metal cations with high affinity, zinc is the predom-inant heavy metal contained in and released by hippocampalmossy fibers.

Our finding that CaEDTA slows the development of mossyfiber-evoked granule cell bursts suggests that released zincfacilitates the recruitment of granule cells into such bursts.The development of population bursting has been studiedmost thoroughly in area CA3 of the hippocampus. In thatregion, the synchronous discharge of a few unusuallyexcitable pyramidal cells can trigger bursting of the entirepyramidal cell population through a progressive recruitmentprocess that depends on synaptic connections among theneurons in the population (Traub et al., 1987a,b). Thedevelopment of synaptic connections among granule cellsprobably promotes a similar recruitment process in thedentate gyrus.

Further studies are required to identify the mechanismby which released zinc facilitates granule cell recruitmentand increases the magnitude of AMPA/kainate receptor-mediated mossy fiber-evoked granule cell bursts. It seemsunlikely that zinc could have acted by reducing GABA

231B R A I N R E S E A R C H 1 0 7 8 ( 2 0 0 6 ) 2 2 7 – 2 3 4

inhibition, because all experiments were performed in thepresence of bicuculline. In addition, a previous study foundno evidence for a significant action of released zinc ongranule cell GABAA receptors during population bursts(Molnár and Nadler, 2001a). Two alternative mechanismsthat might be considered are that zinc enhances AMPAchannel function and that it facilitates recurrent mossyfiber synaptic transmission by altering signal transductionmechanisms. AMPA receptors mediate fast synaptic trans-mission at recurrent mossy fiber synapses. Thus, evokedAMPA currents are critical for mossy fiber-driven granulecell bursts (Okazaki and Nadler, 2001) and presumably alsofor the progressive recruitment of granule cells into thebursts. At low micromolar concentrations, zinc enhancesthe channel function of some AMPA receptors, specificallyof homomeric GluR3 receptors (Dreixler and Leonard, 1994,1997). Flip splice variants are more sensitive than flopvariants (Shen and Yang, 1999). Receptors of this typeappear to be expressed on CA3 pyramidal cells (Lin et al.,2001) and could be present on granule cells as well. Aprevious study demonstrated that CaEDTA does not affectthe AMPA/kainate receptor-mediated component of therecurrent mossy fiber EPSC, suggesting that zinc-sensitiveAMPA receptors are not present at these synapses (Molnárand Nadler, 2001b). However, mossy fiber stimulationreleased much more glutamate and zinc under the condi-tions used in the present study. Glutamate and zinc mayhave overflowed the synapse and reached extrasynaptic(possibly zinc-sensitive) AMPA receptors in sufficient con-centration for glutamate to activate them and for zinc toamplify the evoked current. Some of the released zinc couldalso have entered the postsynaptic granule cell and/or themossy fiber bouton (Li et al., 2001a), where it can potentiallyinteract with several signal transduction mechanisms(Beyersmann and Haase, 2001) to enhance synaptic trans-mission. Zinc enters cells through various Ca2+ channels, ofwhich Ca2+-permeable AMPA/kainate receptors have thegreatest permeability (Yin et al., 2002). Permeation of zincthrough pre- and/or postsynaptic membranes may berequired for LTP at mossy fiber synapses on CA3 pyramidalcells (Li et al., 2001a). The development of mossy fiber LTPand mossy fiber-evoked granule cell bursts may thus havemechanistic similarities.

The facilitatory action of released zinc on epileptiformactivity in the synaptically reorganized dentate gyrus con-trasts with evidence that endogenous zinc elevates theseizure threshold (Cole et al., 2000; Takeda et al., 2003).Even in the dentate gyrus, some actions of zinc tend toreduce epileptiform activity. We confirmed that its NMDAreceptor antagonism can reduce the magnitude of mossyfiber-evoked granule cell bursts, at least when such burstsdepend entirely on the activation of NMDA receptors.Released zinc can also reduce glutamate release frommossy fiber terminals (Bancila et al., 2004; Quinta-Ferreiraand Matias, 2004; Takeda et al., 2004; Quinta-Ferreira andMatias, 2005). However, the facilitatory action(s) of zincovercomes these inhibitory actions to promote granule cellpopulation bursting. Thus, the net effect of endogenous zincin the reorganized dentate gyrus appears to differ from itsaction in the brain generally.

4. Experimental procedure

4.1. Pilocarpine-induced status epilepticus

Adult male Sprague–Dawley rats (175–200 g; Zivic Laborato-ries, Pittsburgh, PA, USA) were injected intraperitoneally withpilocarpine hydrochloride (350–380 mg/kg) 30 min afterpretreatment with scopolamine methyl bromide and terbuta-line hemisulfate (both 2 mg/kg, i.p.). Status epilepticus,defined as a continuous limbic motor seizure of stage 2 orhigher (Racine, 1972), was terminated 3.5 h after onset withsodium phenobarbital (50 mg/kg, i.p.) on some occasions, butwas normally allowed to self-terminate. Pilocarpine-inducedstatus epilepticus destroys neurons in many regions of thebrain, followed by recurrent mossy fiber sprouting (Okazaki etal., 1995,1999), the establishment of mossy fiber-granule cellsynapses (Wuarin and Dudek, 1996; Molnár and Nadler, 1999;Okazaki et al., 1999; Scharfman et al., 2003) and spontaneousseizures (Mello et al., 1993; Lemos and Cavalheiro, 1996).Electrophysiological studies were performed 10–20 weeksafter pilocarpine administration. All protocols were approvedin advance by the Duke University Institutional Animal Careand Use Committee.

4.2. Hippocampal slice preparation, stimulation andrecording

The brain was removed, chilled in ice-cold high-Mg2+ artificialcerebrospinal fluid (aCSF; 122 mM NaCl, 25 mM NaHCO3,3.1 mM KCl, 1.8 mM CaCl2, 12 mM MgSO4, 0.4 mM KH2PO4,10 mM D-glucose, pH 7.4) and quartered. A high concentrationof Mg2+ was used to minimize excitotoxicity. Transverse400 μM-thick slices were prepared from the caudal hippocam-pus with a vibratome. Slices corresponded to horizontal plates98–100 of Paxinos andWatson (1986). They were transferred toa net in a beaker of high-Mg2+ aCSF at room temperature andgassed continuously with 95% O2/5% CO2. Beginning 45 minlater, individual slices were transferred to a small experimen-tal chamber maintained at 35 °C, barely submerged inconventional aCSF (1.2 mM Mg2+), and superfused at 2 ml/min. A knife cut was placed across area CA3 to preventreverberating excitation in the hippocampal–entorhinocorti-cal circuit. To activate the mossy fibers, a monopolarstimulating electrode was placed in stratum lucidum of areaCA3b N100 μM from the opening of the dentate hilus. Toactivate perforant path fibers, the electrode was placed in thesubiculum. The stimulating electrode was fashioned from25 μM-diameter nichrome wire insulated to the tip withpolymerized polyvinyl resin (Formvar). To activate themossy fibers, constant-current rectangular pulses (0.1 msduration) were presented every 30 s at an intensity thatevoked an antidromic population spike ≥85% of maximalamplitude (mean intensity 452 μA; range 260–620 μA). Toactivate the perforant path, the stimulus intensity was set toevoke an EPSP of ∼3 mV peak amplitude when recordedextracellularly in the granule cell body layer (mean intensity540 μA; range 310–660 μA). An extracellular glass recordingelectrode filled with 1 M NaCl (resistance 5–7 mΩ) was placedin the granule cell body layer where the antidromic population

232 B R A I N R E S E A R C H 1 0 7 8 ( 2 0 0 6 ) 2 2 7 – 2 3 4

spike or the response to perforant path stimulation was ofgreatest amplitude. Responses were amplified with ACcoupling, filtered below 3 kHz, digitized at 10 kHz and storedto disk with use of a Digidata 1200 and PClamp8 software(Axon Instruments, Foster City, CA, USA).

After a 5 min baseline recording period, granule cellbursts were evoked by stimulating the mossy fibers orperforant path in aCSF that contained 6 mM KCl and 30 μMbicuculline (Hardison et al., 2000; Okazaki and Nadler, 2001;Tu et al., 2005). In some experiments on mossy fibers, westudied evoked bursts mediated by just NMDA or AMPA/kainate receptors. NMDA receptor-mediated evoked burstswere isolated by omitting MgSO4 from the superfusionmedium and adding 10 μM 2,3-dihydroxy-6-nitro-7-sulfa-myl-benzo(F)quinoxaline (NBQX) to block AMPA and kainatereceptors. We isolated evoked bursts mediated by AMPA/kainate receptors by adding 50 μM 2-amino-5-phosphono-pentanoate (D-AP5) to block NMDA receptors. The pathwaywas stimulated and extracellular recordings were acquiredevery 30 s for 70 min. An epileptiform response wasidentified by the appearance of multiple sharp waves onthe trace that were at least three times larger in amplitudethan the background activity and b7 ms in duration. In someexperiments, 1 mM CaEDTA was added to the superfusionmedium when the other modifications were made, that is atthe end of the 5 min baseline recording period. In otherexperiments, it was added at the end of the 70 minposttreatment recording period and recording continued foran additional 30 min.

4.3. Data analysis

The magnitude of granule cell epileptiform activity wasquantified by measurement of the coastline index (Dingledineet al., 1986), as previously described (Hardison et al., 2000;Okazaki and Nadler, 2001; Tu et al., 2005). The coastline indexallows for statistical comparisons among treatment groupsregardless of the mechanism responsible for those differ-ences. For this purpose, the data files were transferred toMicrosoft Excel and the Pythagorean theorem was used todetermine the change in response between successive datapoints. These changes were then summed for the 387 msperiod beginning at the end of the antidromic population spikein the case of mossy fiber-evoked activity and beginning at theend of the stimulus artifact in the case of perforant path-evoked activity. To account for baseline electrical noise, thecoastline index was also determined for the 108 ms periodimmediately before the stimulus. This value was extrapolatedto 387 ms and subtracted from the poststimulus coastlineindex.

The effect of CaEDTA on the time course for developmentof mossy fiber-evoked bursts was assessed initially with atwo-way ANOVA (treatment × time). The ANOVA revealedsignificant effects of both variables (P b 0.001) with nosignificant interaction between them. We then analyzed theeffects of CaEDTA on the rate at which epileptiform activitydeveloped and the greatest magnitude of that activityobserved during the recording period. The rate of developmentwas assessed from measurement of the time required for theresponse to reach half of its final magnitude. Differences

between the treatment and control groups were evaluatedwith a two-tailed Student's t test. The effect of CaEDTA addedafter evoked epileptiform activity had developedwas assessedby averaging the coastline indices for the 10 responsesrecorded 15–19.5 min after the addition of CaEDTA andcomparing that value to the average coastline index of thelast 10 responses recorded before CaEDTA was added to thesuperfusion medium. The significance of any apparentdifference was evaluated with a two-tailed paired t test.Effects of CaEDTA on the slow wave that underlay the evokedburst were determined after refiltering the traces at 75–200 Hzto remove population spikes. P values b0.05 were regarded assignificant. Grouped data are presented in the text and figuresasmeans ± SEM. The standard error of the ratio was calculatedas described by McLean and Welch (1971).

4.4. Timm histochemistry

To verify the presence of recurrent mossy fiber growth, slicesthat had been used for electrophysiological recording wereprocessed for histochemical detection of heavy metals. Theywere immersed in 0.1% (wt/vol) Na2S, 0.1 M sodium phosphatebuffer, pH 7.3 for 1.5 h followed by fixation in phosphate-buffered 10% formalin at 4 °C for 1–2 days. Slices were thenembedded in albumin-gelatin, and 30 μM-thick sections wereprepared with a vibratome. Slide-mounted sections wereprocessed as described by Danscher (1981) and lightly counter-stained with cresyl violet.

4.5. Materials

Some of the NBQX used in this study was a gift from NovoNordisk (Måløv, Denmark) and the rest was purchased fromTocris Cookson (Ballwin, MO, USA). D-AP5 was purchasedfrom Tocris Cookson and bicuculline methiodide from RBI(Natick,MA, USA). CaEDTA, phenobarbital sodium, pilocarpinehydrochloride, (−)scopolamine methyl bromide and terbulinehemisulfate were obtained from Sigma Chemical (St. Louis,MO, USA).

Acknowledgments

We thank Y. Jiao for technical assistance and K. Gorham forsecretarial help. This study was supported by NationalInstitutes of Neurological Disorders and Stroke grants NS17771 and NS 38108.

R E F E R E N C E S

Aniksztejn, L., Charton, G., Ben-Ari, Y., 1987. Selective release ofendogenous zinc from the hippocampal mossy fibers in situ.Brain Res. 404, 58–64.

Assaf, S.Y., Chung, S-H., 1984. Release of endogenous Zn2+ frombrain tissue during activity. Nature 308, 734–735.

Bancila, V., Nikonenko, I., Dunant, Y., Bloc, A., 2004. Zinc inhibitsglutamate release via activation of pre-synaptic KATPchannels and reduces ischaemic damage in rat hippocampus.J. Neurochem. 90, 1243–1250.

233B R A I N R E S E A R C H 1 0 7 8 ( 2 0 0 6 ) 2 2 7 – 2 3 4

Beyersmann, D., Haase, H., 2001. Functions of zinc in signaling,proliferation and differentiation of mammalian cells.BioMetals 14, 331–341.

Brooks-Kayal, A.R., Shumate, M.D., Jin, H., Rikhter, T.Y., Coulter,D.A., 1998. Selective changes in single cell GABAA receptorsubunit expression and function in temporal lobe epilepsy.Nat. Med. 4, 1166–1172.

Budde, T., Minta, A., White, J.A., Kay, A.R., 1997. Imaging free zincin synaptic terminals in live hippocampal slices. Neuroscience79, 347–358.

Buhl, E.H., Otis, T.S., Mody, I., 1996. Zinc-induced collapse ofaugmented inhibition by GABA in a temporal lobe epilepsymodel. Science 271, 369–373.

Chen, N., Moshaver, A., Raymond, L.A., 1997. Differentialsensitivity of recombinant N-methyl-D-aspartate receptorsubtypes to zinc inhibition. Mol. Pharmacol. 51, 1015–1023.

Chen, N., Murphy, T.H., Raymond, L.A., 2000. Competitiveinhibition of NMDA receptor-mediated currents byextracellular calcium chelators. J. Neurophysiol. 84, 693–697.

Cohen, A.S., Lin, D.D., Quirk, G.L., Coulter, D.A., 2003. Dentategranule cell GABAA receptors in epileptic hippocampus:enhanced synaptic efficacy and altered pharmacology. Eur. J.Neurosci. 17, 1607–1616.

Cole, T.B., Robbins, C.A., Wenzel, H.J., Schwartzkroin, P.A.,Palmiter, R.D., 2000. Seizures and neuronal damage in micelacking vesicular zinc. Epilepsy Res. 39, 153–169.

Collins, R.C., Tearse, R.G., Lothman, E.W., 1983. Functionalanatomy of limbic seizures: focal discharges from medialentorhinal cortex in rats. Brain Res. 280, 25–40.

Danscher, G., 1981. Histochemical demonstration of heavymetals. A revised version of the sulphide silver methodsuitable for both light and electron-microscopy.Histochemistry 71, 1–16.

Dingledine, R., Hynes, M.A., King, G.L., 1986. Involvement ofN-methyl-D-aspartate receptors in epileptiform bursting in therat hippocampal slice. J. Physiol. 380, 175–189.

Dreixler, J.C., Leonard, J.P., 1994. Subunit-specific enhancement ofglutamate receptor responses by zinc. Mol. Brain Res. 22,144–150.

Dreixler, J.C., Leonard, J.P., 1997. Effects of external calcium on zincmodulation of AMPA receptors. Brain Res. 752, 170–174.

Frederickson, C.J., Danscher, G., 1990. Zinc containing neurons inhippocampus and related CNS structures. In: Storm-Mathisen,J., Zimmer, J., Ottersen, O.P. (Eds.), Understanding the Brainthrough the Hippocampus. Prog. Brain Res., vol. 83.Amsterdam, Elsevier, pp. 71–84.

Gibbs, J.W., Shumate, M.D., Coulter, D.A., 1997. Differentialepilepsy-associated alterations in postsynaptic GABAA

receptor function in dentate granule and CA1 neurons.J. Neurophysiol. 77, 1924–1938.

Hardison, J.L., Okazaki, M.M., Nadler, J.V., 2000. Modest increase inextracellular potassium unmasks effect of recurrent mossyfiber growth. J. Neurophysiol. 84, 2380–2389.

Howell, G.A., Welch, M.G., Frederickson, C.J., 1984.Stimulation-induced uptake and release of zinc inhippocampal slices. Nature 308, 736–738.

Lee, J.-Y., Cole, T.B., Palmiter, R.D., Koh, J.-Y., 2000. Accumulationof zinc in degenerating hippocampal neurons of ZnT3-nullmice after seizures: evidence against synaptic vesicle origin.J. Neurosci. 20, RC79.

Lemos, T., Cavalheiro, E.A., 1996. Status epilepticus and the latedevelopment of spontaneous seizures in the pilocarpinemodelof epilepsy. Epilepsy Res., Suppl. 12, 137–144.

Li, Y., Hough, C.J., Frederickson, C.J., Sarvey, J.M., 2001a. Inductionof mossy fiber → CA3 long-term potentiation requirestranslocation of synaptically released Zn2+. J. Neurosci. 21,8015–8025.

Li, Y., Hough, C.J., Suh, S.W., Sarvey, J.M., Frederickson, C.J., 2001b.Rapid translocation of Zn2+ from presynaptic terminals into

postsynaptic hippocampal neurons after physiologicalstimulation. J. Neurophysiol. 86, 2597–2604.

Lin, D.D., Cohen, A.S., Coulter, D.A., 2001. Zinc-inducedaugmentation of excitatory synaptic currents and glutamatereceptor responses in hippocampal CA3 neurons.J. Neurophysiol. 85, 1185–1196.

Lopantsev, V., Wenzel, H.J., Cole, T.B., Palmiter, R.D.,Schwartzkroin, P.A., 2003. Lack of vesicular zinc in mossyfibers does not affect synaptic excitability of CA3 pyramidalcells in zinc transporter 3 knockout mice. Neuroscience 116,237–248.

Lothman, E.W., Stringer, J.L., Bertram, E.H., 1992. The dentategyrus as a control point for seizures in the hippocampus andbeyond. Epilepsy Res., Suppl. 7, 301–313.

Lu, Y.-M., Taverna, F.A., Tu, R., Ackerley, C.A., Wang, Y.-T., Roder,J., 2000. Endogenous Zn2+ is required for the induction oflong-term potentiation at rat hippocampal mossy fiber-CA3synapses. Synapse 38, 187–197.

McLean, R.A., Welch, B.L., 1971. A common error in assessing thesignificance of percentage change in neuropharmacology.J. Pharm. Pharmacol. 23, 643–645.

Mello, L.E.A.M., Cavalheiro, E.A., Tan, A.M., Kupfer, W.R., Pretorius,J.K., Babb, T.L., Finch, D.M., 1993. Circuit mechanisms ofseizures in the pilocarpine model of chronic epilepsy: cell lossand mossy fiber sprouting. Epilepsia 34, 985–995.

Molnár, P., Nadler, J.V., 1999. Mossy fiber-granule cell synapses inthe normal and epileptic rat dentate gyrus studied withminimal laser photostimulation. J. Neurophysiol. 82,1883–1894.

Molnár, P., Nadler, J.V., 2001a. Lack of effect of mossyfiber-released zinc on granule cell GABAA receptors in thepilocarpine model of epilepsy. J. Neurophysiol. 85, 1932–1940.

Molnár, P., Nadler, J.V., 2001b. Synaptically-released zinc inhibitsN-methyl-D-aspartate receptor activation at recurrent mossyfiber synapses. Brain Res. 910, 205–207.

Nadler, J.V., 2003. The recurrent mossy fiber pathway of theepileptic brain. Neurochem. Res. 28, 1649–1658.

Okazaki, M.M., Nadler, J.V., 2001. Glutamate receptor involvementin dentate granule cell epileptiform activity evoked by mossyfiber stimulation. Brain Res. 15, 58–69.

Okazaki, M.M., Evenson, D.A., Nadler, J.V., 1995. Hippocampalmossy fiber sprouting and synapse formation after statusepilepticus in rats: visualization after retrograde transport ofbiocytin. J. Comp. Neurol. 352, 515–534.

Okazaki, M.M., Molnár, P., Nadler, J.V., 1999. Recurrent mossy fiberpathway in rat dentate gyrus: synaptic currents evoked inpresence and absence of seizure-induced growth.J. Neurophysiol. 81, 1645–1660.

Paoletti, P., Ascher, P., Neyton, J., 1997. High-affinity zinc inhibitionof NMDA NR1–NR2A receptors. J. Neurosci. 17, 5711–5725.

Patrylo, P.R., Dudek, F.E., 1998. Physiological unmasking of newglutamatergic pathways in the dentate gyrus of hippocampalslices from kainate-induced epileptic rats. J. Neurophysiol. 79,418–429.

Paxinos, G., Watson, C., 1986. The Rat Brain in StereotaxicCoordinates, 2nd edition. Academic, Orlando.

Perez-Clausell, J., Danscher, G., 1985. Intravesicular localization ofzinc in rat telencephalic boutons: a histochemical study. BrainRes. 337, 91–98.

Peters, S., Koh, J., Choi, D.W., 1987. Zinc selectively blocks theaction of N-methyl-D-aspartate on cortical neurons. Science236, 589–593.

Quinta-Ferreira, M.E., Matias, C.M., 2004. Hippocampalmossy fibercalcium transients are maintained during long-termpotentiation and are inhibited by endogenous zinc. Brain Res.1004, 52–60.

Quinta-Ferreira, M.E., Matias, C.M., 2005. Tetanically released zincinhibits hippocampal mossy fiber calcium, zinc and synapticresponses. Brain Res. 1047, 1–9.

234 B R A I N R E S E A R C H 1 0 7 8 ( 2 0 0 6 ) 2 2 7 – 2 3 4

Racine, R.J., 1972. Modification of seizure activity by electricalstimulation: II. Motor seizure. Electroencephalogr. Clin.Neurophysiol. 32, 281–284.

Reece, L.J., Dhanjal, S.S., Chung, S.-H., 1994. Zinc induceshyperexcitability in the hippocampus. NeuroReport 5,2669–2672.

Ruiz, A., Walker, M.C., Fabian-Fine, R., Kullmann, D.M., 2004.Endogenous zinc inhibits GABAA receptors in a hippocampalpathway. J. Neurophysiol. 91, 1091–1096.

Scharfman, H.E., Sollas, A.L., Berger, R.E., Goodman, J.H., 2003.Electrophysiological evidence of monosynaptic excitatorytransmission between granule cells after seizure-inducedmossy fiber sprouting. J. Neurophysiol. 90, 2536–2547.

Shen, Y., Yang, X.-L., 1999. Zinc modulation of AMPA receptorsmay be relevant to splice variants in carp retina. Neurosci. Lett.259, 177–180.

Stringer, J.L., Williamson, J.M., Lothman, E.W., 1989. Induction ofparoxysmal discharges in the dentate gyrus: frequencydependence and relationship to afterdischarge production.J. Neurophysiol. 62, 126–135.

Tauck, D.L., Nadler, J.V., 1985. Evidence of functional mossy fibersprouting in the hippocampal formation of kainic acid-treatedrats. J. Neurosci. 5, 1016–1022.

Takeda, A., Hirate, M., Tamano, H., Nisibaba, D., Oku, N., 2003.Susceptibility to kainate-induced seizures under dietary zincdeficiency. J. Neurochem. 85, 1575–1580.

Takeda, A., Minami, A., Seki, Y., Oku, N., 2004. Differential effectsof zinc on glutamatergic and GABAergic neurotransmittersystems in the hippocampus. J. Neurosci. Res. 75, 225–229.

Traub, R.D., Miles, R., Wong, R.K.S., 1987a. Models of synchronizedhippocampal bursts in the presence of inhibition: I. Singlepopulation events. J. Neurophysiol. 58, 739–751.

Traub, R.D., Miles, R., Wong, R.K.S., Schulman, L.S., Schneiderman,J.H., 1987b. Models of synchronized hippocampal bursts in thepresence of inhibition: II. Ongoing spontaneous populationevents. J. Neurophysiol. 58, 752–764.

Tu, B., Timofeeva, O., Jiao, Y., Nadler, J.V., 2005. Spontaneousrelease of neuropeptide Y tonically inhibits recurrent mossyfiber synaptic transmission in epileptic brain. J. Neurosci. 25,1718–1729.

Ueno, S., Tsukamoto, M., Hirano, T., Kikuchi, K., Yamada, M.K.,Nishiyama, N., Nagano, T., Matsuki, N., Ikegaya, Y., 2002. Mossyfiber Zn2+ spillover modulates heterosynapticN-methyl-D-aspartate receptor activity in hippocampal CA3circuits. J. Cell Biol. 158, 215–220.

Vogt, K., Mellor, J., Tong, G., Nicoll, R., 2000. The action ofsynaptically released zinc at hippocampal mossy fibersynapses. Neuron 26, 187–196.

Westbrook, G.L., Mayer, M.L., 1987. Micromolar concentrations ofZn2+ antagonize NMDA and GABA responses of hippocampalneurons. Nature 328, 640–643.

Wuarin, J.-P., Dudek, F.E., 1996. Electrographic seizures and newrecurrent excitatory circuits in the dentate gyrus ofhippocampal slices from kainate-treated epileptic rat.J. Neurosci. 16, 4438–4448.

Xu, H., Mitchell, C.L., 1993. Chelation of zinc bydiethyldithiocarbamate facilitates bursting induced by mixedantidromic plus orthodromic activation of mossy fibers inhippocampal slices. Brain Res. 624, 162–170.

Yin, H.Z., Sensi, S.L., Ogoshi, F., Weiss, J.H., 2002. Blockade ofCa2+-permeable AMPA/kainate channels decreasesoxygen-glucose deprivation-induced Zn2+ accumulation andneuronal loss in hippocampal pyramidal neurons. J. Neurosci.22, 1273–1279.