Proc. Natl. Acad. Sci. USAVol. 93, pp. 11962-11967, October 1996Neurobiology

Role of cortical N-methyl-D-aspartate receptors in auditorysensory memory and mismatch negativity generation:Implications for schizophrenia

(phencyclidine/monkey/intracortical/cognitive/event-related potential)

DANIEL C. JAVITT*t, MITCHELL STEINSCHNEIDERtt, CHARLES E. SCHROEDERtt, AND JOSEPH C. AREZZOttDepartments of *Psychiatry, tNeuroscience, and tNeurology, Jack and Pearl Resnick Campus, 1300 Morris Park Avenue, Albert Einstein College of Medicine,Bronx, NY 10461

Communicated by Dominick P. Purpura, Albert Einstein College of Medicine, Bronx, NY, July 10, 1996 (received for review March 28, 1996)

ABSTRACT Working memory refers to the ability of thebrain to store and manipulate information over brief timeperiods, ranging from seconds to minutes. As opposed tolong-term memory, which is critically dependent upon hip-pocampal processing, critical substrates for working memoryare distributed in a modality-specific fashion throughoutcortex. N-methyl-D-aspartate (NMDA) receptors play a cru-cial role in the initiation oflong-term memory. Neurochemicalmechanisms underlying the transient memory storage re-quired for working memory, however, remain obscure. Audi-tory sensory memory, which refers to the ability of the brainto retain transient representations of the physical features(e.g., pitch) of simple auditory stimuli for periods of up toapproximately 30 sec, represents one of the simplest compo-nents of the brain working memory system. Functioning of theauditory sensory memory system is indexed by the generationof a well-defined event-related potential, termed mismatchnegativity (MMN). MMN can thus be used as an objectiveindex of auditory sensory memory functioning and a probe forinvestigating underlying neurochemical mechanisms. Mon-keys generate cortical activity in response to deviant stimulithat closely resembles human MMN. This study uses a com-bination of intracortical recording and pharmacological mi-cromanipulations in awake monkeys to demonstrate that bothcompetitive and noncompetitive NMDA antagonists block thegeneration ofMMN without affecting prior obligatory activityin primary auditory cortex. These findings suggest that, on aneurophysiological level, MMN represents selective currentflow through open, unblocked NMDA channels. Furthermore,they suggest a crucial role of cortical NMDA receptors in theassessment of stimulus familiarity/unfamiliarity, which is akey process underlying working memory performance.

Working memory refers to the ability of the brain to store andmanipulate information over brief time periods, ranging fromseconds to minutes. As compared with long-term memory,which is critically dependent upon hippocampal long-termpotentiation, critical substrates for working memory are dis-tributed in a modality specific fashion throughout cortex (1, 2).On a neurophysiological level, working memory has beenlinked to transient, task-related alterations in the firing ratesof neurons in modality-specific brain regions, permitting theuse of intracortical recordings to investigate neurochemicalmechanisms underlying functioning of the brain working mem-ory system (3-5).

Cortical information processing is critically dependent uponthe interplay between glutamatergic and y-aminobutyric acid(GABA)ergic neurotransmission. Glutamate is the primaryexcitatory amino transmitter in mammalian cortex, being

present in approximately 60% of cortical neurons and 100% ofcortical pyramidal neurons. Glutamatergic fibers mediate allthalamocortical and corticortical projections within cortex, aswell as corticofugal projections from cortex to subcorticalstructures. Within cortex, the interplay of excitatory neuro-transmission is regulated by local circuit GABAergic inhibitoryneurons that define the borders and functioning of individualcortical columns. Glutamate mediates its actions at distinctreceptor types that are differentiated based upon their sensi-tivity to the synthetic glutamate derivative N-methyl-D-aspartate (NMDA). NMDA receptor activation plays a criticalrole in learning and the initiation of long-term memoryformation (reviewed in ref. 6). This study investigates thehypothesis that NMDA receptors play a crucial role in corticalworking memory as well.

This study focuses on auditory sensory ("echoic") memory,a preattentive component of the brain working memory sys-tem, as a model for more complex attention-dependent formsof working memory. Echoic memory refers to the ability of thebrain to store representations of the physical characteristics ofsimple auditory stimuli for seconds to minutes even when suchstimuli are ignored at the time of presentation (7). Functioningof the echoic memory system is indexed by the generation ofa specific auditory event-related potential component, termedmismatch negativity (MMN), which can, therefore, be used asa probe of the auditory sensory memory system (8, 9). MMNis elicited most frequently in an auditory oddball paradigm inwhich a sequence of repetitive standard stimuli is interruptedinfrequently by a physically deviant "oddball" stimulus. Crit-ical substrates for echoic memory and MMN generation residein the region of primary auditory cortex (Al) (9-12), andintracortical recordings from Al show prominent contribu-tions to surface-recorded MMN in both cats (13) and monkeys(14). In humans, MMN to pitch and intensity deviants occurswith an approximate peak latency of 150 msec (9). In monkeys,comparable activity occurs with an approximate peak latencyof 90 msec (15), reflecting the smaller brain sizes and corre-spondingly shorter conduction latencies.To investigate neurochemical mechanisms underlying MMN

generation in awake monkeys, intracortical recordings throughAI were combined with focal infusion of NMDA and non-NMDA antagonists into the microregion of the recording elec-trode. Recordings were obtained using linear array, multichannelelectrodes acutely inserted into the cortical region of interest. Thisstudy shows that NMDA antagonists abolish local generation ofMMN within Al without altering the generation of prior oblig-atory event-related potential components, indicating thatNMDAreceptors play a crucial role in the neurophysiological operationsunderlying cortical working memory.

Abbreviations: NMDA, N-methyl-D-aspartate; GABA, y-aminobu-tyric acid; ERP, event-related potential; MMN, mismatch negativity;Al, primary auditory cortex; PCP, phencyclidine; CSD, current sourcedensity.

11962

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Oct

ober

30,

202

0

Proc. Natl. Acad. Sci. USA 93 (1996) 11963

MATERIALS AND METHODS

Subjects consisted of five cynomolgus monkeys (Macaca fas-cicularis), weighing between 2.5 and 4.0 kg. Monkeys wereprepared for intracortical studies using previously describedsurgical procedures (14). Briefly, under general anesthesia(sodium pentobarbital) and using aseptic techniques, portionsof the skull were resected and matrices of adjacently placedepidural guide tubes were implanted, providing access tounderlying brain structures. The matrices were oriented ster-eotaxically to target Al and were embedded in a mound ofdental acrylic that was affixed to the skull using machinescrews. Plexiglas bars were included in the implant to permitpainless head fixation during recording sessions. Animals werehoused in an American Association for the Accreditation ofLaboratory Animal Care accredited animal facility, were

monitored daily for assessment of their well-being, and were

allowed to recover from surgery prior to initiation of experi-ments. Also, prior to experimentation, monkeys were accus-

tomed to handling and restraint in customized, individuallyfitted primate chairs. On the day of experiment, monkeys were

placed in a recording chamber and electrodes were insertedacutely into Al through the epidural guide tubes. Monkeyswere not given any systemic agents prior to recordings, andremained awake and alert throughout. Monkeys were fed withpreferred foods (e.g., apples) at regular intervals during therecording sessions.For each experiment, a 14-16 channel linear array multi-

electrode with 200 ,um of intercontact spacing and an attached30-gauge injection catheter was inserted acutely by microdriveinto Al. Electrodes were centered about the point of inversionof the initial intracortical negativity (N8), which has beenshown to correspond to the base of lamina IV (16). Subsequentlamina assignments were made relative to this position. Elec-trode location was verified histologically for each penetrationfollowing the sacrifice of each monkey.

Recordings were made using either intensity deviant or pitchdeviant sequences. For intensity deviants, infrequently occur-

ring soft [60-65 dB sound pressure level (SPL)] clicks were

interspersed among more frequent loud (80-85 dB) clicks. Forpitch deviants, infrequent 150-msec duration tones, chosen tocorrespond to the best frequency of the area of penetration,were presented against a background of more frequent tonesof either higher or lower pitch. Stimuli were presented atinterstimulus intervals of 0.6-1 sec; deviant stimuli occurredevery 3-5 sec (5-15% sequential probability). Differenceprofiles were constructed by subtracting responses to a stim-ulus presented as a repetitive standard from responses to thesame stimulus presented as a deviant in a subsequent "odd-ball" condition. In the oddball condition, stimuli were pre-sented infrequently against a background of more frequentstimuli of different intensity or pitch. Response profiles to25-100 separate standard or deviant stimuli were averagedon-line. Responses greater than 4+325 ,uV were automaticallyrejected. Laminar response profiles were then subjected toone-dimensional current source density (CSD) analyses todifferentiate locally generated from volume-conducted event-related potential components. Summed, rectified CSD (SUM-REC) waveforms were constructed by combining absoluteCSD amplitudes across contacts within either lamina IV or

supragranular cortex.Within each experiment, the response to standard and

deviant stimuli was characterized prior to pharmacologicalmanipulation. Specific agent was infused through a canullaattached to the recording electrode. For within-run statistics,responses were combined in bins, each reflecting the averageof 25-50 successive presentations. For between-run statistics,mean percentage change scores were combined across runs.

Separate repeated-measure ANOVAs were conducted to an-

alyze the effects of NMDA antagonists on responses to

standard stimuli, and on responses to stimulus deviance asreflected in the amplitude of the deviant minus standarddifference waveform.

RESULTSEffects ofNMDA blockers and control agents on intracorticalMMN generation were determined in the course of 14 histo-logically verified penetrations through Al involving 5 separatemonkeys. During each penetration, data were sampled from 15separate locations across the depth of cortex. An additional10-20 locations were sampled during each run to position theelectrode and ensure adequate sampling of active tissue.

Intracortical profiles elicited by standard and deviant clicks(Fig. 1) consisted of an initial brief activation within lamina IVduring the 5-15 msec latency range, followed by early and latesupragranular activation during the 15-60 and 60-150 msecranges. The early supragranular activation occurred primarilywithin lower supragranular laminae and was associated withsuperficial current return, whereas the later supragranular sinkoccurred within higher laminae and was associated with deepreturn. Deviant stimuli elicited a similar degree of initiallamina IV activation as did standards, but elicited a markedlyincreased degree of supragranular activation (Fig. 1 Middle).Intracortical MMN was defined as the difference waveformobtained by subtracting intracortical responses to standardstimuli from responses to deviants (Fig. 1 Bottom).

Statistical comparisons were performed using SUMRECwaveforms, which provide a temporal index of total current flowwithin identified laminae (Fig. 2). Separate SUMREC wave-forms were constructed from CSD profiles to standard stimuli,deviant stimuli, and deviant minus standard difference wave-forms (Figs. 2 and 3). MMN amplitude was defined as theintegrated area under the deviant minus standard supragranularSUMREC waveform during the 15-200 msec latency range (Fig.3 Inset). Pharmacological effects of PCP, MK-801, and CGS-19755 were determined by comparing the mean amplitudes of thesupragranular SUMREC waveform during the 15-200 mseclatency range prior to and after focal microinfusions of each agentthrough a canulla attached to the recording electrode.

Focal microinjection ofup to 10 ,ug of PCP did not significantlyalter the intracortical response profile to standard stimuli (Fig. 1Top). These drugs also did not alter the initial slope of theintracortical SUMREC waveform, which provides an index ofinitial stimulus-induced cortical excitation (14) (Fig. 2). In con-trast, PCP significantly reduced the memory-dependent enhance-ment in supragranular activity elicited by deviant stimuli (Fig. 1Middle), as reflected in a dose-dependent reduction in thedifferential activity obtained by subtracting the profile elicited bystandard stimuli from the profile elicited by deviants (Fig. 1Bottom). SUMREC analysis of the difference protocol revealeda significant reduction of deviance-related enhancement of su-pragranular activity to approximately 50% of control levelsfollowing injection of 10 ,ug of PCP (F1,2 = 23.0, P < 0.05) acrossstudies (Fig. 3). Similar results were obtained with MK-801, whichled to a 32 ± 7% decrease in intracortical MMN amplitudefollowing injection of an intracortical dose of 0.5 ,ug (F1,3 = 18.2,P < 0.025) and CGS-19755, which led to a significant (F3,8 = 9.37,P < 0.015) dose-dependent 49 ± 6% decrease, following doses of20-100 ,ug (Figs. 4 and 5). No significant effect of these agents wasobserved on initial cortical activation to standard or deviantstimuli, as reflected by integrated SUMREC amplitude withinlamina IV during the 0-15-msec latency range, or on supragranu-lar response to standard stimuli, as reflected by integratedsupragranular SUMREC activity during the 15-200-msec range.Two comparison agents were employed to assess the spec-

ificity of the NMDA effect: (i) kynurenic acid, which blocksnon-NMDA as well as NMDA-type glutamate receptors and(ii) bicuculline, which blocks GABAA receptors. Injection ofkynurenic acid significantly blocked local response generationto standard stimuli within supragranular laminae (F3,11 = 33.7,

Neurobiology: Javitt et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

30,

202

0

11964 Neurobiology: Javitt et al.

STANDARDPre-PCP

X IVlv

iI I I I I , I AI I I0 120 240

DEVIANT

II/

III

EcIVV

-v

0 120 240

DIFFERENCE(x2)

II/

- IVIII-M IV -- ---- v

0 120 240JIlL!0 120 240

Proc. Natl. Acad. Sci. USA 93 (1996)

10 [lg PCP

_ _

-_0I I I I 2 I -I

0 120 240

0 120 240

I I I I I I I I I0 120 24()

TIME (msec)

P < 0.0001) and showed a tendency to induce such inhibitionwithin lamina IV (F3,11 = 2.71, P < 0.1). Bicuculline had anaction opposite to that of kynurenic acid in that it enhanced alllocally generated activity within both lamina IV and supra-granular laminae (Fig. 6). The increase was most pronouncedwithin supragranular laminae, and was reversed by systemicinjection of MK-801, as has been reported previously for visualresponses within Vl (17).

DISCUSSIONNMDA-type glutamate receptors are unique among channelforming receptors in that current flow through NMDA chan-nels is both ligand- and voltage-dependent. In all other iden-tified channel-forming receptors (e.g., nicotinic, GABAA,a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid/kainate), binding of neurotransmitter leads obligatorily tocurrent flow through the open receptor channel. In contrast,NMDA channels are blocked in a voltage-dependent fashionby physiological concentrations of Mg2+. As a consequence ofthe Mg2+ blockade, presynaptic glutamate release does notlead to current flow through postsynaptic NMDA channelsunless the postsynaptic membrane is significantly depolarizedfrom resting levels. The postsynaptic depolarization may occureither as a consequence of decreased recurrent GABAergicinhibition or independent depolarization via modulatory neu-

Lt I IJ I I I J. iL..lL..0 120 240

V

FIG. 1. Effect of the noncompetitiveNMDA antagonist phencyclidine (PCP)

liiiIl111l IIllIon CSD profiles from Al. Profiles were0 120 240 obtained to soft (65 dB SPL) clicks pre-

sented either as standards (100% sequen-tial probability, Top) or as deviants (15%probability, Middle) against a backgroundof more frequent loud (85 dB) clicks in an

~------------- "oddball" paradigm. PCP (5 ,Lg/I,u) was-~--------~------~.-----------*-- infused into the local region of cortex in 1

,ul increments via a canulla attached to therecording electrode. Each infusion oc-curred over 2-3 min. (Bottom) DifferenceCSD profiles obtained by subtracting the

_______ CSD profile to soft clicks presented asstandards from the CSD profile to the

I I1!I1J III I I same stimulipresented as deviants prior to0 120 240 and following PCP injection. Calibration

bar corresponds to 1 mV/mm2 for stan-source dard and deviant profiles, 0.5 mV/mm2Slink | for difference profiles.

rotransmitters or non-NMDA glutamate receptors. Theunique dual voltage- and ligand-sensitivity of NMDA recep-tors permits them to function in a "Hebbian" fashion tointegrate information from multiple, independent input path-ways (reviewed in ref. 6).NMDA receptors have been most extensively studied in

relation to hippocampal long-term potentiation, which is acritical step in the initiation of long-term memory formation.Neocortex contains NMDA-receptor densities similar to thoseobserved in hippocampus. However, the role played by corticalNMDA receptors in normal adult brain functioning andinformation processing is relatively unknown (18). This studyindicates that NMDA receptors may play a limited role in theprocessing of repetitive stimuli, but that such receptors arecrucial for the~processing of the relationship between stimuliin a working memory task. The ability of bicuculline to induceMK-801-reversible epileptiform activity in the same corticallaminae that generate intracortical MMN indicates thatNMDA receptors within these laminae are maintained undertonic inhibition by GABAergic interneurons. When GABAAreceptors are blocked, therefore, NMDA receptors are re-leased from inhibition giving rise to large-amplitude, spike-likedischarges (17).The findings of this study suggest that, like other compo-

nents of working memory (19), auditory sensory memory may

lw

lll---W-------...

I

Dow

nloa

ded

by g

uest

on

Oct

ober

30,

202

0

Proc. Natl. Acad. Sci. USA 93 (1996) 11965

SUMRECPCP (5 ug) PCP (1 0 ug)

IV

0 1 20 240 0 120 241TIME (msec

depend upon synergism between excitatory and inhibitoryprocesses in cortex. Furthermore, a particular model thatcould account for the effects of NMDA and GABAA antag-onists on memory-related ensemble response within auditorycortex is as follows: First, repetitively presented stimuli of onepitch or intensity may increase the level of tonic inhibition ofsupragranular neurons that are sensitive to that pitch or

intensity while simultaneously decreasing the level of inhibi-tion of neurons sensitive to other pitches or intensity levels.Deviant stimuli, presented following a sequence of repetitivestandards, would then activate neurons that are already par-tially disinhibited and on which NMDA channels are alreadypartially unblocked. Activation of such neurons, therefore,would recruit greater levels of postsynaptic current flowthrough NMDA channels than if those same stimuli were

presented without preceding standards. MMN, which is de-fined as the differential response to deviant stimuli comparedwith standards, would thus represent the additional currentflow mediated by open, unblocked NMDA channels. Thememory trace underlying MMN generation appears to persistfor approximately 10-20 sec following stimulus presentation(9). Given that GABAA-mediated responses persist only on theorder of milliseconds, this study does not address the mecha-nisms by which alterations in tonic inhibition levels may bemaintained over periods of tens of seconds. The present studydoes, however, suggest that expression of working memory, asreflected by experience-dependent alterations in neuronalfiring patterns, depends upon the functional interaction be-tween excitatory and inhibitory processes within cortex medi-ated by NMDA and GABAA receptors, respectively.

It has been reported that MMN generation is diminished insubjects with dorsolateral prefrontal lesions (20), suggestingthat the interplay between auditory and prefrontal corticesmay be important for MMN generation. However, in thosestudies, the decrease in MMN amplitude was associated withan accompanying increase in the amplitude of P1, an earlycortical obligatory component. Moreover, the MMN deficitwas greatest when stimuli were presented ipsilateral to thelesioned hemisphere, indicating that the frontal lobe may beprimarily involved in the gating of information entering theauditory cortex, rather than processing information within theauditory cortex. In that study also, the lesion-related deficit inMMN generation was accompanied by decreased responseaccuracy, supporting the concept that MMN indexes a preat-tentive process that contributes to subsequent attention-dependent mnemonic performance.

J FIG. 2. Effect of PCP on_ SUMREC CSD profiles ob-

tained from supragranularlaminae (II/III) and lamina

~~0 120 240 ~~IV of AIlto deviant (DEV)0 0 1 20 240 and standard (STD) stimuliprior to and following focalPCP infusion.

The finding that NMDA receptors play a crucial role inworking memory is consistent with behavioral studies ofworking memory in rats and monkeys. In monkeys, NMDAantagonists from a variety of chemical classes impair the abilityto perform complex responses chains, whereas a variety ofcompounds that interact with dopaminergic, serotonergic, andopiate neurotransmitter systems are ineffective (21, 22). Im-paired intracortical MMN generation induced by NMDAantagonists may be relevant to the pathophysiology of disor-ders associated with working memory impairments, such as

schizophrenia (23). Schizophrenic subjects show significantdeficits in MMN generation (24-27), along with impairedperformance of auditory sensory memory tasks (28). It hasbeen suggested that impairments in NMDA receptor activity(29) may contribute significantly to the pathophysiology ofschizophrenia. Furthermore, it has been shown that glycine, anNMDA augmenting agent, may ameliorate symptoms ofschizophrenia when given at doses sufficient to increase brain

C. 20 ~

LU0Pre-PCP 5,.gg 1 Ogg

PRE-PCP

PCP (10gg)

0 120 240TIME (msec)

FIG. 3. Effect of PCP on supergranular SUMREC waveformsobtained from CSD difference profiles prior to and after focal PCPinfusion. Calibration bar corresponds to 0.1 mV/mm2. (Inset) Summedarea under the SUMREC curve during the 151-200-msec latency range,expressed in msec*mV/mm2.

11/111

cuc._

ct

Neurobiology: Javitt et al.

h',

Dow

nloa

ded

by g

uest

on

Oct

ober

30,

202

0

Proc. Natl. Acad. Sci. USA 93 (1996)

STANDARDPre-CGS 20 tg 60 tg 1 00 plg

11c.E

IV411 LLLLLLI10 120 240

DEVIANT

LLX11I" -

Iv.,LILI ILLLLL.I II0 120 240

DIFFERENCE(X2)

IV

.11 T I II ,.A

0 120 240

:o; 1 2L 240.. 1-1:10 1 20 240

-f>

0LLL I 24I0I0 120 240

0 120 240

-AII-L[IJ I I II0 120 240

.-B/

. I I 1 1111-11.1Ad!0 120 240

. 120IIIIIII III0 120 2

I,dI!,IrL,.H J.i.l,Ll,,0 120 240

0Lu1.L 12 I24I0III0 120 240

L W I ll "I I I III240 0 120

sourcesink

FIG. 4. Effect of the competitiveNMDA antagonist CGS-19755 on intra-cortical CSD profiles to 9 kHz, 150 msec

r"W', tones presented either as standards (100%_j1-4 sequential probability) or as deviants

(16.7% sequential probability) against abackground of more frequent 7 kHz tonesin an "oddball" paradigm. Interstimulusinterval (ISI) of 9 kHz tones in the stan-

240 dard runs (4.8 sec) was matched to ISI ofdeviants in the oddball run. CGS-19755

1 (10 ,tg/,ll) was infused at increments of1-2 ,ul over 2-3 min each.

glycine levels (30). This study, demonstrating schizophrenia-like inhibition of MMN generation by NMDA antagonists,provides important support to the PCP/NMDA model ofschizophrenia.As we have reported (14), the intracortical events underlying

MMN generation begin earlier than is apparent in surfacerecordings (15). The discrepancy is due to the fact that theearlier (15-60 msec) events occur primarily within deep cor-tical laminae. Furthermore, the pattern of current return isvariable across sites, so that activity cancels across adjacentcortical columns and is not apparent at surface recording sites.Finally, the most common pattern observed during the 15-60-msec range consists of a deep sink and superficial source.To the extent that this activity does propogate to the surface,it would be expected to correspond to a surface positivity,rather than a surface negativity. Small positivities precedingMMN are frequently observed in human recordings. Intracor-tical activity during the latency range of surface MMN (60-150msec in monkeys) consists of a superficial current sink anddeeper source. This configuration is constant across recordingsites, and would be expected to produce the observed surfacenegativity. An important finding of this paper is that NMDAantagonists block both the early and late intracortical re-sponses to stimulus deviance, indicating that both may bemanifestations of the same underlying process.

A limitation of this study is that it does not address thedegree to which modulatory neurotransmitters, such as dopa-mine, serotonin, norepinephrine, acetylcholine or neuropep-tides, might affect memory-dependent functioning within au-ditory cortex. Injection of Dl antagonists impairs prefrontallymediated visuospatial working memory and related alterationsin neuronal firing rate, raising the possibility that dopaminereceptors may play a significant role in the functioning of thecortical working memory system (31, 32). However, MMNgeneration in schizophrenic subjects appears to be relativelyunaffected by antipsychotic agents, which work by blockingdopamine (D2) receptors (25). Similarly, MMN generation inchildren with attention-deficit disorder appears to be relativelyunaffected by methylphenidate, a dopamine-releasing agent,

0la

E

(5m

m

100

80

60

400 FIG. 5. Dose response of

20 40 60 80 100 CGS-19755-induced inhibition0 20 40 60 80 100 of supragranular SUMREC ac-CGS-1 9755 dose (pg) tivity across runs (n = 4).

TIME (msec)

11966 Neurobiology: Javitt et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

30,

202

0

Proc. Natl. Acad. Sci. USA 93 (1996) 11967

~ -

-Y\--ll2ll2l4l0 lll2

0 120 240 0 120 240

sourceTIME (msec) sink

FIG. 6. Effect of the GABAA antagonist bicuculline on intracor-tical responses to repetitive 0.1-msec clicks. Profiles were obtained bypreinfusion (A), following infusion of cumulative doses of 0.2 nmol(0.07 ,ug) each (B and C), and following subsequent subcutaneousadministration of 0.1 mg/kg MK-801 (D). Each profile represents themean of 2-4 independent averages of 75 sweeps each. Treatmenteffects on supragranular activation during the 15-200-msec range werehighly significant (F3,12 = 25.2, P < 0.0001), with Student-Newman-Keals post-hoc tests showing significant (P < 0.05) augmentation ofactivity by high-dose bicuculline relative to both control and low-doseconditions, and inhibition of this increase by peripherally administeredMK-801. There was a less significant effect of treatment on initiallamina IV response (F3,12 = 4.1, P < 0.05) with no significantdifferences detectable by post-hoc examination. Calibration represents1 mV/mm2. This experiment represents one of two similar studies.

indicating that the dopaminergic system probably does not playa major role in echoic memory (33). AI receives relativelysparse dopaminergic input, compared with the frontal regionsthat mediate visuospatial memory (34). However, Al doesreceive prominent serotonergic and noradrenergic innervation(35). A goal of future studies will be to examine the relativeroles of these systems in cortical working memory perfor-mance. A second limitation of this study is that it targets a

preattentive working memory component using untrainedmonkeys. Desimone (36) has suggested a division of workingmemory into four prototypic processes. Echoic memory isprototypic of "adaptive filter" type working memory in whichpresentation of a stimulus of one type alters the response to arelated but different stimulus. Future studies will address thedegree to which NMDA receptors participate in other types ofworking memory and controlled cortical processing.

Despite the above limitations, this study has several signif-icant findings. First, it demonstrates the feasibility of usingintracortical ensemble recording, combined with focal micro-infusion, for the study of neurochemical mechanisms under-lying cognitive processing. Use of such a combination oftechniques has only recently been accomplished in single-cellrecordings of monkey cognitive functioning (32). Second, itdemonstrates the utility of CSD approach for detecting dif-ferential drug affects on activity deriving from distinct corticallaminae. Finally, it demonstrates that the relative contributions

of different receptors to cortical processing can be differen-tiated in awake, behaving monkeys and that the pattern ofauditory dysfunction observed in schizophrenia can be repro-duced by microinfusion of NMDA receptor antagonists intoAl. A major limitation in schizophrenia research is the lack ofappropriate animal models, especially for the cognitive dys-function domain of symptoms. The present findings suggestthat analysis of intracortical mechanisms underlying MMNgeneration may provide important insights into the pathophys-iology of cognitive dysfunction in schizophrenia.We acknowledge the technical assistance of Shirley Seto and Mona

Litwak. The research was supported by U.S. Public Health ServiceGrants R03 MH46962 to D.C.J. and R01 MH06723 to J.C.A., and bygrants from the Scottish Rite Research Foundation and Zara andBernad Jakubovitz Fund for Brain Research to D.C.J.

1. Goldman-Rakic, P. S. (1988) Annu. Rev. Neurosci. 11, 137-156.2. Wilson, F. A. W., O'Scalaidhe, S. P. & Goldman-Rakic, P. S. (1993) Science

260, 1955-1958.3. Funahashi, S., Bruce, C. J. & Goldman-Rakic, P. S. (1989) J. Neurophysiol.

61, 331-349.4. Miller, E. K., Li, L. & Desimone, R. (1993) J. Neurosci. 13, 1460-1478.5. Goldman-Rakic, P. S. (1995) Neuron 14, 477-485.6. Cotman, C. W., Monaghan, D. T. & Ganong, A. H. (1988) Annu. Rev.

Neurosci. 11, 61-80.7. Cowan, N. (1988) Psychol. Bull. 104, 163-191.8. Cowan, N., Winkler, I., Teder, W. & Naatanen, R. (1993) J. Exp. Psychol.

Learning Memory Cognit. 19, 909-921.9. Naatanen, R. (1990) Behav. Brain Sci. 13, 201-288.

10. Scherg, M., Vajsar, J. & Picton, T. W. (1989)J. Cognit. Neurosci. 1, 336-355.11. Hari, R., Hamalainen, M., Ilmoniemi, R., Kaukoranta, E., Reinikainen, K.,

Salminen, J., Alho, K., Naatanen, R. & Sams, M. (1984) Neurosci. Lett. 50,127-132.

12. Ritter, W., Deacon, D., Gomes, H., Javitt, D. C. & Vaughan, H. G., Jr.(1995) Ear Hear. 16, 51-66.

13. Csepe, V., Karmos, G. & Molnar, M. (1987) Electroencephalogr. Clin.Neurophysiol. 66, 571-578.

14. Javitt, D. C., Steinschneider, M., Schroeder, C. E., Vaughan, H. G., Jr., &Arezzo, J. C. (1994) Brain Res. 667, 192-200.

15. Javitt, D. C., Schroeder, C. E., Steinschneider, M., Arezzo, J. C. &Vaughan, H. G., Jr. (1992) Electroencephalogr. Clin. Neurophysiol. 83,87-90.

16. Steinschneider, M., Tenke, C. E., Schroeder, C. E., Javitt, D. C., Simpson,G. V., Arezzo, J. C. & Vaughan, H. G., Jr. (1992) Electroencephalogr. Clin.Neurophysiol. 84, 196-200.

17. Schroeder, C. E., Tenke, C. E., Givre, S. J., Arezzo, J. C. & Vaughan, H. G.,Jr. (1990) Brain Res. 515, 326-330.

18. Daw, N. W., Stein, P. S. & Fox, K. (1993)Annu. Rev. Neurosci. 16,207-222.19. Wilson, F. A., O'Scalaidhe, S. P. & Goldman-Rakic, P. S. (1994) Proc. Natl.

Acad. Sci. USA 91, 4009-4013.20. Alho, K., Woods, D. L., Algazi, A., Knight, R. T. & Naatanen, R. (1994)

Electroencephalogr. Clin. Neurophysiol. 91, 353-362.21. Thompson, D. M., Winsauer, P. J. & Mastropaolo, J. (1987) Pharmacol.

Biochem. Behav. 26, 401-405.22. Moerschbaecher, J. M. & Thompson, D. M. (1980) Pharmacol. Biochem.

Behav. 13, 887-894.23. Goldman-Rakic, P. S. (1994) J. Neuropsychiatry Clin. Neurosci. 6, 348-357.24. Javitt, D. C., Doneshka, P., Zylberman, I., Ritter, W. & Vaughan, H. G., Jr.

(1993) Biol. Psychiatry 33, 513-519.25. Javitt, D. C., Doneshka, P., Grochowski, S. & Ritter, W. (1995)Arch. Gen.

Psychiatry 52, 550-558.26. Shelley, A. M., Ward, P. B., Catts, S. V., Michie, P. T., Andrews, S. &

McConaghy, N. (1991) Biol. Psychiatry 30, 1059-1062.27. Catts, S. V., Shelley, A. M., Ward, P. B., Liebert, B., McConaghy, N.,

Andrews, S. & Michie, P. T. (1994) Am. J. Psychiatry 152, 213-219.28. Strous, R., Cowan, N., Ritter, W. & Javitt, D. C. (1995) Am. J. Psychiatry

152, 1517-1519.29. Javitt, D. C. & Zukin, S. R. (1991) Am. J. Psychiatry 148, 1301-1308.30. Javitt, D. C., Zylberman, I., Zukin, S. R., Heresco-Levy, U. & Linden-

mayer, J. P. (1994) Am. J. Psychiatry 151, 1234-1236.31. Funahashi, S., Bruce, C. J. & Goldman-Rakic, P. S. (1993) J. Neurosci. 13,

1479-1497.32. Williams, G. V. & Goldman-Rakic, P. S. (1995) Nature (London) 376,

572-57533. Winsberg, B. G., Javitt, D. C., Silipo, G. S. & Doneshka, P. (1993) Psycho-

pharmacol. Bull. 29, 229-233.34. Lewis, D. A., Campbell, M. J., Foote, S. L., Goldstein, M. & Morrison, J. H.

(1987) J. Neurosci. 7, 279-290.35. Campbell, M. J., Lewis, D. A., Foote, S. L. & Morrison, J. H. (1987)

J. Comp. Neurol. 261, 209-220.36. Desimone, R. (1992) Science 258, 245-246.

A B c D

11

III1 --

IVIE

1b1i±L1LUJ1AJJ1J1l LUJJJJJIJJJ111l0 120 240 0 120 240

Neurobiology: Javitt et al.

V -A-... -..-

--

I

Dow

nloa

ded

by g

uest

on

Oct

ober

30,

202

0