uhlhaas2010abnormal.pdf

Embed Size (px)

Citation preview

  • 7/25/2019 uhlhaas2010abnormal.pdf

    1/14

    Schizophrenia is characterized by prominent psychoticsymptoms that include the false attribution of perceptualexperience to an external source (hallucinations), grosslydistorted thinking (delusions), reduction in affect andbehaviour ( negative symptoms ) and disorganization ofthought and language (thought disorder). In addition,patients with schizophrenia exhibit impairments in bothbasic sensory processing and higher cognitive functions,such as language, reasoning and planning.

    Despite more than 100 years of research, the causes ofschizophrenia are still unknown. Efforts to understandthe pathophysiology of schizophrenia have concentratedon the identification of abnormalities in specific corticalregions that are related to the symptoms of the disorder. Itis becoming increasingly clear, however, that the psychoticphenomena and cognitive dysfunctions that character-ize this disorder are not due to a circumscribed deficitbut rather represent a distributed impairment involving

    many cortical areas and their connectivity. Recent theoriestherefore highlight the possible role of a disconnectionsyndrome and/or disturbed dynamic coordination in thepathophysiology of schizophrenia 1,2. Accordingly, mecha-nisms that mediate the generation of coherent and coordi-nated activity in cortical circuits are prime candidates forunderstanding the pathophysiology of schizophrenia.

    Neural oscillations are a fundamental mechanismfor enabling coordinated activity during normal brainfunctioning 36 and are therefore a crucial target forschizophrenia research. Neural oscillations in the low(theta and alpha) and high (beta and gamma) fre-quency ranges establish precise temporal correlations

    between distributed neuronal responses. Oscillationsin the beta and gamma range establish synchroniza-tion with great precision in local cortical networks 7,8 (FIG. 1), whereas lower frequencies preferentially establishsynchronization over longer distances 9.

    These temporal correlations are functionally relevantas there is abundant evidence for a close relationshipbetween the occurrence of oscillations and cognitiveand behavioural responses, such as perceptual grouping ,attention-dependent stimulus selection, working mem-ory and consciousness (TABLE 1) (for a recent review seeREF. 10 ). Schizophrenia is associated with disturbancesin all these functions 1113 and over the past decade ithas been recognized that cognitive impairments, whichremain largely stable throughout the course of the disor-der, could provide a more direct target for efforts to iden-tify basic pathophysiological mechanisms than psychoticsymptoms. Importantly, cognitive deficits are not modi-

    fied by current pharmacological treatments and underliethe poor functional outcome in most patients 14.

    Furthermore, synchronized oscillations have beenshown to establish the precision in spike timing thatis crucial for use-dependent synaptic plasticity 1518.Although there is not yet direct evidence that synapticplasticity is impaired in schizophrenia, indirect evidencefor this hypothesis comes from studies that have demon-strated impairments in motor circuit reorganization aftertranscranial magnetic stimulation (TMS)-mediated distur-bance of the motor cortex 19 and from impairments inthe mismatch negativity event-related potential (ERP) 20, aphenomenon thought to depend on synaptic plasticity 21.

    *Department ofNeurophysiology, Max-PlanckInstitute for Brain Research,Deutschordenstrasse 46,Frankfurt am Main, 60528,Germany.Laboratory for

    Neurophysiology andNeuroimaging, Department ofPsychiatry, Johann WolfgangGoethe-Universitt, Heinrich-Hoffman-Strasse 10, Frankfurtam Main, 60528, Germany.Frankfurt Institute for

    Advanced Studies, JohannWolfgang Goethe-Universitt,Ruth-Moufang-Strasse 1,60438 Frankfurt am Main,Germany.Correspondence to P.J.U.e-mail: [email protected]:10.1038/nrn2774

    Negative symptomsAn absence of behaviour,characterized by flat or bluntedaffect and emotion, poverty ofspeech (alogia), inability to

    experience pleasure(anhedonia) and lack ofmotivation (avolition).

    Abnormal neural oscillations andsynchrony in schizophrenia Peter J. Uhlhaas* and Wolf Singer*

    Abstract | Converging evidence from electrophysiological, physiological and anatomicalstudies suggests that abnormalities in the synchronized oscillatory activity of neurons mayhave a central role in the pathophysiology of schizophrenia. Neural oscillations are afundamental mechanism for the establishment of precise temporal relationships betweenneuronal responses that are in turn relevant for memory, perception and consciousness. Inpatients with schizophrenia, the synchronization of beta- and gamma-band activity isabnormal, suggesting a crucial role for dysfunctional oscillations in the generation of thecognitive deficits and other symptoms of the disorder. Dysfunctional oscillations may ariseowing to anomalies in the brains rhythm-generating networks of GABA ( -aminobutyricacid) interneurons and in cortico-cortical connections.

    R E V I E W S

    100 | FEBRUARY 2010 | VOLUME 11 www.nature.com/reviews/neuro

    20 Macmillan Publishers Limited. All rights reserved10

    mailto:[email protected]:[email protected]:[email protected]:[email protected]
  • 7/25/2019 uhlhaas2010abnormal.pdf

    2/14

    N u m

    b e r o f s p

    i k e s

    N u m

    b e r o

    f s p

    i k e s

    N u m

    b e r o f s p

    i k e s

    N u m

    b e r o

    f s p

    i k e s

    Spike arriving at peak excitabilitySpike missing peak excitability

    0

    4

    8

    0

    4

    8

    0

    4

    8

    0

    4

    8

    0

    4

    8

    0

    4

    8

    100 0 +100 100 0 +100 100 0 +100

    100 0 +100 100 0 +100 100 0 +100

    Neuron 1 Neuron 2 Cross-correlation

    |

    High-frequency oscillations

    a

    b

    Time

    No high-frequency oscillations

    N u m

    b e r o

    f c o

    i n c i

    d e n

    t e v e n

    t s

    N u m

    b e r o

    f c o

    i n c

    i d e n

    t e v e n

    t s

    Temporal shift ( ms) Temporal shift ( ms) Temporal shift ( ms)

    Temporal shift ( ms) Temporal shift ( ms) Temporal shift ( ms)

    In this Review, we highlight the role of dysfunc-tional neural oscillations in schizophrenia by review-ing the evidence from studies that have examinedoscillatory activity and its synchronization in patientswith schizophrenia using electroencephalography(EEG) and magnetoencephalography (MEG) ( BOX 1 ).Furthermore, we examine the possible neurobiologicalcauses of impaired oscillations and the involvement ofaberrant oscillatory activity in the neurodevelopmentof schizophrenia.

    Neural oscillations in schizophreniaAs noted above, neural oscillations are thought to be afundamental mechanism for the coordination of neu-ronal responses throughout the cortex, and impair-ments in these oscillations are a candidate mechanismfor a pervasive network impairment in schizophrenia.This is supported by the results of EEG and MEG stud-ies (FIG. 2; see Supplementary information S1 (table))that have examined neural oscillations at differenttemporal and spatial scales during cognitive tasks and

    at rest. Studies investigating task-related oscillationshave measured both evoked oscillations, which reflectsensory-driven oscillatory activity and self-generatedoscillations (induced oscillations) and their large-scalesynchronization.

    Steady-state evoked potentials. Steady-state evokedpotentials (SSEPs) are a basic neural response to a tem-porally modulated stimulus to which SSEPs are synchro-nized in frequency and phase. Steady-state paradigmscan probe the ability of neuronal networks to generateand maintain oscillatory activity in different frequencybands. Consistent evidence for a deficit in the SSEPsevoked by auditory stimuli in patients with schizophre-nia has been obtained from eight studies 2229, althoughone study 30 demonstrated impaired auditory SSEPs infirst-degree relatives of patients with chronic schizophre-nia but not in the patients themselves. Dysfunctions inthe auditory SSEP to trains of clicks presented at gammafrequency, in particular at 40 Hz, have been shown tobe pronounced, but deficits in SSEPs in response to thepresentation of stimuli at lower frequency bands havealso been shown 25,26. Deficits have also been reportedfor visual SSEPs, in particular to stimuli in the betafrequency range 31.

    Initially, it was unclear whether the auditory SSEPis an intrinsic oscillatory process or whether it reflects

    the temporal overlap of potentials elicited by singleevents32. However, recent evidence does not support theconcept of superimposed evoked responses. For exam-ple, a perturbation in the auditory SSEP can be inducedby omitting a click in a stimulus series, an observationthat cannot be explained in terms of transient responsesto individual clicks 33. In addition, the temporal pro-file of the response to stimuli at 40 Hz, which begins200 ms after stimulus onset and continues after stimu-lus offset 34, and the frequency-specific modulation ofthe 40 Hz auditory SSEPs by attention 35 support thenotion that the 40 Hz response is indeed reflecting anoscillatory process.

    Figure 1 | Neural oscillations and synchrony in cortical networks. a | The timing ofrhythmic activity in cortical networks influences communication between neuronalpopulations. Three groups of interconnected neurons, each of which is rhythmicallyactive, are shown on the left. On the right are local field potential oscillations and actionpotentials (spikes; indicated by vertical lines) in the three populations. Spikes eitherarrive at the postsynaptic neuron during a peak in its local field potential (arrows),corresponding to a peak in its excitability, or miss these peaks (blunt arrows). Thetiming of the activity of two groups of neurons is thus either aligned, enablingeffective com munication (red and blue group), or not aligned (blue and grey group),preventing communication. b | Synchronization between neurons in local corticalnetworks depends on the occurrence of gamma oscillations 7. The panels show auto-(left-hand panels) and cross-correlograms (right-hand panels) of the responses of twoneurons (green and blue) in cortical area 17 in anaesthetized cats to a drifting gratingstimulus recorded at different times. Cross-correlograms are an index of the temporalcorrelation between neuronal responses, whereas auto-correlograms reflect thetemporal structure of a single channel. Autocorrelograms in the upper and lower rowsrespectively indicate phases with weak and strong oscillatory modulation of responses.The cross-correlograms indicate synchronization only in the presence of oscillations of~25 Hz (bottom row). Part a is modified, with permission, from REF. 137 (2005) Elsevier.Data in part b are courtesy of D. Nikoli, Max-Planck Institute for Brain Research,Frankfurt am Main, Germany.

    R E V I E W S

    NATURE REVIEWS | NEUROSCIENCE VOLUME 11 | FEBRUARY 2010 | 101

    20 Macmillan Publishers Limited. All rights reserved10

    http://www.nature.com/nrn/journal/v11/n2/suppinfo/nrn2774.htmlhttp://www.nature.com/nrn/journal/v11/n2/suppinfo/nrn2774.html
  • 7/25/2019 uhlhaas2010abnormal.pdf

    3/14

    Perceptual groupingThe ability of perceptualsystems to organize sensoryinformation into coherentrepresentations that can serveas the basis of our phenomenalexperience of the world.

    Transcranial magneticstimulation

    (TMS). A non-invasive methodto excite neurons in the brainby inducing weak electriccurrents in the tissue usingrapidly changing magneticfields.

    Mismatch negativityAn event-related potential thatis elicited when a sequence ofrepeated stimuli (standards) isinterrupted by stimuli thatdeviate in sensorycharacteristics such asintensity, frequency or duration(deviants).

    Evoked oscillations. Consistent with the evidence thatearly sensory processes are impaired in schizophre-nia13, several studies3640 have demonstrated abnormali-ties in the stimulus-locked activity that occurs within50150 ms after a stimulus is presented. For example,reductions in the amplitude and phase locking of evokedoscillations have been shown during the processing of

    visual information 37,40, suggesting an impaired ability toprecisely align oscillatory activity with incoming sensoryinformation.

    The evidence for deficits in evoked activity in theauditory domain is less consistent. Several studieshave shown that patients with schizophrenia are char-acterized by reduced amplitude and phase locking ofthe early (50150 ms) evoked beta- and gamma-bandresponse 36,38,39, but a recent study 40 did not confirm thisfinding. In addition, another study 41 observed reductionsin evoked gamma oscillations only in a latency range of220350 ms over frontal electrodes.

    Reductions in evoked gamma-band oscillations havealso been demonstrated in frontal regions, an area thathas been a traditional focus of schizophrenia research,through measurement of EEG responses following theapplication of TMS to the premotor cortex 42. Relative tohealthy controls, schizophrenia patients had a markeddecrease in gamma oscillations within the first 100 msafter TMS, particularly in a cluster of electrodes locatedin a fronto-central region. Source analyses revealed that inschizophrenia patients the gamma-band oscillationstriggered by TMS did not propagate beyond the area ofstimulation, whereas in controls activity was found inseveral motor and sensorimotor areas.

    Induced oscillations. Patients with schizophrenia also

    demonstrate reduced amplitude and synchronization ofself-generated, rhythmic activity in several cortical regions.Preliminary evidence for a deficit in high-frequency(60120 Hz) gamma-band activity comes from a recentstudy that tested gamma-band oscillations with MEGduring a perceptual organization task 43. Impaired per-formance in patients with schizophrenia was accom-panied by a widespread reduction in the power ofgamma-band oscillations in the right temporal lobe in atime window of 50300 ms after stimulus onset.

    The finding that there are intrinsic deficits in neuraloscillations in frontal circuits in schizophrenia is com-patible with EEG studies that have tested frontal gamma

    and theta oscillations during executive and workingmemory tasks. Patients with schizophrenia were char-acterized by a reduced amplitude of gamma and thetaoscillations in frontal regions 4446 and an impaired stim-ulus-induced phase resetting of ongoing oscillations atlow and high frequencies 47.

    Reductions in the amplitude of neural oscillations

    during cognitive tasks are accompanied by reducedphase synchronization of induced oscillatory activity.Phase synchronization has been proposed to providean effective mechanism for the integration of neuralresponses in distributed local cortical networks 48 (seefigure part a in Supplementary information S2 (figure)).Several studies have shown that in patients with schizo-phrenia the phase synchronization of oscillations inthe beta and gamma frequency bands during visuo-perceptual organization and auditory processing isreduced 4951. These findings suggest that impaired syn-chronization of beta- and gamma-band oscillationsunderlies the proposed functional disconnectivity of cor-tical networks in schizophrenia 1,2. It is currently unclear,however, to what extent impairments in local circuitscontribute to long-range synchronization impairmentsor whether these are two independent phenomena.

    Resting-state oscillations. Changes in neural oscillationshave also been demonstrated during rest in schizophre-nia: studies have reported an increase in low-frequency 52 activity and a reduction in high-frequency activity 53. Inaddition, a decrease in the amplitude of oscillations hasbeen shown to be accompanied by reductions in thecoherence of oscillations at theta frequency 54.

    Are there medication effects? An important question is

    to what extent the impaired neural oscillations seen inpatients with schizophrenia might be related to the effectsof medication. There is preliminary evidence that patientstreated with atypical antipsychotic medication may haveintact that is, within the normal range gamma-bandoscillations in the auditory SSEP paradigm 30. In addition,other studies have shown that deficits in neural oscillationsare present regardless of medication status 41,43. For exam-ple, preliminary evidence suggests that the reductions ingamma-band oscillations in MEG data during perceptualorganization are also present in never-medicated, first-episode patients with schizophrenia, albeit to a lesserdegree than in chronic, medicated patients 43. Another

    Table 1 | Neural oscillations in cortical networks

    Frequency band Anatomy Function

    Theta (47 Hz) Hippocampus 134 , sensory cortex 140 andprefrontal cortex 141

    Memory 142,143 , synaptic plasticity 18, top-downcontrol 9 and long-range synchronization 9

    Alpha (812 Hz) Thalamus 144 , hippocampus 145 , reticularformation 145 , sensory cortex 146 and motorcortex 147

    Inhibition 148 , attention 149 , consciousness 150 ,top-down control 9 and long-rangesynchronization 151

    Beta (1330 Hz) All cortical structures, subthalamic nucleus 152 ,basal ganglia 152 and olfactory bulb 153

    Sensory gating 154 , attention 155 , motor control 156 and long-range synchronization 157

    Gamma (30200 Hz) All brain structures, retina 158 and olfactorybulb 159

    Perception 7, attention 160 , memory 161 ,consciousness 162 and synaptic plasticity 16

    R E V I E W S

    102 | FEBRUARY 2010 | VOLUME 11 www.nature.com/reviews/neuro

    20 Macmillan Publishers Limited. All rights reserved10

    http://www.nature.com/nrn/journal/v11/n2/suppinfo/nrn2774.htmlhttp://www.nature.com/nrn/journal/v11/n2/suppinfo/nrn2774.htmlhttp://www.nature.com/nrn/journal/v11/n2/suppinfo/nrn2774.html
  • 7/25/2019 uhlhaas2010abnormal.pdf

    4/14

    0

    1

    V 2

    V 2

    |

    On Off t

    t = 50 ms

    F r e q u e n c y

    ( H z

    )

    20406080

    0.2 0.0 0.2 0.4 0.6 0.8 1.00

    1

    100 ms

    a

    b

    100 ms

    200 0 200 400 600 800Time (s)

    050

    100150

    200300

    F r e q u e n c y

    ( a v e r a g e

    )

    200 0 200 400 600 800

    200 0 200 400 600 800

    Time (s)

    Time (s)

    04080120160200240

    0

    1

    V 2

    0

    1

    V 2

    Trial 1

    Trial 2

    Trial 3

    ERP/ERF

    Evoked InducedF(trial 1)

    F(trial 2)

    F(trial i)

    sum(F(trial i))/NF(ERP/ERF)

    Time (s)

    study 41 showed a reduction in evoked gamma-bandactivity during an auditory oddball paradigm in medi-cation-naive first-episode patients with schizophrenia.In addition, several studies have reported abnormalities

    in the amplitude and phase of gamma-band oscillations inpatients who were treated only for a brief period withantipsychotic medication 27,51. Thus these data suggest thatdysfunctions in neural oscillations and synchronization are

    Box 1 | Measuring neural oscillations in EEG and MEG signals

    To measure the amplitude and synchrony of oscillations in electroencephalography (EEG) and magnetoencephalography(MEG) data the electrophysiological signal must be transformed into the frequency domain 136 . Measures of the amplitudeof oscillations can be further differentiated into steady-state evoked potentials as well as evoked and induced components.

    Measurement of steady-state evoked potentialsThe left panel in part a of the figure illustrates a steady-state stimulation at a frequency of 20 Hz. Each vertical linecorresponds to a stimulus. Below this is a voltage trace recorded from an EEG or MEG electrode. The amplitude of the signalis modulated by the stimulation frequency. The right-hand panel shows the spectral power of oscillations (indicated by thecolour scale). A peak of spectral power that corresponds to the stimulation frequency (20 Hz) and a harmonic response at40 Hz are shown. Steady-state evoked potentials can probe the ability of neuronal networks to generate and maintainoscillatory activity in different frequency bands.

    Measuring evoked and induced oscillatory activityEvoked oscillations occur at a consistent time lag after the onset of an external event and can be identified by averagingthe responses of several trials. By contrast, induced oscillations are not locked to the onset of a stimulus. Analysis ofinduced oscillations must therefore be performed on a single-trial basis because averaging across trials would cancel outoscillations owing to random phase shifts. The top left panel of part b of the figure shows the EEG or MEG signal recordedacross individual trials and the average of the signal across trials (the event-related potential (ERP) or event-related field(ERF)). The bottom left panel is a timefrequency map of the ERP/ERF, showing the spectral power (indicated by the colourscale). The peak of spectral power corresponds to the onset of the evoked oscillations. The top right panel shows threesingle-trial timefrequency maps. This reveals two peaks of spectral power, corresponding to the evoked and the inducedoscillations. The bottom right panel shows an average of the single-trial timefrequency maps.

    Evoked and induced oscillations thus reflect different aspects of information processing in cortical networks. Owing toits close temporal relationship with the incoming stimulus, evoked activity reflects bottom-up sensory transmission.Induced oscillations represent the internal dynamics of cortical networks and have been related to higher cognitivefunctions. Images courtesy of F. Roux, Max-Planck Institute for Brain Research, Frankfurt am Main, Germany.

    R E V I E W S

    NATURE REVIEWS | NEUROSCIENCE VOLUME 11 | FEBRUARY 2010 | 103

    20 Macmillan Publishers Limited. All rights reserved10

  • 7/25/2019 uhlhaas2010abnormal.pdf

    5/14

    |

    ControlsControls

    20

    40

    60

    80

    100

    20

    40

    60

    80

    100Patients with ScZ

    Patients with ScZ

    0 300

    0300

    300

    300P L F

    0.13

    0.04

    ba

    c

    00

    10

    20

    20 40 60 00

    10

    20

    20 40 60

    00

    10

    20

    20 40 60 00

    10

    20

    20 40 60

    00

    10

    20

    20 40 60 00

    10

    20

    20 40 60

    M e a n p o w e r

    ( V 2 )

    F r e q u e n c y

    ( H z )

    F r e q u e n c y

    ( H z )

    Time (ms)Time (ms)

    S D

    1.50.50.51.52.53.54.55.56.57.5

    S D

    1.50.50.51.52.53.54.55.56.57.5

    152025303540

    4550

    152025303540

    4550

    400 200 200 400 600 8000400 200 200 400 600 8000

    Controls Patients with ScZ

    Difference map

    0100 ms 100200 ms 200300 ms 300400 ms

    Time (ms)

    Time (ms)

    40 Hz 40 Hz

    30 Hz 30 Hz

    20 Hz 20 Hz

    Frequency (Hz)

    F r e q u e n c y

    ( H z )

    F r e q u e n c y

    ( H z )

    EndophenotypeA neurophysiological,neuroanatomical, cognitive orneuropsychological markerthat points to the geneticunderpinnings of a clinicalsyndrome. An endophenotypemust be heritable and stateindependent, and withinfamilies the endophenotypeand illness must co-segregate.

    Positive symptomsA range of psychotic symptomsthat most individuals do notnormally experience. Typicalsymptoms are hallucinations invarious modalities (auditory,visual and tactile) anddelusions (paranoid delusionsand delusions of reference).

    present at illness onset and are not due to the confoundinginfluences of medication. Nevertheless, more research isnecessary to address this important issue.

    Oscillations as an endophenotype. Recent evidence indi-cates that dysfunctional neural oscillations represent anendophenotype of schizophrenia. Work in healthy twinshas demonstrated that the power and temporal correla-tions of oscillations during the resting state are highlyheritable 55, indicating that neural oscillations can beexploited in the search for genetic contributions to schizo-phrenia. Indeed, a recent study 56 has provided importantevidence for the relationship between impaired neuraloscillations and genetic predisposition to schizophrenia.The authors examined the heritability of deficits in thetime-frequency-transformed ERP during sensory gatingin a large sample of patients with schizophrenia, theirfirst-degree relatives and healthy control participants.Theta- and alpha-band oscillations were impaired inpatients and first degree-relatives and this impairmentwas more heritable than the traditional P50 measure. The

    same group also reported a deficit in the auditory SSEP infirst-degree relatives of patients with schizophrenia 30.

    Neural oscillations and the core symptoms of schizophrenia. There is also evidence that aberrant oscillatory activitycould be related to the core symptoms of schizophre-nia, such as hallucinations, thought disorder and nega-tive symptoms. Several studies have found that positivesymptoms of schizophrenia are correlated with enhancedamplitude and phase synchronization of evoked andinduced beta- and gamma-band activity in circum-scribed brain regions 27,28,37,49,50,57,58, whereas disorganiza-tion and negative symptoms have been related to bothenhanced 37 and reduced high-frequency oscillations 43,57.This association is particularly robust for the presence ofauditory and visual hallucinations. There is evidence ofgreater high-frequency activity during the resting stateas well as during auditory and visual sensory processingin sensory areas in patients with hallucinations than inthose without 27,28,37,57,58. These findings suggest that thecortical areas involved in generating hallucinations might

    Figure 2 | Neural oscillations and synchrony inschizophrenia. a | Auditory steady-state responses inpatients with schizophrenia (ScZ). The left-hand panelsshow the spectral power over a midline frontal electrodesite in controls (n = 15) and patients with schizophrenia(n = 15) during the presentation of click trains at 40 Hz,30 Hz and 20 Hz. Patients with schizophrenia show lowerpower to stimulation at 40 Hz than control subjects, but nodifference at lower frequencies of stimulation. b | Sensoryevoked oscillations during a visual oddball task in patientswith schizophrenia. The coloured scale indicates the phaselocking factor (PLF; see figure part b Supplementaryinformation S2 (figure)) a measure of the degree of

    phase locking that can range from 0 (random distribution)to 1 (perfect phase locking) of oscillations in the20100 Hz frequency range in the occipital cortex(electrode O1) for healthy controls and patients withschizophrenia. Control participants show an increase inphase locking for gamma oscillations ~ 100 ms afterstimulus presentation. However, this is significantly smallerin patients with schizophrenia, indicating a dysfunction inearly sensory processes. c | Dysfunctional phase synchronyduring Gestalt perception in schizophrenia. Mooney faceswere presented in an upright and inverted orientation andparticipants indicated whether a face was perceived. Thetop right panels show the average phase synchrony (indicated by the coloured scale) over time for allelectrodes. In patients with schizophrenia, phasesynchrony between 200300 ms was significantly reducedrelative to controls. In addition, patients with schizophreniashowed a desynchronization in the gamma band(3055 Hz) in the 200280 ms interval. The bottom panelshows differences in the topography of phase synchrony inthe 2030 Hz frequency range between groups. Red linesindicate less synchrony between two electrodes in patientswith schizophrenia than in controls. Green lines indicategreater synchrony for patients with schizophrenia. SD,standard deviation. Part a is modified, with permission,from REF. 24 (1999) American Medical Association. Part b is modified, with permission, from REF. 41 (2004) Elsevier.Part c is modified, with permission, from REF. 49 (2006)Society for Neuroscience.

    R E V I E W S

    104 | FEBRUARY 2010 | VOLUME 11 www.nature.com/reviews/neuro

    20 Macmillan Publishers Limited. All rights reserved10

  • 7/25/2019 uhlhaas2010abnormal.pdf

    6/14

    Phase synchronyPhase synchrony andcoherence are estimates of thesynchrony of brain oscillations.Phase synchrony provides anestimate of synchronyindependent of the amplitudeof oscillations. This contrastswith measures of coherence, inwhich synchrony andamplitude are intertwined.

    Corollary dischargeThe estimate of sensoryfeedback that is derived fromthe internal copy of the motorsignal (the efference copy).

    Diffusion tensor imaging(DTI). An MRI technique usedto map three-dimensionaldiffusion of water in braintissue. It provides informationabout the microstructuralintegrity of the white matter,including axonal density andthickness, myelination andaxonal fibre direction.

    be hyperexcitable, a hypothesis that is consistent withthe increased haemodynamic responses that have beenobserved in the respective primary sensory areas 59.

    The local increase in neural oscillations seen inpatients with positive symptoms is accompanied bya deficit in the precise synchronization of oscillationsbetween cortical areas, which might lead to an impair-ment in corollary discharge . Several authors have arguedthat a failure of corollary discharge underlies theimpaired ability of patients with schizophrenia to distin-guish self-generated and externally generated perceptsand actions 12,60,61. One study 62 examined the coherenceof theta oscillations between frontal and temporal lobesin patients with and without auditory hallucinationsand in healthy controls as participants either listened totheir own played back speech or were instructed to talkaloud. In the controls and patients with schizophreniawithout hallucinations, talking was associated with anincrease in theta coherence between left frontal andtemporal electrodes relative to the listening condition.In patients with schizophrenia with hallucinations, this

    modulation was absent, suggesting a failure in the prepa-ration of temporal areas for speech production that couldlead to the misattribution of self-generated speech to anexternal source. The same group also reported reducedgamma-band oscillations before movement initiation inschizophrenia 63, suggesting that reduced neural oscilla-tions could also underlie dysfunctions in sensorimo-tor communication and associated impairments in theinitiation of willed action 12.

    These data highlight the fact that, although the powerand synchrony of neural oscillations are decreased in spe-cific frequency bands in schizophrenia, positive symp-toms may be associated with circumscribed increases inoscillatory activity that may reflect the read-out of storedexperiences, consistent with the role of high-frequencyactivity in the generation of internal representations 64.The increase in the power and synchrony of oscilla-tions in local circuits may be accompanied by impairedcorollary discharge mechanisms, as described above.However, more research is required to provide directevidence for this hypothesis.

    Neurobiology of abnormal oscillationsThe neuronal mechanisms responsible for generatingoscillatory activity and its synchronization have beenstudied extensively both in vivo and in vitro (FIG. 3) . Thishas enabled the identification of anatomical deficits and

    abnormalities in neurotransmitter systems in schizo-phrenia (FIG. 4) that may underlie the abnormalities seenin EEG and MEG studies (FIG. 2).

    Anatomical def ici ts. Neural oscillations and theirsynchronization are dependent on the integrity of thesynaptic contacts in local cortical circuits. Schizophreniais associated with widespread reductions in the volume ofgrey matter 65 that are thought to reflect a reductionof synaptic connectivity, whereas the overall number ofneurons is largely preserved 66. This may explain theobserved reductions in the power of neural oscillations,as previous studies have demonstrated relationships

    between the degree of grey matter reduction anddecreases in the amplitude of ERPs 6769.

    Synchronization of oscillatory activity in the beta andgamma frequency range is dependent on cortico-corticalconnections that reciprocally link cells situated in thesame cortical area, in different areas or even in differ-ent hemispheres 70,71 (FIG. 3b) . Accordingly, disruptions inthe volume and organization of anatomical connectiv-ity should impair long-range synchronization. Evidencefrom in vivo and post-mortem studies suggests that whitematter volume and integrity are altered in patients withschizophrenia. These studies have found reduced vol-ume of white matter in several brain regions as well asreduced organization of cortico-cortical connectionsas disclosed by diffusion tensor imaging (DTI) 7274 (FIG. 4a) .

    Neurotransmitter systems. Several neurotransmittersystems that are abnormal in schizophrenia are also cru-cially involved in the generation and synchronization ofcortical beta and gamma oscillations. Of particular impor-tance is the network of GABA (-aminobutyric acid)-ergic

    interneurons that acts as a pacemaker in the generationof high-frequency oscillations by producing rhythmicinhibitory postsynaptic potentials (IPSPs) in pyramidalneurons. IPSPs generated by a single GABAergic neu-ron may be sufficient to synchronize the firing of a largepopulation of pyramidal neurons 75, and the duration ofthe IPSP can determine the dominant frequency of oscil-lations in a network 76 (FIG. 3a) . Interneurons that expressthe Ca2+-binding protein parvalbumin are of particularrelevance as these are fast-spiking cells and their activ-ity has been demonstrated to be causally related to thegeneration of gamma oscillations in mice in vivo 77.

    A large body of work suggests that schizophreniainvolves alterations in GABAergic neurotransmission 78(FIG. 4b) . One widely replicated finding is a reduction inthe mRNA of GAD67 (also known as glutamate decar-boxylase 1), an enzyme that synthesizes GABA, in sev-eral cortical areas in patients with schizophrenia 79,80.The decrease is specific to layers 35 (REFS 79,80) and isaccompanied by reduced expression of the GABA mem-brane transporter 1 ( GAT1 ; also known as SLC6A1)81,indicating that there is impaired synthesis and reuptake ofGABA in interneurons in schizophrenia. Further studiesrevealed that these deficits are particularly pronouncedin parvalbumin-positive interneurons: GAD67 mRNAwas not detectable in 50% of these cells in patients withschizophrenia, whereas the overall number of cells was

    unchanged 82.Several studies have provided direct links between

    abnormal GABAergic neurotransmission and alteredneural oscillations. A recent study 23 showed that in a net-work simulation of the auditory cortex an increase in thedecay time of IPSPs produced a pronounced oscillatoryresponse at 20 Hz stimulation, whereas neural oscilla-tions were reduced at 40 Hz. MEG data from patientswith chronic schizophrenia revealed a similar profile ofabnormal neural oscillations. The authors suggested thatthe reduced availability of GAT1 is a candidate mecha-nism of the extended IPSPs that in turn results in theauditory SSEP deficits in schizophrenia.

    R E V I E W S

    NATURE REVIEWS | NEUROSCIENCE VOLUME 11 | FEBRUARY 2010 | 105

    20 Macmillan Publishers Limited. All rights reserved10

    http://www.uniprot.org/uniprot/Q99259http://www.uniprot.org/uniprot/P30531http://www.uniprot.org/uniprot/P30531http://www.uniprot.org/uniprot/Q99259
  • 7/25/2019 uhlhaas2010abnormal.pdf

    7/14

    |

    ms0100 +100

    a

    18 1819 19

    17 17

    CC

    Intact corpuscallosum

    Sectioned corpuscallosum

    b

    80 +800 N u m

    b e r o

    f c o

    i n c

    i d e n

    t e v e n

    t s

    N u m

    b e r o

    f c o

    i n c

    i d e n

    t e v e n

    t s

    50

    Time (ms) Time (ms)

    RH+LH

    40 ms

    Electrical coupling

    Pyramidal cell

    Basket cell(parvalbumin-

    expressing cell)80 +800

    50

    LFP electrode

    Data from two animal models of schizophreniafurther support a link between abnormal parvalbuminexpression and impairments in gamma-band oscilla-tions. Treatment of rats with methylazoxymethanolacetate led to decreased expression of parvalbuminin interneurons in the medial prefrontal cortex and inthe ventral subiculum of the hippocampus thatwas accompanied by a reduction in gamma-band

    responses to a conditioned tone during a latentinhibition paradigm 83 (FIG. 4c) . Similarly, lysophospha-tidic acid receptor 1-deficient mice, which display arange of cognitive and neurochemical deficits similarto those seen in schizophrenia, are characterized by areduction in gamma oscillations and in the numbersof parvalbumin-positive interneurons in the medialentorhinal cortex 84.

    Figure 3 | Mechanisms underlying the generation of gamma oscillations and synchrony. a | A neocortical circuitinvolved in the generation of gamma-band oscillations. Generation of synchronized neural activity in neocortical circuits isdependent on negative feedback inhibition of pyramidal cells by GABA ( -aminobutyric acid)-ergic interneurons that expressthe Ca 2+-binding protein parvalbumin. These receive glutamate receptor-mediated feedforward excitatory inputs, whichmakes them susceptible to changes in glutamatergic drive. Transient excitation of parvalbumin-expressing interneuronsleads to a depolarization of many interneurons, which are themselves reciprocally interconnected through gap junctions andchemical GABAergic synapses. Electrical synapses are important for the synchronization of network activity because theyrapidly propagate activity. Conversely, mutual inhibition through chemical synapses is a crucial determinant of the networkfrequency, as the duration of inhibitory postsynaptic potentials determines the dominant oscillation frequency. The resultingrhythmic inhibitory postsynaptic potentials can synchronize the firing of a large population of pyramidal neurons as the axonof an individual GABAergic neuron makes multiple postsynaptic contacts onto several pyramidal cells. This phasic inhibitionleads to the synchronization of spiking activity that can be recovered with a cross-correlogram. A local field potential (LFP)recorded with an extracellular electrode reflects the average of the transmembrane currents that fluctuate at gamma-bandfrequency. Its extracranial counterpart can be reflected in electroencephalography (EEG) or magnetoencephalography(MEG) signals. b | Cortico-cortical connections mediate long-distance synchronization. The relationship between theintegrity of the corpus callosum and interhemispheric synchronization of gamma-band oscillations in the cat visual cortex isillustrated. Recording electrodes were placed in the vicinity of the border of areas 17 and 18 of the right (RH) and left (LH)cortical hemispheres during stimulation with a light bar. In the bottom panels are cross-correlograms between responsesfrom different recording sites in the LH and RH that indicate the degree of interhemispheric synchronization. When thecorpus callosum was intact (left-hand panel), strong interhemispheric synchronization occurred with no phase lag betweenthe LH and RH recording sites. Sectioning of the corpus callosum (right-hand panel) abolished interhemisphericsynchronization while leaving synchronization within hemispheres intact. These data show that synchronization can occurover long distances with high precision and is crucially dependent on the integrity of cortico-cortical connections. The upperpanel of part b is modified with permission, from REF. 163 (1972) Elsevier. The lower panel of part b is modified, withpermission, from REF. 70 (1991) American Association for the Advancement of Science.

    R E V I E W S

    106 | FEBRUARY 2010 | VOLUME 11 www.nature.com/reviews/neuro

    20 Macmillan Publishers Limited. All rights reserved10

  • 7/25/2019 uhlhaas2010abnormal.pdf

    8/14

    |

    Rostrum

    Inferior

    genu

    Superiorgenu

    Posteriorgenu

    Anteriormidbody

    Middlemidbody

    Posteriormidbody

    Isthmus

    Splenium

    ControlSchizophrenia

    Cortical layer

    n C i p e r g

    01 2 3 4 5/6

    20

    40

    60

    80

    100

    120

    *

    P V c e

    l l s p e r m m

    3 (

    % )

    mPFC ACg vSub dSub0

    50

    100

    150

    **

    *

    ControlMAM

    c

    a b

    500 2 4 6 8 10

    100

    150

    200

    250

    Time (s)

    % B a s e

    l i n e

    Saline

    500 2 4 6 8 10

    100

    150

    200

    250

    Time (s)

    % B a s e

    l i n e

    MAM

    A crucial issue is whether the observed impairmentsin GABAergic interneurons are primary or second-ary to alterations in other neurotransmitter systems.Parvalbumin-expressing interneurons receive excitatoryinputs through NMDA ( N -methyl- -aspartate) recep-

    tors85, particularly the NR2A/NR2B subtype, which makesthem susceptible to changes in glutamatergic drive. Severallines of evidence support the notion that NMDA receptordysfunction may be related to altered GABA neurotrans-mission. Application of NMDA antagonists that producepsychosis in healthy participants 86 also causes changes ininhibitory synaptic transmission. Similarly, acute admin-istration of ketamine to mice reduces the amplitude andfrequency of IPSPs87 and decreases the power of gamma-band oscillations in superficial layers of the medialentorhinal cortex 84. Furthermore it seems that the acuteeffects of NMDA antagonists on parvalbumin-positiveinterneurons are mediated by oxidative stress 88,89.

    Recent evidence from animal models suggests thatthe effects of NMDA receptor blockade on neural oscil-lations may be region specific and in some instancescan also lead to an increase in high-frequency oscilla-tions90. For example, application of NMDA antagonists

    produced an increase in gamma-band oscillations inlocal circuits in the auditory cortex in humans 91 and inthe neocortex of rats 92. This paradoxical result could bedue to the disinhibition of principal cells as a result ofreduced interneuron excitation 93, which would facilitatethe transient and uncoordinated generation of gammaoscillations. NMDA receptors are candidates for the long-range synchronization of local circuits as they are promi-nent in the superficial cortical layers that are the mainrecipients of long cortico-cortical connections. Thus,disruption of NMDA receptors may lead to a decouplingof local gamma oscillators from the controlling influ-ence of extended networks, resulting in a pathological

    Figure 4 | Neurobiological correlates of deficits in neural oscillations and synchrony in schizophrenia.a | Connectivity of the corpus callosum and its abnormalities in schizophrenia as reflected in diffusion tensor imagingdata. Changes in connectivity were measured by fractional anisotropy in the corpus callosum of 24 patients withschizophrenia and 24 healthy controls. Fractional anisotropy values estimate the presence and coherence of orientedstructures, such as myelinated axons. Regions marked in dark red were significantly different between patients andcontrols at Bonferroni corrected p < 0.0055. Regions marked with light red were significantly different only at theuncorrected level of p < 0.05. Patients with schizophrenia show significantly less organization in subdivisions of the corpuscallosum than controls. b | Expression of parvalbumin (PV) mRNA in layers 34 of the dorsolateral prefrontal cortex isreduced in patients with schizophrenia. Together with other data showing that the number of parvalbumin-positiveinterneurons is unchanged in schizophrenia 138 , these findings suggest that the ability of parvalbumin-positive interneuronsto express important genes is impaired in schizophrenia. c | Reduction in gamma oscillations and parvalbumin-positiveneurons in the medial prefrontal cortex (mPFC) in an animal model of schizophrenia. Rats are treated with methyla -zoxymethanol acetate (MAM) in utero at gestational day 17. This model reproduces the anatomical changes, behaviouraldeficits and altered neuronal information processing observed in patients 139 . Treated rats display a regionally specificreduction in the density of parvalbumin-positive neurons throughout the mPFC and ventral subiculum (vSUB). As shown inthe middle and right-hand panels, the presentation of a tone induces a mild increase in prefrontal gamma (3055 Hz)oscillations in saline- but not MAM-treated rats. * indicates statistically significant difference from control (p=0.05). indicates significance level p=0.005. ACg, anterior cingulate cortex. dSub, dorsal subiculum. Part a is modified, withpermission, from REF. 73 (2008) Academic Press. Part b is modified, with permission, from REF. 82 2003) Society forNeuroscience. Part c is modified, with permission, from REF. 83 (2009) Society for Neuroscience.

    R E V I E W S

    NATURE REVIEWS | NEUROSCIENCE VOLUME 11 | FEBRUARY 2010 | 107

    20 Macmillan Publishers Limited. All rights reserved10

  • 7/25/2019 uhlhaas2010abnormal.pdf

    9/14

    and short-term increase of gamma-band power. Thisscenario resembles the changes observed during theproduction of hallucinations in schizophrenia 58,62.

    In addition to the involvement of GABA- andNMDA-receptor mediated neurotransmission, datafrom animal models have suggested that alterations inthe dopaminergic and cholinergic systems may contrib-ute to abnormal oscillations 94. However, this hypothesishas not been thoroughly explored. Recent evidence indi-cates that cholinergic modulation has a crucial role in thefast, state-dependent facilitation of high-frequency oscil-lations and the associated response synchronization inanimal models 95,96. Accordingly, alterations in choliner-gic neurotransmission may be relevant for dysfunctionsin neural oscillations in schizophrenia 97. In particular,there is evidence that the 7 nicotinic receptor that isexpressed by GABAergic interneurons 98 is abnormal inpatients with schizophrenia 99,100.

    Dopamine is a neuromodulator that has tradition-ally been implicated in the pathophysiology of schiz-ophrenia. However, evidence for a direct impact of

    dopaminergic transmission on neural oscillations inschizophrenia is lacking. Interestingly, there is evidencesuggesting that dopamine agonists decrease pathologi-cal beta-band oscillations and increase gamma-bandoscillations in cortical and subcortical networks 101 in patients with Parkinsons disease, highlighting theimpact of dopaminergic signalling on neural oscil-lations. Dopamine could affect neural oscillations incortical networks by modulating frequency-dependentsignal transmission, as has been recently demonstratedin the hippocampus 102.

    Neurodevelopmental hypothesis and oscillations Neurodevelopment and schizophrenia. Schizophrenia ischaracterized by abnormal brain maturation at severalstages of development 103. Children who are later diag-nosed with schizophrenia have cognitive and behav-ioural impairments 104, suggesting that an early pre- orperinatal event may contribute to the pathogenesis ofthe disorder. Several environmental risk factors, such asobstetric complication and viral infections, in addition tothe genetic contribution may lead to altered developmentof neural circuits 103. Schizophrenia, however, typicallymanifests during late adolescence and early adulthood,raising the question of the contribution of later develop-mental processes. For example, it has been proposed thatthe appearance of psychosis is related to overpruning of

    synaptic contacts during adolescence 105.

    Oscillations and synchrony in the development ofcortical networks. Neural oscillations are involved inthe maturation and plasticity of cortical networks atseveral developmental stages. During early pre- andperinatal periods, spontaneous correlated neural activ-ity is a hallmark of the developing nervous system 106109.For example, patterned retinal activity synchronizes theactivity of neurons in the neonatal visual cortex 110111 andis essential for organizing connections in the visual cir-cuitry 112,113. Similarly, whisker-triggered oscillations actas a topographic template by synchronizing the activity

    of neurons in columns in the neonatal barrel field of thecortex before the emergence of cortical barrels 114.

    At later stages, experience-dependent modification ofcortical circuits contributes to the shaping and develop-ment of cortical networks. Modification of synaptic con-tacts is dependent on the precise temporal coordinationof neural activity 115. For spike timing-dependent plasticityto occur, pre- and postsynaptic spiking is required withina critical window of tens of milliseconds 116. Stimulation ofneurons during the depolarizing peak of the theta cyclein the hippocampus favours long-term potentiation,whereas stimulation in the trough causes depression 18.The same relationship holds for oscillations in the betaand gamma frequency range 16, indicating that oscillationsallow the precise alignment of the amplitude and tem-poral relations of presynaptic and postsynaptic activitythat determine whether a synaptic contact is strength-ened or weakened 16. Accordingly, aberrant neural oscil-lations during early critical periods may lead to imprecisetemporal coordination of neural activity and result in thepathological modification of cortical circuits.

    Neural oscillations and adolescent brain development. Recent evidence117 (FIG. 5a,b) points to profound changesin neural oscillations during late adolescence and earlyadulthood that could provide important clues to theemergence of psychosis. During early adulthood theta-,beta- and gamma-band oscillations and their long-rangesynchronization increase dramatically. Interestingly, thisincrease is preceded by a significant reduction of betaand gamma oscillations during late adolescence, sug-gesting that a transient destabilization occurs before theemergence of mature cortical networks. This highlightslate adolescence as a critical developmental period thatis associated with a rearrangement of functional net-works and with an increase in the temporal precisionand spatial focusing of neuronal interactions. We there-fore suggest that in schizophrenia cortical circuits thatare characterized by imprecise temporal dynamics areunable to support the neural coding regime that emergesduring the late adolescent period, leading to a break-down of coordinated neural activity and the emergenceof psychosis and cognitive dysfunctions.

    Several facts support the proposed expression ofhigh-frequency oscillations during late adolescence. Oneis the continued maturation of cortico-cortical connec-tions, involving increased myelination of long axonaltracts 118,119. As a result, transmission delays between

    brain regions are reduced during adolescence 120, sup-porting precise temporal coordination of distributedneural activity. Development of cortico-cortical connec-tions73 and maturation of grey matter have been foundto be abnormal in patients with schizophrenia 121,122 andmay lead to abnormalities in the occurrence of neuraloscillations and synchrony at different spatial scales.

    Recent evidence points to important changes inGABAergic neurotransmission during the adolescentperiod that could affect the development and synchro-nization of gamma-band oscillations (FIG. 5c) . A pre-dominance of GABA 2 subunits was observed in themonkey dorsolateral prefrontal cortex during early

    R E V I E W S

    108 | FEBRUARY 2010 | VOLUME 11 www.nature.com/reviews/neuro

    20 Macmillan Publishers Limited. All rights reserved10

  • 7/25/2019 uhlhaas2010abnormal.pdf

    10/14

    F r e q u e n c y

    ( H z

    )

    S D

    F r e q u e n c y

    ( H z

    )

    S D

    00

    2

    4

    67

    1

    3

    5

    Gamma-spectral power (3075 Hz)a

    Beta-phase synchrony (1330 Hz)b

    c

    1

    1

    3

    56

    0

    2

    4

    AdulthoodLate adolescenceEarly adolescence

    Late childhoodEarly childhood

    AdulthoodLate adolescenceEarly adolescence

    Late childhoodEarly childhood

    20

    200200 0 200200

    3040506070

    2030

    40506070

    Early adolescence(1214 yrs)

    Late adolescence(1517 yrs)

    Adulthood(1821 yrs)

    Early adolescence(1214 yrs)

    Late adolescence(1517 yrs)

    Adulthood(1821 yrs)

    P r o

    b a

    b i l i t y

    mIPSP decay time constant (ms)706050403020100

    0.0

    0.2

    0.4

    0.6

    0.8

    1.0

    |

    Prepubertal Postpubertal1 mV100 msec

    200 0 200 400

    0.0

    0.5

    1.0

    1.5

    2.0

    2.5

    S D

    Time (ms)

    Time (ms) Time (ms) Time (ms)

    Time (ms) Time (ms) Time (ms)0.5

    200 0 200 4001

    0

    1

    2

    3

    4

    Time (ms)

    S

    D

    400 400 0 200200 400

    0 200200 0 200200400 400 0 200200 400

    Figure 5 | Emergence of high-frequency oscillations and synchrony during the transition from adolescence toadulthood. a | The graph shows the spectral power of oscillations in the 3075 Hz range 100300 ms after thepresentation of Mooney faces at different ages. The right-hand panels show timefrequency maps for early adolescent,late adolescent and adult participants. Gamma oscillations increase significantly during the transition from adolescenceto adulthood. b | The graph shows phase synchrony in the 1330 Hz frequency range for all electrode pairs 100300 msafter stimulus presentation at different ages. The right-hand panels show phase synchrony (indicated by the colouredscale) of oscillations in the beta and gamma bands averaged across all electrodes for early adolescent, late adolescent andadult participants. The phase synchrony of beta-band oscillations increased until early adolescence and was thensubstantially reduced during late adolescence, suggesting that cortical networks reorganize during the transition fromadolescence to adulthood. c | Changes in GABA A (-aminobutyric acid type A) receptor-mediated neurotransmission inthe monkey dorsolateral prefrontal cortex during adolescence. The left-hand panel shows postnatal development ofminiature inhibitory postsynaptic potentials (mIPSPs) recorded from pyramidal neurons of prepubertal and postpubertalmonkeys in the dorsolateral prefrontal cortex. The right-hand panel shows cumulative probability distribution curves ofthe mIPSP decay time constant in prepubertal (blue) and postpubertal (green) animals. The left shift of the curve fromprepubertal to postpubertal animals indicates a higher fraction of shorter mIPSPs in postpubertal animals thanin prepubertal animals. As the decay time of IPSPs is a critical factor for the dominant frequency of oscillations in anetwork 68, these data provide one mechanism for the late maturation of high-frequency oscillations in electroencepha-lography data 117. SD, standard deviation. Parts a and b are modified, with permission, from REF. 117 (2009) NationalAcademy of Sciences. Part c is modified, with permission, from REF. 123 (2009) Elsevier.

    R E V I E W S

    NATURE REVIEWS | NEUROSCIENCE VOLUME 11 | FEBRUARY 2010 | 109

    20 Macmillan Publishers Limited. All rights reserved10

  • 7/25/2019 uhlhaas2010abnormal.pdf

    11/14

    development, whereas in adult animals 1 subunits areexpressed 123. This was accompanied by marked changesin the kinetics of GABA transmission, including a sig-nificant reduction in the duration of miniature IPSPs inpyramidal neurons. The shift in -subunit expressioncould provide a direct correlate of the observed increasein gamma-band oscillations, as 1 subunits predominateat synapses of parvalbumin-positive basket cells 124, whichare crucially involved in the generation of gamma-bandoscillations77. Recent data 125 also suggest that there aresignificant changes in the signalling properties of bas-ket cells during later developmental periods, as indicatedby a decrease in action potential duration, propagationtime, duration of the release period and the decay timeconstant of inhibitory postsynaptic currents.

    ConclusionsCurrent evidence points to a crucial role for altered neuraloscillations and synchrony in the pathophysiology ofschizophrenia. Reductions of beta and gamma oscilla-tions and their synchronization have been demonstrated

    during cognitive tasks and at rest, suggesting that thereis an intrinsic deficit in the temporal coordination ofdistributed neural activity. Correlations with the coresymptoms of schizophrenia furthermore highlight thepotential role of neural oscillations in the productionof psychotic symptoms. Finally, these impairments arepresent at illness onset, are likely to be independent ofmedication status and are highly genetically heritable,raising the possibility that abnormal oscillations andsynchrony in schizophrenia directly reflect the biologicalprocesses that underlie the syndrome.

    We posit that the genetic vulnerability for schizophre-nia is translated into imprecise temporal coordinationof neural activity. Several risk genes for schizophrenia,including the neuregulin genes, CHRNA7, dysbin-din (also known as DTNBP1 ) and GAD1 , modulateGABAergic, cholinergic and NMDA receptor-mediatedneural transmission 94, and these transmitter systems arein turn crucially involved in the generation of neural oscil-lations. Recent evidence has furthermore provided directlinks between polymorphisms of risk genes and changesin neural oscillations. Polymorphisms in neuregulin 1 strongly modulate the amplitude of gamma oscillationsin rat hippocampal slices 126. Similarly, polymorphisms inthe genes that encode dopamine receptor D4 ( DRD4 ) anddopamine transporter ( SLC6A3 ) modulate the pattern ofevoked gamma responses in humans 127.

    Abnormal temporal dynamics of cortical circuits mayresult in impairments in synaptic plasticity (see REF. 21 for a similar perspective). Impaired plasticity is a candi-date mechanism for the enduring cognitive deficits andaberrant neurodevelopment observed in schizophrenia,and there is evidence that neural oscillations and syn-chrony may have a crucial role in synaptic modifications.As behavioural impairments are already detectable inchildren who later develop the disorder, dysfunctionalneural oscillations and plasticity are likely to cause aber-rant early pre- and perinatal development, leading tomaladaptive formation of cortical networks and faultyprogramming of synaptic connections. Accordingly,

    we propose that this fundamental impairment remainssilent until the late adolescent period when corticalnetworks fully exploit neural oscillations, in particularin the beta and gamma range, for the coordination ofdistributed brain processes.

    Impaired neural oscillations in schizophrenia maylead to functional disconnections between and withincortical regions, as previously proposed by several theo-ries1,2. Here, we note that most studies in schizophreniahave used functional MRI (fMRI) to study functionalconnectivity. fMRI, however, lacks the required temporalresolution, as dynamic neural interactions occur in thetime range of milliseconds. EEG and MEG are appro-priate tools with which to test functional connectivityanomalies, and several groups have demonstrated thefeasibility of this approach. Recent advances in sourcelocalization 128 have also improved the spatial resolutionof EEG and MEG measurements.

    Research into neural oscillations is likely to contrib-ute to the further delineation of the biological causesand mechanisms of schizophrenia and to the eventual

    development of pathophysiologically based treatmentinterventions. Neural oscillations and the molecularmechanisms and circuits that underlie them are highlyconserved in insects, birds and mammals. This allowshypotheses regarding the biological mechanisms thatunderlie impaired neural oscillations to be directlytested in animal models and in in vitro preparations.Indeed, such work is already under way 84 and mayhelp to link data from EEG and MEG experimentswith patient populations to alterations in neurotrans-mitter systems. This possibility may not be offered byother imaging techniques, such as fMRI, for which thebiological mechanisms of signal generation are lessclear and the direct t ranslation of findings from dataobtained with human experiments to animal modelsis more difficult to accomplish. Neural oscillationscould therefore be considered an ideal intermediatephenotype that potentially allows the direct map-ping of genetic mechanisms of schizophrenia onto theneurobiology 129.

    Despite this enthusiasm, numerous unresolved issuesneed to be addressed. One is the diagnostic specificity ofdeficits in neural oscillations. A generalized impairmentof neural oscillations across diverse brain disorders willstrongly reduce their utility as a biomarker and inter-mediate phenotype. Thus, at present we cannot discardthe possibility that altered neural oscillations and syn-

    chrony are nonspecific features of several brain disordersreflecting diverse pathophysiological processes. There isalready evidence that several disorders are associatedwith abnormal neural oscillations 130, the genetic andbehavioural phenotypes of some of which overlap withthose of schizophrenia. For example, patients with bipolardisorder display impairments in auditory SSEPs similarto those of patients with schizophrenia 131. Furthermore,adults with autism spectrum disorders are also charac-terized by an impairment in gamma oscillations duringperceptual organization 132. However, comparison withfirst-episode schizophrenia patients revealed differencesbetween the two disorders: gamma-band dysfunctions

    R E V I E W S

    110 | FEBRUARY 2010 | VOLUME 11 www.nature.com/reviews/neuro

    20 Macmillan Publishers Limited. All rights reserved10

    http://www.ncbi.nlm.nih.gov/gene/1139?ordinalpos=1&itool=EntrezSystem2.PEntrez.Gene.Gene_ResultsPanel.Gene_RVDocSumhttp://www.ncbi.nlm.nih.gov/gene/84062?ordinalpos=3&itool=EntrezSystem2.PEntrez.Gene.Gene_ResultsPanel.Gene_RVDocSumhttp://www.ncbi.nlm.nih.gov/gene/84062?ordinalpos=3&itool=EntrezSystem2.PEntrez.Gene.Gene_ResultsPanel.Gene_RVDocSumhttp://www.ncbi.nlm.nih.gov/gene/2571?ordinalpos=3&itool=EntrezSystem2.PEntrez.Gene.Gene_ResultsPanel.Gene_RVDocSumhttp://www.ncbi.nlm.nih.gov/gene/112400?ordinalpos=2&itool=EntrezSystem2.PEntrez.Gene.Gene_ResultsPanel.Gene_RVDocSumhttp://www.ncbi.nlm.nih.gov/gene/1815?ordinalpos=2&itool=EntrezSystem2.PEntrez.Gene.Gene_ResultsPanel.Gene_RVDocSumhttp://www.ncbi.nlm.nih.gov/gene/6531?ordinalpos=1&itool=EntrezSystem2.PEntrez.Gene.Gene_ResultsPanel.Gene_RVDocSumhttp://www.ncbi.nlm.nih.gov/gene/6531?ordinalpos=1&itool=EntrezSystem2.PEntrez.Gene.Gene_ResultsPanel.Gene_RVDocSumhttp://www.ncbi.nlm.nih.gov/gene/1815?ordinalpos=2&itool=EntrezSystem2.PEntrez.Gene.Gene_ResultsPanel.Gene_RVDocSumhttp://www.ncbi.nlm.nih.gov/gene/112400?ordinalpos=2&itool=EntrezSystem2.PEntrez.Gene.Gene_ResultsPanel.Gene_RVDocSumhttp://www.ncbi.nlm.nih.gov/gene/2571?ordinalpos=3&itool=EntrezSystem2.PEntrez.Gene.Gene_ResultsPanel.Gene_RVDocSumhttp://www.ncbi.nlm.nih.gov/gene/84062?ordinalpos=3&itool=EntrezSystem2.PEntrez.Gene.Gene_ResultsPanel.Gene_RVDocSumhttp://www.ncbi.nlm.nih.gov/gene/84062?ordinalpos=3&itool=EntrezSystem2.PEntrez.Gene.Gene_ResultsPanel.Gene_RVDocSumhttp://www.ncbi.nlm.nih.gov/gene/1139?ordinalpos=1&itool=EntrezSystem2.PEntrez.Gene.Gene_ResultsPanel.Gene_RVDocSum
  • 7/25/2019 uhlhaas2010abnormal.pdf

    12/14

    1. Phillips, W. A. & Silverstein, S. M. Convergence ofbiological and psychological perspectives on cognitivecoordination in schizophrenia. Behav. Brain Sci. 26 ,6582 (2003).

    2. Friston, K. J. Schizophrenia and the disconnectionhypothesis. Acta Psychiatr. Scand. Suppl. 395 , 6879(1999).

    3. Buzsaki, G. & Draguhn, A. Neuronal oscillations incortical networks. Science 304 , 19261929 (2004).

    4. Fries, P. Neuronal gamma-band synchronization as afundamental process in cortical computation. Annu.Rev. Neurosci. 32 , 209224 (2009).

    5. Buzsaki, G. Rhythms of the Brain (Oxford Univ. Press,New York, 2006).

    A brilliant and comprehensive overview of the roleof neural oscillations in cortical processing.

    6. Singer, W. Neuronal synchrony: a versatile code for thedefinition of relations? Neuron 24 , 4965, 111125(1999).

    7. Gray, C. M., Konig, P., Engel, A. K. & Singer, W.Oscillatory responses in cat visual cortex exhibit inter-columnar synchronization which reflects globalstimulus properties. Nature 338 , 334337 (1989).The first study to show a close correspondencebetween synchronized gamma-band oscillationsand Gestalt properties of visual stimuli in theprimary visual cortex of anaesthetized cats.

    8. Womelsdorf, T. et al. Modulation of neuronalinteractions through neuronal synchronization.Science 316 , 16091612 (2007).

    9. von Stein, A., Chiang, C. & Konig, P. Top-downprocessing mediated by interareal synchronization.Proc. Natl Acad. Sci. USA 97 , 1474814753 (2000).

    10. Uhlhaas, P. J. et al. Neural synchrony in corticalnetworks: history, concept and current status. Front.Integr. Neurosci. 3 , 17 (2009).

    11. Park, S. & Holzman, P. S. Schizophrenics show spatialworking memory deficits. Arch. Gen. Psychiatry 49 ,975982 (1992).

    12. Frith, C. D. Neuropsychology of Schizophrenia (Taylor

    and Francis, Hove, 1992).13. Uhlhaas, P. J. & Silverstein, S. M. Perceptual

    organization in schizophrenia spectrum disorders:empirical research and theoretical implications.Psychol. Bull. 131 , 618632 (2005).

    14. Green, M. F. What are the functional consequences ofneurocognitive deficits in schizophrenia? Am.

    J. Psychiatry 153 , 321330 (1996).15. Singer, W. & Gray, C. M. Visual feature integration and

    the temporal correlation hypothesis. Annu. Rev.Neurosci. 18 , 555586 (1995).

    16. Wespatat, V., Tennigkeit, F. & Singer, W. Phasesensitivity of synaptic modifications in oscillating cellsof rat visual cortex. J. Neurosci. 24 , 90679075(2004).

    17. Pavlides, C., Greenstein, Y. J., Grudman, M. & Winson, J.Long-term potentiation in the dentate gyrus is inducedpreferentially on the positive phase of theta-rhythm.Brain Res. 439 , 383387 (1988).

    18. Huerta, P. T. & Lisman, J . E. Heightened synapticplasticity of hippocampal CA1 neurons during acholinergically induced rhythmic state. Nature 364 ,723725 (1993).

    19. Daskalakis, Z. J., Christensen, B. K., Fitzgerald, P. B. &Chen, R. Dysfunctional neural plasticity in patientswith schizophrenia. Arch. Gen. Psychiatry 65 ,378385 (2008).

    20. Shelley, A. M. et al. Mismatch negativity: an index of apreattentive processing deficit in schizophrenia. Biol.Psychiatry 30 , 10591062 (1991).

    21. Stephan, K. E., Friston, K. J. & Frith, C. D.Dysconnection in schizophrenia: from abnormalsynaptic plasticity to failures of self-monitoring.Schizophr. Bull. 35 , 509527 (2009).

    22. Light, G. A. et al. Gamma band oscillations revealneural network cortical coherence dysfunction inschizophrenia patients. Biol. Psychiatry 60 ,12311240 (2006).

    23. Vierling-Claassen, D., Siekmeier, P., Stufflebeam, S. &Kopell, N. Modeling GABA alterations inschizophrenia: a link between impaired inhibitionand altered gamma and beta range auditoryentrainment. J. Neurophysiol. 99 , 26562671(2008).

    24. Kwon, J. S. et al. Gamma frequency-rangeabnormalities to auditory stimulation inschizophrenia. Arch. Gen. Psychiatry 56 , 10011005(1999).The first demonstration of a deficit in theentrainment of gamma-band oscillations to clicktrains presented at 40 Hz in the auditory cortex inpatients with schizophrenia.

    25. Krishnan, G. P. et al. Steady state and inducedauditory gamma deficits in schizophrenia. Neuroimage 47 , 17111719 (2009).

    26. Brenner, C. A., Sporns, O., Lysaker, P. H. & ODonnell,B. F. EEG synchronization to modulated auditory tonesin schizophrenia, schizoaffective disorder, andschizotypal personality disorder. Am. J. Psychiatry 160 , 22382240 (2003).

    27. Spencer, K. M., Salisbury, D. F., Shenton, M. E. &McCarley, R. W. Gamma-band auditory steady-stateresponses are impaired in first episode psychosis. Biol.Psychiatry 64 , 369375 (2008).

    28. Spencer, K. M., Niznikiewicz, M. A., Nestor, P. G.,Shenton, M. E. & McCarley, R. W. Left auditory cortexgamma synchronization and auditory hallucinationsymptoms in schizophrenia. BMC Neurosci. 10 , 85(2009).

    29. Wilson, T. W. et al. Cortical gamma generators suggestabnormal auditory circuitry in early-onset psychosis.Cereb. Cortex 18 , 371378 (2008).

    30. Hong, L. E. et al. Evoked gamma band synchronizationand the liability for schizophrenia. Schizophr. Res. 70 ,293302 (2004).

    31. Krishnan, G. P. et al. Steady state visual evokedpotential abnormalities in schizophrenia. Clin.Neurophysiol. 116 , 614624 (2005).

    32. Galambos, R., Makeig, S. & Talmachoff, P. J. A 40-Hzauditory potential recorded from the human scalp.Proc. Natl Acad. Sci. USA 78 , 26432647 (1981).

    33. Ross, B., Herdman, A. T. & Pantev, C. Stimulus induceddesynchronization of human auditory 40-Hz steady-state responses. J. Neurophysiol. 94 , 40824093(2005).

    34. Ross, B., Picton, T. W. & Pantev, C. Temporalintegration in the human auditory cortex asrepresented by the development of the steady-statemagnetic field. Hear. Res. 165 , 6884 (2002).

    35. Skosnik, P. D., Krishnan, G. P. & ODonnell, B. F. Theeffect of selective attention on the gamma-bandauditory steady-state response. Neurosci. Lett. 420 ,223228 (2007).

    36. Johannesen, J. K., Bodkins, M., ODonnell, B. F.,Shekhar, A. & Hetrick, W. P. Perceptual anomalies inschizophrenia co-occur with selective impairments inthe gamma frequency component of midlatencyauditory ERPs. J. Abnorm. Psychol. 117 , 106118(2008).

    37. Spencer, K. M. et al. Neural synchrony indexesdisordered perception and cognition in schizophrenia.Proc. Natl. Acad. Sci. USA 101 , 1728817293 (2004).

    38. Hirano, S. et al. Abnormal neural oscillatory activity tospeech sounds in schizophrenia: amagnetoencephalography study. J. Neurosci. 28 ,48974903 (2008).

    39. Roach, B. J. & Mathalon, D. H. Event-related EEGtime-frequency analysis: an overview of measures andan analysis of early gamma band phase locking inschizophrenia. Schizophr. Bull. 34 , 907926 (2008).

    40. Spencer, K. M., Niznikiewicz, M. A., Shenton, M. E. &McCarley, R. W. Sensory-evoked gamma oscillations inchronic schizophrenia. Biol. Psychiatry 63 , 744747(2008).

    41. Gallinat, J., Winterer, G., Herrmann, C. S. &Senkowski, D. Reduced oscillatory gamma-bandresponses in unmedicated schizophrenic patientsindicate impaired frontal network processing. Clin.Neurophysiol. 115 , 18631874 (2004).

    42. Ferrarelli, F. et al. Reduced evoked gamma oscillationsin the frontal cortex in schizophrenia patients: a TMS/EEG study. Am. J. Psychiatry 165 , 9961005(2008).

    43. Tillmann, C. et al. Source localization of high-frequencyoscillations reveals widespread reductions in gamma-band activity during perceptual organisation in chronicand first-episode schizophrenia. Soc. Neurosci. Abstr. 54.2 (2008).

    44. Haenschel, C. et al. Cortical oscillatory activity is criticalfor working memory as revealed by deficits in earlyonset schizophrenia J. Neurosci. 29 , 94819489(2009).

    45. Schmiedt, C., Brand, A., Hildebrandt, H. & Basar-Eroglu, C. Event-related theta oscillations duringworking memory tasks in patients with schizophreniaand healthy controls. Brain Res. Cogn. Brain Res. 25 ,936947 (2005).

    in participants with autism spectrum disorders extendedto low gamma-band oscillations (3060 Hz), which wereintact in patients with schizophrenia.

    These data highlight the need to carefully character-ize neural oscillations in specific disorders across differentspatial and temporal scales. Further research into neuraloscillations should also take into account the possibilitythat the impairments in high-frequency oscillations arerelated to alterations in low-frequency activity, in partic-ular in the theta-frequency range, which have been lessexplored so far. Neural oscillations across different fre-quencies form a hierarchical system in which dif ferentfrequencies interact by coupling the amplitude and phaseof ongoing oscillations 133. For example, the amplitude ofgamma-band oscillations is tightly linked to the phaseof theta oscillations 134, and this has been proposed toprovide a general coding scheme in cortical networks.

    Although greater understanding of these aspectsof neural oscillations is needed to establish a complete

    picture of the nature of brain functions and theirimpairment in schizophrenia, we firmly believe thatneural oscillations are a crucial factor in the patho-physiology of schizophrenia and that further investi-gation will eventually lead to an understanding of thesyndrome as a disorder of temporal coordination andto evidenced-based pharmacological interventions thatcorrect these alterations. This reconceptualization ofschizophrenia is part of a wider paradigm shift in theneurosciences in which the brain is increasingly viewedas a self-organizing system with complex and nonlineardynamics in which cognitive processes arise out of thedynamic interaction between multiple brain regions 135.Thus, understanding the mechanisms that give rise tothis enigmatic disorder and that have now been elu-sive for over a century may also provide importantinsights into the biological mechanisms that underliethe mental processes that are fundamentally altered inschizophrenia.

    R E V I E W S

    NATURE REVIEWS | NEUROSCIENCE VOLUME 11 | FEBRUARY 2010 | 111

    20 Macmillan Publishers Limited. All rights reserved10

  • 7/25/2019 uhlhaas2010abnormal.pdf

    13/14

    46. Cho, R. Y., Konecky, R. O. & Carter, C. S. Impairmentsin frontal cortical gamma synchrony and cognitivecontrol in schizophrenia. Proc. Natl Acad. Sci. USA 103 , 1987819883 (2006).

    47. Winterer, G. et al. Schizophrenia: reducedsignal-to-noise ratio and impaired phase-lockingduring information processing. Clin. Neurophysiol. 111 , 837849 (2000).

    48. Varela, F., Lachaux, J. P., Rodriguez, E. & Martinerie,J. The brainweb: phase synchronization and large-scale integration. Nature Rev. Neurosci. 2 , 229239

    (2001).49. Uhlhaas, P. J. et al. Dysfunctional long-rangecoordination of neural activity during Gestalt perceptionin schizophrenia. J. Neurosci. 26 , 81688175 (2006).

    50. Spencer, K. M. et al. Abnormal neural synchrony inschizophrenia. J. Neurosci. 23 , 74077411 (2003).

    51. Symond, M. P., Harris, A. W., Gordon, E. & Williams,L. M. Gamma synchrony in first-episodeschizophrenia: a disorder of temporal connectivity?

    Am. J. Psychiatry 162 , 459465 (2005).52. Boutros, N. N. et al. The status of spectral EEG

    abnormality as a diagnostic test for schizophrenia.Schizophr. Res. 99 , 225237 (2008).

    53. Rutter, L. et al. Magnetoencephalographic gammapower reduction in patients with schizophrenia duringresting condition. Hum. Brain Mapp. 30 , 32543264(2009).

    54. Koenig, T. et al. Decreased functional connectivity ofEEG theta-frequency activity in first-episode, neuroleptic-naive patients with schizophrenia: preliminary results.Schizophr. Res. 50 , 5560 (2001).

    55. Linkenkaer-Hansen, K. et al. Genetic contributions tolong-range temporal correlations in ongoingoscillations. J. Neurosci. 27 , 1388213889 (2007).

    56. Hong, L. E. et al. Sensory gating endophenotypebased on its neural oscillatory pattern and heritabilityestimate. Arch. Gen. Psychiatry 65 , 10081016(2008).

    An important study that examined auditorysensory gating in patients with schizophrenia andtheir first-degree relatives, highlighting the utilityof neural oscillations as an endophenotype inschizophrenia research.

    57. Lee, K. H., Williams, L. M., Haig, A. & Gordon, E.Gamma (40 Hz) phase synchronicity and symptomdimensions in schizophrenia. Cogn. Neuropsychiatry 8 , 5771 (2003).

    58. Lee, S. H. et al. Quantitative EEG and low resolutionelectromagnetic tomography (LORETA) imaging ofpatients with persistent auditory hallucinations.Schizophr. Res. 83 , 111119 (2006).

    59. Dierks, T. et al. Activation of Heschls gyrus duringauditory hallucinations. Neuron 22 , 615621 (1999).

    60. Ford, J. M. & Mathalon, D. H. Corollary dischargedysfunction in schizophrenia: can it explain auditoryhallucinations? Int. J. Psychophysiol. 58 , 179189(2005).

    61. Feinberg, I. Efference copy and corollary discharge:implications for thinking and its disorders. Schizophr.Bull. 4 , 636640 (1978).

    62. Ford, J. M., Mathalon, D. H., Whitfield, S., Faustman,W. O. & Roth, W. T. Reduced communication betweenfrontal and temporal lobes during talking inschizophrenia. Biol. Psychiatry 51 , 485492 (2002).

    63. Ford, J. M., Roach, B. J., Faustman, W. O. & Mathalon,D. H. Out-of-synch and out-of-sorts: dysfunction ofmotor-sensory communication in schizophrenia. Biol.Psychiatry 63 , 736743 (2007).

    64. Gross, J., Schnitzler, A., Timmermann, L. & Ploner, M.Gamma oscillations in human primary somatosensorycortex reflect pain perception. PLoS Biol. 5 , e133

    (2007).65. Shenton, M. E., Dickey, C. C., Frumin, M. & McCarley,

    R. W. A review of MRI findings in schizophrenia.Schizophr. Res. 49 , 152 (2001).

    66. Selemon, L. D. & Goldman-Rakic, P. S. The reducedneuropil hypothesis: a circuit based model ofschizophrenia. Biol. Psychiatry 45 , 1725 (1999).

    67. Onitsuka, T. et al. Functional and structural deficits inbrain regions subserving face perception inschizophrenia. Am. J. Psychiatry 163 , 455462(2006).

    68. McCarley, R. W. et al. Association between smaller leftposterior superior temporal gyrus volume on magneticresonance imaging and smaller left temporal P300amplitude in first-episode schizophrenia. Arch. Gen.Psychiatry 59 , 321331 (2002).

    69. Salisbury, D. F., Kuroki, N., Kasai, K., Shenton, M. E. &McCarley, R. W. Progressive and interrelatedfunctional and structural evidence of post-onset brain

    reduction in schizophrenia. Arch. Gen. Psychiatry 64 ,521529 (2007).

    70. Engel, A. K., Konig, P., Kreiter, A. K. & Singer, W.Interhemispheric synchronization of oscillatoryneuronal responses in cat visual cortex. Science 252 ,11771179 (1991).

    71. Lwel, S. & Singer, W. Selection of intrinsic horizontalconnections in the visual cortex by correlated neuronalactivity. Science 255 , 209212 (1992).

    72. Kubicki, M. et al. A review of diffusion tensor imagingstudies in schizophrenia. J. Psychiatr. Res. 41 , 1530

    (2007).73. Rotarska-Jagiela, A. et al. The corpus callosum inschizophrenia-volume and connectivity changes affectspecific regions. Neuroimage 39 , 15221532 (2008).

    74. Lim, K. O.et al. Compromised white matter tractintegrity in schizophrenia inferred from diffusiontensor imaging. Arch. Gen. Psychiatry 56 , 367374(1999).

    75. Cobb, S. R., Buhl, E. H., Halasy, K., Paulsen, O. &Somogyi, P. Synchronization of neuronal activity inhippocampus by individual GABAergic interneurons.Nature 378 , 7578 (1995).

    76. Wang, X. J. & Buzsaki, G. Gamma oscillation bysynaptic inhibition in a hippocampal interneuronalnetwork model. J. Neurosci. 16 , 64026413 (1996).

    77. Sohal, V. S., Zhang, F., Yizhar, O. & Deisseroth, K.Parvalbumin neurons and gamma rhythms enhancecortical circuit performance. Nature 459 , 698702(2009).

    An elegant demonstration of the functionalsignificance of parvalbumin-containing interneuronsin the generation of gamma-band oscillations thatused a novel combination of optogenetictechnologies in mice to selectively modulatemultiple distinct circuit elements in the neocortex.

    78. Lewis, D. A., Hashimoto, T. & Volk, D. W. Corticalinhibitory neurons and schizophrenia. Nature Rev.Neurosci. 6 , 312324 (2005).

    An excellent review that highlights the roleof inhibitory neurons in the pathophysiology ofschizophrenia.

    79. Akbarian, S. et al. Gene expression for glutamic aciddecarboxylase is reduced without loss of neurons inprefrontal cortex of schizophrenics. Arch. Gen.Psychiatry 52 , 258266 (1995).

    80. Volk, D. W., Austin, M. C., Pierri, J. N., Sampson, A. R.& Lewis, D. A. Decreased glutamic aciddecarboxylase67 messenger RNA expression in asubset of prefrontal cortical gamma-aminobutyric acidneurons in subjects with schizophrenia. Arch. Gen.Psychiatry 57 , 237245 (2000).

    81. Volk, D. W., Austin, M. C., Pierri, J. N., Sampson, A. R.& Lewis, D. A. GABA transporter-1 mRNA in theprefrontal cortex in schizophrenia: decreasedexpression in a subset of neurons. Am. J. Psychiatry 158 , 256265 (2001).

    82. Hashimoto, T. et al. Gene expression deficits in asubclass of GABA neurons in the prefrontal cortex ofsubjects with schizophrenia. J. Neurosci. 23 ,63156326 (2003).

    83. Lodge, D. J., Behrens, M. M. & Grace, A. A. A loss ofparvalbumin-containing interneurons is associatedwith diminished oscillatory activity in an animal modelof schizophrenia. J. Neurosci. 29 , 23442354(2009).

    84. Cunningham, M. O. et al. Region-specific reduction inentorhinal gamma oscillations and parvalbumin-immunoreactive neurons in animal models ofpsychiatric illness. J. Neurosci. 26 , 27672776(2006).

    85. Kinney, J. W. et al. A specific role for NR2A-containing

    NMDA receptors in the maintenance of parvalbuminand GAD67 immunoreactivity in culturedinterneurons. J. Neurosci. 26 , 16041615 (2006).

    86. Krystal, J. H. et al. Subanesthetic effects of thenoncompetitive NMDA antagonist, ketamine, inhumans. Psychotomimetic, perceptual, cognitive, andneuroendocrine responses. Arch. Gen. Psychiatry 51 ,199214 (1994).

    87. Zhang, Y., Behrens, M. M. & Lisman, J. E. Prolongedexposure to NMDAR antagonist suppresses inhibitorysynaptic transmission in prefrontal cortex.

    J. Neurophysiol. 2 , 959965 (2008).88. Behrens, M. M. et al. Ketamine-induced loss of

    phenotype of fast-spiking interneurons is mediated byNADPH-oxidase. Science 318 , 16451647 (2007).

    89. Do, K. Q., Cabungcal, J. H., Frank, A., Steullet, P. &Cuenod, M. Redox dysregulation, neurodevelopment,and schizophrenia. Curr. Opin. Neurobiol. 19 ,220230 (2009).

    90. Roopun, A. K. et al. Region-specific changes in gammaand beta2 rhythms in NDMA receptor dysfunctionmodels of schizophrenia. Schizophr. Bull. 34 ,962973 (2008).

    91. Plourde, G., Baribeau, J. & Bonhomme, V. Ketamineincreases the amplitude of the 40-Hz auditory steady-state response in humans. Br. J. Anaesth. 78 ,524529 (1997).

    92. Pinault, D. N -methyl D -aspartate receptor antagonistsketamine and MK-801 induce wake-related aberrantgamma oscillations in the rat neocortex. Biol.

    Psychiatry 63 , 730735 (2008).93. Homayoun, H. & Moghaddam, B. NMDA receptorhypofunction produces opposite effects on prefrontalcortex interneurons and pyramidal neurons.

    J. Neurosci. 27 , 1149611500 (2007).94. Lisman, J. E. et al. Circuit-based framework for

    understanding neurotransmitter and risk geneinteractions in schizophrenia. Trends Neurosci. 31 ,234242 (2008).

    95. Rodriguez, R., Kallenbach, U., Singer, W. & Munk,M. H. Short- and long-term effects of cholinergicmodulation on gamma oscillations and responsesynchronization in the visual cortex. J. Neurosci. 24 ,1036910378 (2004).

    96. Steriade, M., Dossi, R. C., Pare, D. & Oakson, G.Fast oscillations (2040 Hz) in thalamocorticalsystems and their potentiation by mesopontinecholinergic nuclei in the cat. Proc. Natl Acad. Sci. USA 88 , 43964400 (1991).

    97. Sarter, M., Nelson, C. L. & Bruno, J. P. Corticalcholinergic transmission and cortical informationprocessing in schizophrenia. Schizophr. Bull. 31 ,117138 (2005).

    98. Krenz, I. et al. Parvalbumin-containing interneurons ofthe human cerebral cortex express nicotinicacetylcholine receptor proteins. J. Chem. Neuroanat. 21 , 239246 (2001).

    99. Martin, L. F. & Freedman, R. Schizophrenia and the 7nicotinic acetylcholine receptor. Int. Rev. Neurobiol. 78 , 225246 (2007).

    100. Guan, Z. Z., Zhang, X., Blennow, K. & Nordberg, A.Decreased protein level of nicotinic receptor 7subunit in the frontal cortex from schizophrenic brain.Neuroreport 10 , 17791782 (1999).

    101. Brown, P. et al. Dopamine dependency of oscillationsbetween subthalamic nucleus and pallidum inParkinsons disease. J. Neurosci. 21 , 10331038(2001).

    102. Ito, H. T. & Schuman, E. M. Frequency-dependentgating of synaptic transmission and plasticity bydopamine. Front. Neural Circuits 1 , 1 (2007).

    103. Lewis, D. A. & Levitt, P. Schizophrenia as a disorder ofneurodevelopment. Annu. Rev. Neurosci. 25 ,409432 (2002).

    104. Walker, E. F., Savoie, T. & Davis, D. Neuromotorprecursors of schizophrenia. Schizophr. Bull. 20 ,441451 (1994).

    105. Feinberg, I. Schizophrenia: caused by a fault inprogrammed synaptic elimination duringadolescence? J. Psychiatr. Res. 17 , 319334(1982).

    106. Khazipov, R. & Luhmann, H. J. Early patterns ofelectrical activity in the developing cerebral cortex ofhumans and rodents. Trends Neurosci. 29 , 414418(2006).

    107. Ben-Ari, Y. Developing networks play a similar melody.Trends Neurosci. 24 , 353360 (2001).

    108. Katz, L. C. & Shatz, C. J. Synaptic activity and theconstruction of cortical circuits. Science 274 ,11331138 (1996).

    109. Khazipov, R. et al. Early motor activity drives spindle

    bursts in the developing somatosensory cortex.Nature 432 , 758761 (2004).

    110. Hanganu, I.L., Ben-Ari, Y. & Khazipor, R. Retinal wavestrigger spindle bursts in the neonatal rat visual cortex.

    J. Neurosci . 26 , 67286736 (2006)111. Chiu, C. & Weliky, M. Spontaneous activity in

    developing ferret visual cortex in vivo . J. Neurosci. 21 ,89068914 (2001).

    112. Stellwagen, D. & Shatz, C. J. An instructive role forretinal waves in the development of retinogeniculateconnectivity. Neuron 33 , 357367 (2002).

    113. Cang, J. et al. Development of precise maps in visualcortex requires patterned spontaneous activity in theretina. Neuron 48 , 797809 (2005).

    114. Yang, J. W., Hanganu-Opatz, I. L., Sun, J. J. &Luhmann, H. J. Three patterns of oscillatory activitydifferentially synchronize developing neocorticalnetworks in vivo . J. Neurosci. 29 , 90119025(2009).

    R E V I E W S

    112 | FEBRUARY 2010 | VOLUME 11 www.nature.com/reviews/neuro

    20 Macmillan Publishers Limited. All rights reserved10

  • 7/25/2019 uhlhaas2010abnormal.pdf

    14/14

    115. Hebb, D. O. The Organization of Behavior: ANeuropsychological Theory (Wiley, New York, 1949).

    116. Markram, H., Lubke, J., Frotscher, M. & Sakmann, B.Regulation of synaptic efficacy by coincidence ofpostsynaptic APs and EPSPs. Science 275 , 213215(1997).

    An important paper showing that modifications insynaptic connections between neurons in theneocortex are dependent on the precise temporalrelation between postsynaptic action potentialsand unitary excitatory postsynaptic potentials.

    117. Uhlhaas, P. J. et al. The development of neuralsynchrony reflects late maturation and restructuringfunctional networks in humans. Proc. Natl Acad. Sci.USA 106 , 98669871 (2009).The first study to comprehensively examine thedevelopment of task-related neural synchrony inhumans, highlighting the role of late brainmaturation for the shaping and reorganization offunctional networks.

    118. Yakovlev, P. I.