positive and negative symptoms in schizophrenia: the nmda receptor hypofunction hypothesis,...

http://anp.sagepub.com/Psychiatry

Australian and New Zealand Journal of

http://anp.sagepub.com/content/43/8/711The online version of this article can be found at:

DOI: 10.1080/00048670903001943

2009 43: 711Aust N Z J PsychiatryMaxwell Bennett

Neuregulin/ErbB4 and Synapse RegressionPositive and Negative Symptoms in Schizophrenia: The NMDA Receptor Hypofunction Hypothesis,

Published by:

http://www.sagepublications.com

On behalf of:

The Royal Australian and New Zealand College of Psychiatrists

can be found at:Australian and New Zealand Journal of PsychiatryAdditional services and information for

http://anp.sagepub.com/cgi/alertsEmail Alerts:

http://anp.sagepub.com/subscriptionsSubscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

What is This?

- Aug 1, 2009Version of Record >>

at TEXAS SOUTHERN UNIVERSITY on October 28, 2014anp.sagepub.comDownloaded from at TEXAS SOUTHERN UNIVERSITY on October 28, 2014anp.sagepub.comDownloaded from

http://anp.sagepub.com/

http://anp.sagepub.com/content/43/8/711

http://www.sagepublications.com

http://www.ranzcp.org

http://anp.sagepub.com/cgi/alerts

http://anp.sagepub.com/subscriptions

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsPermissions.nav

http://anp.sagepub.com/content/43/8/711.full.pdf

http://online.sagepub.com/site/sphelp/vorhelp.xhtml



Positive and negative symptoms in schizophrenia:the NMDA receptor hypofunction hypothesis,neuregulin/ErbB4 and synapse regression

Maxwell Bennett

Carlsson has put forward the hypothesis that the positive and negative symptoms ofschizophrenia are due to failure of mesolimbic and mesocortical projections consequent onhypofunction of the glutamate N-methyl-d-aspartate (NMDA) receptor. The hypothesis hasbeen recently emphasized in this Journal that the loss of synaptic spines with NMDAreceptors, which can be precipitated by stress, can explain the emergence of positivesymptoms such as hallucinations and that this synapse regression involves moleculessuch as neuregulin and its receptor ErbB4 that have been implicated in schizophrenia. Inthis essay these two hypotheses are brought together in a single scheme in whichemphasis is placed on the molecular pathways from neuregulin/ErbB4, to modulation of theNMDA receptors, subsequent changes in the synaptic spine’s cytoskeletal apparatus andso regression of the spines. It is suggested that identification of the molecular constituentsof this pathway will allow synthesis of suitable substances for removing the hypofunction ofNMDA receptors and so the phenotypic consequences that flow from this hypofunction.Key words: Dopamine, ErbB4, F-actin, neuregulin, NMDA, schizophrenia, synapse loss.

Australian and New Zealand Journal of Psychiatry 2009; 43:711�721

Carlsson has put forward the hypothesis that the

basic nervous system failure in schizophrenia is

hypofunction of the glutamate receptor, N-methyl-

d-aspartate (NMDA) [1,2]. This hypothesis is attrac-

tive because it integrates failure of neural networks

involving the prefrontal cortex, the ventral tegmental

area, the nucleus accumbens and the thalamus with

the positive and negative symptoms in schizophrenia,

as well as with changes in dopamine and serotonin

in these structures. One of the clearest concomitants

of schizophrenia, however, namely the regression of

synapses [3], with resulting positive symptoms such as

hallucinations [4], has not yet been integrated into the

NMDA hypofunction hypothesis. In the present

essay the efficacy of NMDA receptors is shown to

determine the integrity of synaptic spines within

the neural networks. It is suggested that it is the

loss of these synaptic spines that leads to the failure

of the neural networks responsible for the positive

and negative symptoms in schizophrenia. The cytos-

keletal system within the spines determines their

growth, stability and regression. This system is under

the control of a range of membrane receptors, in

particular NMDA receptors. Attention is given to

those receptors that determine NMDA receptor

efficacy, such as the ErbB4 receptors. Consideration

is then given to how failure of the ErbB4 receptors is

responsible for NMDA receptor decline, with con-

comitant regression of synaptic spines resulting in

changes in neural network function that underlie the

positive and negative symptoms of schizophrenia. It

is suggested that fast through-put bioassays are now

required to assess the effects of substances that will

Maxwell Bennett AO, Professor of Neuroscience; University Chair;Scientific Director

Brain and Mind Research Institute, University of Sydney, Camperdown,NSW 2050, Australia. Email: [email protected]

Received 23 December 2008; accepted 15 April 2009.

# 2009 The Royal Australian and New Zealand College of Psychiatrists at TEXAS SOUTHERN UNIVERSITY on October 28, 2014anp.sagepub.comDownloaded from


restore the functioning of the NMDA receptor andtherefore re-establish synaptic spines.

NMDA receptor control of synaptic spineformation and regression

The main mechanisms of spine enlargement involveincreasing F-actin polymerization at the apex ofthe spine head (Figure 1). The two principal waysof achieving this is to either increase Arp2/3 and soF-actin proliferation and branching, or to inhibitadenosine diphosphate (ADF)/cofilin and so preventdepolymerization of F-actin while allowing prolifera-tion to proceed. Increases in profilin will also assistspine growth through accelerating adenosine dipho-sphate (ADP)-profilin to adenosine-5?-triphosphate(ATP)-profilin, so making the G-actin available forattachment at the ends of F-actin at the apex of thespine. The principal means of obtaining spine short-ening is to activate myosin in the spine, so increasingthe retrograde movement of F-actin from the apex ofthe spine head.

The principal receptors at the plasmalemma thatactivate signalling cascades that lead to changes inF-actin treadmilling and spine shape are neurotrans-mitter receptors as well as voltage-dependent calcium-ion channels on the one hand, and surface membranereceptors that activate intracellular signalling cascadesthat control activity of Rho GTPases on the other(Figure 2) [7]. The main transmitter-activated calcium-permeable channels are the NMDA receptors, whichare modulated by ErbB4 receptors, and the mainreceptors controlling the RhoGTPases are ephrin [8],brain-derived neurotrophic factor (BDNF) [9] andplexin (Figure 3).

NMDA receptor activation by glutamate leads tothe largest calcium influx that any receptor cangenerate. This calcium can in turn activate calcium-calmodulin kinase (CaMKII), which in turn phos-phorylates and activates calmodulin-dependent kinasekinase (CaMKK�CaMK1). These kinases form amultimolecular complex with the guanine-nucleotideexchange factor bPIX, as shown in Figure 3 [10,11].Phosphorylation of this results in activation of Rac1and Pak1. Pak1-mediated phosphorylation then acti-vates LIM kinase (LIMK), which inactivates actin-depolymerizing protein ADP/cofilin, so promotingstabilization of F-actin and giving spine enlargement(Figure 3).

In hippocampal neurons an actin-regulated path-way is directed by Rho small GTPases, such as RhoA [12]. Rho A recruits and activates its specific Rho-

associated kinase (ROCK), which in turn complexeswith profilin Iia, so modifying actin stability througha RhoA/ROCK/profiling IIa pathway (Figure 3) [13].It is known that RhoA associates with ionotropicglutamate receptors (iGluRs) and metabotropic glu-tamate receptors (mGluRs) at the plasma membraneof synaptic spines, with activation of the iGluRsleading to detachment of RhoA from these receptorsand their recruitment to the mGluRs. This triggerslocal reduction of RhoA activity at the iGluRs,resulting in inactivation of RhoA-specific kinaseROCK there and so disruption of the ROCK-profilinIIa, producing rapid changes in the underlying actincytoskeleton and remodelling of spine shape [14].

Activation of the NMDA receptors can also lead toincreases in activity of LIMK, and hence inhibition ofADP/cofilin and spine enlargement, by the GTPaseRhoA through RhoA-specific kinase (ROCK; (Fig-ure 3). This pathway can also increase profilin andhence availability of G-ATP-actin for F-actin poly-merization. ROCK can also activate myosin regula-tory light chain (MLC) and simultaneously inhibitmyosin light chain phosphatase (MLCP), so leadingto myosin motor activity and increased retrogradeflow of F-actin [15] with concomitant spine short-ening. Collectively, glutamate stimulation of NMDAreceptors can, through both CaKM11 and RhoA,lead to inhibition of ADF/cofilin (with perhaps asmall enhancement from the phosphatase calcineurin)and increase profilin, which probably more thanoffsets in most instances the increased retrogradeflow of F-actin due to enhanced myosin motoractivity contingent on the increase in ROCK.

Neuregulin-ErbB4 receptor control of NMDAreceptor efficacy

Neuregulin (NRG) and its ErbB4 receptor havebeen implicated in schizophrenia. It is then of specialinterest to inquire into how the ErbB4 receptorregulates the NMDA receptor leading to its hypo-function with subsequent spine loss (Figure 3).

Neuregulin transphosporylation of receptor kinase

ErbB4 and recruitment of Src family kinases

The ErbB receptors are found in the plasmalemma(detergent-soluble) and membrane caveoli (detergent-insoluble), with translocation between these occur-ring on activation of the receptors with NRG [16].Thus NRG promotes both the translocation ofErbB4 to the caveoli as well as its heterodimerization

712 SYNAPSE REGRESSION AND SCHIZOPHRENIA

at TEXAS SOUTHERN UNIVERSITY on October 28, 2014anp.sagepub.comDownloaded from


Figure 1. The actin nucleation model of growth and regression of synaptic spines. (1) Receptors activate signallingpathways that lead to Rho GTPases (2). (3) These activate WASP proteins, which, in turn, in (4), lead toactivation of the Arp2/3 complex and the subsequent binding of adenosine-5?-triphosphate (ATP)-actin into thecomplex. (5) This leads to the formation of new actin filaments through attachment of the Wiskott�Aldrichsyndrome protein (WASP)-Arp2/3-ATP-actin complex to an existing filament to form a side branch. This newfilament grows as a consequence of being supplied from a high concentration of profilin bound actin (11), with theresult that the spine membrane is pushed out (6). (7) Capping proteins bind to the growing barbed ends of thefilaments, so terminating elongation or they can be protected from capping by plasmalemma-bound vasodilator-stimulated phosphoprotein (VAMP). Actin-polymerizing factor (ADF)-cofilin severs and depolymerizes the

adenosine diphosphate (ADP)-actin in the older regions of the filaments (8, 9). Profilin then promotes dissociationof ADP and binding of ATP to the dissociated subunits (10). (11) ATP-actin then binds profilin, making them

available for assembly into actin filaments. (Diagram and description after Figure 1 in [5]).

M. BENNETT 713



partner ErbB2, which is there tyrosine-phosphory-

lated to a level fourfold greater than ErbB2 in the

non-caveoli regions of the membrane. NRG binds

ErbB receptors to induce formation of ErbB4/ErbB2

heterodimers (Figure 4). This then activates intrinsic

kinases, which leads to the phosphorylation of

specific tyrosines located in the ErbB’s cytoplasmic

region. Such phosphorylated residues are docking

sites for non-receptor protein kinases (Src) family

signalling molecules, the recruitment of which stimu-

lates the intracellular signalling cascades leading

to upregulation of NMDA-receptor function [17]

(Figure 4).The crystal structure of ErbB4 kinase has

been described and shown, on activation by NRG, to

adopt an asymmetric dimer confirmation [18].NRGs, encoded by four different genes, namely

NRG-1, NRG-2, NRG-3 and NRG-4, are ligands for

ErbB4 and promote its autophosphorylation [19].

Receptor dimerization occurs following formation

of ligand�receptor complexes, so initiating signalling.

ErbB4 most frequently forms a heterodimer with

ErbB2. Autophosphorylation of ErbB receptors takes

place at 18 tyrosine residues within the carboxy-terminal region [20,21], of which approximately fiveare potential autophosphorylation sites (Figure 4).When phosphorylated, these conserved tyrosine re-sidues can function as docking sites for downstreamsignalling molecules (Figure 4). Different downstreammolecules are recruited, depending on the ErbB4ligand, because this determines the pattern of itsphosphorylation [22].

ErbB4 interacts directly with the first two PDZdomains of PSD-95 [23�25], through its carboxy-terminal sequence TVV.

NMDA receptor modulation by phosphorylation

through Src family kinases

Src and other Src family kinases, such as Fyn,regulate the activity of NMDA receptors, with Srcand Fyn phosphorylating and thereby upregulatingthe function of NMDA receptors in long-termpotentiation [26,27], increasing currents through thereceptor as well as intracellular calcium [28]. The Srcadaptor protein NADH dehydrogenase subunit 2(ND2) is required for Src to be anchored to theNMDA receptor (Figure 5) [6,29].

Activation of Src depends on the focal adhesionnon-receptor protein kinase CaKb (Figure 5) [30].Activation of CaKb is required for inducing long-term potentiation, and such activation is indirectlydependent on calcium, probably through stimulationof protein kinase C and perhaps calcium�calmodulin-dependent kinases protein (CaMKII) [31].

Activation of CaKb leads to its autophosphoryla-tion at Tyr-402 on the linker between the catalyticdomain (CD) and the band 4.1, Janus Kinase family,ERM protein family, focal adhesion kinase familydomain (band 4.1-JEF domain) as well as on Tyr-579/580 in the CD (Figure 5) [6]. Phosphorylation onTyr-402 creates an SH2 ligand through which CAKbbinds to the SH2 domain of Src and activates thiskinase by relieving autoinhibition [32]. When thisoccurs, autophosphorylation of Y416 in the CDresults in a conformational change of the activationloop, producing a fully active kinase. The SH2domain binds to peptide motifs that contain phos-phorylated tyrosine. The CD contains phosphory-lated tyrosine. The CD contains the tyrosine kinaseactivity that phosphorylates the NR2A subunit of theNMDA receptor in its carboxy (C)-terminal tail atleast at sites Y1292, Y1325 and Y1387 (Figure 5) [32].

Fyn is associated with PSD-95 that, like the adaptorprotein for Src, brings the kinase into close proximitywith the NMDA receptors (Figure 2) [6,33]. This leads

Figure 2. Upregulation of N-methyl-d-aspartate(NMDA) receptors by Src family kinase Fyn actingon the NR2B receptor subunit. Fyn, like Src, hasdomains UD (unique domain), SH3, SH2 and CD(catalytic domain) and also requires phosphorylation(P) in the CD loop in order to undergo a conforma-tional change of an activation loop in order to producea fully active Fyn kinase. Fyn is associated with post-synaptic density (PSD)-95 through its protein PDZ3domain and when the kinase is brought into closeproximity with the NR2B subunit of NMDA itphosphorylates it at Y1472, Y1336 and Y1252

(Fig. 4), leading to increased efficacy of the NMDAreceptor. (After Figure 1 in [6].)




to phosphorylation of the NR2B subunit ofN-methyl-D-aspartate receptor (NMDAR) at Y1472, Y1336and Y1252 [34]. During long-term potentiation (LTP)induction there is increased phosphorylation ofNR2bsubunit [35] and Fyn mutant mice have im-paired LTP that can be rescued by postnatal expres-sion of Fyn [36].

The kinetics of phosphorylation by the SH3domain of Fyn have been described [37], as has the

kinetics of phosphorylation by Src of the G protein-coupled receptor kinase GRK2 [38] as well as ofdynamin [39].

It has been noted here that Src facilitates NMDAreceptor function by increasing tyrosine phosphory-lation of the NR2A subunit [40], whereas Fyn doesso by phosphorylating the NR2B subunit (Figure 2,Figure 5) [32]. The Src-dependent regulationof NMDA receptor activity is regulated by an

Figure 3 (Continued)

M. BENNETT 715



interaction between C-terminal Src kinase (Csk)and the NMDA receptor, probably through directassociation of Csk with the Src-phosphorylated NR2subunits [41]. In this way Csk maintains constant theexcitatory synaptic transmission at impulse rates thatdo not evoke long-term potentiation by inhibitingcalcium increases of Src kinase-dependent increases.

Kinetics of Erb (EGFR) transphosporylation and of

subsequent Src family kinases

A detailed kinetic model of epidermal growthfactor receptor (EGFR)/ErbB signalling pathways,describing the activation, dimerization, and tyrosinephosphorylation of EGFR/ErbB, followed by thebinding and activation/phosphorylation of an adap-tor/target protein such as Src, has been developed byKholodenko et al. [42,43]. Two kinetic modules maybe considered here, one concerning the dimerizationof EGFR/ErbB on activation with its subsequentphosphorylation, and the other module the bindingand activation/phosphorylation of the target protein,

in this case Src. Considering first the second moduleconcerning the target protein: this is bound andphosphorylated by EGFR/ErbB through activatedEGFR/ErbB kinase and a coupled cycle of phos-phorylated and dephosphorylation of the targetprotein by specific tyrosine phosphatase. Other con-siderations of these kinetic schemes are given in[42,44�46].

Kinetics of Src family kinases (Src, Fyn) interaction

with NR2B subunit of the NMDA receptor

There have been no quantitative models of Src andFyn interaction and subsequent tyrosine phosphoryla-tion of the NMDA receptors, although such sites areidentified. Kinetic parameter constants have beendetermined for calcium/calmodulin-dependent proteinkinase (CaMKII) binding to specific residues onNR2B subunit [47]. An appropriate kinetic model,however, has probably been provided by [48]. Thisgives the following kinetic modules: dimerizationtransphosphorylation of ErbB to give the activated

Figure 3. Membrane receptors and voltage-gated ionic channel control of synaptic spine actin cytoskeleton and sosynapse formation and regression. The following receptor and ion channel pathways are indicated on the spines

plasmalemma (read counter-clockwise around the spine). (1) Sem 3A acting on plexin A receptors to decrease RAS,ERK and soWASP nucleation of G-actin around Arp2/3. (2) BDNF acting on TrkB receptors to activate RAS, ERKand so WASP nucleation of G-actin around Arp2/3. (3) NRG-1 on ErbB4/ErbB2 receptors, anchored to the PSD-95scaffolding protein and RhoGTPase to increase WASP as well as release of SFKs to act on the NR2B subunit of theNMDA receptor to phosphorylate it, so enhancing calcium influx through NMDA. (4) Glutamate acting on NMDAreceptors to enhance calcium entry into the cytosol and via CaMKK, CaMK2, BPIX, GIT1, RAC, PAK, LIMK1 toinhibit cofilin depolymerization of ADP-actin in the F-actin filaments. (5) Voltage-dependent calcium channel,

CAV1.3a, which is anchored by PDZ and SHANK in the cytosol and through HOMER releases calcium from calciumstores. (6) Dopamine acting on D2 dopamine receptors, which, through PP1, activates calcium calmodulin kinase 2 tomodulate CAV1.3a and phosphorylates the NR1 subunit of NMDA. (7) Dopamine acting on D2 dopamine receptors,which through Gp and PLCb, releases IP3 to activate calcium release from internal stores; this calcium activatespCREB. (8) Glutamate acting on NMDA receptors activates the RhoA, ROCK, profilin pathway to provide G-actinbound to profilin. (9) Ephrin binding to the EphB receptor activates Kalirin, which then acts on RAC1/PAK to exciteLIMK1 and so inhibits cofilin depolymerization of F-actin. BDNF, brain derived neurotrophic factor; BPIX, guaninenucleotide exchange factor for RAC; CaMK2, calcium calmodulin-dependent kinase 2; CaMKK, calcium calmodulin-dependent protein kinase kinase; cofilin, severs and depolymerizes ADP-actin; D2, dopamine D2 receptor; EphB,ephrin receptor; ErbB2, receptors for neuregulin; ErbB4, receptors for neuregulin; ERK, extracellular signal-regulatedkinases; GKAP, guanylate kinase-associated protein; Gp, G-protein; HOMER, scaffolding protein; IP3, inositoltriphosphate; kalirin, Rho guanine nucleotide exchange protein (GEF); LIMK, LIM kinase, phosphorylates ADF/cofilin; NMDA, N-methyl-d-aspartate; NR1, NR2A, NR2B, subunits of the NMDA receptor; NRG-1, neuregulin 1;PAK, downstream effector of RAC (sometimes called P21-activated kinase); pCREB, phosphorylated cyclic AMPresponse element-binding protein; PDZ, protein domain; PLCb, protein lipase Cb; plexin A, receptor for Sema 3A;PP1, protein phosphatase 1; profilin, actin regulatory molecule; PSD-95, post-synaptic density 95, a scaffoldingprotein; RAC, Rho-GTPase; RAS, Rho-GTPase; RhoA, Rho-GTPase; Rho-GTPase, Rho-family GTPases, a

subgroup of the superfamily of GTPases; ROCK, Rho-associated kinase; sema 3A, semaphorin 3A; SFK, src familykinase; SHANK, scaffolding molecule; TrkB, BDNF receptor; WASP, Wiskott�Aldrich syndrome protein that

triggers actin polymerization via Arp 2/3 complex.




receptor (Figure 5); subsequent phosphorylation (de-phosphorylation) of the target/adaptor protein givenby Src (or Fyn or other Src family kinase (SFK)); thenbinding of Shc (or Fyn or other SFK) to the NR2Bsubunit of the NMDA receptor (Figures 2,5).

Quantitative measures of ErbB4 control of NMDA

receptor activity

Early studies examined the acute effects of NRG-1application on NMDA receptors and reached theconclusions that NRG-1 inhibits synaptic plasticity[49�51] and reduces NMDA receptor activity andsynaptic transmission. These studies used tonic ap-plication of a functional domain, such as the epider-mal growth factor (EGF) domain, of particularisoforms of recombinant NRG-1 [49,51�53], but thefunction of endogenous NRG-1 at synapses may bedifferent. The location or duration or physiologicalconcentration of these functional domains is notknown [54], nor their action on different ErbBreceptors. Indeed, NRG-1-deficient mice show reduc-tion in NMDA receptor activity, compared with wild

type, rather than the opposite effects that would beexpected from the acute studies [55]. Recently, Liet al. have shown that preventing NRG-1/ErbB4signalling by using either ErbB4 knockout mice orErbB4 RNAi leads to loss of NMDA receptorcurrents in hippocampal neurons and this is accom-panied by a decrease in spine size and number [56].

If Fyn is the SFK under consideration, then theFyn kinase activity as a consequence of the NRG-1action on ErbB4 increases to a peak in approximately7 min, well after the ErbB4 kinase activity reaches amaximum at 1 min. The dose�response curve forNRG-1 increasing kinase activity of ErbB and of Fynindicates that they have similar ED50s but quitedifferent activation levels [57].

Changes in the ErbB4-NMDA receptor pathwayin schizophrenia

ErbB4 hyperphosphorylation in schizophrenia

In schizophrenia there is a large increase in activa-tion of ErbB4 in the prefrontal cortex by NRG-1,

Figure 4. ErbB-receptor tyrosine family members ErbB4 and ErbB2, after heterodimerization on activation byneuregulin, can phosphorylate N-methyl-d-aspartate (NMDA) receptor subunits NR2A and NR2B through

mediation by interaction partners Src and Fyn of the ErBB-receptor tyrosine residues. These upregulate the functionof NMDA receptors through Fyn acting at Y1252, Y1363 and Y1353 sites to phosphorylate NR2B and through Src

acting at Y1492, Y1353 and Y1387 to phosphorylate NR2A. Src and Fyn are family kinases.

M. BENNETT 717



with such hyperactivation traced to a substantial

increase in ErbB4�PSD-95 interaction [58]. Although

those authors reported that, in post-mortem brain

slices, NRG-1 stimulation suppressed NMDA recep-

tor activation in human prefrontal cortex, such

experimental observations are open to criticisms

concerning interpretation of the effects of exogenous

application of NRG, as recently presented by Li et al.

[56].PSD-95 is a multiprotein complex containing

various protein kinases with the PD21, PD22,

PD23, SH3 and GuK domains. The PSD-95 binds

SFKs such as Fyn, as well as ERbB4/ ErbB2 and the

NR2B subunit of the NMDA receptor, so enhancing

their interaction. It is claimed that in subjects withschizophrenia ErbB in this complex is hyperpho-sphorylated and this leads to hypophosphorylation ofthe NMDA receptor by the SFKs (Figure 6) [58,59].The question arises as to how increases in ErbBhyperphosphorylation leads to decreases in phos-phorylated SFKs, which in turn lead to decreases inphosphorylation of the NR2B subunit of the NMDAreceptor (Figure 6). It will be noted in this schemethat it is not NRG-1 that is abnormal, but the state ofphosphorylation of the ErbB4 receptor because, it issuggested, of its enhanced interaction with the PSD-95 complex, although it might be claimed thatenhanced or reduced delivery of normal NRG-1 leadsto a long-standing change in ErbB4 �PSD-95 inter-action.

Changes in splice variants of ErbB4 in schizophrenia

Stimulation of ErbB4 by NRG-1 promotes receptorectodomain cleavage by the metalloprotease tumornecrosis factor-a-converting enzyme (TACE) [60].The intracellular fragment containing the transmem-brane and cytoplasmic domains may be subsequentlycleaved by presenilin/g-secretase, so releasing itsintracellular domain (E41CD) into the cytosol fromwhere it can be translocated into the nucleus [61].

Alternative splicing of the ErbB4 gene generatestwo isoforms that differ in their extracellular juxta-membrane (JM) regions as well as their sensitivity toproteolytic cleavage, thus having different capacitiesto signal through regulated intramembrane proteoly-sis. ErbB4 JMa is cleaved by TACE and presenilin,whereas ErbB4 JMb isoform is not [62,63]. Interest-ingly, mRNA for ErbB4 JMa/CYTI is elevated inpatients with schizophrenia [64], suggesting thatdysregulation of splice-variant specific expressionof ErbB4 is the basis of the association of this genewith schizophrenia. Indeed, consideration of thecleavage products of the ErbB4 protein (at full length,180 kDa), namely at 21, 55 and 60 kDa in theprefrontal cortex of the brains of patients withschizophrenia, shows that the ratios 21 kDa/180 kDaand 55 kDa/180 kDa are significantly reduced, withthe full-length protein at 180 kDa increased by 30%[65]. Whether these ratios are reduced simply becauseof the elevation of the full-length 180 kDa protein isnot clear.

Polymorphisms in Fyn and Src in schizophrenia

In Fyn knockout mice, there are very few spines onthe dendrites of cortical neurons [66]. There is a

Figure 5. Upregualtion of N-methyl-D-aspartate(NMDA) receptors by Src family kinase Src acting onNR2A receptor subunits. Src has domains UD (uniquedomain), SH3, SH2 and CD (catalytic domain).Phosphorylation (P) in the CD results in a confor-mational change of the activation loop, producing afully active Src kinase. This can now act, via its

tyrosine kinase activity, through an adaptor protein, tophosphorylate the NR2A subunit of the NMDAreceptor in its carboxy (C)-terminal tail at least atsites Y1292, Y1325 and Y1387 (Fig. 4). Besides Srcbecoming a fully active kinase through its interactionwith the ErbB receptor, it can also be activated by thenon-receptor protein kinase cell adhesion kinase beta(CaKb), which possesses three domains, BAND4.1JEF (Janus kinase/ERM/FAK), CD (catalyticdomain) and FAT (focal-adhesion targeting). CaKb isactivated by protein kinase C (PKC) and thereforeindirectly by calcium and by G-protein coupled recep-

tors (GPCRs). (After Figure 1 in [6]).




relationship between particular polymorphisms of theFyn gene and performance on neuropsychologicaltests of prefrontal cortex activity, namely the Wis-consin Card Sorting Test, in patients with schizo-phrenia [67] (although in another study it was claimedthat there is no evidence for involvement of genomicFyn gene mutations in schizophrenia [68]). It isinteresting in this context that Src inhibitions protectthe injury to retrospinal cortical neurons by NMDAantagonists, such as ketamine and MK-801, and thiseffect of Src inhibitors is mimicked by atypicalantipsychotics such as clozapine [69].

Clozapine modulation of ErbB4 and NMDA receptors

The atypical neuroleptic clozapine increases NMDAreceptor-mediated excitatory currents in most prefron-tal cortical neurons, facilitating long-term potentia-tion, whereas the typical neuroleptic, haloperidol, doesnot [70]. This potentiation is selective for the NR2Bsubtype-containing NMDA receptors in the nucleusaccumbens and is blocked by selective SFK inhibitors[71]. Interestingly, clozapine increases the proteinlevels of ErbB-4 receptors in the hippocampus [72].

Conclusion

The NMDA hypofunction hypothesis for schizo-phrenia suggests that agents should be sought that

enhance the activity of this receptor. NMDA agonistsare not likely to be useful because of their leading,among other things, to excitotoxic cell death of theneurons consequent on excess calcium entry. Ap-proaches using the allosteric site on the NMDAreceptor complex that bind glycine, a necessarycondition for operation of the receptor, are promis-ing. Indeed, chronic treatment of rodents with glycinedoes not produce excitotoxicity [73]. Multiple smalltrials with this approach have in general producedpositive cognitive enhancement [8,74]. Another pos-sible approach is to prevent reuptake of glycine, soenhancing its extracellular concentration and NMDAreceptor activity [75]. Of particular interest is theNRG-ErbB4 pathway in NMDA regulation. Ratherthan directly attempting to upregulate the NMDAreceptor, it is suggested that compounds should besought that prevent hyperphosphorylation of ErbB4or the subsequent hypophosphorylation of NMDAreceptors through changes in the recruitment of theSFKs. This paper and previous ones emphasize thathallucinations, delusions and cognitive decline can allbe related to loss of synaptic spines and hencefunctioning synapses in particular parts of the brain[3,4,76]. NMDA hypofunction, with subsequent lossof spines, emphasizes the need for fast-throughputassays to identify compounds that will restore thefunctioning of NMDA receptors and hence synapseswith the resultant re-establishment of neural circuitry,

Figure 6. The ErbB hyperphosphorylation hypothesis for schizophrenia. This suggests that ErbB phosphotyrosinesare hyperphosphorylated following the action of neuregulin leading to failure of phosphorylation of sites on the NR2subunits of the N-methyl-d-aspartate (NMDA) receptor. The sites shown as no longer phosphorylated here (�), are

purely speculative. Compare this Figure with Figure 4.

M. BENNETT 719



holding out the possibility of some rehabilitation ofthose suffering from schizophrenia.

References

1. Carlsson A. The neurochemical circuitry of schizophrenia.Pharmacopsychiatry 2006; 39 (Suppl 1):S10�S14.

2. Carlsson M, Carlsson A. Schizophrenia: a subcorticalneurotransmitter imbalance syndrome? Schizophr Bull 1990;16:425�432.

3. Bennett MR. Dual constraints on synapse formation andregression in schizophrenia: neuregulin, neuroligin, dysbindin,DISC1, musk and agrin. Aust N Z J Psychiatry 2008; 42:662�677.

4. Bennett MR. Consciousness and hallucinations inschizophrenia: the role of synapse regression. Aust N Z JPsychiatry 2008; 42:915�931.

5. Pollard TD. The cytoskeleton, cellular motility and thereductionist agenda. Nature 2003; 422:741�745.

6. Ali DW, Salter MW. NMDA receptor regulation by Srckinase signalling in excitatory synaptic transmission andplasticity. Curr Opin Neurobiol 2001; 11:336�342.

7. Ethell IM, Pasquale EB. Molecular mechanisms of dendriticspine development and remodeling. Prog Neurobiol 2005;75:161�205.

8. Gray JA, Roth BL. The pipeline and future of drugdevelopment in schizophrenia. Mol Psychiatry 2007;12:904�922.

9. Schubert V, Dotti CG. Transmitting on actin: synapticcontrol of dendritic architecture. J Cell Sci 2007; 120(Pt2):205�212.

10. Cingolani LA, Goda Y. Actin in action: the interplay betweenthe actin cytoskeleton and synaptic efficacy. Nat Rev Neurosci2008; 9:344�356.

11. Saneyoshi T, Wayman G, Fortin D et al. Activity-dependentsynaptogenesis: regulation by a CaM-kinase kinase/CaM-kinase 1/betaPIX signaling complex. Neuron 2008; 57:94�107.

12. Nakayama AY, Luo L. Intracellular signaling pathways thatregulate dendritic spine morphogenesis. Hippocampus 2000;10:582�586.

13. Da Silva JS, Medina M, Zuliani C, Di Nardo A, Witke W,Dotti CG. RhoA/ROCK regulation of neuritogenesis viaprofilin 11a-mediated control of actin stability. J Cell Biol2003; 162:1267�1279.

14. Schubert V, Da Silva JS, Dotti CG. Localized recruitmentand activation of RhoA underlies dendritic spine morphologyin a glutamate receptor-dependent manner. J Cell Biol 2006;172:453�467.

15. Schmidt JT, Morgan P, Dowell N, Leu B. Myosin light chainphosphorylation and growth cone motility. J Neurobiol 2002;52:175�188.

16. Zhou W, Carpenter G. Heregulin-dependent translocationand hyperphosphorylation of ErbB-2. Oncogene 2001;20:3918�3920.

17. Holbro T, Hynes NE. ErbB receptors: directing key signalingnetworks throughout life. Annu Rev Pharmacol Toxicol 2004;44:195�217.

18. Qiu C, Tarrant MK, Choi SH et al. Mechanism of activationand inhibition of the HER4/ErbB4 kinase. Structure 2008;16:460�467.

19. Zhou W, Carpenter G. Heregulin-dependent trafficking andcleavage of ErbB-4. J Biol Chem 2000; 275:34 737�34 743.

20. Chattopadhyay A, Vecchi M, Ji Q, Mernaugh R, CarpenterG. The role of individual SH2 domains in mediatingassociation of phospholipase C-gamma1 with the activatedEGF receptor. J Biol Chem 1999; 274:26 091�26 097.

21. Schulze WX, Deng L, Mann M. Phosphotyrosine interactomeof the ErbB receptor kinase family. Mol Syst Biol 2005;1:2005�2008.

22. Sweeney C, Lai C, Riese DJ II, Diamonti AJ, Cantley LC,Carraway KL III. Ligand discrimination in signaling throughan ErbB4 receptor homodimer. J Biol Chem 2000; 275:19 803�19 807.

23. Murphy SP, Bielby-Clarke K. Neuregulin signaling inneurons depends on ErbB4 interaction with PSD-95. BrainRes 2008; 1207:32�35.

24. Garcia RA, Vasudevan K, Buonanno A. The neuregulinreceptor ErbB4 interacts with PDZ-containing proteins atneuronal synapses. Proc Natl Acad Sci USA 2000; 97:3596�3601.

25. Huang YZ, Won S, Ali DW et al. Regulation of neuregulinsignaling by PSD-25 interacting with ErbB4 at CNS synapses.Neuron 2000; 26:443�455.

26. Yu XM, Askalan R, Keil GJ II, Salter MW. NMDA channelregulation by channel-associated protein tyrosine kinase Src.Science 1997; 275:674�678.

27. Kalia LV, Gingrich JR, Salter MW. Src in synaptictransmission and plasticity. Oncogene 2004; 23:8007�8016.

28. Wang YT, Salter MW. NMDA receptor regulation bytyrosine kinases and phosphatases. Nature 1994; 369:233�235.

29. Gingrich JR, Pelkey, KA, Fam SR et al. Unique domainanchoring of Src to synaptic NMDA receptors via themitochondrial protein NADH dehydrogenase subunit 2. ProcNatl Acad Sci USA 2004; 101:6237�6242.

30. Huang Y, Lu W, Ali DW et al. CAKbeta/Pyk2 kinase is asignaling link for induction of long-term potentiation in CA1hippocampus. Neuron 2001; 29:485�496.

31. Guo J, Meng F, Fu X, Song B, Yan X, Zhang G. N-methyl-D-aspartate receptor and L-type voltage-gated Ca2� channelactivation mediate proline-rich tyrosine kinase 2phosphorylation during cerebral ischaemia in rats. NeurosciLett 2004; 355:177�180.

32. Salter MW, Kalia LV. Src kinases: a hub for NMDA receptorregulation. Nat Rev Neurosci 2004; 5:317�328.

33. Tezuka T, Umemori H, Akiyama T, Nakanishi S, YamamotoT. PSD-95 promotes Fyn-mediated tyrosine phosphorylationof the N-methyl-D-aspartate receptor subunit NR2A. ProcNatl Acad Sci USA 1999; 96:435�440.

34. Cheung HH , Gurd JW. Tyrosine phosphorylation of the N-methyl-D-aspartate receptor by exogenous and postsynapticdensity-associated Src-family kinases. J Neurochem 2001;78:524�534.

35. Rostas JA, Brent VA, Voss K, Errington ML, Bliss TV, GurdJW. Enhanced tyrosine phosphorylation of the 2B subunit ofthe N-methyl-D-aspartate receptor in long-term potentiation.Proc Natl Acad Sci USA 1996; 93:10 452�10 456.

36. Kojima N, Wang J, Mansuy IM, Grant SG, Mayford M,Kandel ER. Rescuing impairment of long-term potentiationin fyn-deficient mice by introducing Fyn transgene. Proc NatlAcad Sci USA 1997; 94:4761�4765.

37. Bhaskar K, Yen SH, Lee G. Disease-related modifications intau affect the interaction between Fyn and Tau. J Biol Chem2005; 280:35 119�35 125.

38. Sarnago S, Elorza A, Mayor F Jr. Agonist-dependentphosphorylation of the G protein-coupled receptor kinase 2(GRK2) by Src tyrosine kinase. J Biol Chem 1999; 274:34411�34416.

39. Solomaha E, Szeto FJ, Yousef MA, Palfrey HC. Kinetics ofSrc homology 3 domain association with the proline-richdomain of dynamins: specificity, occlusion, and the effects ofphosphorylation. J Biol Chem 2005; 280:23 147�23 156.

40. Lu YM, Roder JC, Davidow J, Salter MW. Src activation inthe induction of long-term potentiation in CA1 hippocampalneurons. Science 1998; 279:1363�1367.




41. Xu J, Weerapura M, Ali MK et al. Control of excitatorysynaptic transmission by C-terminal Src kinase. J Biol Chem2008; 283:17 503�17 514.

42. Kholodenko BN, Demin OV, Moehren G, Hoek JB.Quantification of short term signaling by the epidermalgrowth factor receptor. J Biol Chem 1999; 274:30 169�30 181.

43. Moehren G, Markevich N, Demin O et al. Temperaturedependence of the epidermal growth factor receptor signalingnetwork can be accounted for by a kinetic model.Biochemistry 2002; 41:306�320.

44. Hatakeyama M, Kimura S, Naka T et al. A computationalmodel on the modulation of mitogen-activated protein kinase(MAPK) and Akt pathways in heregulin-induced ErbBsignalling. Biochem J 2003; 373(Pt 2):451�463.

45. Nakakuki T, Yumoto N, Naka T, Shirouzu M, Yokoyama S,Hatakeyama M. Topological analysis of MAPK cascade forkinetic ErbB signaling. PLoS ONE 2008; 3:e1782.

46. Schoeberl B, Eichler-Jonsson C, Gilles ED, Muller G.Computational modeling of the dynamics of the MAP kinaseactivated by surface and internalized EGF receptors. NatBiotechnol 2002; 20:370�375.

47. Mayadevi M, Praseeda M, Kumar KS, Omkumar RV.Sequence determinants on the NR2A and NR2B subunitsof NMDA receptor responsible for specificity ofphosphorylation by CaMK11. Biochem Biophys Acta 2002;15 9840�15 9845.

48. Castellani GC, Quinlan EM, Bersani F, Cooper LN, ShouvalHZ. A model of bidirectional synaptic plasticity: fromsignaling network to channel conductance. Learn Mem 2005;12:423�432.

49. Huang YZ, Won S, Ali DW et al. Regulation of neuregulinsignaling by PSD-95 interacting with ErbB4 at CNS synapses.Neuron 2000; 26:443�455.

50. Ma L, Huang YZ, Pitcher GM et al. Ligand-dependentrecruitment of the ErbB4 signaling complex into neuronallipid rafts. J. Neurosci 2003; 23:3164�3175.

51. Kwon OB, Longart M, Vullhorst D, Hoffman DA, BuonannoA. Neuregulin-1 reverses long-term potentiation at Ca1hippocampal synapses. J Neuroscience 2005; 25:9378�9383.

52. Gu Z, Jiang Q, Fu AK, Ip NY, Yan Z. Regulation of NMDAreceptors by neuregulin signaling in prefrontal cortex. JNeurosci 2005; 25:4974�4984.

53. Roysommuti S, Carroll SL, Wyss JM. Neuregulin-1betamodulates in vivo entorhinal-hippocampal synaptictransmission in adult rats. Neuroscience 2003; 121:779�785.

54. Woo RS, Li XM, Tao Y et al. Neuregulin-1 enhancesdepolarization-induced GABA release. Neuron 2007; 54:599�610.

55. Stefansson H, Sigurdsson E, Steinthorsdottir S et al.Neuregulin 1 and susceptibility to schizophrenia. Am J HumGenet 2002; 71:877�892.

56. Li B, Woo RS, Mei L, Malinow R. The neuregulin-1 receptorerbB4 controls glutamatergic synapse maturation andplasticity. Neuron 2007; 54:583�597.

57. Bjarnadottir M, Misner DL, Haverfield-Gross S et al.Neuregulin1 (NRG1) signaling through Fyn modulatesNMDA receptor phosphorylation: differential synapticfunction in NRG1�/� knock-outs compared with wild-typemice. J Neurosci 2007; 27:4519�4529.

58. Hahn CG, Wang HY, Cho DS et al. Altered neuregulin1-ErbB4 signaling contributes to NMDA receptorhypofunction in schizophrenia. Nat Med 2006; 12:824�828.

59. Fischbach GD. Schizophrenia: signals from the other side.Nat Med 2006; 12:734�735.

60. Rio C, Buxbaum JD, Peschon JJ, Corfas G. Tumor necrosisfactor-alpha-converting enzyme is required for cleavage ofErbB4/HER4. J Biol Chem 2000; 275:10 379�10 387.

61. Ni CY, Murphy MP, Golde TE, Carpenter G. Gamma-secretase cleavage and nuclear localization of ErbB4 receptortyrosine kinase. Science 2001; 294:2179�2181.

62. Plowman GD, Green JM, Culouscou JM, Cartlon GW,Rothwell VM, Buckley S. Heregulin induces tyrosinephosphorylation of HER4/p180erbB4. Nature 1993; 366:473�475.

63. Plowman GD, Culouscou JM, Whitney GS et al. Ligand-specific activation of HER4/p180erbB4, a fourth member ofthe epidermal growth factor receptor family. Proc Natl AcadSci USA 1993; 90:1746�1750.

64. Law AJ, Kleinman JE, Weinberger DR, Weickert CS.Disease-associated intronic variants in the ErbB4 gene arerelated to altered ErbB4 splice-variant expression in the brainin schizophrenia. Hum Mol Genet 2007; 16:129�141.

65. Chong VZ, Thompson M, Beltaifa S, Webster MJ, Law AJ,Weickert CS. Elevated neuregulin-1 and ErbB4 protein in theprefrontal cortex of schizophrenic patients. Schizophr Res2008; 100:270�280.

66. Morita A, Yamashita N, Sasaki Y et al. Regulation ofdendritic branching and spine maturation by semaphorin3A-Fyn signaling. J Neurosci 2006; 26:2971�2980.

67. Rybakowski JK, Borkowska A, Skibinska M, Hauser J.Polymorphisms of the Fyn kinase gene and a performance onthe Wisconsin Card Sorting Test in schizophrenia. PsychiatrGenet 2007; 17:201�204.

68. Ishiguro H, Saito T, Shibuya H, Toru M, Arinami T.Mutation and association analysis of the Fyn kinase gene withalcoholism and schizophrenia. Am J Med Genet 2000; 96:16�20.

69. Dickerson J, Sharp FR. Atypical antipsychotics and a Srckinase inhibitor (PP1) prevent cortical injury produced by thepsychomimetic, noncompetitive NMDA receptor antagonistMK-801. Neuropsychopharmacology 2006; 31:1420�1430.

70. Gemperle AY, Enz A, Pozza MF, Luthi A, Olpe HR. Effectsof clozapine, haloperidol, and iloperidone onneurotransmission and synaptic plasticity in prefrontal cortexand their accumulation in brain tissue: an in vitro study.Neuroscience 2003; 117:681�695.

71. Wittmann M, Marino MJ, Henze DA, Seabrook GR, ConnPJ. Clozapine potentiation of N-methyl-D-aspartate receptorcurrents in the nucleus accumbens: role of NR2B and proteinkinase A/Src kinases. J Pharmacol Exp Ther 2005; 313:594�603.

72. Wang XD, Su YA, Guo CM, Yang Y, Si TM. Chronicantipsychotic drug administration alters the expression ofneuregulin 1beta, ErbB2, ErbB3, and ErbB4 in the ratprefrontal cortex and hippocampus. Int JNeuropsychopharmacol 2008; 11:553�5561.

73. Shoham S, Javitt DC, Heresco-Levy U. Chronic high-doseglycine nutrition: effects on rat brain cell morphology. BiolPsychiatry 2001; 49:876�885.

74. Gray JA, Roth BL. Molecular targets for treating cognitivedysfunction in schizophrenia. Schizophr Bull 2007; 33:1100�1119.

75. Javitt DC, Duncan L, Balla A, Sershen H. Inhibition ofsystem A-mediated glycine transport in cortical synaptosomesby the therapeutic concentrations of clozapine: implicationsfor mechanisms of action. Mol Psychiatry 2005; 10:275�287.

76. Bennett MR. Stress and anxiety in schizophrenia anddepression: glucocorticoids, corticotropin-releasing hormoneand synapse regression. Aust N Z J Psychiatry 2008; 42:995�1002.

M. BENNETT 721



positive and negative symptoms in schizophrenia: the nmda receptor hypofunction hypothesis,...

Documents