Integrin signaling cascades are operational in adult hippocampal synapses and modulate NMDA receptor physiology

Integrin signaling cascades are operational in adult hippocampalsynapses and modulate NMDA receptor physiology

Joie A. Bernard-Trifilo,* Eniko A. Kramar�, Reidun Torp,� Ching-Yi Lin,* Eduardo A. Pineda,*Gary Lynch� and Christine M. Gall*,§

*Department of Anatomy & Neurobiology, University of California, Irvine, California, USA

�Department of Psychiatry and Human Behavior, University of California, Irvine, California, USA

�Center for Molecular Biology and Neuroscience, University of Oslo, Blindern, Oslo, Norway

§Department of Neurobiology and Behavior, University of California, Irvine, California, USA

Abstract

Integrin class adhesion proteins are concentrated at adult

brain synapses. Whether synaptic integrins engage kinase

signaling cascades has not been determined, but is a question

of importance to ideas about integrin involvement in functional

synaptic plasticity. Accordingly, synaptoneurosomes from

adult rat brain were used to test if matrix ligands activate

integrin-associated tyrosine kinases, and if integrin signaling

targets include NMDA-class glutamate neurotransmitter

receptors. The integrin ligand peptide Gly-Arg-Gly-Asp-Ser-Pro

(GRGDSP) induced rapid (within 5 min) and robust increases

in tyrosine phosphorylation of focal adhesion kinase, proline-

rich tyrosine kinase 2 and Src family kinases. Increases were

similarly induced by the native ligand fibronectin, blocked with

neutralizing antibodies to b1 integrin, and not obtained with

control peptides, indicating that kinase activation was integrin-

mediated. Both GRGDSP and fibronectin caused rapid Src

kinase-dependent increases in tyrosine phosphorylation of

NMDA receptor subunits NR2A and NR2B in synaptoneuro-

somes and acute hippocampal slices. Tests of the physiolo-

gical significance of the latter result showed that ligand

treatment caused a rapid and b1 integrin-dependent increase

in NMDA receptor-mediated synaptic responses. These

results provide the first evidence that, in adult brain, synaptic

integrins activate local kinase cascades with potent effects on

the operation of nearby neurotransmitter receptors implicated

in synaptic plasticity.

Keywords: adhesion, fibronectin, glutamate receptor,

integrin, phosphorylation, synaptoneurosome.

J. Neurochem. (2005) 93, 834–849.

Integrins are transmembrane, heterodimeric (ab) receptorsthat mediate cell adhesion by crosslinking extracellularmatrix ligands with the actin cytoskeleton (Hynes 1992;Miranti and Brugge 2002). Studies of non-neural cells haveshown that integrins also initiate acute signaling (Lafrenieand Yamada 1996; Schlaepfer and Hunter 1998) involvingtyrosine kinase cascades that modulate functional propertiesof neighboring transmembrane proteins including ion chan-nels (Wu et al. 1998; Davis et al. 2002) and growth factorreceptors (Porter and Hogg 1998; Miranti and Brugge 2002;Moro et al. 2002). Through these interactions integrinsinfluence a diverse array of basic activities, including cellcycling, phenotypic differentiation, gene expression, processoutgrowth and cell survival (Schwartz et al. 1995; Yamadaand Miyamoto 1995). Localization studies have shown thatseveral integrin proteins (a3, a5, a8, av, b1, b3, b8) arepresent in hippocampal and/or cortical synapses in adult

brain (Bahr and Lynch 1992; Einheber et al. 1996; Capaldiet al. 1997; Nishimura et al. 1998; Rodriguez et al. 2000;Kramar et al. 2002; Dong et al. 2003). Importantly, thebalance of integrin subunits expressed varies markedly

Received September 10, 2004; revised manuscript received December21, 2004; accepted December 22, 2004.Address correspondence and reprint requests to Christine Gall, Pro-

fessor, Department of Anatomy and Neurobiology, Gillespie Neuro-science Research Facility, University of California, Irvine, CA 92697-4292, USA. E-mail: [email protected] used: aCSF, artificial cerebrospinal fluid; AMPA,

a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid; AP5, D-(–)-ami-nophosphonopentanoic acid; CNQX, 6-cyano-7-nitroquinoxaline-2,3-dione; EPSC, excitatory postsynaptic current; fEPSP, field excitatorypostsynaptic potential; FAK, focal adhesion kinase; GABA, c-aminobu-tyric acid; LTP, long-term potentiation; PIC, Protease Inhibitor Cocktail;Pyk2, proline-rich tyrosine kinase 2; SFK, Src family kinase.

Journal of Neurochemistry, 2005, 93, 834–849 doi:10.1111/j.1471-4159.2005.03062.x

834 � 2005 International Society for Neurochemistry, J. Neurochem. (2005) 93, 834–849

across regions and neuronal cell types (Pinkstaff et al. 1999),and even between different dendritic domains on the sameneuron (Bi et al. 2001). As integrin subunit composition isknown to constrain signaling activities of the receptor (Hynes1992; Huhtala et al. 1995; Giancotti and Ruoslahti 1999), itis tempting to speculate that integrin functions vary betweenbrain areas as well. Before addressing this, it is necessary tofirst consider the more basic, and largely unexplored,question of whether integrins trigger signaling cascades inadult brain and, more specifically, at adult synapses.

Most studies of integrins in the central nervous systemhave focused on early stages of development (Milner andCampbell 2002) and have demonstrated important roles forthe adhesion receptors in mechanisms of cellular migration(Huttenlocher et al. 1997; Sheetz et al. 1998; Calderwoodet al. 2001), process outgrowth (McKerracher et al. 1996;Grabham et al. 2000), and cellular positioning relative tomatrix cues (Schmid and Anton 2003). In the adult, there isan emerging body of evidence that integrins modulatesynaptic transmission as well. Peptides containing the Arg-Gly-Asp (RGD) binding sequence recognized by a largeproportion of forebrain integrins (Pinkstaff et al. 1999)increase synaptic responses mediated by NMDA- (Lin et al.2003) and a-amino-3-hydroxy-5-methyl-4-isoxazoleprop-ionic acid (AMPA)-type (Kramar et al. 2003) glutamatereceptors. The enhancement of NMDA receptor gatedcurrents begins within 15 min, is unaffected by the GABA-A receptor antagonist picrotoxin, and is not accompanied byevident changes in paired pulse facilitation. This pattern ofresults indicates that the integrin effect is not secondary tochanges in GABAergic transmission or presynaptic glutam-ate release. Presumably related to this, manipulation ofintegrin activity with peptide ligands, disintegrins andfunction blocking antibodies to specific integrin subunits(a3, a5) prevents the stabilization of long-term potentiation(LTP) (Staubli et al. 1990, 1998; Bahr et al. 1997; Chunet al. 2001; Kramar et al. 2002; LaBaron et al. 2003);similar effects have been obtained in mutant mice withreduced expression of the same integrins proteins (a3, a5,and a8) (Chan et al. 2003). These findings indicate thatintegrin binding has important consequences to synapticfunction, but leave open the question of whether these effectsare achieved via activation of local integrin signaling orinstead arise from disturbances to integrin-mediated adhe-sion. The present studies tested if integrin binding activatessignaling cascades at mature synapses and if these cascadesthen modify synaptic glutamate receptors.

In non-neural systems, integrin ligation initiates signalingin part by activating focal adhesion kinase (FAK) and itshomologue proline-rich tyrosine kinase 2 (Pyk2); FAK andPyk2 then act through a diverse array of secondaryproteins, prominent among which are the Src familykinases (SFKs) (Hanks et al. 1992; Lipfert et al. 1992;Schlaepfer and Hunter 1998). Accordingly, we first

determined if matrix ligand sequences activate theseintegrin-related kinases in synaptoneurosomes preparedfrom forebrain of adult rats. Tests were then carried outwith the same preparation to determine if integrin signalingis followed by tyrosine phosphorylation of NMDA-classglutamate receptors. Finally, we asked if integrin ligandshave similar effects when applied to hippocampal slicesand, importantly, if these effects are suppressed byantibodies that block b1 integrin function. The resultsprovide the first direct evidence that integrin kinasesignaling is operational in adult forebrain synapses andmediates integrin regulation of neurotransmitter receptorfunction at excitatory glutamatergic synapses.

Materials and methods

Synaptoneurosome preparation and treatment

Synaptoneurosomes were prepared as previously described (Lai

and Clark 1979) with minor modifications. Briefly, male Sprague

Dawley rats (1–2 months of age) were anesthetized with

methoxyflurane (Pitman-Moore, Mundelein, IL, USA) and decap-

itated. The telencephalon was homogenized in a glass/Teflon

homogenizer in 0.25 M sucrose, 20 mM HEPES (pH 7.2), 1 mM

EDTA, and 1 · Complete� Protease Inhibitor Cocktail (PIC;

Roche Diagnostics GmbH, Mannheim, Germany; PIC has broad

inhibitory specificity including inhibition of serine, cysteine, and

metalloproteases as well as calpains). The homogenate was

centrifuged for 3 min at 1300 g and the resulting pellet was

diluted, homogenized, and centrifuged for 3 min at 1300 g. Thesupernatant was centrifuged at 13 000 g for 10 min and the pellet

was resuspended, layered onto a two-step Ficoll� (F-5415;

Sigma, St Louis, MO, USA) gradient (7.5–10%), and centrifuged

at 100 000 g for 30 min. The synaptosomal fraction at the 7.5%

and 10% interface was removed and suspended in artificial

cerebrospinal fluid (aCSF) containing (in mM) 124 NaCl, 3 KCl,

1.25 KH2PO4, 2.5 MgSO4, 3.4 CaCl2, 26 NaHCO3, and 10

glucose. Synaptoneurosomes were then centrifuged at 9000 g for

10 min and the pellet was resuspended in aCSF containing PIC

and Phosphatase Inhibitor Cocktails 1 and 2 (P2850 and P5726,

Sigma) as per kit instructions. Volume was adjusted to normalize

the final protein concentration to 1–2 mg/mL as determined by

the Bio-Rad Protein Assay (Bio-Rad, Hercules, CA, USA). For

each experiment, 6–10 aliquots (600–800 lL) from a common

synaptoneurosome preparation were used.

For treatment, synaptoneurosomal fractions were incubated with

the integrin ligand peptide Gly-Arg-Gly-Asp-Ser-Pro (GRGDSP;

Calbiochem, La Jolla, CA, USA), full-length human plasma

fibronectin (33016-015; Invitrogen, Carlsbad, CA, USA), or aCSF

vehicle. Experiments with the Src tyrosine kinase inhibitor, PP2, the

control compound PP3 (Calbiochem), or function blocking b1antibody (22631D; Pharmingen, Palo Alto, CA, USA) involved

pretreatment of the synaptoneurosomes with the antagonist for

5 min before cotreatment with GRGDSP or aCSF. Aliquots were

lightly agitated at room temperature during incubation periods.

Reactions were terminated with the addition of reducing sodium

dodecyl sulfate–polyacrylamide gel electrophoresis sample buffer

(Kramar et al. 2002).

Synaptic integrins regulate NMDA receptors 835

� 2005 International Society for Neurochemistry, J. Neurochem. (2005) 93, 834–849

Western blot analyses

For straight western blots (without immunoprecipitation), samples

were separated on 4–12% gradient gels, transferred to polyvinylid-

ene difluoride membrane (Amersham Pharmacia Biotech, Piscata-

way, NJ, USA) and processed as described elsewhere (Kramar et al.2002) with primary antisera used at a 1 : 1000 dilution in Tris-

buffered saline with 5% bovine serum albumin overnight at 4�C.Immunoreactive bands were visualized using the enhanced chemi-

luminescence ECL Plus kit (Amersham Pharmacia Biotech). For

analyses of signaling kinases, blots were probed with anti-phospho

FAK Tyr397 (#44-624) (Laser et al. 2000; Sanders and Basson

2000; Bongiorno-Borbone et al. 2002), anti-phospho Pyk2 Tyr402

(#44-618) (Katagiri et al. 2000; Keely et al. 2000; Shi et al. 2000;Bongiorno-Borbone et al. 2002), or anti-phospho Src Tyr418 (#44-

660) (Laser et al. 2000; Sanna et al. 2000; Sussman et al. 2000;Thodeti et al. 2000) from BioSource Int. (Camarillo, CA, USA); in

some instances this was followed by stripping and reprobing with

anti-FAK (#05-537), anti-Pyk2 (#05-488), or anti-Src (#05-184),

respectively, from Upstate Biotech (Lake Placid, NY, USA).

Immunoblots for NMDA receptor subunit proteins were probed

with antisera to NR2A (MAB5216 or AB1555P, Chemicon Int.,

Temecula, CA, USA) and NR2B (AB1557P, Chemicon).

For immunoprecipitation analyses, synaptoneurosomes were

centrifuged at 10 000 g for 10 min and then lysed by resuspending

the pellet in ristocetin-induced platelet agglutination buffer (50 mM

Tris-HCl, pH 7.4, 1% NP-40, 1% Triton X-100, 1 mM EDTA,

150 mM NaCl, PIC, and Phosphatase Inhibitor Cocktails 1 and 2).

For analyses of FAK, Pyk2, and NMDA receptor subunit

phosphorylation, and for differentiating Src and Fyn activation,

samples were immunoprecipitated with anti-phosphotyrosine anti-

body 4G10 (Upstate Biotech). For the kinases, immunoprecipitates

were then run for western blot analysis using (in separate blots)

anti-Src (#44-656, BioSource Int.), anti-Fyn (#06-133, Upstate

Biotech), anti-FAK (#05-537, Upstate Biotech), or anti-Pyk2

(#05-488, Upstate Biotech). For reverse immunoprecipitation

analyses of signaling kinases, samples were immunoprecipitated

with anti-FAK (#05-537, Upstate Biotech) or anti-Pyk2 (#05-488,

Upstate Biotech) and then western blots were probed with 4G10.

For reverse immunoprecipitation analyses of NMDA receptor

proteins, samples were precipitated with anti-NR2A, anti-NR2B, or

anti-NR2A/B (AB1555P, AB1557 and AB1548, respectively;

Chemicon) and then immunoblots were probed with anti-

phosphotyrosine 4G10. In all cases, samples were incubated with

the precipitating antibody at 1 : 1000 dilutions in Tris-buffered

saline with 5% bovine serum albumin overnight at 4�C. Protein G

Agarose (80 lL; #16-266, Upstate Biotech) was added for the final

2 h and then complexes were pelleted by centrifugation. Immune

complexes were washed twice with ristocetin-induced platelet

agglutination buffer and three times with Tris-buffered saline

(containing PIC and Phosphatase Inhibitor Cocktails 1 and 2),

eluted with sample buffer, and processed for western blot analysis

as described above.

Immunoreactive bands were evaluated by densitometry of ECL

films using the AIS imaging system (Imaging Res., St Catherines,

Ontario, Canada). Significance of effects of treatment was deter-

mined by one-way ANOVA followed by the Student–Neuman–Keuls

post hoc test for paired comparisons. Effects with p £ 0.05 were

considered statistically significant.

Electron microscopy

To assess the integrity of synaptoneurosomal preparations, samples

were collected from the ficoll gradient and pelleted by centrifugation

at 9000 g for 20 min. The pellet was fixed by immersion in 4%

paraformaldehyde plus 0.1% glutaraldehyde, rinsed in 0.1 M sodium

phosphate buffer (pH 7.2), treated with 1% osmium tetroxide,

dehydrated, embedded in Durcupan (ACM Fluka) and sectioned for

electron microscopy. Following contrasting with uranyl acetate and

lead citrate, the sections were photographed using a Philips CM10

microscope.

Electrophysiology

For electrophysiological studies, transverse slices (350-lm thick)

through the mid-septotemporal arc of the hippocampus were

prepared as described elsewhere (Kramar et al. 2002), with the

exception that after sectioning the slices were transferred to a

shallow ice-chilled high magnesium (5.0 mM) low calcium (3.4 mM)

aCSF bath and, under visual inspection with a dissecting micro-

scope, individually blocked to include only the CA1 field. These

CA1-mini-slices were then transferred to an interface recording

chamber containing aCSF (including in mM, 124 NaCl, 3 KCl, 1.25

KH2PO4, 2.5 MgSO4, 3.4 CaCl2, 26 NaHCO3, and 10 glucose) and

maintained at 31 ± 1�C. Throughout the experiment, slices were

continuously perfused with aCSF (75 mL/h) and upper slice

surfaces were exposed to warm, humidified 95% O2/5% CO2.

Recordings began at least 1 h after sectioning.

Field excitatory postsynaptic potentials (fEPSPs) were elicited

by orthodromic stimulation (twisted nichrome wires, 65 lm) of

the Schaffer collateral-commissural projections to CA1 stratum

radiatum and recorded from CA1b stratum radiatum using a

single glass pipette filled with 0.15 M NaCl (2–3 MW resistance).

Pulses were delivered to the stimulation electrode at 0.05 Hz with

current test intensity adjusted to obtain 50–60% of the maximum

fEPSP. After establishing a 10–20 min stable baseline, test

compounds were introduced by switching the bath-perfusate from

control aCSF to drug-containing aCSF. Measurements of fEPSP

amplitude and half-width were recorded and digitized by NAC

2.0 Neurodata Acquisition Systems (Theta Burst Corp., Irvine,

CA, USA). In all cases, response size measures given in the text

represent group means ± standard error of the mean (SEM). The

sample size for all experiments represents the number of animals

used. The significance of effects of treatment was determined

using the Students t-test with p £ 0.05 considered statistically

significant.

Hippocampal slice/drug application

The NMDA antagonist, D-(–)-aminophosphonopentanoic acid (AP5;

Sigma), the AMPA type glutamate receptor blocker 6-cyano-7-

nitroquinoxaline-2,3-dione (CNQX – salt form, Sigma) and the

integrin ligand peptide, GRGDSP were prepared in aCSF as a stock

solution (100 mM, 10 mM and 200 mM, respectively) prior to being

added to the bath via an infusion line. The function blocking

monoclonal antibodies to b1 or a2 (control antibody) integrin

subunits (anti-b1 integrin, MAB1987Z; anti-a2 integrin,

MAB1950Z; Chemicon) were diluted in aCSF (containing 2 mM

GRGDSP) for a pipette concentration of 0.2 mg/mL; aliquots were

stored in )25�C. Prior to application, anti-b1 or anti-a2 integrin

samples were allowed to thaw and then loaded into a glass

836 J. A. Bernard-Trifilo et al.

� 2005 International Society for Neurochemistry, J. Neurochem. (2005) 93, 834–849

micropipette (tip diameter 25 lm). Local perfusion of the integrin

antibodies or aCSF (the antisera vehicle) was achieved by

positioning the infusion pipette in the bath immediately adjacent

and slightly upstream from the hippocampal slice that was being

monitored. Pipette contents were ejected using a Picospritzer

(General Valve, Fairfield, NJ, USA) with ejection pressure set at

8–12 psi: a 10 ms pulse, estimated to deliver 3 nL, was applied

every 20 s. The infusion pipette contained red dye so that the

diffusion of the ejected solution could be monitored using a

dissection microscope. Antibody treatment began 20 min before

bath application of 2 mM GRGDSP, and continued for 30 min until

the end of dual application with GRGDSP. Vehicle-control

experiments involved pipette ejection of aCSF containing 2 mM

GRGDSP by the same technique and schedule used for antisera

application.

Hippocampal slice western blots

Effects of GRGDSP treatment on NMDA receptor phosphorylation

in hippocampal slices were assessed with two preparations. First,

field CA1 mini-slices were prepared and used for electrophysio-

logical studies (as described above) and then harvested 1 h after

the onset of GRGDSP treatment. These samples were processed

for phosphotyrosine immunoprecipitation followed by western blot

analysis of NR2A and p-NR2B subunits as described above. The

1-h time point was chosen because prior work with acute

hippocampal slices demonstrated effects of GRGDSP treatment

on AMPA and NMDA receptor-mediated synaptic responses at this

interval and because previous (Kramar et al. 2003) and preliminary

studies indicated increases in receptor subunit protein phosphory-

lation were evident at this time point. Second, for some

assessments full hippocampal slices were prepared (as above),

placed on the membrane of millicel inserts (Lauterborn et al. 2000)and maintained at 31 ± 1�C in static aCSF in an interface

configuration: the membrane-opposed surface of the slice was in

aCSF while the upper surface was exposed to warm, humidified

95% O2/5% CO2 that was continuously supplied to the chamber.

After 1-h stabilization, the media was exchanged with fresh aCSF

(for control slices) or aCSF containing 3 mM GRGDSP (Gall et al.2003). Media was gently pipetted over the tissue every 15 min

until slices were harvested for immunoblot analysis as described

above.

Results

Characterization of synaptoneurosomes

The integrity of synaptoneurosomes used in signaling studieswas assessed in samples prepared as described in Methodsand evaluated by transmission electron microscopy. Thepreparations contained membrane-enclosed heterogeneousprofiles including lucent figures of modest size, largerprofiles with vesicles, and free mitochondria. Fully elabor-ated synaptic complexes were easily recognized and presentin large numbers (Fig. 1). These consisted of vesicle-containing presynaptic elements (boutons) apposed, via athickened synaptic membrane specialization, to enclosed,less dense profiles. The presynaptic elements often contained

one or more mitochondria with regularly spaced lamella,along with homogenous populations of round vesicles;presynaptic profiles with flattened vesicles were less fre-quently observed. The distribution of the vesicles varied frominstances in which they evenly filled almost the entire extentof the bouton to cases in which they were clustered into arestricted space. The postsynaptic side of the junctionfrequently included a conspicuous postsynaptic densityfading into a light or patchy ground substance within theprofile. The morphological integrity of the synaptic com-plexes was comparable to that of published descriptions ofsynaptoneurosomal preparations from other laboratories(Dong et al. 2003; Williams et al. 2003).

Fig. 1 Morphological assessment of synaptoneurosomal prepara-

tions. High magnification electron micrographs showing representative

synaptic complexes contained within synaptoneurosomal samples

prepared as described in Methods. Each can be seen to contain a

vesicle filled presynaptic bouton (B), a synaptic membrane special-

ization, and an enclosed postsynaptic element with a clear postsy-

naptic density. Mitochondria are found within both presynaptic

boutons. Bar ¼ 0.050 lm for (a) and 0.038 lm for (b).

Synaptic integrins regulate NMDA receptors 837

� 2005 International Society for Neurochemistry, J. Neurochem. (2005) 93, 834–849

Integrin ligand peptide activates synaptoneurosomal

focal adhesion kinase, proline-rich tyrosine kinase 2

and Src kinase

Focal adhesion kinase (FAK) and its homologue, proline-richtyrosine kinase 2 (Pyk2; also referred to as cell adhesionkinase b or CAKb, RAFTK, and Ca2+-dependent tyrosinekinase or CADTK) attach to integrin b-subunit cytoplasmicdomains (Schlaepfer and Hunter 1998; Liu et al. 2000).Integrin ligation activates tyrosine kinase signaling throughFAK and Pyk2 and, secondary to them, Src kinases (Lipfertet al. 1992; Schlaepfer and Hunter 1998; Vuori 1998; DellaMoret et al. 2000; Parsons et al. 2000). Synaptoneurosomeswere treated with the integrin ligand peptide GRGDSP at10 lM to 2 mM across multiple experiments and processedfor western blot analysis using phosphospecific antibodies tothe activation sites of FAK, Pyk2, and SFK. Figure 2(a–c)illustrates a representative experiment using 10 lMGRGDSP. As shown, levels of phosphorylated SFK(Y418),Pyk2(Y402), and FAK(Y397) were markedly increased by a5 min GRGDSP treatment, whereas total SFK, Pyk2 andFAK protein levels were not affected. Phospho (p)-Pyk2

remained elevated at 30 min, whereas levels ofp-FAK(Y397) and p-SFK(Y418) returned to near controllevels at this time point. Figure 2(d) summarizes results fromseveral experiments employing 10 lM GRGDSP, withdensitometric measures expressed as a function of pairedcontrol values (i.e. control samples processed within thesame western blot). Comparable effects were obtained atGRGDSP concentrations of 100 lM, 500 lM, 1 mM, and2 mM.

To corroborate the above results, additional tests for kinaseactivation with GRGDSP treatment were carried out usingimmunoprecipitation. Similar to results obtained with straightwestern blots using phospho-specific antisera, RGD-inducedincreases in phosphorylated Pyk2 were evident when sam-ples, from control and RGD-treated synaptoneurosomes,were immunoprecipitated with anti-Pyk2 (Fig. 2e) or anti-phosphotyrosine (Fig. 2f) and western blots were probedwith anti-p-Pyk2(Y402). For FAK, increases in thephosphoprotein were demonstrated when samples wereimmunoprecipitated with anti-FAK and probed with anti-p-FAK(Y397) (Fig. 2g) or were immunoprecipitated with

(a) (d)

(e)

(f)

(g)

(h)

(b)

(c)

Fig. 2 The GRGDSP integrin ligand increases SFK(Y418),

Pyk2(Y402), and FAK(Y397) phosphorylation. Synaptoneurosomes

were treated with aCSF (CON) or 10 lM GRGDSP (RGD) for

5–45 min and then run for western blot analysis. (a–c) Upper blots

show levels of immunoreactivity for phospho (p)-Src family kinase

(SFK)(Y418) (a), p-Pyk2(Y402) (b), and p-FAK(Y397) (c) and the

lower blots show immunoreactivity for total Src (a), Pyk2 (b), and FAK

(c) in synaptoneurosomal samples collected after 5 and 30 min

GRGDSP treatment (15 lg protein/lane): At 5 min ligand treatment

there is an increase in levels of the activated (phosphorylated) form of

each kinase but no increase in total kinase protein levels. (d) Bar

graph shows quantification of phospho-specific western blots with

experimental band densities expressed as a function (fold-increase) of

paired aCSF-control band values (means ± SEM of five to nine

measures from independent synaptoneurosomal preparations shown).

One-way ANOVA of raw data demonstrated significant effects of

GRGDSP treatment on all three kinases (p ¼ 0.001, p ¼ 0.002 and

p ¼ 0.007 for p-SFK, p-Pyk2 and p-FAK, respectively; **p < 0.01 for

comparison to aCSF-control values, Students–Neuman–Keuls (SNK)

post hoc test). (e) Synaptoneurosomes were collected after 5 min and

45 min of GRGDSP treatment (or after 5 min aCSF treatment for

controls) and then 400 lg of each sample was immunoprecipitated

with anti-Pyk2 and blotted for p-Pyk2(Y402). (f) Synaptoneurosomal

samples collected at 5 min were immunoprecipitated with anti-phos-

photyrosine 4G10 and then precipitates were processed for western

blot analysis of p-Pyk2(Y402). (g) Samples collected at 5 min were

immunoprecipitated with anti-FAK and then precipitates were proc-

essed for western blot analysis of p-FAK(Y397). (h) Samples collected

at 15 and 45 min were immunoprecipitated with anti-phosphotyrosine

antibody 4G10; precipitates were processed for western blot analysis

of FAK protein. The different preparations demonstrate that GRGDSP

treatment consistently increases Pyk2 and FAK phosphorylation.

838 J. A. Bernard-Trifilo et al.

� 2005 International Society for Neurochemistry, J. Neurochem. (2005) 93, 834–849

anti-phosphotyrosine and blots were probed with anti-FAK(Fig. 2h).

Some effects of RGD peptides may not be mediated byintegrin binding (Buckley et al. 1999) and there is evidencefor FAK and Pyk2 activation independent of integrinsignaling (Derkinderen et al. 1998; Heidinger et al. 2002).Therefore, two additional tests were carried out to determineif integrin binding is responsible for the results described inFig. 2. Synaptoneurosomes were treated with (a) smallpeptides that do not bind to integrins or (b) with GRGDSPin the presence of function blocking antisera to the b1integrin subunit. A large proportion of integrins expressed inrat forebrain are members of the b1 integrin family (Pinkstaffet al. 1999) and signaling though these would be blocked orsubstantially attenuated in the anti-b1 treated preparations.Tyrosine-isoleucine-glycine-serine-arginine (YIGSR), a fre-quently used control peptide in integrin studies (Pierschba-cher and Ruoslahti 1984; Buckley et al. 1999), had no effecton SFK, Pyk2, or FAK activation at 10 lM (Fig. 3a).Similarly, a control peptide with the core sequence replacedwith RAD (i.e. GRADSP) (10 lM-1 mM), did not increaseFAK and SFK autophosphorylation in synaptoneurosomes(data not shown). Incubating synaptoneurosomes with anti-b1 (1 lg/100 lL) 5 min before and throughout GRGDSPapplication blocked GRGDSP-induced phosphorylation of

FAK(Y397) and SFK(Y418) as assessed at 5 min (Fig. 3b),as expected if the phosphorylation events were mediated byRGD-binding, b1-containing integrins.

The SFK antibody used in the above experiments recog-nizes the tyrosine 418 site conserved across multiple Srckinases including Src and Fyn, two family members that arelocalized at synapses (Boxall and Lancaster 1998; Sala andSheng 1999; Ali and Salter 2001). To determine if one orboth of these kinases is targeted by integrins, synaptoneu-rosomes were treated with GRGDSP (10 lM) for 5 min,immunoprecipitated with anti-phosphotyrosine antibody4G10, and probed with anti-Src or anti-Fyn. GRGDSPinduced comparable increases in p-Src and p-Fyn levels(mean 3.8- and 3.4-fold increases relative to paired controlvalues, respectively).

Fibronectin initiates integrin signaling

Among the extracellular matrix proteins expressed in brain,fibronectin is a likely ligand for integrin-mediated effects onglutamatergic transmission in hippocampus. It contains theRGD sequence (Potts and Campbell 1996) recognized byseveral integrin dimers (a5b1, a3b1, avß1, avß3 and avß8)(Milner and Campbell 2002) expressed in forebrain (Pink-staff et al. 1999) and, from immunochemical studies, appearsto be enriched in synaptic complexes (Bahr et al. 1997;Nishimura et al. 1998; Rodriguez et al. 2000; Chun et al.2001; Kramar et al. 2002; Dong et al. 2003). Moreover, the‘classic’ fibronectin receptor (a5b1 integrin) is concentratedat PSD-95-immunoreactive dendritic spines in dissociatedhippocampal neurons (Bernard-Trifilo and Gall, unpublisheddata). Therefore, to test if native ligands activate synaptictyrosine kinase signaling, full length, soluble human fibro-nectin (1.25 lM) or GRGDSP (10 lM) was applied tosynaptoneurosomes, which were then processed for westernblot analysis using phosphospecific antibodies recognizingp-SFK(Y418), p-Pyk2(Y402), or p-FAK(Y397) (as above).Figure 4 shows (at top) representative western blots from anindividual experiment and (at bottom) quantification ofresults from five to nine independent GRGDSP experiments(i.e. separate synaptoneurosomal preparations and separatewestern blots) and four to nine independent fibronectinexperiments. The blots show that both fibronectin andGRGDSP induced rapid and robust increases in tyrosinephosphorylation of all three kinases: in this particularinstance the levels of phosphorylation of all three is reducedfrom the 5 to 30 min treatment intervals. Quantification ofband densities demonstrated that fibronectin significantlyincreased SFK, Pyk2, and FAK tyrosine phosphorylation,with mean increases of 2.3-, 3.1-, 6.5-fold control values,respectively, at the 5 min time point. For treatments withboth ligands, levels of pSFK and pFAK immunoreactivitieswere not significantly different from control values at30 min, whereas p-Pyk2 levels declined from 5 to 30 minbut were still significantly elevated at the later time point.

(a)

(b)

Fig. 3 GRGDSP effects are sequence- and integrin-specific. Repre-

sentative western blots of synaptoneurosomes treated with aCSF

(CON), 10 lM GRGDSP (RGD), 10 lM YIGSR, or 10 lM

GRGDSP + 1 lg/100 lL anti-b1 (RGD + b1). Blots show levels of

immunoreactivity for p-SFK(Y418), p-Pyk2(Y402), and p-FAK(Y397)

in two separate experiments, presented in panels (a) and (b) (15 lg of

protein/lane). (a) Blots show that in contrast to the effects of GRGDSP,

YIGSR did not increase p-SFK(Y418), p-Pyk2(Y402), or p-FAK(Y397)

levels at 5 or 30 min. (b) Western blots show that GRGDSP (RGD)-

induced increases in p-FAK(Y397) and p-Src(Y418) are blocked by

treatment with neutralizing antibody to the b1 integrin subunit (5 min

GRGDSP treatment interval shown).

Synaptic integrins regulate NMDA receptors 839

� 2005 International Society for Neurochemistry, J. Neurochem. (2005) 93, 834–849

NMDA receptor phosphorylation after integrin binding

Given that integrins and glutamate receptors are bothconcentrated in synapses, the question arises as to whetherkinase signaling initiated by new integrin binding results intyrosine phosphorylation of glutamate receptor and, inparticular, NMDA receptor proteins. This was tested byincubating synaptoneurosomes with GRGDSP (10 lM) for15 and 45 min; samples were then immunoprecipitated withanti-phosphotyrosine 4G10 and processed for immunoblotanalysis of NR2A and NR2B subunits of the NMDA receptor(Sala and Sheng 1999; Ali and Salter 2001; Cheung andGurd 2001; Heidinger et al. 2002). Figure 5(a) shows blotsfrom an individual experiment, and quantification of blotsfrom four separate experiments, and demonstrates that therewere significant increases in tyrosine phosphorylation of bothNR2A and NR2B after 15 min of GRGDSP treatment (mean7.2- and 2.8-fold increases relative to paired control values,respectively); phospho-NR2A and -NR2B levels returned to

control, or in some instances below control, values after45 min. These effects were replicated with GRGDSP con-centrations of 500 lM and 2 mM (Fig. 5b) and with recip-rocal immunoprecipitation strategies (Figs 5c and d). Asshown in Fig. 5(c), immunoprecipitation with an antiserumthat detects both NR2A and NR2B subunits (i.e. anti-NR2A/B) followed by western analysis with anti-phosphotyrosine4G10 demonstrated marked increases in phosphoproteinlevels after 15 min GRGDSP treatment. Similarly, westernblots of samples precipitated with anti-NR2B (Fig. 5d) oranti-NR2A (not shown) and probed with 4G10 demonstratedrobust increases in phosphorylation of both subunits withoutchanges in total subunit protein levels in the immunoprecip-itates (Fig. 5c) or the whole homogenates (Fig. 5d).

Ligand effects on NMDA receptor phosphorylation

are Src-dependent

Several lines of evidence indicate that SFK signaling iscritical for full integrin and FAK activation (Schlaepfer and

Fig. 4 Fibronectin and GRGDSP induce similar increases in

SFK(Y418), Pyk2(Y402), and FAK(Y397) phosphorylation. Synapto-

neurosomes were treated with aCSF (CON), 10 lM GRGDSP (RGD),

or 1.25 lM fibronectin (FN) for 5 and 30 min. Immunoblots show levels

of p-SFK(Y418), p-Pyk2(Y402), and p-FAK(Y397) 15 lg protein/lane.

Bar graph shows quantification of western blots with experimental

band densities expressed as a function of paired (i.e. within blot)

control values (mean ± SEM values shown from five to nine inde-

pendent GRGDSP experiments and four to nine independent fibro-

nectin experiments). One-way ANOVA of raw data demonstrated

significant effects of both GRGDSP (p ¼ 0.001, 0.002 and 0.007) and

fibronectin (p ¼ 0.004, 0.005 and 0.0001) treatment on p-SFK, p-Pyk2

and p-FAK, respectively (*p < 0.05, **p < 0.01, and ***p < 0.001 for

comparison to yoked control values, SNK).

(a) (c)

(d)

(e)

(b)

Fig. 5 GRGDSP increases phosphorylation of NMDA receptor sub-

units NR2A and NR2B. Synaptoneurosomes were treated with aCSF

(CON) or GRGDSP (RGD) at 10 lM for (a), (c), (d), and (e) and at

doses indicated for (b). (a) Samples were collected at 15 and 45 min,

then 400 lg of each was processed for immunoprecipitation with anti-

phosphotyrosine antibody 4G10 followed by western blot analysis of

NR2A and NR2B. As shown in the representative immunoblot (left),

GRGDSP induced a marked increases in tyrosine phosphorylation of

NR2A and NR2B. The graph (right) shows the magnitude of this effect

as assessed by densitometric analysis of immunoblots from four

independent experiments (mean ± SEM shown; p ¼ 0.035 for NR2A,

p ¼ 0.024 for NR2B, one-way ANOVA; *p < 0.05 for comparison to

control values, SNK). (b) similar increases in NR2B tyrosine phos-

phorylation were observed with GRGDSP treatment at 10 lM, 0.5 mM,

and 2 mM (c) and (d) Samples were collected at 15 min, then 400 lg

of each was processed for immunoprecipitation with anti-NR2A/B (c)

or anti-NR2B (d); immunoprecipitates were processed for western

blots probed with anti-phosphotyrosine (4G10) or anti-NR2B [the two

blots in (d) are from the same precipitates]. As shown, treatment with

GRGDSP increased levels of tyrosine-phosphorylated NR2A/B (c) and

NR2B (d) immunoreactivities but had no effect on total NR2B levels

within the precipitates (d). (e) Western blots of synaptoneurosomes

treated with GRGDSP for 15 min show no change in levels of total

NR2B immunoreactivity with treatment. (15 lg protein/lane, all

panels).

840 J. A. Bernard-Trifilo et al.

� 2005 International Society for Neurochemistry, J. Neurochem. (2005) 93, 834–849

Hunter 1996; Vuori 1998; Avraham et al. 2000; Huang et al.2001). The specific Src tyrosine kinase inhibitor PP2 wasused to test if this is also the case for integrin signaling atsynapses. Synaptoneurosomes were pretreated with PP2 at0.5–10 lM for 5 min, followed by GRGDSP (10 lM) plusPP2 for 5 or 30 min. As shown in Fig. 6(a), PP2 (10 lM)completely blocked RGD-induced phosphorylation ofSFK(Y418) at 5 min and, at this dose, reduced SFKactivation below control levels. The antagonist also reducedFAK activation by the integrin ligand but did not detectablyaffect Pyk2 phosphorylation. At 1 lM, but not 0.5 lM, PP2blocked RGD-induced increases in p-SFK and attenuatedligand-induced increases in p-FAK(Y397) (Fig. 6b). Thecontrol compound PP3 (10 lM) had no effect on RGD-induced activation of SFK or FAK (not shown).

The above results opened the way to tests of whethertyrosine kinase cascades mediate the effects of integrinbinding on NMDA receptor phosphorylation. Synaptoneu-rosomes were pretreated with PP2 (1 or 10 lM) for 5 min,followed by GRGDSP (10 lM) plus PP2 for an additional 15

or 45 min. Samples were immunoprecipitated with anti-phosphotyrosine and immunoblotted for NMDA receptorsubunits NR2A and NR2B. The SFK inhibitor completelyblocked RGD-induced increases in tyrosine phosphorylationof NR2A and NR2B (Fig. 7) whereas the control compoundPP3 (1 and 10 lM) had no effect; similar complete elimin-ation of RGD-induced increases in NR2A and NR2B subunitphosphorylation were obtained in three separate experiments.

GRGDSP increases tyrosine phosphorylation

and synaptic responses of NMDA receptors

in hippocampal slices

The above evidence for GRGDSP-induced increases inNMDA receptor phosphorylation is consistent with previousresults of whole cell clamp analyses showing that the peptideligand causes Src-dependent increases in NMDA receptor-mediated excitatory postsynaptic currents (EPSCs) in thehippocampal Schaffer collateral system (Lin et al. 2003).However, prior electrophysiological analyses did not estab-lish that GRGDSP acts through integrins to achieve thiseffect. To test this, and the specific prediction that b1-containing integrins are involved, the effects of GRGDSPand b1-neutralizing antisera on NMDA receptor-mediatedevoked potentials were examined using hippocampal fieldCA1 mini-slices.

The AMPA receptor antagonist CNQX was used topharmacologically isolate NMDA receptor-mediatedcomponents of field EPSPs (fEPSP) (Muller and Lynch1988). As shown in Fig. 8(a), CNQX caused a markedreduction in fEPSPs recorded from the CA1b stratumradiatum in response to stimulation of the Schaffer collaterals,leaving a small response that was completely eliminated by

(a)

(b)

Fig. 6 Src tyrosine kinase inhibitor PP2 blocks GRGDSP effects on

SFK and FAK, but not Pyk2, phosphorylation. (a) Synaptoneurosomes

were treated with aCSF (CON), 10 lM GRGDSP, or 10 lM GRGDSP

plus PP2 (following 5 min PP2 pre-incubation) for 5 and 30 min. Blots

show treatment effects on levels of p-SFK(Y418), p-Pyk2(Y402), and

p-FAK(Y397) in a representative experiment using PP2 at 10 lM

(15 lg of protein/lane). These effects were replicated across four and

two independent experiments with PP2 at 10 and 1 lM, respectively.

(b) Immunoblot showing PP2-dose effects on synaptoneurosome

levels of p-SFK(Y418) and p-FAK(Y397). Synaptoneurosomes were

treated with aCSF (CON), 10 lM GRGDSP alone (RGD), GRGDSP in

combination with different PP2 doses (as above) or 1 lM PP2 alone;

immunoblots were processed with anti-SFK(Y418) or anti-FAK(Y397).

As shown, 1 lM PP2 blocked effects of GRGDSP on both

p-SFK(Y418) and p-FAK(Y397) but, when applied alone, had no effect

on basal phospho-kinase levels.

Fig. 7 Src family kinase inhibitor PP2 blocks GRGDSP effects on

NMDA receptor phosphorylation. Synaptoneurosomes were treated

with aCSF (CON), 10 lM GRGDSP, or 10 lM GRGDSP plus 10 lM

PP2 (following a 5 min PP2 pre-incubation). Samples were collected

at 15 and 45 min and 400 lg of each homogenate was processed for

immunoprecipitation with anti-phosphotyrosine antibody 4G10 fol-

lowed by western blot analysis of levels of tyrosine-phosphorylated

NR2A and NR2B. GRGDSP induced marked increases in levels of

tyrosine-phosphorylated NR2A and NR2B at 15 min and this effect

was completely blocked by PP2. Similar results were obtained in 4

independent experiments with PP2 at 10 lM and, in individual

experiments, with PP2 at 0.5 and 1 lM.

Synaptic integrins regulate NMDA receptors 841

� 2005 International Society for Neurochemistry, J. Neurochem. (2005) 93, 834–849

the NMDA receptor antagonist AP5 (Fig. 8a). GRGDSP at2 mM caused a steady increase in both the amplitude(Fig. 8a) and half-width (Fig. 8b) of this CNQX-insensitivefEPSP (Figs 8a and b). Figures 8(c and d) summarize results

collected during the last 70 min of the experiments with thedata plotted as percentage change in fEPSP size relative tobaseline (pre-GRGDSP) responses in the presence of CNQX.GRGDSP caused a significant increase in evoked response

(a)

(c)

(e)

(f) (g)

(b)

(d)

Fig. 8 GRGDSP increases NMDA receptor-mediated synaptic

responses and NMDA receptor tyrosine phosphorylation in hippo-

campal slices. (a, b) After establishing a stable 10 min baseline for the

CA1-Schaffer collateral response, 10 lM CNQX was infused into the

bath causing a dramatic drop in response (mean ± SEM) amplitude

(a) and half-width (b) that reached a plateau 50 min postinfusion (n ¼4). Infusion of 2 mM GRGDSP in the presence of CNQX significantly

increased the NMDA receptor-mediated synaptic response, an effect

that was substantially greater for half-width than for amplitude. The

enhanced NMDA receptor-mediated synaptic response was com-

pletely blocked by the NMDA receptor antagonist AP5. (c) and (d)

show data from the last 70 min of the experiments plotted in (a) and

(b), respectively, but with values expressed as the mean (± SEM)

percentage of the ‘CNQX baseline’; the increases in field EPSP

amplitude and duration occur over the same time period but are

markedly different in size. (e) Representative traces collected during

(i) initial baseline recordings, (ii) 50 min after the start of CNQX

infusion, (iii) 60 min after the co-application of GRGDSP and CNQX,

and (iv) 20 min after the introduction of AP5 in the presence of CNQX

and GRGDSP. (f, g) Western blots showing that GRGDSP increases

NMDA receptor tyrosine phosphorylation in acute hippocampal slices.

(f) Hippocampal mini-slices (field CA1) were treated with GRGDSP

(RGD) or normal aCSF (CON), tested physiologically and, after 1 h,

harvested for phosphotyrosine immunoprecipitation followed by

western blot analysis of tyrosine-phosphorylated NR2A (left) and

NR2B (right). (g) Conventional hippocampal slices were treated with

normal aCSF (CON) or 3 mM GRGDSP (RGD) in aCSF for 1, 2, or 3 h

and then harvested for immunoprecipitation analysis of tyrosine-

phosphorylated NR2B (procedures as in f). Quantification of phospho-

NR2B band densities, normalized to total NR2B protein level per

sample, demonstrated that tyrosine-phosphorylated NR2B levels were

elevated to 2.3-, 1.7- and 1.2-fold control values with 1, 2, and 3 h

GRGDSP treatments, respectively.

842 J. A. Bernard-Trifilo et al.

� 2005 International Society for Neurochemistry, J. Neurochem. (2005) 93, 834–849

amplitude (+241 ± 85% for n ¼ 4; p ¼ 0.009, paired t-test;Fig. 8c) and half-width (+842 ± 126%, p < 0.0001, Fig. 8d)as compared to baseline (all in the presence of CNQX). Theenhanced responses were completely blocked 20 min afterintroducing 50 lM AP5 into the bath, thereby demonstratingthat they were NMDA receptor-mediated.

To test if GRGDSP treatment elicits NMDA receptorphosphorylation in hippocampal slices, as demonstratedabove for synaptoneurosomes, CA1 mini-slices were har-vested at the end of 1 h exposure to GRGDSP (2 mM), atwhich time the ligand effect on NMDA receptor-mediatedfEPSPs was maximal and stable; the tissue was thenprocessed for phosphotyrosine immunoprecipitation fol-lowed by western blot analysis of NR2A and NR2B levels.GRGDSP increased tyrosine-phosphorylated NR2A andNR2B levels as compared to levels in paired control mini-slices treated with aCSF alone (Fig. 8f). Similar increases inNMDA receptor tyrosine phosphorylation were observed inthree experiments each for p-NR2A and p-NR2B. Separateexperiments evaluated NMDA receptor tyrosine phosphory-lation in conventional acute hippocampal slices treated instatic aCSF bath with 3 mM GRGDSP. The ligand increasedtyrosine-phosphorylated NR2B across 1–3 h of treatment(Fig. 8g); normalization of p-NR2B band densities to totalNR2B protein levels per sample (determined from westernblots of aliquots from whole homogenates collected prior tophosphotyrosine immunoprecipitation) demonstrated thatrelative phospho-NR2B levels were increased over twofoldwith 1 h GRGDSP treatment and remained elevated at 1.7-fold control levels at 2 h.

Local application of function-blocking antisera to a2 andb1 integrin subunits was used to test if GRGDSP-inducedincreases in NMDA receptor synaptic responses are integrin-mediated. The great majority of integrins expressed inhippocampus include the b1 subunit, whereas a2 integrinexpression has not been detected in this region (Pinkstaffet al. 1999); thus if GRGDSP effects on synaptic responsesare indeed integrin-mediated, one would expect them to beblocked by function-blocking antisera to the b1 subunit butnot by antisera to the a2 subunit.

Following a 50 min bath infusion of CNQX to fully blockAMPA receptor responses, an infusion pipette filled with

(a)

(c)

(b)

Fig. 9 Function blocking antisera to b1 integrin blocks GRGDSP-

induced increases in NMDA receptor-mediated synaptic responses.

Following a 50 min infusion of 10 lM CNQX, a pipette was used to

infuse 2 mM GRGDSP in aCSF (vehicle control; open circles), anti-a2

integrin plus 2 mM GRGDSP in aCSF (gray filled circles), or anti-b1

integrin plus 2 mM GRGDSP in aCSF (black filled circles) within the

bath slightly upstream from the test slice. Local perfusion of pipette

contents began 20 min before introduction of 2 mM GRGDSP into the

bath. (a, b) Function blocking antisera to b1 integrin completely

inhibited GRGDSP-induced increases in the amplitude (a) and half-

width (b) of the NMDA receptor-mediated synaptic response,

respectively. In contrast, response size was not significantly different

in a comparison of slices treated with GRGDSP alone and GRGDSP

plus anti-a2 integrin. (c) Representative field traces recorded from

slices treated with aCSF (top two traces), anti-a2 integrin (middle two

traces) and anti-b1 integrin (bottom two traces) collected 30 min after

the start of CNQX (baseline) infusion and 30 min after adding bath

GRGDSP.

Synaptic integrins regulate NMDA receptors 843

� 2005 International Society for Neurochemistry, J. Neurochem. (2005) 93, 834–849

aCSF plus 2 mM GRGDSP (vehicle control group), anti-a2integrin plus 2 mM GRGDSP in aCSF, or anti-b1 integrinplus 2 mM GRGDSP in aCSF was positioned in the bathupstream from the recording electrode. GRGDSP wasincluded in the pipette with the antisera to assure thatejection of pipette contents did not wash away GRGDSPbeing applied within the bath. In all cases, pipette reagentinfusion began 20 min before 2 mM GRGDSP was intro-duced into the bath and continued for the remaining periodover which recordings were collected. As illustratedin Fig. 9(a and b), bath and pipette infusion ofaCSF + GRGDSP alone or in combination with pipetteapplied anti-a2 integrin induced comparable increases inNMDA receptor-mediated synaptic response amplitude(aCSF ¼ +63 ± 33% vs. anti-a2 ¼ +58 ± 16%; p ¼ 0.77)and half-width (aCSF ¼ +150 ± 67% vs. anti-a2 ¼+216 ± 73%; p ¼ 0.32). In marked contrast, experimentalslices treated with pipette infusion of anti-b1 had no increasein response amplitude ()20 ± 17.5%, p ¼ 0.14 forcomparison to baseline) or half-width ()4 ± 18.7%) aftermore than 50 min of GRGDSP treatment. Differences in theeffects of bath GRGDSP between anti-a2- and anti-b1-treated slices were highly significant (p ¼ 0.002 for ampli-tude; p ¼ 0.002 for half-width, unpaired t-test). These resultsindicate that GRGDSP effects on synaptic responses are b1integrin-mediated and show that effects of antibody infusionare antigen-specific.

Discussion

The results presented here constitute the first direct evidencethat synaptic integrins activate local tyrosine kinase signalingand NMDA receptor phosphorylation in association withenhanced NMDA receptor-mediated function in maturehippocampal synapses. Treatment of adult rat forebrainsynaptoneurosomes with the broad-spectrum, integrin ligandGRGDSP induced rapid activation of FAK, Pyk2 and Srckinases and Src-dependent phosphorylation of NMDAreceptor subunits NR2A and NR2B. These results werereplicated with fibronectin, a native RGD-containing matrixprotein expressed in adult brain (Hoffman et al. 1998b) andknown to bind integrins that are localized at synapses (Biet al. 2001; Schuster et al. 2001; Kramar et al. 2002; Chanet al. 2003; Dong et al. 2003). GRGDSP activation of FAKand SFK was blocked by b1 integrin neutralizing antisera,and kinase activation was not observed following treatmentswith the non-integrin binding YIGSR control peptide(Pierschbacher and Ruoslahti 1984), thereby supporting theconclusion that these synaptic phosphorylation events wereintegrin-mediated.

Studies of acute hippocampal slices verified that GRGDSPsimilarly increases NMDA receptor tyrosine phosphorylationin intact tissue and that these changes are associated withb1-integrin-dependent enhancement of NMDA receptor-

mediated synaptic responses in hippocampal field CA1. Inthe electrophysiological analysis, changes in NMDA receptorresponses were slow to emerge, becoming evident within 15–20 min of treatment onset and reaching maximal levels20 min later. This is consistent with prior studies ofGRGDSP effects on synaptic physiology and likely reflectsboth the time needed for the ligand to penetrate the sliceparenchyma and for signaling events to modify NMDAreceptor function. In contrast to NR2 phosphorylation insynaptoneurosomes, RGD-induced increases in phospho-NR2 levels within hippocampal slices were present beyond1 h of treatment onset. This difference in time course mayalso reflect the staggered latency to drug effect withGRGDSP movement into the slice tissue as well asdifferences in the cellular machinery available to sustainincreased receptor phosphorylation in the slice preparation.Regarding these possibilities it is noteworthy that we havesimilarly observed more protracted increases in the phos-phorylation of signaling kinases in slice as opposed tosynaptoneurosomal preparations (e.g. elevated pSFK levelsafter 30 and 60 min GRGDSP treatment; Pineda and Gall,unpublished observations). The present electrophysiologicalresults complement earlier findings showing that GRGDSPinduces Src-dependent increases in NMDA receptor-medi-ated EPSCs in hippocampal slices (Lin et al. 2003) and areconsistent with the hypothesis that ligation of synapticintegrins initiates local tyrosine kinase signaling and regu-lates functional properties of NMDA receptors in adultsynapses. However, the present studies leave open thequestion of which native integrin ligand or ligands canactivate this regulatory pathway in situ. Although fibronectinwas shown to activate synaptic integrin signaling, animportant goal of future studies will be to test the effectsof this and other b1 integrin ligands expressed in hippocam-pus on NMDA receptor-mediated synaptic responses.

It is well established in non-neuronal systems that integrinbinding activates non-receptor tyrosine kinases includingFAK, Pyk2 and SFKs (Vuori 1998; Avraham et al. 2000).The present results demonstrate that similar patterns ofintegrin-mediated signaling are present at adult synapses.The specific FAK and Pyk2 phosphorylation sites evaluatedare Src-family SH2 binding domains (Avraham et al. 2000)and, in line with this, FAK(Y397) and Pyk2(Y402) phos-phorylation was associated with similarly rapid and robustSFK activation. The Src inhibitor PP2 blocked and attenu-ated SFK and FAK activation, respectively, but did notsignificantly attenuate Pyk2(Y402) phosphorylation. Thelatter result is not proof of the direction of the signalingstream but is consistent with Pyk2 activation preceding thatof the SFKs in the synaptic signaling cascade in this system,as has been proposed elsewhere (Salter and Kalia 2004), andindicates that recurrent phosphorylation by Src does not playa major role in the sustained (> 30 min) Pyk2(Y402)phosphorylation observed here. Enduring Pyk2 activation

844 J. A. Bernard-Trifilo et al.

� 2005 International Society for Neurochemistry, J. Neurochem. (2005) 93, 834–849

with integrin ligation has been noted in other systems(Duong and Rodan 2000) and may reflect the diverse signals(e.g. protein kinase C activities, elevated calcium levels,G-protein signaling) that activate this kinase (Avraham et al.2000; Duong and Rodan 2000).

Several studies have shown that phosphorylation alters thefunctional properties of NMDA receptors (Boxall andLancaster 1998; Salter and Kalia 2004). Tyrosine phos-phorylation of the receptor occurs in conjunction with LTP(Rosenblum et al. 1996; Rostas et al. 1996), ischemia (Liuet al. 2003), seizures (Moussa et al. 2001), neurotrophinsignaling (Lin et al. 1998) and metabotropic glutamatereceptor stimulation (Heidinger et al. 2002); binding ofRGD-ligands to integrins can now be added to this list.Among the various kinases implicated in NMDA receptorphosphorylation (Suen et al. 1998; Cheung and Gurd 2001),the SFKs are of particular interest because they areimplicated in NMDA receptor-dependent processes in LTPand learning (Sala and Sheng 1999; Ali and Salter 2001;Salter and Kalia 2004). Src and Fyn phosphorylate similarsites on NR2A and NR2B, and SFK phosphorylationpotentiates NMDA receptor-mediated EPSCs (Chen andLeonard 1996; Kohr and Seeburg 1996; Yu and Salter 1999)either through direct effects on channel properties orindirectly, through regulation of protein interactions thatcould have effects on NMDA receptor signaling or stabilitywithin the membrane (Salter and Kalia 2004; for discussion).Thus, integrin activation of Src and Fyn likely contributes to,and may fully account for, PP2-sensitive increases in NMDAreceptor tyrosine phosphorylation (present study) andGRGDSP-induced increases in NMDA receptor-mediatedsynaptic responses. However, Src and Fyn may be involvedin separate signaling streams involving Pyk2 and FAK,respectively. Although FAK and Pyk2 share some sequenceidentity and functions, there are differences in their regula-tion and signaling (Schaller and Sasaki 1997; Girault et al.1999) and, in particular, in their Src family kinase associ-ations following LTP induction (Lauri et al. 2000). More-over, evidence for SFK phosphorylation of specific tyrosinesites on the NR2 subunits most strongly implicate Src andFyn in the phosphorylation of NR2A and NR2B, respectively(Salter and Kalia 2004). Additional studies are needed todetermine if synaptic integrin ligation activates multipleSFK-signaling pathways that influence different NMDAreceptor targets and functions.

It is noteworthy that a number of neuromodulatorreceptors employ the same signaling pathways used byintegrins to influence NMDA receptor phosphorylation stateand function. For example, EphB receptors are transmem-brane tyrosine kinases localized to the postsynaptic elementat glutamatergic synapses. As with integrins, EphB receptorbinding to its ligand (ephrinB) leads to SFK activation, Src-dependent NR2 phosphorylation and increased NMDAreceptor channel activity (Takasu et al. 2002). Similarly,

stimulation of group 1 metabotropic glutamate receptors(Baude et al. 1993) results in sequential activation of Pyk2,Src and Fyn, Src-dependent phosphorylation of NR2A andNR2B, and increased NMDA receptor-mediated currents(Heidinger et al. 2002). It will be of interest to determine ifcontrol of this signaling pathway by a particular receptor issituationally dependent or if signals from the differentreceptors sum at the level of Src activation (Salter and Kalia2004) and, beyond this, in the regulation of NMDA receptoractivities.

Although the specific integrins regulating synaptic tyro-sine kinase signaling and NMDA receptor function are notknown, the list of possibilities is significantly narrowed byevidence that signaling is activated by both GRGDSP andfibronectin, and that both kinase activation and increases inNMDA receptor-mediated synaptic responses are blocked byb1 integrin neutralizing antisera. The 18 known a and nineknown b integrin subunits form a limited number of dimercombinations; a subset of these are both RGD- andfibronectin-binding, and a further subset (a3b1, a5b1,avß1, a8b1) contains b1 (Pinkstaff et al. 1999; Plow et al.2000; Milner and Campbell 2002). Thus, these four dimers,which are all expressed by cortical and hippocampal neurons(Einheber et al. 1996; Pinkstaff et al. 1999), are leadingcandidates for mediating integrin ligand effects reported here.Among these there is independent evidence that a3b1, a5b1,and a8b1 influence glutamatergic synapse function. All threeare localized at synapses (Bahr and Lynch 1992; Einheberet al. 1996; Kramar et al. 2002) and studies using subunit-specific neutralizing antisera have shown that both a5 and a3contribute to consolidation of LTP (Chun et al. 2001; Kramaret al. 2002). More recent work has demonstrated impair-ments in LTP and spatial memory with reduced a3, a5, anda8 expression (Chan et al. 2003). Finally, the a3b1 ligandreelin, activates mRNA translation in cortical synaptoneuro-somes (Dong et al. 2003).

The findings reported here accord with the broad hypo-thesis that synapses in adult brain are dynamic adhesionjunctions. Past experiments using brief exposures to neutral-izing antibodies (Chun et al. 2001; Kramar et al. 2002) led tothe conclusion that changes in integrin binding occur withinminutes of intense afferent activity and are essential forstable synaptic modifications (e.g. LTP). The present resultsdescribe signaling cascades set in motion within minutes ofnew integrin binding in adult synaptoneurosomes andincreased NMDA receptor function within 15 min of excessligand presentation to acute hippocampal slices. The ques-tions now arise as to whether the transient, activity-depend-ent events that recruit integrin involvement in synapticplasticity (such as LTP-inducing stimulation) entail newintegrin binding and, if so, does this occur via increasedligand availability, integrin modifications (e.g. membraneinsertion, activation), or both? There is evidence that intenseneuronal activity (seizures) in adult hippocampus increases

Synaptic integrins regulate NMDA receptors 845

� 2005 International Society for Neurochemistry, J. Neurochem. (2005) 93, 834–849

levels of fibronectin and other matrix ligands and isassociated with the release of proteases that degrade matrixproteins (Hoffman et al. 1998a; Endo et al. 1999; Komaiet al. 2000; Gall and Lynch 2004). These activities couldincrease concentrations of RGD-ligands and exposure ofextant matrix binding domains to synaptic integrins (Daviset al. 2000). Moreover, postsynaptic responses to bursts ofafferent activity have components known to regulate integrinactivation and/or surface expression (Kolanus and Seed1997; Hughes and Pfaff 1998); prominent among these areelevations in intracellular calcium and in activities of proteinkinase C and the cysteine protease calpain (Sacktor et al.1993; Vanderklish et al. 1995; Rock et al. 1997; Yan et al.2001). Studies in our laboratories have shown that AMPAreceptor stimulation increases a5 and b1 integrin surfaceexpression (Gall and Lin, unpublished observations) butfurther analyses are needed to determine if natural activationof synaptic AMPA receptors has similar effects.

Finally, there is the question of how fluctuations in integrinsignaling, and in particular signaling through to the NMDAreceptor, affects neuronal operations beyond the proposedcontribution to LTP stabilization. Integrin ligation modulatesgene expression in mature neurons (Gall et al. 2003) as innon-neural cells (Giancotti and Ruoslahti 1999) but theseeffects do not appear to depend upon Src-mediated NMDAreceptor phosphorylation (i.e. they are not blocked by PP2;Gall, unpublished data). The known contributions of NMDAreceptors to synaptic potentiation occur as part of themilliseconds long dendritic response to high frequencyactivity (Malenka and Nicoll 1999); there is no evident role,at least in LTP induction, for the more enduring increases inNMDA receptor phosphorylation and synaptic functiondescribed here. An alternative and speculative possibilityconcerns the reversal of LTP that is obtained when lowfrequency stimulation is applied during the period, lastingtens of minutes, after the induction of potentiation. LTPreversal depends on NMDA receptors (Bear and Malenka1994; Kramar and Lynch 2003), and thus would presumablybe facilitated by the enhancement of NMDA receptorcurrents that is attendant to integrin signaling. Similarly,synaptic activity can effect enduring (minutes to hour long)NMDA receptor-dependent changes that lower the thresholdfor long-term depression (Mockett et al. 2002). Theseobservations suggest that integrin binding could have theseemingly paradoxical consequences of both stabilizing LTPand, through integrin-mediated NMDA receptor modifica-tions, increasing the likelihood of its erasure.

Acknowledgements

The authors thank Drs Patrick Sullivan and Edward Modafferi for

technical advice and Dr Julie Lauterborn for comments on the

manuscript. This work was supported by NINDS grant NS37799

and NIA grant AG00358 to CMG and NIMH grant MH61077 to

GL. JAB was supported by NIA Training Program in the

Neurobiology of Aging (AG00096-18).

References

Ali D. W. and Salter M. W. (2001) NMDA receptor regulation by Srckinase signalling in excitatory synaptic transmission and plasticity.Curr. Opin. Neurobiol. 11, 336–342.

Avraham H., Park S. Y., Schinkmann K. and Avraham S. (2000) RAFTK/Pyk2-mediated cellular signalling. Cell. Signal. 12, 123–133.

Bahr B. A. and Lynch G. (1992) Purification of an Arg-Gly-Aspselective matrix receptor from brain synaptic plasma membranes.Biochem. J. 281, 137–142.

Bahr B. A., Staubli U., Xiao P., Chun D., Ji Z.-X., Esteban E. T. andLynch G. (1997) Arg-Gly-Asp-Ser-Selective adhesion and thestabilization of long-term potentiation: Pharmacological studiesand the characterization of a candidate matrix receptor. J. Neurosci.17, 1320–1329.

Baude A., Nusser Z., Roberts J. D., Mulvihill E., McIlhinney R. A. andSomogyi P. (1993) The metabotropic glutamate receptor (mGluR1alpha) is concentrated at perisynaptic membrane of neuronal sub-populations as detected by immunogold reaction. Neuron 11, 771–787.

Bear M. F. and Malenka R. C. (1994) Synaptic Plasticity: LTP and LTD.Curr. Opin. Neurobiol. 4, 389–399.

Bi X., Lynch G., Zhou J. and Gall C. M. (2001) Polarized distribution ofa5 integrin in dendrites of hippocampal and cortical neurons. J.Comp. Neurol. 435, 184–193.

Bongiorno-Borbone L., Onofri F., Giovedi S., Ferrari R., Girault J.-A.and Benfenati F. (2002) The translocation of focal adhesion kinasein brain synaptosomes is regulated by phosphorylation and actinassembly. J. Neurochem. 81, 1212–1222.

Boxall A. R. and Lancaster B. (1998) Mini Review: Tyrosine kinasesand synaptic transmission. Eur. J. Neurosci. 10, 2–7.

Buckley C. D., Pilling D., Henriquez N. V., Parsonage G., Threlfall K.,Scheel-Toellner D., Simmons D. L., Akbar A. N., Lord J. M. andSalmon M. (1999) RGD peptides induce apoptosis by directcaspase-3 activation. Nature 397, 534–539.

Calderwood D. A., Huttenlocher A., Kiosses W. B., Rose D. M.,Woodside D. G., Schwartz M. A. and Ginsberg M. H. (2001)Increased filamin binding to beta-integrin cytoplasmic domainsinhibits cell migration. Nat. Cell Biol. 3, 1060–1068.

Capaldi D., Rosario R., Esteban E. T. and Bahr B. A. (1997) A 27-kDamatrix receptor from rat brain synaptosomes: Selective recognitionof the Arg-Gly-Asp-Ser domain and unique resistance to calcium-dependent. Neurosci. Res. 28, 275–279.

Chan C.-S., Weeber E. J., Kurup S., Sweatt J. D. and Davis R. L. (2003)Integrin requirement for hippocampal synaptic plasticity and spa-tial memory. J. Neurosci. 23, 7107–7116.

Chen S. J. and Leonard J. P. (1996) Protein tyrosine kinase-mediatedpotentiation of currents from cloned NMDA receptors. J. Neuro-chem. 67, 194–2000.

Cheung H. H. and Gurd J. W. (2001) Tyrosine phosphorylation of theN-methyl-D-aspartate receptor by exogenous and postsynapticdensity-associated Src-family kinases. J. Neurochem. 78, 524–534.

Chun D., Gall C. M., Bi X. and Lynch G. (2001) Evidence that integrinscontribute to multiple stages in the consolidation of long termpotentiation. Neuroscience 105, 815–829.

Davis G. E., Bayless K. J., Davis M. J. and Meininger G. A. (2000)Regulation of tissue injury responses by the exposure of matri-cryptic sites within extracellular matrix molecules. Am. J. Pathol.156, 1489–1498.

846 J. A. Bernard-Trifilo et al.

� 2005 International Society for Neurochemistry, J. Neurochem. (2005) 93, 834–849

Davis M. J., Wu X., Nurkiewicz T. R., Kawasaki J., Gui P., Hill M. A.and Wilson E. (2002) Regulation of ion channels by integrins. CellBiochem. Biophys. 36, 41–66.

Della Moret R., Squillacioti C., Garbi C., Derkinderen P., Belisario M.A., Girault J. A., Di Natale P., Nitsch L. and Staiano N. (2000)Echistatin inhibits pp125FAK autophosphorylation, paxillin phos-phorylation and pp125FAK–paxillin interactions in fibronectin-adherent malanoma cells. Eur. J. Biochem. 267, 5047–5054.

Derkinderen P., Siciliano J. C., Toutant M. and Girault J.-A. (1998)Differential regulation of FAK+ and PYK2/Cakb, two relatedtyrosine kinases, in rat hippocampal slices: effects of LPA, carba-chol, depolarization and hyperosmolarity. Eur. J. Neurosci. 10,1667–1675.

Dong E., Caruncho H., Liu W. S., Smalheiser N. R., Grayson D. R.,Costa E. and Guidotti A. (2003) A reelin–integrin receptor inter-action regulates Arc mRNA translation in synaptoneurosomes.Proc. Natl Acad. Sci. USA 100, 5479–5484.

Duong L. T. and Rodan G. A. (2000) PYK2 is an adhesion kinase inmacrophages, localized to podosomes and activated by b2-integrinligation. Cell Motil. Cytoskel. 47, 174–188.

Einheber S., Schnapp L. M., Salzer J. L., Cappiello Z. B. and Milner T.A. (1996) Regional and ultrastructural distribution of the a8 inte-grin subunit in developing and adult rat brain suggests a role insynaptic function. J. Comp. Neurol. 370, 105–134.

Endo A., Nagai N., Urano T., Takada Y., Hashimoto K. and Takada A.(1999) Proteolysis of neuronal cell adhesion molecule by the tissueplasminogen activator-plasmin system after kainate injection in themouse hippocampus. Neurosci. Res. 33, 1–8.

Gall C. M. and Lynch G. (2004) Integrins, synaptic plasticity, and epi-leptogenesis. Adv. Exp. Med. Biol. 548, 12–33.

Gall C. M., Pinkstaff J. K., Lauterborn J. L., Xie Y. and Lynch G. (2003)Integrins regulate neuronal neurotrophin gene expression througheffects on voltage sensitive calcium channels. Neuroscience 118,925–940.

Giancotti F. G. and Ruoslahti E. (1999) Integrin signaling. Science 285,1028–1032.

Girault J.-A., Costa A., Dirkinderen P., Studler J.-M. and Toutant M.(1999) FAK and PYK2/CAKb in the nervous system: a linkbetween neuronal activity, plasticity, and survival. Trends Neurosci.22, 257–263.

Grabham P. W., Foley M., Umeojiako A. and Goldberg D. J. (2000)Nerve growth factor stimulates coupling of beta1 integrin to dis-tinct transport mechanisms in the filopodia of growth cones. J. CellSci. 113, 3003–3012.

Hanks S. K., Calalb M. B., Harper M. C. and Patel S. K. (1992) Focaladhesion protein-tyrosine kinase phosphorylated in response to cellattachment to fibronectin.Proc.Natl. Acad. Sci.USA89, 8487–8491.

Heidinger V., Manzerra P., Wang X. Q., Strasser U., Yu S.-P., Choi D. W.and Behrens M. M. (2002) Metobotropic glutamate receptor1-induced upregulation of NMDA receptor current: Mediationthrough Pyk2/Src-family kinase pathway in cortical neurons.J. Neurosci. 22, 5452–5461.

Hoffman K. B., Martinez J. and Lynch G. (1998a) Proteolysis of celladhesion molecules by serine proteases: a role in long termpotentiation? Brain Res. 811, 29–33.

Hoffman K. B., Pinkstaff J. K., Gall C. M. and Lynch G. (1998b) Seizureinduced synthesis of fibronectin is rapid and age dependent:implications for long-term potentiation and sprouting. Brain Res.812, 209–215.

Huang Y.-Q., Lu W. Y., Ali D. W. et al. (2001) CAKb/PYK2 kinase is asignaling link for induction of long-term potentiation in CA1hippocampus. Neuron 29, 485–496.

Hughes P. E. and Pfaff M. (1998) Integrin affinity modulation. TrendsCell Biol. 8, 359–364.

Huhtala P., Humphries M. J., McCarthy J. B., Tremble P. M., Werb Z.and Damsky C. H. (1995) Cooperative signaling by a5b1 anda4b1 integrins regulated metalloproteinase gene expression infibroblasts adhering to fibronectin. J. Cell Biol. 129, 867–879.

Huttenlocher A., Palecek S. P., Lu Q., Zhang W., Mellgren R. L.,Lauffenburger D. A., Ginsberg M. H. and Horwitz A. F. (1997)Regulation of cell migration by the calcium-dependent proteasecalpain. J. Biol. Chem. 272, 32 719–32 722.

Hynes R. O. (1992) Integrins: Versatility, modulation, and signaling incell adhesion. Cell 69, 11–25.

Katagiri T., Takahashi T., Sasaki T., Nakamura S. and Hattori S. (2000)Protein-tyrosine kinase Pyk2 is involved in interleukin-2 produc-tion by Jurkat T cells via its tyrosine 402. J. Biol. Chem. 275,19 645–19 652.

Keely S. J., Calandrella S. O. and Barrett K. E. (2000) Carbachol-sti-mulated transactivation of epidermal growth factor receptor andmitogen-activated protein kinase in T(84) cells is mediated byintracellular Ca(2+), PYK-2, and P60(src). J. Biol. Chem. 275,12 619–12 625.

Kohr G. and Seeburg P. H. (1996) Subtype-specific regulation ofrecombinant NMDA receptor-channels by protein tyrosine kinasesof the src family. J. Physiol. 492, 445–452.

Kolanus W. and Seed B. (1997) Integrins and inside-out signal trans-duction: converging signals from PKC and PIP3. Cur. Opin. CellBiol. 9, 725–731.

Komai S., Matsuyama T., Matsumoto K., Kato K., Kobayashi M.,ImamuraK.,YoshidaS.,UgawaS. andShiosakaS. (2000)Neuropsinregulates an early phase of Schaffer-collateral long term potentiationin the murine hippocampus. Eur. J. Neurosci. 12, 1479–1486.

Kramar E. A. and Lynch G. (2003) Developmental and regional differ-ences in the consolidation of long term potentiation. Neurosci. 118,387–398.

Kramar E. A., Bernard J. A., Gall C. M. and Lynch G. (2002) Alpha3integrin receptors contribute to the consolidation of long-termpotentiation. Neurosci. Lett. 110, 29–39.

Kramar E. A., Bernard J. A., Gall C. M. and Lynch G. (2003) Integrinsmodulate fast excitatory transmission at hippocampal synapses. J.Biol. Chem. 278, 10 722–10 730.

LaBaron R. G., Hernandez R. V., Orfila J. E. and Martinez J. L. J. (2003)An integrin binding peptide reduced rat CA1 hippocampal long-term potentiation during the first few minutes following theta burststimulation. Neurosci. Lett. 339, 199–202.

Lafrenie R. and Yamada K. (1996) Integrin-dependent signal transduc-tion. J. Cell. Biochem. 61, 543–553.

Lai and Clark (1979) Isolation of brain mitochondria. Meth. Enzymol.55, 55–60.

Laser M., Willey C. D., Jiang W., Cooper G.4th, Menick D. R., Zile M.R. and Kuppuswamy D. (2000) Integrin activation and focalcomplex formation in cardiac hypertrophy. J. Biol. Chem. 275,35 624–35 630.

Lauri S. E., Taira T. and Rauvala H. (2000) High-frequency synapticstimulation induces association of fyn and c-src to distinct phos-phorylated components. Neuroreport 11, 997–1000.

Lauterborn J. C., Lynch G., Vanderklish P., Arai A. and Gall C. M.(2000) Positive modulation of AMPA receptors increases neurot-rophin expression by hippocampal and cortical neurons. J. Neu-rosci. 20, 8–21.

Lin S. Y., Wu K., Levine E. S., Mount H. T. J., Cuen P. C. and Black I.B. (1998) BDNF acutely increases tyrosine phosphorylation of theNMDA receptor subunit 2B in cortical and hippocampal postsy-naptic densities. Mol. Brain Res. 55, 20–27.

Lin B., Arai A. C., Lynch G. and Gall C. M. (2003) Integrins regulateNMDA receptor-mediated synaptic currents. J. Neurophysiol. 89,2874–2878.

Synaptic integrins regulate NMDA receptors 847

� 2005 International Society for Neurochemistry, J. Neurochem. (2005) 93, 834–849

Lipfert L., Haimovich B., Schaller M. D., Cobb B. S., Parsons T. J. andBrugge J. S. (1992) Integrin-dependent phosphorylation and acti-vation of the protein tyrosine kinase pp125FAK in platelets. J. CellBiol. 119, 905–912.

Liu S., Calderwood D. A. and Ginsberg M. H. (2000) Integrin cyto-plasmic domain-binding proteins. J. Cell Sci. 113, 3563–3571.

Liu Y., Hou X. Y., Zhang G. Y. and Xu T. L. (2003) 1-type voltage-gatedcalcium channel attends regulation of tyrosine phosphorylation ofNMDA receptor subunit 2A induced by transient brain ischemia.Brain Res. 972, 142–148.

Malenka R. C. and Nicoll R. A. (1999) Long-term potentiation – adecade of progress. Science 285, 1870–1874.

McKerracher L., Chamoux M. and Arregui C. O. (1996) Role of lamininand integrin interactions in growth cone guidance. Mol. Neurobiol.12, 95–116.

Milner R. and Campbell I. L. (2002) The integrin family of cell adhesionmolecules has multiple functions within the CNS. J. Neurosci. Res.69, 286–291.

Miranti C. K. and Brugge J. S. (2002) Sensing the environment: a his-torical perspective on integrin signal transduction. Nat. Cell Biol. 4,83–90.

Mockett B., Coussens C. and Abraham W. C. (2002) NMDA receptor-mediated metaplasticity during the induction of long-term depres-sion by low-frequency stimulation. Eur. J. Neurosci. 15, 1819–1826.

Moro L., Dolce L., Cabodi S. et al. (2002) Integrin-induced epidermalgrowth factor (EGF) receptor activation requires c-Src andp130Cas and leads to phosphorylation of specific EGF receptortyrosines. J. Biol. Chem. 277, 9405–9414.

Moussa R. C., Ikeda-Douglas C. J., Thakur V., Milgram N. W. and GurdJ. W. (2001) Seizure activity results in increased tyrosine phos-phorylation of the N-methyl-D-aspartate receptor in the hippo-campus. Mol. Brain Res. 95, 36–47.

Muller D. and Lynch G. (1988) Long-term potentiation differentiallyaffects two components of synaptic responses in hippocampus.Proc. Natl Acad. Sci. USA 85, 9346–9350.

Nishimura S. L., Boylen K. P., Einheber S., Milner T. A., Ramos D. M.and Pytela R. (1998) Synaptic and glial localization of the integrinavb8 in mouse and rat brain. Brain Res. 791, 271–282.

Parsons J. T., Martin K. H., Slack J. K., Taylor J. M. and Weed S. A.(2000) Focal adhesion kinase; a regulator of focal adhesiondynamics and cell movement. Oncogene 19, 5606–5613.

Pierschbacher M. and Ruoslahti E. (1984) Cell attachment activity offibronectin can be duplicated by small synthetic fragments of themolecule. Nature 304, 30–33.

Pinkstaff J. K., Detterich J., Lynch G. and Gall C. (1999) Integrinsubunit gene expression is regionally differentiated in adult brain.J. Neurosci. 19, 1541–1556.

Plow E. F., Haas T. A., Zhang L., Loftus J. and Smith J. W. (2000)Ligand binding to integrins. J. Biol. Chem. 275, 21 785–21 788.

Porter J. C. and Hogg N. (1998) Integrins take partners: cross-talkbetween integrins and other membrane receptors. Trends Cell Biol.8, 390–396.

Potts J. R. and Campbell I. D. (1996) Structure and function of fibro-nectin molecules. Matrix Biol. 15, 313–320.

Rock M. T., Brooks W. H. and Roszman T. L. (1997) Calcium-dependent signaling pathways in T cells: potential role of calpain,protein tyrosine phosphatase 1b, and p130Cas in integrin-mediatedsignaling events. J. Biol. Chem. 272, 33 377–33 383.

Rodriguez M. A., Pesold C., Liu W. S., Kriho V., Guidotti A., Pappas G.D. and Costa E. (2000) Colocalization of integrin receptors andreelin in dendritic spine postsynaptic densities of adult nonhumanprimate cortex. Proc. Natl Acad. Sci. USA 97, 3550–3555.

Rosenblum K., Dudai Y. and Richter-Levin G. (1996) Long-termpotentiation increases tyrosine phosphorylation of the N-methyl-

D-aspartate receptor subunit 2B in rat dentate gyrus in vivo. Proc.Natl Acad. Sci. USA 93, 10 457–10 460.

Rostas J. A. P., Brent V. A., Voss K., Errington M. L., Bliss T. V. P. andGurd J. W. (1996) Enhanced tyrosine phosphorylation of the 2Bsubunit of the N-methyl-D-aspartate receptor in long-term potenti-ation. Proc. Natl Acad. Sci. USA 93, 10 452–10 456.

Sacktor T. C., Osten P., Valsamis H., Jiang X., Niak M. U. and SubletteE. (1993) Persistent activation of the zeta isoform of protein kinaseC in the maintenance of long-term potentiation. Proc. Natl Acad.Sci. USA 90, 8342–8346.

Sala C. and Sheng M. (1999) The fyn art of N-methyl-D-aspartatereceptor phosphorylation. Proc. Natl Acad. Sci. USA 96, 335–337.

Salter M. W. and Kalia L. V. (2004) Src kinases: a hub for NMDAreceptor regulation. Nat. Rev. Neurosci. 5, 317–328.

Sanders M. A. and Basson M. D. (2000) Collagen IV-dependent ERKactivation in human Caco-2 intestinal epithelial cells requires focaladhesion kinase. J. Biol. Chem. 275, 38 040–38 047.

Sanna P. P., Berton F., Cammalleri M., Tallent M. K., Siggins G. R.,Bloom F. E. and Francesconi W. (2000) A role of Src kinase inspontaneous epileptiform activity in the CA3 region of the hip-pocampus. Proc. Natl Acad. Sci. USA 97, 8653–8657.

Schaller M. D. and Sasaki T. (1997) Differential signaling by the focaladhesion kinase and cell adhesion kinase b. J. Biol. Chem. 272,25 319–25 325.

Schlaepfer D. D. and Hunter T. (1996) Evidence for in vivo phos-phorylation of the Grb2 SH2-domain binding site on focal adhe-sion kinase by Src-family protein-tyrosine kinases. Mol. Cell. Biol.16, 5623–5633.

Schlaepfer D. and Hunter T. (1998) Integrin signalling and tyrosinephosphorylation: just the FAKs? Trends Cell Biol. 8, 151–157.

Schmid R. S. and Anton E. S. (2003) Role of integrins in the develop-ment of the cerebral cortex. Cereb. Cortex 13, 219–224.

Schuster T., Krug M., Stalder M., Hackel N., Gerardy-Schahn R. andSchachner M. (2001) Immunoelectron microscopic localization ofthe neural recognition molecules L1, NCAM, and its isoformNCAM180, the NCAM-associated polysialic acid, b1 integrin andthe extracellular matrix molecule tenascin-R in synapses of theadult rat hippocampus. J. Neurobiol. 49, 142–158.

Schwartz M. A., Schaller M. D. and Ginsberg M. H. (1995) Integrins:emerging paradigms of signal transduction. Annu. Rev. Cell Dev.Biol. 11, 549–599.

Sheetz M. P., Felsenfeld D. P. and Galbraith C. G. (1998) Cell migration:Regulation of force on extracellular-matrix-integrin complexes.Trends Cell Biol. 8, 51–54.

Shi C. S., Sinnarajah S., Cho H., Kozasa T. and Kehrl J. H. (2000)G13alpha-mediated PYK2 activation. PYK2 is a mediator ofG13alpha-induced serum response element-dependent transcrip-tion. J. Biol. Chem. 275, 24 470–24 476.

Staubli U., Chun D. and Lynch G. (1998) Time-dependent reversal oflong-term potentiation by an integrin antagonist. J. Neurosci. 18,3460–3469.

Staubli U., Vanderklish P. W. and Lynch G. (1990) An inhibitor ofintegrin receptors blocks LTP. Behav. Neural Biol. 53, 1–5.

Suen P.-C., Wu K., Xu J.-L., Lin S.-Y., Levine E. S. and Black I. B.(1998) NMDA receptor subunits in the postsynaptic density of ratbrain: expression and phosphorylation by endogenous proteinkinases. Mol. Brain Res. 59, 215–228.

Sussman M. A., Welch S., Walker A., Klevitsky R., Hewett T. E., PriceR. L., Schaefer E. and Yager K. (2000) Altered focal adhesionregulation correlates with cardiomyopathy in mice expressingconstitutively active rac1. J. Clin. Invest. 105, 875–886.

Takasu M. A., Dalva M. B., Zigmond R. E. and Greenberg M. E. (2002)Modulation of NMDA receptor-dependent calcium influx and geneexpression through EphB receptors. Science 295, 491–495.

848 J. A. Bernard-Trifilo et al.

� 2005 International Society for Neurochemistry, J. Neurochem. (2005) 93, 834–849

Thodeti C. K., Adolfsson J., Juhas M. and Sjolander A. (2000) Leuko-triene D(4) triggers an association between gbetagamma subunitsand phospholipase C-gamma1 in intestinal epithelial cells. J. Biol.Chem. 275, 9849–9853.

Vanderklish P., Saido T. C., Gall C., Arai A. and Lynch G. (1995)Proteolysis of spectrin by calpain accompanies theta-burst stimu-lation in cultured hippocampal slices. Mol. Brain Res. 32, 25–35.

Vuori K. (1998) Integrin signaling: tyrosine phosphorylation events infocal adhesions. J. Membr. Biol. 165, 191–199.

Williams J. M., Guevremont D., Kennard J. T., Mason-Parker S. E.,Tate W. P. and Abraham W. C. (2003) Long-term regulation ofN-methyl-D-aspartate receptor subunits associated synaptic proteinsfollowing hippocampal synaptic plasticity. Neuroscience 118,1003–1013.

Wu X., Mogford J. E., Platts S. H., Davis G. E., Meininger G. A. andDavis M. J. (1998) Modulation of calcium current in arteriolarsmooth muscle by alphav beta3 and alpha5 beta1 integrin ligands.J. Cell Biol. 143, 241–252.

Yamada K. M. and Miyamoto S. (1995) Integrin transmembranesignaling and cytoskeletal control. Cur. Opin. Cell Biol. 7, 681–689.

Yan B., David A., Calderwood D. A., Yaspan B. and Ginsberg M. H.(2001) Calpain cleavage promotes talin binding to the b3 integrincytoplasmic domain. J. Biol. Chem. 276, 28 164–28 170.

Yu X.-M. and Salter M. W. (1999) Src, a molecular switch governinggain control of synaptic transmission mediated by N-methyl-D-aspartate receptors. Proc. Natl Acad. Sci. USA 96, 7697–7704.

Synaptic integrins regulate NMDA receptors 849

� 2005 International Society for Neurochemistry, J. Neurochem. (2005) 93, 834–849