614 WISSENSCHAFT

BIOspektrum | 06.07 | 13. Jahrgang

NILS BROSE

MAX-PLANCK-INSTITUT FÜR EXPERIMENTELLE MEDIZIN, GÖTTINGEN

Erkrankungen aus dem Autismus-Spektrum – autism spectrum disordersoder ASDs – sind häufig auftretende Entwicklungsstörungen des Nerven-systems. Epidemiologische und humangenetische Studien lassen auf einegenetische Prädisposition als wesentliche Ursache von ASDs schließen.In einer Reihe jüngst charakterisierter Fälle monogen erblicher ASDs sindGene betroffen, deren Proteinprodukte an der Ausbildung glutamatergerSynapsen zwischen Nervenzellen beteiligt sind. In Mausmodellen mitDeletionen dieser Gene ist die synaptische Signalübertragung fundamen-tal gestört. Offenbar sind also zumindest einige ASD-Formen Synapto-pathien.

Epidemiologie und Genetik von ASDs

ó Unter dem Begriff autism spectrum disor-ders oder ASDs wird eine Gruppe von Krank-heiten zusammengefasst, zu der neben demprototypischen Autismus das Asperger-Syn-drom, der atypische Autismus (pervasive de-velopmental disorder not otherwise specified –PDD-NOS) sowie syndromische Formen desAutismus gehören. Die Prävalenz von typi-schem Autismus liegt bei etwa 0,2 Prozent.Betrachtet man das gesamte ASD-Spektrum,liegt die entsprechende Rate bei 0,6 Prozent[1].

Dass ASDs ganz wesentlich durch geneti-sche Faktoren verursacht werden, konntein Zwillingsstudien, Familienanalysen undhumangenetischen Untersuchungen nachge-wiesen werden. Die Konkordanzrate beimonozygoten Zwillingen etwa liegt je nachDefinition der Phänotypen bei bis zu 92 Pro-zent[1]. Genomweite Suchen nach ASD-Sus-zeptibilitätsgenen und zytogenetischenAbnormalitäten bei ASDs ergaben Hinweisefür ASD-Lozi auf allen Chromosomen. Auchdie Assoziation von ASD-verwandten Symp-

tomen mit komplexeren monogen erblichenSyndromen wie etwa dem Rett-Syndrom, demFragile-X Syndrome, der tuberösen Hirn-sklerose oder der Neurofibromatose vom Typ1 weist auf einen elementaren genetischenBeitrag zur Ätiologie von ASDs hin[2–4].

Monogen erbliche ASD-Formen

Monogen erbliche, nicht-syndromale Formenvon ASDs waren lange Zeit unbekannt. EineReihe neuester Studien hat diese Situationjedoch grundlegend geändert. Die erste Ent-deckung zweier klar abgrenzbarer Mutatio-nen als Ursachen für ASDs gelang einem fran-zösisch-schwedischen Konsortium ausHumangenetikern und Psychiatern im Rah-men einer Studie an Geschwisterpaaren mittypischem Autismus oder Asperger-Syn-drom[5]. Bei zwei Brüderpaaren wurden dabeiMutationen in den X-chromosomalen GenenNLGN3 beziehungsweise NLGN4 entdeckt.Die beiden Gene kodieren für zwei neurona-le synaptische Adhäsionsproteine, Neuroli-gin-3 und Neuroligin-4 (NL-3, NL-4), die alsRegulatoren der Synapsenentwicklung undSynapsenfunktion gelten[6–9]. Neuroligine(NLs) sind Typ-1-Transmembranproteine miteiner extrazellulären, katalytisch inaktivenSerinesterase-Domäne und einer kurzenintrazellulären Domäne, die unter andereman PDZ-Domänen verschiedener Gerüstpro-teine binden kann (Abb. 1 und 2). Diebeschriebenen Mutationen führen zu einemFunktionsverlust von NL-3 und NL-4[10, 11].Dass Mutationen in NLGN3 und NLGN4 ASDsund geistige Behinderung verursachen kön-nen, wurde in zwei Folgestudien bestä-tigt[12, 13]. Allerdings sind durch NL-Mutatio-nen verursachte ASD-Formen eher selten,denn in mehreren systematischen Analysenan ASD-Patientengruppen konnten keineMutationen in NLGN-Genen nachgewiesenwerden[14–17].

In einer zweiten Studie entdeckte dasselbeKonsortium französischer und schwedischerForscher drei verschiedene ASD-verursa-chende Mutationen im SHANK3-Gen[18].SHANK3 gehört zu einer Familie wichtiger

Neurogenetik

Autismus – wenn Nervenzellenkontaktscheu sind

˚ Abb. 1: Domänenstrukturen von NL-3, NL-4, NX-1α, NX-1β und SHANK3. Pfeile unter denjeweiligen Domänenstrukturen zeigen die Sequenzbereiche von NL-4 und SHANK3 an, in denendie in autistischen Patienten identifizierten Mutationen zu Abbrüchen führen[9, 22]. Die Position deridentifizierten missense-Mutation R451C in NL-3[9] ist durch eine Pfeilspitze gekennzeichnet. ANK:Ankyrin-repeats, CH: O-Glykosylierungsstellen, EGF: EGF-ähnliche Domäne, LNS: laminin-A/neu-rexin/sex-hormone-binding-globulin repeats, PDZ: postsynaptic-density-95/discs-large/zona-occludens-1-Domäne, PDZ-BM: PDZ-Domänen-Bindungsmotiv, ppI: Cortactin-Bindungsdomäne,PRC: prolinreiche Cluster, SAM: sterile alpha motif, SH3: Src-Homologiedomäne 3.

589_621_BIOsp_0607.qxd:589_621 18.09.2007 8:21 Uhr Seite 614

Gerüstproteine glutamaterger Synapsen undinteragiert über Ankyrin-repeats, SH3-, PDZ-und sterile-alpha-motif-Domänen sowie pro-linreiche Regionen mit verschiedenen Gerüst-proteinen, Zytoskelettproteinen und Rezep-toren (Abb. 2). Bis auf eine Ausnahme füh-ren die beschriebenen Mutationen zu einemFunktionsverlust von SHANK3. Besondersinteressant ist ein Mutationstyp, bei dem esbei einem Geschwisterpaar entweder zu einerterminalen Deletion in 22qter oder zu einerentsprechenden 22qter-Trisomie kommt,wobei die 22qter-Deletion zu einem hemi-zygoten Verlust des SHANK3-Gens undAutismus mit schwerer Sprachstörung führt,während die entsprechende Trisomie einAsperger-Syndrom mit besonderer Sprach-begabung verursacht.

NL-3, NL-4 und SHANK3 sind Komponen-ten eines Proteinnetzwerks, das für die mole-kulare Organisation glutamaterger Synapsenin Neuronen verantwortlich ist (Abb. 2). DassFunktionsverlust-Mutationen in allen dreientsprechenden Genen zu ASDs führen, zeigt,dass eine Fehlentwicklung oder eine Fehl-funktion glutamaterger Synapsen ursächlichan der Entstehung von ASDs beteiligt seinkann. Diese Annahme hat jüngst durch einesystematische Studie über Genkopienzahl-Variationen bei ASD überzeugende Unter-stützung erhalten[19]. Neben einer Reihe ande-rer Mutationen wurde im Rahmen dieser Stu-die bei zwei Schwestern eine hemizygoteDeletion identifiziert, die eine Reihe kodie-render Exone von NRXN1 eliminiert. DasNRXN1-Gen kodiert mithilfe zweier Promo-toren zwei Typen von Neurexin-1-Proteinen(NX-1α und NX-1β). Beide NX-1-Variantensind Typ-1-Transmembranproteine mit unter-schiedlich langen extrazellulären Bereichen,die LNS-Domänen und EGF-ähnliche Domä-nen enthalten, und derselben kurzen car-boxyterminalen Domäne, die wie im Fallder NLs unter anderem an PDZ-Domänenverschiedener Gerüstproteine binden kann

(Abb. 1). In diesem Zusammenhang beson-ders bemerkenswert ist der Umstand, dassNXs mit NLs eine transsynaptische adhäsiveVerbindung herstellen, die die Assemblierungund Positionierung des präsynaptischenTransmitter-Sekretionsapparats und des post-synaptischen Signaltransduktionsapparatsan glutamatergen Synapsen koordiniert(Abb. 2). Die beschriebene Deletion imNRXN1-Gen führt zu einem Verlust der NX-1α- und NX-1β-Expression. In derselben[19]

sowie in einer methodisch ähnlichen Studieüber de novo-Variationen der Genkopienzahlin autistischen Patienten[20] wurden eine Rei-he weiterer möglicher ASD-Gene identifiziert,deren Produkte direkt oder indirekt die syn-aptische Transmission an glutamatergen Syn-apsen beeinflussen können[19].

NXs, NLs und SHANKs in

glutamatergen Synapsen

Synapsen sind strukturell und funktionellhoch komplexe Kontaktstellen zwischen Ner-venzellen (Abb. 3). Sowohl der präsynapti-sche Transmitter-Sekretionsapparat als auchder postsynaptische Signalrezeptionsapparatbestehen aus Dutzenden von Proteinen (Abbil-

dung 2 zeigt eine Auswahl), die für die räum-liche und zeitliche Genauigkeit synaptischerÜbertragungsprozesse und ihre Modulationetwa bei Adaptations- und Gedächtnispro-zessen verantwortlich sind. Das synaptischetrafficking, die prä- und postsynaptische Ver-ankerung und die Funktion dieser Proteinewerden durch eine Kombination aus Zellad-häsionsproteinen und Gerüstproteinengewährleistet, die die prä- und postsynapti-schen Kompartimente einer Synapse struk-turell zusammenhalten und funktionell auf-einander abstimmen. NXs, NLs und SHANKssind für den Aufbau und die Funktion diesesProteinnetzwerks von elementarer Bedeutung(Abb. 2).

NXs und NLs – auch die als Produkte vonASD-Genen identifizierten Proteine NX-1, NL-

3 und NL-4 – bilden eine transsynaptischeAdhäsionsbrücke, die als „Kristallisations-punkt“ für die Rekrutierung prä- und post-synaptischer Proteine dient. MausgenetischeStudien zeigen, dass eine Deletion aller α-NXs zu einer präsynaptischen Fehlfunktionspannungsabhängiger Ca2+-Kanäle[21] und zueinem partiellen Verlust postsynaptischerNMDA-Rezeptoren[22] führt. Aufgrund dieserVeränderungen, die durch die Unterbrechungder indirekten Interaktion von NXs mit Ca2+-Kanälen (über das Zytoskelett) und NMDA-Rezeptoren (über NLs und PSD-95) (Abb. 2)in NX-defizienten Tieren erklärt werden kann,kommt es zu einer postnatal letalen Störungder synaptischen Signalübertragung. Die Eli-mination von NX-1 in ASD-Patienten führtwahrscheinlich in Analogie zu den Befundenan NX-defizienten Mäusen zu einer ähnlichen,allerdings weniger stark ausgeprägten syn-aptischen Fehlfunktion, die für die ASD-Symp-tome verantwortlich gemacht werden kann.

Die gleichzeitige Elimination von NL-1,NL-2 und NL-3 in mutanten Mäusen verur-sacht ähnlich wie im Fall der α-NX-Deletioneine postnatal letale synaptische Fehlfunk-tion, die zumindest teilweise auf den partiel-len Verlust postsynaptischer Glutamatrezep-toren vom NMDA- und AMPA-Typ zurückzu-führen ist[23]. Die selektive Deletion von NL-1, das den beiden in ASD involvierten NLsfunktionell sehr ähnlich ist[9], verursacht mil-dere phänotypische Veränderungen, diedurch einen partiellen Verlust postsynapti-scher NMDA-Rezeptoren und eine entspre-chend gestörte synaptische Übertragung cha-rakterisiert sind[24]. Der Verlust postsynapti-scher NMDA- und AMPA-Rezeptoren nach NL-Deletion kann durch deren indirekte Inter-aktion erklärt werden, die im Fall von NMDA-Rezeptoren über PSD-95 und im Fall vonAMPA-Rezeptoren über PSD-95 und dasGerüstprotein Stargazin erfolgt und in denMausmutanten gestört ist (Abb. 2). Die phä-notypischen Konsequenzen der Deletion von

¯ Abb. 2: NLs bilden mit NXs eine transsynaptische Adhäsionsbrücke, dieeine zentrale Rolle bei der Ausbildung funktionell essentieller synaptischerProteinnetzwerke spielt. Über intrazelluläre Interaktionen mit Gerüst- undZytoskelettproteinen rekrutieren NXs spannungsabhängige Ca2+-Kanäleund Komponenten des Sekretionsapparats zur Präsynapse. Auf analogeWeise rekrutieren postsynaptische NLs postsynaptische Rezeptoren.SHANK-Proteine sind postsynaptische Gerüstproteine, die ebenfalls überdefinierte Protein-Protein-Interaktionen zur Rekrutierung postsynaptischerRezeptoren und anderer Signalproteine sowie zu deren Verankerung imZytoskelett beitragen. CASK: calcium/calmodulin-dependent serine pro-tein kinase, GKAP: guanylate kinase domain-associated protein, mGlu-Rezeptor: metabotroper Glutamatrezeptor, Mint: Munc18 interacting pro-tein, NMDA: N-Methyl-D-Aspartat, ProSAP: proline-rich synapse-associatedprotein, PSD95: postsynaptic density protein of 95 kDa, SAP90: synapto-some associated protein of 90 kDa, SAPAP: SAP90/PSD95-associatedprotein, SHANK: SH3 domain and ankyrin repeat containing protein, Veli:vertebrate Lin 7.

615

BIOspektrum | 06.07 | 13. Jahrgang

589_621_BIOsp_0607.qxd:589_621 18.09.2007 8:21 Uhr Seite 615

NL-3 oder NL-4 in autistischen Patienten soll-ten denen der NL-1-Deletion in Mäusen ähn-lich sein. Tatsächlich zeigen Pilotexperimen-te, dass NL-4-defiziente Mäuse mit ASD ver-gleichbare Defizite im Sozialverhalten auf-weisen.

Hinsichtlich der Funktion von SHANK3existieren derzeit keine Daten aus mausge-netischen Experimenten. Für alle bekanntenSHANK-Isoformen wurden Interaktionen mitverschiedenen Komponenten des postsynap-tischen Signaltransduktionsapparats nach-gewiesen[25–27]. Funktionelle Studien an kul-tivierten Neuronen legen den Schluss nahe,dass SHANK-Proteine sowohl an der Ausbil-dung postsynaptischer Dornen glutamater-ger Synapsen als auch über die Interaktionmit anderen Gerüst- und Zytoskelettproteinenan der Rekrutierung postsynaptischer Rezep-toren beteiligt sind. Angesichts dieser Befun-de ist davon auszugehen, dass ein Verlust derFunktion von SHANK3 in vivo, wie er bei denbeschriebenen autistischen Patienten auftritt,zu einer Reduktion postsynaptischer Dornenund einer gestörten synaptischen Transmis-sion führt, während eine SHANK-3-Über-expression, wie sie im Fall des Patienten mit22qter-Trisomie und einem zusätzlichenSHANK-3-Gen zu erwarten ist, den gegentei-ligen Effekt haben sollte.

NXs, NLs und SHANKs – Trio infernale

oder Spitze des Eisbergs?

Ob die Befunde aus Studien über NLGN3-,NLGN4-, NRXN1- und SHANK3-Mutationen beiASD generalisierbar sind und zu einer allge-meinen Erklärung der Ätiologie von ASDs

herangezogen werden können, ist derzeitnoch unklar. Offensichtlich ist, dass eine Rei-he anderer Mutationen, die wahrscheinlichnicht direkt den glutamatergen oder irgend-einen anderen Typ synaptischer Signalüber-tragung stören, ebenfalls zu ASD führen kön-nen[4, 19, 20]. Das bedeutet allerdings nicht,dass in diesen Fällen nicht auch die synapti-sche Signalübertragung im Nervensystembeeinflusst ist. Vielmehr ist klar, dass die Stö-rung fast jeden neuronalen Prozesses ulti-mativ auch zu Veränderungen der synapti-schen Übertragungseigenschaften führt.

In der Gesamtschau zeigen jüngste geneti-sche Untersuchungen über monogene Ursa-chen von ASDs und Untersuchungen an ent-sprechenden Mausmutanten, dass eine Fehl-funktion glutamaterger Synapsen ursächlichan der Entstehung von ASDs beteiligt seinkann. NLGN3-, NLGN4-, NRXN1- und SHANK3-Mutationen als Ursachen von ASDs mögennur die Spitze des Eisbergs darstellen. Aller-dings zeigen die entsprechenden Befundedeutlich, dass ein die Funktion glutamater-ger Synapsen regulierendes Proteinnetzwerkunter Beteiligung von NL-3, NL-4, NX-1 undSHANK3 (Abb. 2) im Zentrum eines wichti-gen ASD-Suszeptibilitätsmechanismus steht.Die Verfügbarkeit entsprechend mutierterMauslinien bietet einen idealen Ansatzpunktfür die Entwicklung neuer Diagnose- und The-rapieverfahren für ASDs. ó

Literatur[1] Fombonne, E. (2005): J. Clin. Psychiatry 66 (Suppl. 10): 3–8.[2] Freitag, C. M. (2007): Mol. Psychiatry 12: 2–22.[3] Vorstman, J. A., Staal, W. G., van Daalen, E., vanEngeland, H., Hochstenbach, P. F., Franke, L. (2006): Mol.Psychiatry 11: 18–28.[4] Persico, A. M., Bourgeron, T. (2006): Trends Neurosci. 29:349–358.[5] Jamain, S., Quach, H., Betancur, C., Råstam, M.,Colineaux, C., Gillberg, I. C., Soderstrom, H., Giros, B.,Leboyer, M., Gillberg, C., Bourgeron, T., Paris AutismResearch International Sibpair Study (2003): Nat. Genet. 34:27–29.[6] Ichtchenko, K., Hata, Y., Nguyen, T., Ullrich, B., Missler,M., Moomaw, C., Südhof, T. C. (1995): Cell 81: 435–443.[7] Song, J. Y., Ichtchenko, K., Südhof, T. C., Brose, N. (1999):Proc. Natl. Acad. Sci. U.S.A. 96: 1100–1105.[8] Chih, B., Engelman, H., Scheiffele, P. (2005): Science 307:1324–1328.[9] Graf, E. R., Zhang, X., Jin, S. X., Linhoff, M. W., Craig, A. M.(2004): Cell 119: 1013–1026.

[10] Comoletti, D., De Jaco, A., Jennings, L. L., Flynn, R. E.,Gaietta, G., Tsigelny, I., Ellisman, M. H., Taylor, P. (2004): J.Neurosci. 24: 4889–4893.[11] Chih, B., Afridi, S. K., Clark, L., Scheiffele, P. (2004):Hum. Mol. Genet. 13: 1471–1477.[12] Laumonnier, F., Bonnet-Brilhault, F., Gomot, M., Blanc,R., David, A., Moizard, M. P., Raynaud, M., Ronce, N.,Lemonnier, E., Calvas, P., Laudier, B., Chelly, J., Fryns, J. P.,Ropers, H. H., Hamel, B. C., Andres, C., Barthélémy, C.,Moraine, C., Briault, S. (2004): Am. J. Hum. Genet. 74: 552–557.[13] Talebizadeh, Z., Lam, D. Y., Theodoro, M. F., Bittel, D. C.,Lushington, G. H., Butler, M. G. (2006): J. Med. Genet. 43: e21.[14] Blasi, F., Bacchelli, E., Pesaresi, G., Carone, S., Bailey, A.J., Maestrini, E., International Molecular Genetic Study ofAutism Consortium (IMGSAC) (2006): Am. J. Med. Genet. BNeuropsychiatr. Genet. 141: 220–221.[15] Gauthier, J., Bonnel, A., St-Onge, J., Karemera, L.,Laurent, S., Mottron, L., Fombonne, E., Joober, R., Rouleau,G. A. ( 2005): Am. J. Med. Genet. B Neuropsychiatr. Genet. 132:74–75.[16] Vincent, J. B., Kolozsvari, D., Roberts, W. S., Bolton, P. F.,Gurling, H. M., Scherer, S. W. (2004): Am. J. Med. Genet. BNeuropsychiatr. Genet. 129: 82–84.[17] Ylisaukko-oja, T., Rehnström, K., Auranen, M., Vanhala,R., Alen, R., Kempas, E., Ellonen, P., Turunen, J. A., Makkonen,I., Riikonen, R., Nieminen-von Wendt, T., von Wendt, L.,Peltonen, L., Järvelä, I. (2005): Eur. J. Hum. Genet. 13: 1285–1292.[18] Durand, C. M., Betancur, C., Boeckers, T. M., Bockmann,J., Chaste, P., Fauchereau, F., Nygren, G., Rastam, M.,Gillberg, I. C., Anckarsäter, H., Sponheim, E., Goubran-Botros, H., Delorme, R., Chabane, N., Mouren-Simeoni, M. C.,de Mas, P., Bieth, E., Rogé, B., Héron, D., Burglen, L., Gillberg,C., Leboyer, M., Bourgeron, T. (2007): Nat. Genet. 39: 25–27.[19] Autism Genome Project Consortium et al. (2007): Nat.Genet. 39: 319–328.[20] Sebat, J. et al. (2007): Science 316: 445–449.[21] Missler, M., Zhang, W., Rohlmann, A., Kattenstroth, G.,Hammer, R. E., Gottmann, K., Südhof, T. C. (2003): Nature423: 939–948.[22] Kattenstroth, G., Tantalaki, E., Südhof, T. C., Gottmann,K., Missler, M. (2004): Proc. Natl. Acad. Sci. U.S.A. 101: 2607–2612.[23] Varoqueaux, F., Aramuni, G., Rawson, R. L., Mohrmann,R., Missler, M., Gottmann, K., Zhang, W., Südhof, T. C., Brose,N. (2006): Neuron 51: 741–754.[24] Chubykin, A. A., Atasoy, D., Etherton, M. R., Brose, N.,Kavalali, E. T., Gibson, J. R., Südhof, T. C. (2007): Neuron 54:919–931.[25] Naisbitt, S. Kim, E., Tu, J. C., Xiao, B., Sala, C.,Valtschanoff, J., Weinberg, R. J., Worley, P. F., Sheng, M.(1999): Neuron 23: 569–582.[26] Sala, C., Piëch, V., Wilson, N. R., Passafaro, M., Liu, G.,Sheng, M. (2001): Neuron 31: 115–130.[27] Kreienkamp, H. J., Zitzer, H., Gundelfinger, E. D., Richter,D., Bockers, T. M. (2000): J. Biol. Chem. 275: 32387–32390.

Korrespondenzadresse:

Prof. Dr. Nils BroseAbteilung Molekulare NeurobiologieMax-Planck-Institut für Experimentelle MedizinHermann-Rein-Straße 3D-37075 GöttingenTel.: 0551-3899 725Fax: 0551-3899 715brose@em.mpg.dewww.em.mpg.de/site

˚ Abb. 3: NL-1 in Nervenzellen. Eine dergezeigten Nervenzellen (rot) wurde gene-tisch so manipuliert, dass sie eine grün fluo-reszierende NL-1-Variante bildet. Besondersin der Teilvergrößerung (rechts unten) istsichtbar, dass NL-1 (grün) in postsynapti-schen Dornen zu finden ist, die aus der Den-dritenoberfläche herausragen. Bild: ThomasDresbach (Universität Heidelberg) und NilsBrose.

616 WISSENSCHAFT

BIOspektrum | 06.07 | 13. Jahrgang

AUTORNils Brose1981–1987 Studium der Biochemie, Biologie und Physiologie in Tübingen und Oxford,UK. 1987–1990 Promotion an der Ludwig-Maximilians-Universität in München und amMPI für Psychiatrie in Martinsried. 1991–1995 Postdoc am Salk Institute in La Jolla,CA, USA, und am University of Texas Southwestern Medical Center in Dallas, TX, USA.1995–2001 Leiter einer Forschergruppe am MPI für Experimentelle Medizin in Göttin-gen. Seit 2001 Direktor der Abteilung Molekulare Neurobiologie am MPI für Experi-mentelle Medizin in Göttingen. Seit 2002 Außerplanmäßiger Professor an der Medizi-nischen Fakultät der Georg-August-Universität Göttingen. Seit 2006 Honorarprofessoran der Biologischen Fakultät der Georg-August-Universität Göttingen.

589_621_BIOsp_0607.qxd:589_621 18.09.2007 8:21 Uhr Seite 616

LETTERdoi:10.1038/nature12618

SHANK3 and IGF1 restore synaptic deficits inneurons from 22q13 deletion syndrome patientsAleksandr Shcheglovitov1, Olesya Shcheglovitova1, Masayuki Yazawa1, Thomas Portmann1, Rui Shu1, Vittorio Sebastiano2,3,Anna Krawisz1, Wendy Froehlich4,5, Jonathan A. Bernstein4, Joachim F. Hallmayer5 & Ricardo E. Dolmetsch6

Phelan–McDermid syndrome (PMDS) is a complex neurodevelop-mental disorder characterized by global developmental delay, severelyimpaired speech, intellectual disability, and an increased risk of autismspectrum disorders (ASDs)1. PMDS is caused by heterozygous dele-tions of chromosome 22q13.3. Among the genes in the deletedregion is SHANK3, which encodes a protein in the postsynapticdensity (PSD)2,3. Rare mutations in SHANK3 have been associatedwith idiopathic ASDs4–7, non-syndromic intellectual disability8, andschizophrenia9. Although SHANK3 is considered to be the mostlikely candidate gene for the neurological abnormalities in PMDSpatients10, the cellular and molecular phenotypes associated with thissyndrome in human neurons are unknown. We generated inducedpluripotent stem (iPS) cells from individuals with PMDS and autismand used them to produce functional neurons. We show that PMDSneurons have reduced SHANK3 expression and major defects inexcitatory, but not inhibitory, synaptic transmission. Excitatorysynaptic transmission in PMDS neurons can be corrected by restor-ing SHANK3 expression or by treating neurons with insulin-likegrowth factor 1 (IGF1). IGF1 treatment promotes formation ofmature excitatory synapses that lack SHANK3 but contain PSD95and N-methyl-D-aspartate (NMDA) receptors with fast deactivationkinetics. Our findings provide direct evidence for a disruption in theratio of cellular excitation and inhibition in PMDS neurons, andpoint to a molecular pathway that can be recruited to restore it.

To study cellular and molecular phenotypes associated with PMDSin humans, we generated six iPS cell lines from two patients withPMDS (Supplementary Table 1) that had deletions of approximately1 megabases in chromosome 22 (Supplementary Fig. 1 and Supplemen-tary Table 2) by using standard reprogramming techniques11,12. Ascontrols, we used iPS cell and human embryonic stem (ES) cell linesthat were characterized in previous studies12,13. Control and PMDS iPScell lines had typical human ES-cell-like morphologies and expressedendogenous pluripotency marker proteins, but had suppressed theexpression of the exogenous reprogramming factors, had normal kar-yotypes, and formed teratomas in vivo (Supplementary Fig. 2 andSupplementary Table 3). We differentiated control and PMDS iPS cellsinto neurons using a modified monolayer differentiation protocol(Supplementary Fig. 3) and characterized the population of differen-tiated cells by quantitative PCR with reverse transcription (qRT–PCR)(Supplementary Fig. 4). Similar populations of neuronal cells werefound in cultures of control and PMDS neurons. Analysis of geneexpression with multiplex single-cell qRT–PCR13,14 demonstrated thatapproximately 42% of cells expressed pan-neuronal markers, 15%expressed cortical layer markers, and only 2–5% of the cells expressedthe synaptic markers SHANK1–SHANK3 (Supplementary Fig. 5), indi-cating that only a small fraction of the cells were synaptically matureneurons. To enrich for mature forebrain neurons, we infected cells withenhanced green fluorescent protein (GFP) under the control of theCaMKIIa promoter (CaMKIIa-GFP)15,16 (Fig. 1a). Single-cell profiling

of GFP-expressing cells showed an approximate 12-fold enrichment incells expressing SHANK1–SHANK3 (Fig. 1b). These cells also expressedsynaptic markers and genes typically found in the cortex and striatum,including TBR1, CUX1, SATB2, SOX5, FOXP1, CTIP2 (also known asBCL11B) and FOXP2 (Supplementary Fig. 5) and were stained withantibodies recognizing MAP2 (,99%), TBR1 (,29%), CTIP2 (,66%),SATB2 (,42%), and c-aminobutyric acid (GABA) or GAD67 (,7.5%)(Fig. 1c–h and Supplementary Fig. 6).

We next investigated the electrophysiological properties of controland PMDS neurons. Control and PMDS neurons expressing GFP hadsimilar action potential characteristics, resting membrane potentialand capacitance (Supplementary Table 4). PMDS neurons, however,had an increased input resistance and reduced amplitude and fre-quency of spontaneous synaptic events (Fig. 1i–j and SupplementaryTable 4). A reduction in synaptic transmission in PMDS neuronscould arise from differences in the number of neurons present incultures of PMDS and control neurons. In fact, anti-MAP2 antibodystaining revealed that our PMDS iPS cell lines produced slightly fewerneurons than controls (Supplementary Fig. 7). To control for thisdifference in neuronal density and to study cells of both genotypesunder identical experimental conditions, we infected control and PMDSneurons with either CaMKIIa-GFP or CaMKIIa-mKate2 and grewthem together in the same dish (Fig. 2a). Under these conditions, wefound that the amplitude and frequency of spontaneous excitatory post-synaptic currents (EPSCs) recorded from PMDS neurons were signifi-cantly reduced relative to controls (Fig. 2b, c), indicating that synapticdefects in PMDS neurons are postsynaptic and cell-autonomous. Todetermine if the changes in spontaneous EPSCs were also reflected indifferences in evoked a-amino-3-hydroxy-5-methyl-4-isoxazolepropio-nic acid (AMPA) and NMDA-mediated synaptic responses, we measuredcurrents induced by local field stimulation at 270 mV (AMPA-EPSCs)and 160 mV (NMDA-EPSCs). Smaller AMPA- and NMDA-EPSCswere measured in PMDS neurons in response to different stimulus intens-ities (Fig. 2d–f), indicating that both AMPA- and NMDA-mediated syn-aptic transmissions are significantly impaired.

To determine if inhibitory synaptic transmission was also altered inPMDS neurons, we recorded spontaneous and evoked inhibitory post-synaptic currents (IPSCs) (Fig. 2g–j). There were no differences inIPSCs measured from co-cultured control and PMDS neurons. Todetermine the direction of the voltage shifts induced by IPSCs at thisdevelopmental stage, we measured voltage deflections induced by focalc-aminobutyric acid application (Supplementary Fig. 8). Most controland PMDS neurons responded with hyperpolarizing shifts in mem-brane potential. These results indicate that the defects in PMDS neu-rons are limited to excitatory synaptic transmission and suggest thatthere is an alteration in the balance of cellular excitation and inhibitionin PMDS neurons.

To gain insight into the mechanism underlying the impairment ofexcitatory synaptic transmission in PMDS neurons, we measured the

1Department of Neurobiology, Stanford University, Stanford, California 94305, USA. 2Department of Obstetrics and Gynecology, Stanford University, Stanford, California 94305, USA. 3Institute for StemCell Biology and Regenerative Medicine, Stanford University, Stanford, California 94305, USA. 4Department of Pediatrics, Stanford University, Stanford, California 94305, USA. 5Department of Psychiatryand Behavioral Science, Stanford University, Stanford, California 94305, USA. 6Novartis Institutes for Biomedical Research, Cambridge, Massachusetts 02139, USA.

1 4 N O V E M B E R 2 0 1 3 | V O L 5 0 3 | N A T U R E | 2 6 7

Macmillan Publishers Limited. All rights reserved©2013

expression levels of AMPA and NMDA receptors. PMDS neuronsexpressed significantly less GLUA1 and GLUN1 proteins (Fig. 3a)and generated significantly smaller currents in response to the focalapplication of either AMPA (Fig. 3b) or NMDA (Fig. 3c). In contrast,the amplitude of c-aminobutyric-acid-evoked currents was similar forboth control and PMDS neurons (Supplementary Fig. 9), indicatingthat PMDS neurons express a decreased number of excitatory neuro-transmitter receptors and normal levels of inhibitory receptors.

To determine whether the impairment in excitatory synaptic trans-mission in PMDS neurons is due to a reduced number of synapses,we immunostained co-cultured control and PMDS neurons with anti-bodies that recognize the presynaptic protein synapsin 1 (SYN1) andpostsynaptic protein HOMER1 (Fig. 3d). We observed a significantdecrease in the number of puncta that were SYN11 (Fig. 3e) andHOMER11 SYN11 (Fig. 3f) in PMDS neurons as compared to con-trols, indicating that PMDS neurons have significantly fewer excitatorysynapses than controls.

Although SHANK3 has been proposed to be primarily responsiblefor the neurological phenotype in PMDS patients, the vast majority ofPMDS patients have 10–100 additional genes deleted. To determinewhether the synaptic phenotypes in PMDS neurons are due to reducedSHANK3 expression, we measured the levels of SHANK3 messengerRNA and protein. Whereas MAP2, SHANK1 and SHANK2 wereexpressed at similar levels in control and PMDS neurons (Fig. 4a),SHANK3 mRNA was significantly reduced in PMDS (Fig. 4a).Western blot analysis showed a significant reduction in the expressionof the longest SHANK3 protein isoform (Fig. 4b), which was expressedprimarily in neurons (Supplementary Fig. 10). Immunostaining withanti-SHANK3 antibodies on co-cultured control and PMDS neuronsalso revealed a significantly reduced level of SHANK3 protein expressionin PMDS neurons (Fig. 4c), indicating that loss of a copy of SHANK3 inPMDS significantly downregulates the expression of the protein.

We next investigated whether we could reverse the synaptic pheno-type in PMDS neurons by increasing SHANK3 expression. We infectedPMDS neurons with a lentivirus containing the longest isoform ofSHANK3 fused to EGFP (Fig. 4d and Supplementary Fig. 11). SHANK3expression in PMDS neurons restored both the amplitude and fre-quency of spontaneous EPSCs (Fig. 4e–g). It also completely rescuedevoked AMPA- and NMDA-EPSCs in 43% of recorded cells, and in

a b c

2 s10 p

A

PMDS

Ctrl

0.0

0.5

1.0

0 20 40 60 80Inter-event interval (s)

Cum

ula

tive p

rob

ab

ility

0 20 40 60 80 1000.0

0.5

1.0

Cum

ula

tive p

rob

ab

ility

Amplitude (pA)

g h i j

CtrlPMDS

0

10

20

30

Am

plit

ude (pA

) **

0.0

0.2

0.4

0.6

0.8

1.0

Fre

quency (H

z) **

0

20

40

Am

plit

ud

e (pA

)

0 100 200Amplitude (pA)

0.0

0.2

0.4

0.6

0 50Inter-event interval (s)

100

Ctrl PMDSd e f

AMPA

NMDA

50 ms

50 p

A

AMPA

0.0 0.2 0.4 0.6 0.80

10

20

30

40

50

Stimulus intensity (mA)

Curr

ent

am

plit

ud

e (pA

)

Max a

mp

litud

e (pA

)

0

50

100 NMDA

0.0 0.2 0.4 0.6 0.80

10

20

30

40

50

Stimulus intensity (mA)

Curr

ent

am

plit

ud

e (pA

)

Max a

mp

litud

e (pA

)

0

50

100

**

0.0

0.5

1.0

Cum

ula

tive p

robab

ility

200 ms

20 p

A

Ctrl

0.0 0.2 0.4 0.6 0.80

10

20

30

40

Stimulus intensity (mA)

Curr

ent

am

plit

ud

e (pA

)

PMDS

2 s

50 p

A

PMDS

Ctrl

Fre

quency (H

z)

Figure 2 | PMDS neurons display impaired excitatory synaptictransmission. a, Images of control (green) and PMDS (red) neurons co-cultured on a bed of rat cortical astrocytes (scale bar, 50mm), top. Traces ofspontaneous EPSCs recorded at 270 mV, bottom. b, c, Cumulative distributionand quantification of amplitude (b) and frequency (c) of spontaneous EPSCs(n 5 22 control and 16 PMDS cells). d, Traces of evoked AMPA-(Vh 5 270 mV) and NMDA-EPSCs (Vh 5 160 mV). Arrows indicate the timeof stimulation. Stimulation artefacts were blanked. e, f, Input-output curves ofevoked AMPA- (measured at the peak; n 5 15 control and 23 PMDS cells) andNMDA-EPSCs (measured 50 ms post-stimulus; n 5 14 control and 20 PMDScells) recorded in response to different stimulus intensities. Bar graphs showquantification of maximum current amplitude. g, Traces of spontaneous IPSCsrecorded at 270 mV. h, i, Cumulative distribution and quantification ofamplitude (h) and frequency (i) of spontaneous IPSCs (n 5 21 control and 23PMDS cells). j, Representative traces and input-output curves of evoked IPSCs(n 5 17 control and 16 PMDS cells). Data presented as means 6 s.e.m.;*P , 0.05, **P , 0.01, Mann–Whitney test.

b c

e hgf

a d

0

20

40

60

80

100

Perc

enta

ge o

f cells

h18S

SHANK1

SHANK2

SHANK3

Unlabelled

CaMKII-GFP+

j

PMDS

Ctrl

2 s

10 p

A–70

–70

200 ms

20

mV

i

0

5

10

15

20

25

0.0

0.1

0.2

0.3

0.4

0

1

2

3

4

Inp

ut

resis

tance (GΩ

)

Am

plit

ud

e (p

A)

Fre

quency (H

z)***

***

***

Ctrl

PMDS

Ctrl

PMDS

Ctrl

PMDS

Ctrl

PMDS

MAP2

TBR1

CTI

P2

SATB2

GAB

A/GAD

670

20

40

60

80

100

Perc

enta

ge o

f

CaM

KII-G

FP

+ c

ells

CaMKIIa-GFP CaMKIIa-GFP CaMKIIa-GFP CaMKIIa-GFP MAP2 MAP2

CaMKIIa-GFP CaMKIIa-GFP TBR1 TBR1

CaMKIIa-GFP CaMKIIa-GFP MAP2

CaMKIIa-GFP TBR1

CaMKIIa-GFP CaMKIIa-GFP CTIP2 CTIP2

CaMKIIa-GFP CaMKIIa-GFP SATB2 SATB2

CaMKIIa-GFP CaMKIIa-GFP GAD67 GAD67

CaMKIIa-GFP CTIP2

CaMKIIa-GFP SATB2

CaMKIIa-GFP GAD67

Figure 1 | CaMKIIa-GFP labels functional iPS cell-derived forebrainneurons that express SHANK3. a, Image of iPS cell-derived neurons labelledwith EGFP under the control of CaMKIIa promoter. b, Fraction of cellsexpressing SHANK1-3, selected either randomly (unlabelled, n 5 1 line (total90 cells)) or based on CaMKIIa-GFP expression (CaMKIIa-GFP1, n 5 8 (90))and assessed by multiplex single-cell qRT–PCR (Supplementary Fig. 6). h18S,human 18S ribosomal RNA. c–g, Images of iPS cell-derived neuronsimmunostained with antibodies against GFP and MAP2 (c), TBR1 (d), CTIP2(e), SATB2 (f), and GAD67 (g). h, Fraction of cells expressing MAP2 (n 5 3cover slips (682 cells)), TBR1 (n 5 4 (566)), CTIP2 (n 5 8 (853)), SATB2 (n 5 6(646)), and c-aminobutyric acid (GABA)/GAD67 (n 5 5 (493)) among GFP-expressing cells. i, Representative recordings of action potentials (left, inducedby somatic current injections, DI 5 5 pA) and input resistance measurements(right) obtained from GFP-expressing control (n 5 22 cells from 3 lines) andPMDS neurons (n 5 36 cells from 5 lines). j, Representative recordings (left)and quantification of amplitude and frequency (right) of spontaneous EPSCsobtained from GFP-expressing control (n 5 19 cells from 3 lines) and PMDSneurons (n 5 26 cells from 5 lines) (Supplementary Table 4). Data presented asmeans 6 s.e.m.; ***P , 0.001, Mann–Whitney test. Scale bars, 50mm.

RESEARCH LETTER

2 6 8 | N A T U R E | V O L 5 0 3 | 1 4 N O V E M B E R 2 0 1 3

Macmillan Publishers Limited. All rights reserved©2013

the remaining cells, it rescued only AMPA-EPSCs (Fig. 4h–j). Thisindicates that loss of SHANK3 substantially contributes to the impairedexcitatory synaptic transmission detected in PMDS neurons; however,it also suggests that SHANK3 may act differently on AMPA andNMDA receptors.

We next tested if we could restore synaptic deficits in PMDS neuronspharmacologically, using agents that have been previously reported toincrease the level of SHANK3 expression or to increase synaptic trans-mission, including trichostatin A (TSA)17,18, valproic acid (VPA)17,19,nifedipine (Nif)20, IGF121–23 and IGF224. We treated control and PMDSneurons over the course of 2 to 20 days and measured the numberof HOMER11 SYN11 puncta. Only IGF1 increased the number ofSYN11 HOMER11 puncta to the levels observed in control neuronsin the same culture (Fig. 5a–b). We also found that IGF1 increased theaverage amplitude and frequency of spontaneous EPSCs in PMDSneurons to control levels (Fig. 5c), but had no effect on spontaneousIPSCs (Supplementary Fig. 12). IGF1 also restored the amplitude ofevoked AMPA- and NMDA-EPSCs (Fig. 5d) and reduced the inputresistance of PMDS neurons to that observed in control neurons(Supplementary Fig. 13). Finally, IGF1 completely restored the

whole-cell NMDA receptor currents evoked by focal application ofNMDA, but failed to fully restore whole-cell AMPA receptor currentselicited by application of AMPA (Supplementary Fig. 14). Together,these results indicate that treatment with IGF1 restores synaptic trans-mission in PMDS neurons by increasing the number of synapticAMPA and NMDA receptors.

To gain insight into the mechanism of IGF1 action, we measuredSHANK3 expression in control and PMDS neurons treated with IGF1.Surprisingly, we found that IGF1 caused a marked decrease of SHANK3protein expression in the cell body and processes of both control andPMDS neurons (Supplementary Fig. 15). We next examined SHANK3expression at synapses by infecting cells with GFP-tagged SHANK3and staining neurons with the postsynaptic antibody anti-PSD95 (Sup-plementary Fig. 16a). We found only 50% overlap between SHANK3-and PSD95-containing puncta in human iPS cell-derived neurons. Asimilar pattern of SHANK3 and PSD95 expression was detected in thehuman fetal brain tissue immunostained with anti-SHANK3 and anti-PSD95 antibodies (Supplementary Fig. 16b). This indicates that indeveloping neurons there are three populations of excitatory synapses:those that contain SHANK3 alone, those that contain SHANK3 and

0 20 40 60 80 1000.0

0.5

1.0

Cum

ula

tive p

rob

ab

ility

Amplitude (pA)

NMDAAMPA

0 20 40 60 80

Inter-event interval (s)

d e f g

i j

h

+SHANK3

2 s10

pA

+SHANK3

PMDS

AMPA

NMDA

50 ms

25 p

A

PMDS +SHANK3

Fre

quency (H

z) *

0.0

0.2

0.4

0.6

0.8

1.0

Am

plit

ude (pA

)

***

0

10

20

a b c

ex22

ex2-

3

ex8-

9

SHANK2

SHANK1

MAP2

SHANK3

Ctrl

PMDS

*** *

0.0

0.2

0.4

0.6

0.8

1.0

Perc

ent of GAPDH

+SHANK3 PMDS

Max c

urr

ent

am

plit

ud

e (pA

)

43%

57%

cells

**

170 140

1.0

0.5

0190N

orm

aliz

ed

to

Ctr

l

kDa

Ctrl

PMDS

SHANK3

Ctrl

PMDS

0.0

0.5

1.0***

No

rmaliz

ed

SH

AN

K3

in

ten

sity

Western blotqRT–PCR

Anti-GAPDH

Anti-SHANK3150

250

kDa

37

Map2 GFP–SHANK3 SYN1

Ctrl1 PMDS Ctrl2

Immunostaining

50

0 0

100

150

200

50

100*

**

Map2Map2 GFP-SHANK3GFP-SHANK3SYN1SYN1

MergedMergedPMDSMap2 GFP-SHANK3SYN1

Merged

Figure 4 | Reduced SHANK3 expression contributes to synaptic defects inPMDS neurons. a, Relative expression of SHANK1, SHANK2, SHANK3 andMAP2 mRNAs in iPS cell-derived neurons (day in vitro 30–35; n 5 5 controland 6 PMDS biological replicates; *P , 0.05, **P , 0.01, Student’s t-test).ex, exon. b, Quantification of SHANK3 expression in control and PMDSneurons as assessed by western blot (day in vitro 45–50; n 5 4 pairs of samples;**P , 0.01, one-sample t-test). GAPDH was used as a loading control.c, Images (left) and quantification (right) of SHANK3 expression in co-culturedcontrol (red) and PMDS (green) neurons (scale bar, 20mm). d, Images ofPMDS neurons infected with GFP-tagged SHANK3 lentiviruses andimmunostained with antibodies against MAP2, GFP and SYN1 (scale bars, 20(top) and 5 (bottom)mm). e, Traces of spontaneous EPSCs recorded fromuninfected and SHANK3-infected PMDS neurons at –70 mV. f, g, Cumulativedistribution and quantification of amplitude (f) and frequency (g) ofspontaneous EPSCs (n 5 8 uninfected and 17 SHANK3-infected cells).h, Traces of evoked AMPA- (Vh 5 270 mV) and NMDA-EPSCs(Vh 5 160 mV). i, j, Quantification of maximum amplitudes of evokedAMPA- (i, measured at peak) and NMDA-EPSCs (j, measured 50 ms poststimulus) recorded from uninfected (n 5 6) and SHANK3-infected (n 5 14)PMDS neurons. Data presented as means 6 s.e.m.; *P , 0.05, ***P , 0.001,Mann–Whitney test.

***

cba

–100 –50 0 50 100

–100

–50

0

50

–100 –50 0 50

0

50

100

Curr

ent

am

plit

ud

e (pA

)

Curr

ent

am

plit

ud

e (pA

)

1 s

50 p

A

Ctrl PMDS

1 s20 p

A

Ctrl PMDS

50

100

0

at

+60 m

V (pA

) ***

d

f

e

PMDSCtrl SYN1 HOMER1 Ctrl SYN1 HOMER1

PMDS SYN1 HOMER1

Anti-GLUA1

Anti-GLUA2

Anti-GLUN1

Anti-GAPDH

0

0.5

1.0

No

rma

lize

d t

o c

trl

**

GLUA1 GLUA2 GLUN1

Western blot Focal AMPA application Focal NMDA application

Synaptic staining: HOMER1 SYN1

Ctrl PMDS

100

kDa

100

10037

Vm (mV)

*

50

100

0at

–70 m

V (pA

)

Ctrl1 PMDS Ctrl2

***

*Ctrl

PMDS

Ctrl

PMDS

0 1 2

Normalized

puncta per cell

SYN1

HOMER1 SYN1

0.0

0.5

1.0C

um

ula

tive p

rob

ab

ility

0.0

0.5

1.0

Cum

ula

tive p

rob

ab

ility

Vm (mV)

Figure 3 | PMDS neurons show reduced expression of glutamate receptorsand decreased number of synapses. a, Quantification of expression of AMPAand NMDA receptors in control and PMDS neurons using western blot(day in vitro 45–50; n 5 3–11 pairs of samples; **P , 0.01, ***P , 0.001,one-sample t-test). GAPDH was used as a loading control. b, c, Representativecurrents (top) and peak current versus voltage curves (bottom) recorded incontrol and PMDS neurons in response to focal application of 200mM AMPA(measured at a holding potential of270 to 150 mV, D20 mV, n 5 7 controland 11 PMDS cells) and 100mM NMDA/10mM Gly (260 to 160 mV,D20 mV, n 5 21 control and 30 PMDS cells). Data presented as means 6 s.e.m.;*P , 0.05, ***P , 0.001, Mann–Whitney test. d, Images of co-cultured control(green) and PMDS (red) neurons immunostained for mKate2, GFP, SYN1and HOMER1. Scale bars, 20 (left) and 5mm (right). e, f, Cumulativedistributions of the number of SYN11 (e) and HOMER11 SYN11 (f) punctaon co-cultured control and PMDS neurons; *P , 0.05, ***P , 0.001,Kolmogorov–Smirnov test.

LETTER RESEARCH

1 4 N O V E M B E R 2 0 1 3 | V O L 5 0 3 | N A T U R E | 2 6 9

Macmillan Publishers Limited. All rights reserved©2013

PSD95, and those that contain PSD95 alone. We investigated the effectof IGF1 on these three populations of synapses by staining neuronswith anti-SHANK3 and anti-PSD95 antibodies (Fig. 5e). Treatmentwith IGF1 caused a decrease in the number of puncta expressingSHANK3 alone in control neurons, but had little effect in PMDSneurons, which already had reduced levels of SHANK3 (Fig. 5f). Incontrast, IGF1 caused a 45% and 340% increase in the fraction ofpuncta expressing PSD95 in control and PMDS neurons, respectively,while decreasing the number of puncta containing both markers(Fig. 5f). To determine if synapses that lack SHANK3 are functionallydifferent, we analysed the physiological properties of synapses inPMDS and control neurons. We found that PMDS neurons have asignificantly faster rate of decay of NMDA-EPSCs relative to controlcells and treatment with IGF1 increased the decay rate of NMDA-EPSCs in control neurons to match the rate of decay in PMDS neurons(Fig. 5g). In contrast, SHANK3 expression in PMDS neurons caused adecrease in the rate of decay of NMDA-EPSCs to match that inuntreated control neurons (Fig. 5g). These results are consistent withthe idea that IGF1 treatment rescues synaptic transmission in PMDSneurons by causing loss of SHANK3-containing synapses and gain ofPSD95-containing synapses, which are characterized by rapidly decay-ing NMDA-EPSCs (Fig. 5h).

Here, we show that neurons derived from iPS cells from PMDSpatients have major defects in excitatory synaptic transmission arisingfrom both a failure to form the correct number of excitatory synapsesand a reduction in the expression of glutamate receptors. As no signi-ficant impairments in inhibitory synaptic transmission were found,

this resulted in an imbalance between excitation and inhibition at acellular level. Only some aspects of the phenotype detected in humancells are similar to those found in mouse models of PMDS25–27. Whythere is a discrepancy is unclear. One potentially important issue is thatthere are multiple SHANK3 isoforms. PMDS patients generally lackthe entire gene on one chromosome, whereas mice typically lack onlysome of the isoforms and have generally been studied as homozygoteknockouts. We also demonstrated that SHANK3 and IGF1 largelyrestore synaptic deficits in PMDS neurons. Interestingly, IGF1 decreasesSHANK3 expression and promotes the formation of a class of synapsescontaining PSD95 but lacking SHANK3. One possibility is that SHANK3is required at an early stage of synapse formation, and that IGF1 pro-motes maturation of synapses by triggering the loss of SHANK3 andrecruitment of PSD95. Consistent with this view, SHANK3-containingsynapses have slowly decaying NMDA receptor currents that resemblethose of GLUN2B receptors that occur early in development, whereassynapses that contain PSD95 but lack SHANK3 have rapidly decayingNMDA receptor currents that resemble the GLUN2A-containing synapsesthat appear later in development28. In summary, we have identifiedsynaptic defects in iPS cell-derived neurons from patients with PMDSand we show that these defects depend on loss of SHANK3 and can bereversed by IGF1.

METHODS SUMMARYPrimary fibroblasts from consenting patients with PMDS were acquired by skin-punch biopsy following approved SCRO and IRB protocols. iPS cell lines from thefibroblasts were generated using standard reprogramming techniques11,12. iPS cells

g

h

f

c

e

db

0.0

0.1

0.2

0.3

0.4

0.5

Fre

quency (H

z)

0IGF1w/o IGF1w/o

AMPA NMDA

Max c

urr

ent

am

plit

ud

e (pA

)

50

100

Ctrl IGF1 PMDS IGF1

AMPA

NMDA

50 ms

50

pA

2 s

10

pA

Ctrl IGF1

PMDS IGF1

SHANK3 SHANK3

PSD95

PSD95

Number puncta per cell

PMDSCtrl SYN1 HOMER1

a

w/o

TSAVPA

Nif

IGFI

IIG

FI0.4

0.6

0.8

1.0

No

rmaliz

ed

punc

ta p

er

cell

ratio

(P

MD

S/C

trl)

Ctrl SYN1 HOMER1 PMDS SYN1 HOMER1

**

**

0

5

10

15

20

25

Am

plit

ud

e (p

A) ** *

*

***

***

*

No

rmaliz

ed

to

Ctr

l w

/o t

era

tmen

t

0.0

0.5

1.0

1.5

2.0

IGF1w/o IGF1w/o IGF1w/o

*

CtrlPMDS

IGF1

IGF1w/o IGF1w/o

Synaptic staining: SHANK3 PSD95

w/o

tre

atm

en

tIG

F1

Ctrl/PMDS PSD95 SHANK3Ctrl PMDS

CtrlPMDSPMDS+SHANK3

SHANK3 IGF1

IGF1

PMDS

Ctrl

SHANK3+

PSD95+

**

* ***

***

0

200

400

IGF1w/o w/o

** *****

200 ms

0.0

1.0

CtrlPMDS

w/o

CtrlPMDS

IGF1

PMDS

GFP–SHANK3

No

rmaliz

ed

cu

rren

t

τ decay (m

s)

****

Figure 5 | IGF1 treatment restores excitatorysynaptic transmission in PMDS neurons.a, Quantification of the number ofHOMER11 SYN11 puncta in co-cultured neuronsafter treatment with different agents (w/o : n 5 51control, 43 PMDS cells; TSA: n 5 11/12, VPA:n 5 25/29, Nif: n 5 9/8, IGF2: n 5 31/40, IGF1:n 5 35/32). w/o, without treatment. b, Images ofco-cultured control (green) and PMDS (red)neurons after IGF1 treatment. Scale bars, 20 (top)and 5 (bottom)mm. c, Recordings (top) andquantification of amplitude and frequency(bottom) of spontaneous EPSCs recorded in IGF1-treated cultures at 270 mV (n 5 18 control and25 PMDS cells). d, Recordings (top) andquantification of amplitudes (bottom) of evokedAMPA- (measured at the peak, Vh 5 270 mV) andNMDA-EPSCs (measured 50 ms post-stimulus,Vh 5 160 mV) acquired in IGF1-treated cultures(n 5 14 control and 21 PMDS cells). Dashed linesrepresent data originally presented in Fig. 2.e, Images of co-cultured control (green) and PMDS(red) neurons in untreated (top) and IGF1-treated(bottom) cultures, immunostained with antibodiesagainst GFP, mKate2, SHANK3 and PSD95. Scalebars, 20 (left) and 5 (right)mm. f, Quantification ofthe number of SHANK31, PSD951 andSHANK3 PSD951 puncta in co-cultured control(w/o, n 5 25 cells; IGF1, 21 cells) and PMDS(w/o, n 5 33 cells; IGF1, 25 cells) neurons. Valueswere normalized to the mean values acquired fromcontrol neurons in untreated cultures. g, Meannormalized traces (left) and quantification of thedecay kinetics (right) of NMDA-EPSCs recordedfrom control (w/o, n 5 9; IGF1, 12) and PMDS(w/o, n 5 18; IGF1, 18, GFP–SHANK3, 6) neurons.Data presented as means 6 s.e.m.; *P , 0.05,**P , 0.01, ***P , 0.001, one-way analysis ofvariance with Bonferroni’s test. h, Cartoondepicting possible mechanisms for the recovery ofsynaptic transmission in PMDS neurons.

RESEARCH LETTER

2 7 0 | N A T U R E | V O L 5 0 3 | 1 4 N O V E M B E R 2 0 1 3

Macmillan Publishers Limited. All rights reserved©2013

were differentiated into neurons using a modified monolayer differentiation pro-tocol (Supplementary Fig. 3). Properties of iPS cell-derived neurons were char-acterized using qRT–PCR, multiplex single-cell qRT–PCR, electrophysiology,immunocytochemistry and western blot. A detailed description of all the proce-dures, materials and methods used in this study is available in the Methods section.Performed experiments are also summarized in Supplementary Table 3.

Online Content Any additional Methods, Extended Data display items and SourceData are available in the online version of the paper; references unique to thesesections appear only in the online paper.

Received 8 November 2011; accepted 30 August 2013.

Published online 16 October 2013.

1. Phelan, K. & McDermid, H. E. The 22q13.3 deletion syndrome (Phelan-McDermidsyndrome). Mol. Syndromol. 2, 186–201 (2012).

2. Boeckers, T. M., Bockmann, J., Kreutz, M. R. & Gundelfinger, E. D. ProSAP/Shankproteins – a family of higher order organizing molecules of the postsynapticdensity with an emerging role in human neurological disease. J. Neurochem. 81,903–910 (2002).

3. Sheng, M. & Kim, E. The Shank family of scaffold proteins. J. Cell Sci. 113,1851–1856 (2000).

4. Boccuto, L. et al. PrevalenceofSHANK3 variants inpatients withdifferent subtypesof autism spectrum disorders. Eur. J. Hum. Genet. 21, 310–316 (2013).

5. Durand, C.M.et al.Mutations in the geneencoding the synaptic scaffolding proteinSHANK3 are associated with autism spectrum disorders. Nature Genet. 39, 25–27(2007).

6. Gauthier, J. et al. Novel de novo SHANK3 mutation in autistic patients. Am. J. Med.Genet. B. Neuropsychiatr. Genet. 150B, 421–424 (2009).

7. Moessner, R. et al. Contribution of SHANK3 mutations to autism spectrumdisorder. Am. J. Hum. Genet. 81, 1289–1297 (2007).

8. Hamdan, F. F. et al. Excess of de novo deleterious mutations in genes associatedwith glutamatergic systems in nonsyndromic intellectual disability. Am. J. Hum.Genet. 88, 306–316 (2011).

9. Gauthier, J. et al. De novo mutations in the gene encoding the synaptic scaffoldingprotein SHANK3 in patients ascertained for schizophrenia. Proc. Natl Acad. Sci.USA 107, 7863–7868 (2010).

10. Wilson, H. L. et al. Molecular characterisation of the 22q13 deletion syndromesupports the role of haploinsufficiency of SHANK3/PROSAP2 in the majorneurological symptoms. J. Med. Genet. 40, 575–584 (2003).

11. Takahashi, K. et al. Induction of pluripotent stem cells from adult humanfibroblasts by defined factors. Cell 131, 861–872 (2007).

12. Yazawa, M. et al. Using induced pluripotent stem cells to investigate cardiacphenotypes in Timothy syndrome. Nature 471, 230–234 (2011).

13. Pasca, S. P. et al. Using iPSC-derived neurons to uncover cellular phenotypesassociated with Timothy syndrome. Nature Med. 17, 1657–1662 (2011).

14. Yoo, A. S. et al. MicroRNA-mediated conversion of human fibroblasts to neurons.Nature 476, 228–231 (2011).

15. Dittgen, T. et al. Lentivirus-based genetic manipulations of cortical neurons andtheir optical and electrophysiological monitoring in vivo. Proc. Natl Acad. Sci. USA101, 18206–18211 (2004).

16. Nathanson, J. L., Yanagawa, Y., Obata, K. & Callaway, E. M. Preferential labeling ofinhibitory and excitatory cortical neurons by endogenous tropism ofadeno-associated virus and lentivirus vectors. Neuroscience 161, 441–450(2009).

17. Akhtar, M. W. et al. Histone deacetylases 1 and 2 form a developmental switch thatcontrols excitatory synapse maturation and function. J. Neurosci. 29, 8288–8297(2009).

18. Maunakea, A. K. et al. Conserved role of intragenic DNA methylation in regulatingalternative promoters. Nature 466, 253–257 (2010).

19. Rinaldi, T., Kulangara, K., Antoniello, K. & Markram, H. Elevated NMDA receptorlevels and enhanced postsynaptic long-term potentiation induced byprenatal exposure to valproic acid. Proc. Natl Acad. Sci. USA 104, 13501–13506(2007).

20. Goold, C. P. & Nicoll, R. A. Single-cell optogenetic excitation drives homeostaticsynaptic depression. Neuron 68, 512–528 (2010).

21. Marchetto, M. C. et al. A model for neural development and treatment of Rettsyndrome using human induced pluripotent stem cells. Cell 143, 527–539(2010).

22. O’Kusky, J. R., Ye, P. & D’Ercole, A. J. Insulin-like growth factor-I promotesneurogenesis and synaptogenesis in the hippocampal dentate gyrus duringpostnatal development. J. Neurosci. 20, 8435–8442 (2000).

23. Tropea, D. et al. Partial reversal of Rett syndrome-like symptoms in MeCP2 mutantmice. Proc. Natl Acad. Sci. USA 106, 2029–2034 (2009).

24. Chen, D. Y. et al. A critical role for IGF-II in memory consolidation andenhancement. Nature 469, 491–497 (2011).

25. Bozdagi, O. et al. Haploinsufficiency of the autism-associated Shank3 gene leads todeficits in synaptic function, social interaction, and social communication. Mol.Autism 1, 15 (2010).

26. Peça, J. et al. Shank3 mutant mice display autistic-like behaviours and striataldysfunction. Nature 472, 437–442 (2011).

27. Wang, X.et al.Synaptic dysfunction andabnormalbehaviors inmice lacking majorisoforms of Shank3. Hum. Mol. Genet. 20, 3093–3108 (2011).

28. Paoletti, P., Bellone, C. & Zhou, Q. NMDA receptor subunit diversity: impact onreceptor properties, synaptic plasticity and disease. Nature Rev. Neurosci. 14,383–400 (2013).

Supplementary Information is available in the online version of the paper.

Acknowledgements We are grateful to participants and their families for their support;M. Adam for assistance with recruitment; to X. Jia, A. Cherry, C. Bangs, P. Jones, andJ. Williams for assistance with tissue culture; P. Liao for help with multiplexligation-dependent probe amplification (MLPA); M. Fabian for astrocyte preparations;H.N. Nguyen for consultations on the neural differentiation protocol and spectralkaryotyping (SKY); V. Vu and G. Lin for help with data analysis; T. Sudhof, T. Boeckers,A. Grabruker, C. Garner and C. Sala for antibodies; R. Xavier for SHANK3complementary DNA; R. Reijo-Pera and members of the Dolmetsch laboratory forcommenting on the manuscript; E. Nigh for editing the manuscript. We also thank theStanford Neuroscience Microscopy Service (supported by National Institutes of Health(NIH) NS069375). Support for this study came from the California Institute forRegenerative Medicine CIRM, the Autism Science Foundation and thePhelan-McDermid Syndrome Foundation (to A.S.), the Swiss National ScienceFoundation (to T.P.), the Japan Society for the Promotion of Research Abroad andAmerican Heart Association (to M.Y.), the National Institute of Mental Health (NIMH)grant R33MH087898 (to J.F.H.); NIH Pioneer Award (5DP1OD3889), CIRM (grantRT2-01906) and Simons Foundation (to R.E.D.). We are also grateful for funding fromthe JDH research fund, N. Juaw, B. and F. Horowitz, M. McCafferey, B. and J. Packard,P. Kwan and K. Wang, and the Flora foundation.

Author Contributions A.S. andR.E.D. designedexperiments andwrote the manuscript;A.S. performed iPS cell maintenance, neural differentiation, electrophysiology, cloningand immunocytochemistry; O.S. maintained and characterized iPS cells, performedwestern blot and qRT–PCR; M.Y. generated and characterized iPS cells; T.P. performedmultiplex single-cell qRT–PCR;V.S.performed teratomaassay;R.S. andA.K.performedqRT–PCR and data analysis; W.F., J.A.B. and J.F.H. recruited and characterized patientsand performed the MLPA assay.

Author Information Reprints and permissions information is available atwww.nature.com/reprints. The authors declare no competing financial interests.Readers are welcome to comment on the online version of the paper. Correspondenceand requests for materials should be addressed to R.E.D.(Ricardo.dolmetsch@novartis.com)

LETTER RESEARCH

1 4 N O V E M B E R 2 0 1 3 | V O L 5 0 3 | N A T U R E | 2 7 1

Macmillan Publishers Limited. All rights reserved©2013

METHODSRecruitment of patients with PMDS and skin biopsy procedure. Participantsdiagnosed with 22q13 deletion syndrome were recruited from on-going researchstudies, the medical genetics service at Lucile Packard Children’s Hospital, andinterest groups for patients and families with genetic disorders associated withautistic spectrum manifestations. Consenting participants were evaluated withthe Autism Diagnostic Observation Schedule (ADOS) and, when possible, theStanford-Binet Intelligence Scales, Fifth Edition (SB5). In addition, parents ofparticipants completed the Autism Diagnostic Interview-Revised (ADI-R) andquestionnaires about the participant, including a medical history questionnaireand the Social Responsiveness Scale (SRS). All assessors were trained nationallyand certified as reliable in the administration of ADOS and ADI-R assessments.Skin punch biopsies were performed as follows: after local anaesthesia with topicaland injected xylocaine, a shallow, 3-mm punch biopsy sample was obtained fromeither the forearm or thigh.

This study was approved by both the Stanford University Institutional ReviewBoard (IRB) and the Stem Cell Research Oversight Committee (SCRO).Generation and characterization of iPS cell lines. Detailed procedures used forthe generation, maintenance, and characterization of iPS cells were previouslydescribed12. Briefly, primary fibroblasts from patients with PMDS were acquiredby skin-punch biopsy following approved SCRO and IRB protocols. Cells werepropagated for 2–3 passages, plated at 100,000 cells per well in six-well plates, andinfected with retroviruses carrying SOX2, OCT3/4, c-MYC and KLF4 transcrip-tion factors the next day. iPS cells colonies, normally detected 21–30 days after thefirst infection, were routinely maintained on irradiated CF1 feeder cells in iPS cellmedium (DMEM/F12 (11330-032, Gibco), containing 20% Knockout SR (10828-028, Gibco), 1% MEM (11140, Gibco), 0.5% L-glutamine (25030, Gibco), 1%penicillin streptomycin (15140, Gibco), 0.1 mM b-mercaptoethanol (Sigma) and10 ng ml21 basic fibroblast growth factor (R&D Systems)).

iPS cells were immunostained with antibodies against NANOG (AF1997, R&DSystems) and TRA-2-49/6E (developed by P. W. Andrews and obtained from theDevelopment Studies Hybridoma Bank developed under the auspices of theNational Institute of Child Health and Development (NICHD) and maintainedin the University of Iowa, Department of Biological Sciences).

SKY analysis of all iPS cell lines was performed according to a standard pro-cedure. Briefly, metaphases were extracted after overnight incubation in 0.03mgml21 KaryoMAX Colcemid solution in iPS cell media. Chromosomes were iso-lated and stained with the SKYPaint probe mixture and CAD kit (Applied SpectralImaging), following the manufacturer’s protocol. Spreads were imaged on a Leicamicroscope (DM400B), equipped with SpectraCubeSD300 and a 363 objective,using Case Data Manager 6.0 and HiSKY 6.0 software (Applied Spectral Imaging).Twenty spreads from each iPS cell line with well-separated chromosomes wereimaged and analysed for visible chromosome abnormalities.

For the teratoma formation assay, iPS cells from a confluent 10-cm dish wereenzymatically collected, washed and resuspended in 1.5 ml of PBS. Cells were theninjected into the kidney capsules of SCID mice (Charles River LaboratoriesInternational). Visible tumours were dissected 3–8 weeks post-transplantation andfixed in 4% paraformaldehyde. Fixed samples were sent to the Cureline BiopathologyLaboratory for paraffin embedding, sectioning and staining with haematoxylinand eosin. Sections were then examined for the presence of tissue representatives ofall three germ layers.

Genomic DNA from fibroblasts and iPS cells was isolated using the DNeasyBlood & Tissue kit (Qiagen). Haploinsufficiency of SHANK3 in fibroblasts and alliPS cells of patients, was verified using the SALSA MLPA kit (P188, MRC-Holland),following the manufacturer’s protocol.Neural differentiation. For neural differentiation, a combination of two protocolswas used29,30. The rationale and detailed procedures are presented in Supplemen-tary Fig. 3.qRT–PCR. qRT–PCR was performed in a real-time thermal cycler (Mastercyclerep realplex, Eppendorf) using either FastStart Universal SYBR Green (Rox)(Roche) or RT qPCR master mixes (SABiosciences). Total RNA was isolated usingthe RNeasy Mini kit (Qiagen). cDNA was synthesized with either the SuperScript IIIFirst-Strand Synthesis kit (Invitrogen) or the RT First Strand kit (SABiosciences).All primers are listed in Supplementary Table 5.Multiplex single-cell qRT–PCR. Single cells were collected either by aspirationwith a patch glass pipette filled with 4ml of sterile intracellular solution or by FACSsorting using a BD influx sorter (BD Biosciences) into 10ml of a pre-amplificationmix containing 40 nM of all primers for the genes of interest and the followingcomponents of the CellsDirect One-Step qRT–PCR kit (Invitrogen): 2 3 ReactionMix, SuperScript III RT/Platinum Taq Mix. After collection, samples were reverse-transcribed and pre-amplified for 18 cycles. Pre-amplified samples were diluted(33 ) with TE buffer and stored at 220 uC. Sample and assay (primer pairs) prepara-tion for 96.96 Fluidigm Dynamic arrays was done according to the manufacturer’s

recommendation. Briefly, each sample was mixed with 20 3 DNA binding dyesample-loading reagent (Fluidigm), 20 3 EvaGreen (Biotium) and TaqMan GeneExpression master mix (Applied Biosystems). Assays were mixed with 2 3 assayloading reagent (Fluidigm) and TE to a final concentration of 5mM. The 96.96Fluidigm Dynamic Arrays (Fluidigm) were primed and loaded on an IFC ControllerHX (Fluidigm) and quantitative PCR experiments were run on a Biomark Systemfor Genetic Analysis (Fluidigm). Melting curves were used to determine the speci-ficity of each reaction. In addition to single-cell material, every experiment con-tained four standard dilutions of a mixed human cDNA library. Data were collectedand analysed using Fluidigm Real-Time PCR Analysis software (v.2.1.3 and v.3.0.2).Threshold cycle (CT) as a measure for original template amount was used forquantification. To compare the data acquired from different chips, the Ct valueswere normalized to the median intensity of the entire chip. Because single-cell geneexpression in mammals occurs in bursts and follows a binary distribution, weestablished 32 cycles (at least five standard deviations from the mean of the lowestexpressed genes) as a general threshold for whether or not a gene is expressed.Electrophysiology. All recordings were performed at room temperature (22–25 uC)on GFP- or mKate2-expressing iPS cell-derived neurons on day 31–45 (separatecultures without astrocytes) or day 45–52 (co-cultured on rat cortical astrocytes).Briefly, neurons grown on cover slips or 35-mm plastic dishes were visualized witha 340 air objective on an inverted Nikon microscope (Ellipse TE2000) andrecorded using an EPC10 amplifier (HEKA). The following solutions were used(in mM): extracellular, 140 NaCl, 2.5 KCl, 2.5 CaCl2, 2 MgCl2, 1 NaH2PO4, 20glucose, 10 HEPES, pH 7.4; intracellular, 135 CsMeS, 5 CsCl, 10 HEPES, 0.5 EGTA,1 MgCl2, 4 Mg2ATP, 0.4 NaGTP, 5 QX-314, pH 7.4 with CsOH (for EPSCs,AMPA-, NMDA-, and c-aminobutyric acid-current measurements) and 125CsCl, 10 EGTA, 1 MgCl2, 0.1 CaCl2, 4 MgATP, 0.3 NaGTP, 10 HEPES, 5 QX-314 pH 7.2 with CsOH (for IPSCs measurements). Recording pipettes made ofborosilicate glass (BF150-110-10, Sutter Instruments) had a resistance of 3–6 MVwhen filled with intracellular solution. The following combinations of antagonistswere present in the extracellular solutions to isolate different postsynaptic currents(in mM): 20 bicuculline or 50 picrotoxin (spontaneous and evoked EPSCs), 10NBQX and 50 2-amino-5-phosphonopentanoic acid (APV) (spontaneous andevoked IPSCs), 10 NBQX and 20 bicuculline or 50 picrotoxin (NMDA-current,focal application), 50 APV and 50 picrotoxin (AMPA-current, focal application),10 NBQX and 50 APV (c-aminobutyric acid-current, focal application). SpontaneousEPSCs and IPSCs were recorded for 3 min (1 min after break-in to block sodiumcurrent by QX-314) at holding potential 270 mV. Recordings were filtered at2 kHz and digitized at 4 kHz. Evoked postsynaptic currents were measured afterspontaneous currents and induced by brief (1 ms) unipolar current pulses ofvarious amplitudes (0.1–0.9 mA, D5 0.05–0.1 mA), delivered with a stimulatingelectrode (CBAEC75, FHC) positioned 100–150mm from the cell soma and con-nected to an external stimulator (A365, WPI). Recordings were filtered at 2 kHzand digitized at 10 kHz. NMDA-, AMPA- and c-aminobutyric acid-currents wereinduced by a focal transient application of the following agonists (in mM): 100NMDA and 10 Gly, 200 AMPA, or 100 c-aminobutyric acid. An injection pipette(patch pipette with resistance of 1–2 MV when filled with extracellular solution)was positioned about 20mm from the cell soma and connected to a pico-spritzer(PDES-02DX, npi Electronics) with 10 p.s.i. in-line pressure. In all experiments,access resistance was monitored throughout the experiment (,15 MV), but notcompensated. Cells were included for analysis only if the change of resistance was,25% throughout the experiment.

Data were collected and initially analysed with Patchmaster software (HEKA).Further analysis was performed using Clampfit 10 (Molecular Devices), IgorPRO(Wavemetrics), Excel (Microsoft) and Prism (GraphPad). The baseline noiseamplitude in our experiments was about 2.5 pA. Spontaneous EPSCs were detectedusing a template search procedure in Clampfit 10. A template for automatic EPSCdetection was created by averaging 50 manually picked synaptic events. Spon-taneous IPSCs were detected using MiniAnalysis software. The decay kinetics ofNMDA-EPSCs was estimated using a weighted decay time constant (tdecay)31,calculated from the area under the normalized current on an interval from16.4 ms (3 3 tAMPA decay) to 1.1 s after the peak of AMPA-EPSC.Western blot analysis. Cells grown in 2-cm dishes were lysed in a standard RIPAbuffer containing a protease inhibitor cocktail (Roche Diagnostics). Protein con-centrations were measured using a BCA protein assay kit (Thermo). Proteins weredenatured with Laemmli buffer at 70 uC for 10 min, separated on a NuPAGE4–12% Bis-Tris gel (Novex), and transferred to a polyvinylidene difluoride (PVDF)membrane. Membranes were blocked in 5% milk (1 h at room temperature) andblotted with the following primary antibodies (overnight at 4 uC): guinea pig anti-SHANK3 (1:700, gift from Carlo Sala), mouse anti-GLUN1 (1:1,000, SYSY), rabbitanti-GLUA1 (1:1,000, Upstate), mouse anti-GLUA2 (1:1,000, NeuroMab), andmouse anti-GAPDH (MS, 1:1,000, Ambion). To control loading protein amounts,we first quantified the relative concentrations of GAPDH in all lysates. Afterwards,

RESEARCH LETTER

Macmillan Publishers Limited. All rights reserved©2013

15–25mg of proteins were loaded into a gel and intensities of appropriate bandswere measured and normalized to the intensity of the GAPDH band in the samelane. Bands were visualized and quantified using the FluorChemQ System (AlphaInnotech).Immunocytochemistry, imaging and synaptic assays. Cells were fixed in 4%paraformaldehyde (15 min at room temperature), washed (33) with PBS glycine,permeabilized with 0.5% Nonidet P40 (7 min at room temperature), washed (33)with PBS, and blocked in 3% BSA (overnight at 4 uC). Primary antibodies werediluted in 3% BSA and applied overnight at 4 uC. The following primary antibodieswere used: chicken anti-GFP (Abcam, 1:1,000), rabbit anti-tRFP (Axxora, 1:1,000),guinea pig anti-SHANK3 (produced by TM Boeckers, 1:800), mouse anti-SYN1(SYSY, 1:500), guinea pig anti-HOMER1 (SYSY, 1:200), mouse anti-PSD95 (Abcam,1:200), rat anti-CTIP2 (Abcam, 1:200), mouse anti-SATB2 (Abcam, 1:10), mouseanti-GAD67, and anti-nuclei (Millipore, 1:100). Secondary antibodies were diluted(1:500–1,000) in 3% BSA and applied for 1 h at room temperature. The followingsecondary antibodies were used: Alexa-405, -488, -594 and -647 (Invitrogen).Neurons were visualized using either a Zeiss spinning disc confocal microscope,equipped with 405/488/594/640 lasers (Perkin Elmer) or a laser-scanning confocalmicroscope equipped with 405/488/594/633 lasers (Stanford Neuroscience Micro-scopy Services), using either Volocity (Improvision) or Zenn (Zeiss) software.Z-stack images of fluorescent cells were taken with a 0.2–0.7mm interval using3100 or 363 objectives that cover an 80 3 80 or 150 3 150mm2 area, correspond-ingly. Images within an experiment were acquired with identical acquisition set-tings. Approximately 20 images (10 GFP1 and 10 mKate21 neurons) were collectedfrom each coverslip. Visualization and analysis of all images were performed usingVolocity 5.1 and ImageJ. For quantification of total SHANK3 expression, a three-dimentional mask of a neuron was created based on GFP or mKate2 fluorescenceand the mean SHANK3 intensity within the masked area was calculated. Forquantification of the number of synapses, Z-stacks were merged using maximumintensity readings for each of the channels and a two-dimensional mask of a neuronwas created based on the GFP or mKate2 fluorescence reading. Presynaptic (SYN1)and postsynaptic (SHANK3, HOMER1 or PSD-95) puncta within a mask werefound either manually using a cell counter plug-in for ImageJ or automatically usinga custom measurement protocol created in Volocity. During automatic quantifica-tion, once settings for puncta detection were chosen, they were uniformly appliedfor all images within an experiment. Results of automatic puncta quantificationwere always validated by visual inspection and manual counting for a subset of theimages. All cover slips were coded and processed in a blinded manner. To combinedata acquired from different coverslips in different experiments, we normalized allpuncta counts to the median number of puncta detected on control neurons for aparticular cover slip.Construction of SHANK3 lentiviral vector. The translated region of the ratSHANK3 cDNA (NM_021676.1, generously provided by R. Xavier) was isolatedin pieces by PCR and flanked with NdeI (59) and ClaI (39) restriction enzyme sites,and then subcloned into the pSC-B-amp/kan vector using the StrataClone BluntPCR Cloning kit (Stratagene). Similarly, a DNA fragment of EGFP flanked with

NotI (39 UTR), NdeI (59 TR before stop codon), and ClaI (39 UTR) restrictionenzyme sites was obtained by PCR and subcloned into a pSC-B-amp/kan vector.Afterwards, NotI-EGFP-ClaI (0.7 kb) and NdeI-SHANK3-ClaI (5.2 kb) fragmentswere isolated, gel-purified, and sequentially inserted into the NotI and ClaI sitesand then into an NdeI- and ClaI-digested pHAGE-EF1a lentiviral vector backbone32

to generate the pHAGE-EF1a-EGFP-SHANK3 (13 kb) vector. The sequence wasconfirmed by sequencing.Lentivirus production, titration and infection. Lentiviruses were generated witha five-plasmid transfection system using Fugene 6 transfection reagent (Roche) in293T cells as previously described32. 48 h after transfection, supernatants werecollected at 12-h intervals, 4–5 times in total. Viral particles were concentratedusing Lenti-X concentrator (Clontech), aliquoted (200ml), and stored at 280 uC.

Different amounts of virus (5–100ml) were placed on ,80% confluent HEK293Tcells for 3–4 days. Approximately 80% of the cells in the 35-mm plate were fluor-escent upon addition of 100ml of virus. We used 100ml of GFP–SHANK3 virus totransduce human iPS cell-derived neurons (Day 25–30) growing in 35-mm cultureplates. The medium was replaced 10–20 h later. Neurons were transferred onto ratprimary astrocytes after 5–10 days.Drugs and treatment procedure. The following agents were tested: 0.1mM tri-chostatin A (TSA, Tocris), 1 mM valproic acid (VPA, Calbiochem), 3mM nifedi-pine (Nif, Tocris), 20 ng ml21 insulin growth factor 1 (IGF1, R&D Systems), and20 ng ml21 insulin growth factor 2 (IGF2, R&D Systems). Concentrations werechosen on the basis of previous reports of their use available in the literature.Effects of TSA, VPA, and Nif were assessed after 24–36 h of treatment andIGF1 and IGF2 after 2–3 weeks, with approximately 25% of the media changedevery other day.Data collection and statistics. All experiments were replicated at least three timesusing iPS cell lines indicated in Supplementary Table 3. The sample size anddescription of the sample collection are reported in each figure legend. The num-ber of samples was selected based on the variability of the particular assay. Sampleswere excluded from analysis only when they were clear outliers (identified visuallyand confirmed with the Grubbs’ test) or were erroneously generated as a result of atechnical problem. Wells with neurons were randomly selected for an assay/drugtreatment, coded, and processed in a blinded manner. Statistical tests used forcomparison are reported in each figure legends. For data sets where parametrictests were used for analysis, the normal distribution of data was confirmed usingthe Kolmogorov–Smirnov test.

29. Chambers, S. M. et al. Highly efficient neural conversion of human ES and iPScells by dual inhibition of SMAD signaling. Nature Biotechnol. 27, 275–280(2009).

30. Gaspard, N. et al. Generation of cortical neurons from mouse embryonic stemcells. Nature Protocols 4, 1454–1463 (2009).

31. Bellone, C. & Nicoll, R. A. Rapid bidirectional switching of synaptic NMDAreceptors. Neuron 55, 779–785 (2007).

32. Sommer, C. A. et al. Induced pluripotent stem cell generation using a singlelentiviral stem cell cassette. Stem Cells 27, 543–549 (2009).

LETTER RESEARCH

Macmillan Publishers Limited. All rights reserved©2013