Supporting InformationHsia et al. 10.1073/pnas.1400737111SI Materials and MethodsExpression Vectors and Antibodies. The expression vectors encod-ing lentiviral supplementary proteins (pCMVdeltaR8.2; Addgeneplasmid 12263) and lentiviral envelope protein VSV-G (pMD2.G;Addgene plasmid 12259) were kind gifts from D. Trono (EcolePolytechnique Fédérale de Lausanne, Lausanne, Switzerland).The expression vectors pFUGW (Addgene plasmid 14883) (1)and pFUGWiCre were kind gifts from D. Baltimore (CaliforniaInstitute of Technology, Pasadena, CA) and R. Huganir (JohnsHopkins University School of Medicine, Baltimore, MD). Thefollowing primary antibodies were used: mouse monoclonal anti-PTEN (phosphatase and tensin homolog) (6H2.1; Millipore),rabbit monoclonal anti-PTEN (D4.3; Cell Signaling), mousemonoclonal anti-NeuN (A60; Millipore), mouse monoclonal anti-GFP (7.1 and 13.1; Roche), chicken polyclonal anti-GFP (AvesLabs), rabbit polyclonal anti-Cux1 (M-222; Santa Cruz Biotech-nology), mouse monoclonal anti-Cre (7-23; Sigma-Aldrich), rabbitpolyclonal anti-actin (Sigma-Aldrich), mouse monoclonal anti-actin (AC-40; Sigma-Aldrich), rabbit monoclonal anti-AKT (CellSignaling), rabbit monoclonal anti–phospho-AKT (Ser473) (CellSignaling), rabbit monoclonal anti–phospho-S6 (Ser235/236)(D57.2.2E; Cell Signaling), mouse monoclonal anti-GSK3β (7/GSK-3β; BD Biosciences), rabbit polyclonal anti-AnkyrinG (H-215;Santa Cruz Biotechnology), chicken polyclonal anti-MAP2 (Novus),mouse monoclonal anti–Nedd4-1 (15/Nedd4; BD Biosciences),rabbit polyclonal anti–Nedd4-2 (Cell Signaling), and mousemonoclonal anti-ubiquitin (P4D1; Santa Cruz Biotechnology).

Mass Culture of Hippocampal Neurons and Cell Biology. Hippocampifrom postnatal day (P)0 mice were isolated and digested in papainsolution (25 units/mL; Worthington Biochemical) at 37 °C for1 h. The dissociated neurons were plated on poly–L-lysine(Sigma-Aldrich)–coated coverslips at a density of 30,000 cellsper cm2. Neurons were cultured in Neurobasal-A medium (Gibco)supplemented with B27 (Gibco), 10 units/mL penicillin, 10 μg/mLstreptomycin, and 2 mM GlutaMAX-1 (Gibco). Neurons weremaintained at 37 °C in a cell-culture incubator with 5% CO2(Thermo Scientific) before fixation with 4% paraformaldehyde/PBS. For analyses of neuronal cell polarity (Fig. S9), AnkyrinG-positive but MAP2-negative neurites were scored as axons. Tostudy the outgrowth of axons (Fig. 1 A–C and Fig. S2 A–C),neurons were transfected with an EGFP expression vector at 1 din vitro (DIV1), fixed at DIV7, and immunostained for GFP (7.1and 13.1; Roche), and pictures of EGFP-expressing neuronswere acquired with a 10× objective on an Olympus BX61 fluo-rescence microscope. Analyses were performed using the NeuronJplugin in ImageJ (National Institutes of Health). The longestneurite was taken as the axon, and the branch number was de-termined by counting the number of neurites (>15 μm) extendingfrom the longest neurite. In control experiments, we determinedthat at DIV7 the longest neurite of 96.8% of assayed culturedhippocampal neurons was SMI-312–positive (n = 31 neurons),which validates our choice criterion to identify axons. For studieson the intracellular localization of PTEN, neurons were fixed atDIV7 and subjected to immunostaining for PTEN. The fluores-cence staining was imaged with a 63× objective lens on a Leicaconfocal microscope (TCS SP2 equipped with acousto-opticalbeam splitter). The mean intensity of PTEN signals was scoredusing ImageJ. Nuclei were defined by DAPI staining, and thedimensions of neuronal somata were determined manually. ForSholl analysis (2) of neurite structure, neurons were transfectedwith an EGFP expression vector at DIV1. The neurons were then

treated with 20 nM rapamycin (Cell Signaling) or DMSO at DIV6and fixed at DIV10. The Advanced Sholl Analysis plugin in Im-ageJ was used for analysis, in which concentric circles were drawnat 7.5-μm intervals using the cell body as the common center, andthe numbers of neurites crossing each of the circles were counted.For all morphological analyses on cultured neurons, cells werechosen based on the presence of continuous and extended neu-rites without unusual varicosities (blebbing). Neurons withneurites circling around the soma were excluded because thismorphology confounds the analysis. Acquisition and analyses ofimages were done by a blind observer.

Autaptic Neuron Culture and Electrophysiology. Microisland cul-tures of hippocampal neurons were prepared as described pre-viously (3, 4). In brief, astrocytes for microisland cultures wereobtained from P0 wild-type mouse cortices using digestion with0.25% trypsin (Gibco) for 20 min at 37 °C. The astrocytes wereplated in T75 culture flasks in DMEM (Gibco) containing 10%FBS (PAA) and penicillin/streptomycin (Gibco). The mediumwas exchanged the day after plating, and cells were allowed togrow for 7–10 d. The 32-mm coverslips used for microislandcultures were first coated with agarose (Sigma-Aldrich) andstamped using a custom-made stamp to generate 200 μm × 200 μmsubstrate islands with a coating solution containing poly–D-lysine (Sigma-Aldrich), acetic acid, and collagen (BD Bio-sciences). Astrocytes were collected from the T75 flask usingtrypsin digestion and plated at a density of 12,000 cells per wellon 32-mm coverslips. Hippocampal neurons were prepared asdescribed above and plated in prewarmed complete Neurobasalmedium at a density of 4,000 cells per well on a 32-mm coverslipfor microisland cultures. The cells were infected at DIV1 withlentiviruses to express EGFP alone or EGFP together with Crerecombinase. For electrophysiology, cells were whole-cell volt-age–clamped at −70 mV with the MultiClamp 700B amplifier(Molecular Devices) under the control of Clampex program 10.1(Molecular Devices). All analyses were performed using AxoGraphX (AxoGraph Scientific). Recordings were performed betweenDIV10 and DIV16. Miniature excitatory postsynaptic currentswere recorded in the presence of 300 nM tetrodotoxin (TTX). Theextracellular solution contained 140 mM NaCl, 2.4 mM KCl,10 mM Hepes, 10 mM glucose, 4 mM CaCl2, and 4 mM MgCl2(320 mOsmol/L, pH 7.3). The patch-pipette solution for autapticrecordings contained 136 mM KCl, 17.8 mM Hepes, 1 mMEGTA, 4.6 mMMgCl2, 4 mM NaATP, 0.3 mM Na2GTP, 15 mMcreatine phosphate, and 5 U/mL phosphocreatine kinase (315–320 mOsmol/L, pH 7.4). The series resistance was compensatedby about 60–80%. All chemicals, except for TTX (Tocris Cookson),were purchased from Sigma-Aldrich.

Lentivirus Preparation. Lentiviruses were produced based on apublished procedure with slight modifications (5). HEK293FTcells (1.6 × 107) (Invitrogen) were plated on a poly–L-lysine(Sigma-Aldrich)–coated 15-cm plastic dish and cultured for 24 hin Opti-MEMmedium (Gibco) containing 10% FBS (PAA). Thepackaging vector pCMVdeltaR8.2, the VSV-G expression vectorpMD2.G (6), and the backbone vector were cotransfected toHEK293FT cells with Lipofectamine 2000 (Invitrogen) accord-ing to the manufacturer’s instructions. Cells were incubated for18 h, and afterward the medium was changed to DMEM (Gibco)containing 2% FBS (PAA), 10 units/mL penicillin, 10 μg/mLstreptomycin (Gibco), and 10 mM sodium butyrate (Merck). Theculture medium, which contained lentiviral particles, was harvested

Hsia et al. www.pnas.org/cgi/content/short/1400737111 1 of 9

72 h after transfection. To increase the viral titer, Amicon cen-trifugal filters (100 kDa; Millipore) were used according to themanufacturer’s instructions to concentrate viral particles. Thehigh-titer lentivirus sample was then dialyzed with dialysis buffer(20 mM Tris·HCl, pH 8.0, 150 mM NaCl) overnight, flash-frozenwith liquid nitrogen, and stored at −80 °C until used.

In Utero Electroporation.At embryonic stage 15.5 (E15.5), plasmidDNA was transfected to a subpopulation of neuronal progenitorcells, which generate mainly layer II/III cortical neurons. Theelectroporated brains were fixed at P21, when the axons of layerII/III cortical neurons have reached the contralateral side (cal-losal projections) and branched extensively into the somatosen-sory cortex. Pregnant mice were deeply anesthetized withisoflurane and placed on a warming pad (31–32 °C) throughoutthe surgery. An ∼2-cm midline incision in the abdomen wasmade, and uterine horns were drawn out through the incisionand moistened with warmed PBS containing antibiotics (peni-cillin/streptomycin; Sigma-Aldrich). Plasmid DNA solution pre-pared with the EndoFree Plasmid Maxi Kit (Qiagen) wasinjected into the lateral ventricle through a glass micropipette,and electric pulses (35 V, 50 mA, 50-ms duration, 950-ms in-tervals, eight pulses per embryo) were applied with the ECM830electroporator (BTX Harvard Apparatus). The uterus was placedback into the abdomen, and the incision was closed with surgicalsutures. Animals were kept warm (31–32 °C) until they had re-covered from anesthesia.

Immunoprecipitation of Endogenous PTEN from Mouse Brain. Corti-ces of P7 mice were homogenized in TES buffer (50 mMTris·HCl, pH 7.5, 1 mM EDTA, 150 mM NaCl, 0.2 mM PMSF,1 μg/mL aprotinin, 0.5 μg/mL leupeptin) containing 1% SDS,20 μM MG132, and 20 mM N-ethylmaleimide (E3876; Sigma-Aldrich) and boiled for 5 min. Brain homogenates were diluted10-fold with ice-cold TES buffer containing 1% Triton X-100.After ultracentrifugation at 100,000 × g for 15 min, the solublefraction (5 mg protein) was incubated with 140 ng of a rabbitmonoclonal anti-PTEN antibody (Cell Signaling). Antibody–antigen complexes were bound to protein A Sepharose CL-4Bbeads (GE Healthcare) at 4 °C. After extensively washing withTES buffer containing 1% Triton X-100, proteins were eluted byboiling in Laemmli buffer. The immunoprecipitated PTEN and itsubiquitination level were detected by Western blotting usingmonoclonal mouse anti-PTEN (Millipore) and mouse anti-ubiquitin (P4D1; Santa Cruz Biotechnology) antibodies. For thein vitro PTEN phosphatase activity assay, PTEN was immuno-precipitated from P7 mouse brains as described above with theexception of a different buffer composition (25 mM Tris·HCl,pH 8.0, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, 0.2 mMPMSF, 1 μg/mL aprotinin, 0.5 μg/mL leupeptin). PTEN activitywas assessed using a commercially available phosphatase assaykit (K-1500; Echelon) according to the manufacturer’s instructions.The percentage of PtdInsP3 conversion was calculated accordingto the following formula:

% PtdInsP3   conversion

=f��free  phosphates  in  reaction; pmol�

−�background  phosphate  in  substrate-only  control; pmol

��

÷�input  PtdInsP3;   pmol

�g× 100%:

PTEN Phosphatase Assay. Protein A Sepharose CL-4B beads withimmunoprecipitated PTEN were washed with reaction buffer(25 mM Tris·HCl, pH 7.5, 140 mM NaCl, 2.7 mM KCl, 10 mMDTT). Enzyme reactions were carried out by incubating PTENwith 120 μMwater-soluble phosphatidylinositol-3,4,5-trisphosphate

(diC8; Echelon) in the PTEN reaction buffer for 40 min at 37 °C.Generated free phosphate was detected by the Malachite Greenreagent supplied in the phosphatase assay kit (K-1500; Echelon) ina colorimetric manner. Absorbance at 650 nm was measured andthe free phosphate level (pmol) was calculated using a phosphatestandard.

Immunohistochemistry. Paraffin-embedded mouse brain coronalsections were used for anti-NeuN staining (Fig. S2D–G). Paraffinsections (5-μm) were collected on glass slides. After deparaffi-nation, sections were incubated in blocking buffer (20% goatserum in 10% BSA/PBS) and subsequently incubated with pri-mary antibodies at 4 °C overnight, followed by incubation withthe corresponding fluorescently labeled secondary antibodies.Images were acquired on an Olympus BX61 microscope. Num-bers of cells were counted manually by a blind observer with thehelp of the Cell Counter plugin in ImageJ. For analyses of axonmorphogenesis in vivo (Fig. 1 D–G), immunostaining of 100-μmfloating vibratome sections was performed. Brain sections wereblocked with blocking buffer (5% goat serum, 0.1% Triton X-100in PBS), followed by incubation with chicken anti-GFP (AvesLabs), rabbit anti-Cux1 (Santa Cruz Biotechnology), and mouseanti-Cre (Sigma-Aldrich) antibodies that had been diluted in thesame buffer. The sections were then incubated with fluorescentlylabeled secondary antibodies diluted in 2% goat serum/PBS. Forquantification of axon branching in vivo (Fig. 1 E–G), fluores-cence images were acquired on a Leica SP2 confocal microscope.For each brain section, a z-stack serial scan for 10 images at 2.5-μmintervals was performed at the approximate level of 0.26 mmfrom bregma (7). Images taken from the somatosensory cortex atthe contralateral side were then projected to two dimensionsusing the maximum-intensity projection function of the Leicaconfocal software. Signals from the cortical surface (cortical levelset as 1) to the bottom of the white matter (cortical level set as 0)were quantified using the Plot Profile function in ImageJ.Fluorescence intensity was normalized to the maximum value inthe white matter to compensate for the variability in numbers ofafferent axons originating from the transfected neurons. Thenormalized maximum values from the upper layer (cortical level0.7–1) and lower layer (cortical level 0.25–0.69) of the cortexwere subjected to statistical analyses.

Western Blotting. Proteins separated by SDS/PAGE were trans-ferred to PVDF or nitrocellulose membranes. Nonspecificbinding of antibodies to the membranes was preblocked by in-cubation with 5% skim milk in TBST [20 mM Tris·HCl, pH 7.5,137 mM NaCl, 1% Tween 20 (wt/vol)] containing 5% goat se-rum. Subsequently, the membranes were incubated with primaryantibodies, followed by incubation with the corresponding sec-ondary antibodies diluted in 5% skim milk in TBST. Signals weredetected with the enhanced chemiluminescence (ECL) system(GEHealthcare) or the Odyssey Infrared Imaging System (LI-CORBiosciences). Quantification was done by using Odyssey software,the Bio-Rad ChemiDoc system, or ImageJ.

RNA Preparation. Cultured neurons were harvested at DIV8 withRIPA buffer (50 mM Tris·HCl, pH 7.5, 1 mM EDTA, 150 mMNaCl, 1% Triton X-100, 0.1% SDS), and the NucleoSpin RNAXS Kit (Macherey-Nagel) was used according to the manu-facturer’s instructions. For polysome fractionation and mRNAanalysis, 5 × 106 neurons were harvested at DIV8 with lysisbuffer [10 mM Tris·HCl, pH 7.5, 10 mM NaCl, 10 mM MgCl2,1% Triton X-100, 1% deoxycholate, 1 mM DTT, 30 units/mLRNase OUT (Invitrogen) containing 100 μg/mL cycloheximide(Sigma-Aldrich; 01810), proteinase inhibitors (Sigma-Aldrich;P8340), phosphatase inhibitor II (Sigma-Aldrich; P5726), andphosphatase inhibitor III (Sigma-Aldrich; P0044)]. After 5 minof incubation on ice, the extract was centrifuged for 8 min at

Hsia et al. www.pnas.org/cgi/content/short/1400737111 2 of 9

12,000 × g and 4 °C. The supernatant was loaded onto a 10–60%(wt/vol) sucrose gradient and sedimented by centrifugation at 4 °Cfor 150 min at 169,000 × g in a Beckman SW41 rotor.

Real-Time Quantitative PCR. For the quantitative analysis ofmRNAs (Fig. 3C), 150 ng of total RNA was subjected to onecycle of retrotranscription in a mixture of random nonamer primers,anchored poly-dT primers, and the SuperScript III RNase H re-verse transcriptase (Invitrogen) according to the manufacturer’sinstructions. Real-time PCR [RT-quantitative (q)PCR] was carriedout with SYBR Green Master Mix (Applied Biosystems), cDNAs,and the gene-specific primers. Samples were analyzed using the7500 Fast Real-Time PCR System (Applied Biosystems). The inputconcentration of cDNAs was optimized to ensure that the detectionwas within the linear range. Relative levels of mRNAs of interestwere normalized to the levels of two housekeeping mRNAs, Rpl13aand ATP5b. For the RNA analysis (mRNA translation) afterpolysome fractionation (Fig. 3 D and E), each gradient was col-lected in 12 fractions, followed by the addition of 1% SDS (finalconcentration), 40 pg of exogenous (in vitro transcribed) BC200RNAs, 10 mg glycogen, and 50 μg/mL proteinase K, and incubatedfor 30 min at 37 °C. The exogenous human BC200 RNA (differentsequence from rodent BC1 RNA) was used to monitor possible

RNA loss from each fraction during RNA phenol/chloroform ex-traction and precipitation. RNAs were precipitated with 0.2 MNaOAc and 0.7 volume of isopropanol. Pellets were then re-suspended in 10 μL of ddH2O. RNA fractions 1–7 (polysomalfraction; P) and 1–12 (total RNA) were pooled, and RNA quality/quantity was assessed by 1.8% agarose formaldehyde gel electro-phoresis and spectrophotometry (ND-1000 spectrophotometer;NanoDrop). Following in vitro retrotranscription, RT-qPCRwas performed. The translational efficiency was calculated asfollows: 2–[ΔCt(P)–ΔCt(Total)] = 2–ΔΔCt, where ΔCt equals Ct (spe-cificmRNAs) – Ct (BC200 RNA). Gene-specific primers used forRT-qPCR in the present study were: Nedd4-1 forward, 5′-AG-CATGAACCACCAGGTCA-3′; reverse, 5′-TTTTTCCGAATC-CATCATCC-3′; Nedd4-2 forward, 5′-AATGACCTGGGCCCT-CTT-3′; reverse, 5′-GTAAAACGTGCGGCCATC-3′; Neurofila-ment H forward, 5′-CATTGAGATTGCCGCTTACA-3′; reverse,5′-ACTCGGACCAAAGCCAATC-3′; Rpl13a forward, 5′-ATC-CCTCCACCCTATGACAA-3′; reverse, 5′-GCCCCAGGTAA-GCAAACTT-3′; ATP5b, forward, 5′-GGATCTGCTGGCCCC-ATAC-3′; reverse, 5′-CTTTCCAACGCCAGCACCT-3′; GAP-DH, forward, 5′-GGCTCATGACCACAGTCCA-3′; reverse, 5′-TCCACAGTCTTCTGGGTGG-3′.

1. Lois C, Hong EJ, Pease S, Brown EJ, Baltimore D (2002) Germline transmission andtissue-specific expression of transgenes delivered by lentiviral vectors. Science295(5556):868–872.

2. Sholl DA (1953) Dendritic organization in the neurons of the visual and motor corticesof the cat. J Anat 87(4):387–406.

3. Jockusch WJ, et al. (2007) CAPS-1 and CAPS-2 are essential synaptic vesicle primingproteins. Cell 131(4):796–808.

4. Burgalossi A, et al. (2012) Analysis of neurotransmitter release mechanisms byphotolysis of caged Ca²+ in an autaptic neuron culture system. Nat Protoc 7(7):1351–1365.

5. Naldini L, et al. (1996) In vivo gene delivery and stable transduction of nondividing cellsby a lentiviral vector. Science 272(5259):263–267.

6. Naldini L, Blömer U, Gage FH, Trono D, Verma IM (1996) Efficient transfer,integration, and sustained long-term expression of the transgene in adult ratbrains injected with a lentiviral vector. Proc Natl Acad Sci USA 93(21):11382–11388.

7. Franklin KBJ, Paxinos G (1997) The Mouse Brain in Stereotaxic Coordinates (Academic,San Diego).

Fig. S1. Nedd4-2 regulates neurite growth in mouse hippocampal neurons. (A and B) Representative images of control (A) and Nedd4-2 KO (B) culturedhippocampal neurons transfected with an EGFP-expressing vector. (Scale bars, 50 μm.) (C) Sholl analyses for the two groups in A and B. Note that KO of Nedd4-2slightly reduced neurite growth. (D) Statistical analysis of the total number of crossing neurites obtained in the Sholl analysis shown in C. Nedd4-2f/f, n = 87; NEX-Cre;Nedd4-2f/f, n = 73. *P = 0.0433, unpaired t test. Data are expressed as mean ± SEM.

Hsia et al. www.pnas.org/cgi/content/short/1400737111 3 of 9

Fig. S2. NEX-Cre expression itself does not affect axon morphogenesis, and the thickness of cortices is reduced in NEX-Cre;Nedd4-1f/f;Nedd4-2f/f mice. (A)Representative images of NEX-Cre wild-type (WT; Left) and NEX-Cre heterozygote (Het; Right)-cultured hippocampal neurons transfected with an EGFP-expressing vector. (Scale bars, 100 μm.) (B and C) No difference in length of main axon shafts (B) and in number of primary axonal branches (C) in NEX-Cre Hetneurons compared with NEX-Cre WT neurons was detectable, indicating that the phenotypic change observed in Fig. 1 A–C is not due to the NEX-Cre mutationitself. (D) Anatomical studies on Nedd4-1f/f;Nedd4-2f/f (Left) and NEX-Cre;Nedd4-1f/f;Nedd4-2f/f (Right) mouse brains at P16. Brain paraffin sections (5 μm-thick)were immunostained with an anti-NeuN antibody. Note that the thickness of the cortex was reduced in the NEX-Cre;Nedd4-1f/f;Nedd4-2f/f brain. (Scale bars,500 μm.) (E) High-magnification images of the cortices shown in D revealed a reduction in thickness of the cortex in the NEX-Cre;Nedd4-1f/f;Nedd4-2f/f brain(Right). (Scale bars, 50 μm.) (F) Reduced thickness of the cortex in the NEX-Cre;Nedd4-1f/f;Nedd4-2f/f brain (1.157 ± 0.0173 mm, n = 11 slices from four animals)compared with the control Nedd4-1f/f;Nedd4-2f/f brain (1.459 ± 0.0245 mm, n = 8 slices from four animals). ***P < 0.0001, unpaired t test. (G) The number ofneurons in a cortical region with a width of 100 μm. No significant difference in the number of neurons was observed between the two groups. Nedd4-1f/f;Nedd4-2f/f, 142 ± 2, n = 8 slices from four animals; NEX-Cre;Nedd4-1f/f;Nedd4-2f/f, 140 ± 5, n = 11 slices from four animals. Data are expressed as mean ± SEM.

Hsia et al. www.pnas.org/cgi/content/short/1400737111 4 of 9

Fig. S3. Overexpression of exogenous Cre does not affect Nedd4-1 and PTEN levels in neurons. (A) Representative Western blotting results showing nodifference in PTEN and Nedd4-1 levels when Cre was expressed in wild-type hippocampal neurons using lentiviral infection. Neurons were harvested on DIV8and DIV15. (B) Quantification of Western blotting results for PTEN and Nedd4-1 levels at DIV8 (Left) and DIV15 (Right). No statistical differences were observedbetween groups. n = 3 per group. Data are expressed as mean ± SEM.

Fig. S4. Levels of PTEN-related signaling proteins in NEX-Cre;Nedd4-1f/f;Nedd4-2f/f mouse brains. (A and C) Representative Western blotting results showinglevels of PTEN, AKT, phospho-AKT (p-AKT, Ser473), and GSK3β in brain lysates prepared from P7 mice (A) and 5-wk-old mice (C) with the indicated genotypes.(B and D) Quantification of Western blotting results in A (B) and C (D). Signals from the candidate proteins were normalized to the loading control. Phospho-AKT levels were normalized to total AKT levels and showed a slight down-regulation in P7 but not 5-wk-old NEX-Cre;Nedd4-1f/f;Nedd4-2f/f brains. PTEN, AKT,and GSK3β showed no difference in either P7 or 5-wk-old NEX-Cre;Nedd4-1f/f;Nedd4-2f/f brains compared with controls. The residual expression of Nedd4-1 andNedd4-2 in NEX-Cre;Nedd4-1f/f;Nedd4-2f/f brains is likely from glial cells and inhibitory neurons. Given that PTEN levels and phosphatase activity were notaffected (Fig. 2), the decreased phospho-AKT (Ser473) levels in the P7 NEX-Cre;Nedd4-1f/f;Nedd4-2f/f brain are likely due to misregulation of other substrates ofNedd4-family E3 ligases [e.g., growth factor receptors or phosphatases for phospho-AKT (Ser473)]. **P = 0.002, ***P < 0.001, unpaired t test, n = 5 per group.Data are expressed as mean ± SEM.

Hsia et al. www.pnas.org/cgi/content/short/1400737111 5 of 9

Fig. S5. Normal localization of PTEN upon deletion of Nedd4-1 and Nedd4-2. (A) Specificity of the anti-PTEN antibody (clone D4.3; Cell Signaling) in im-munostaining. (Scale bars, 50 μm.) (B and C) Overview of the subcellular localization of PTEN in hippocampal neurons prepared from Nedd4-1f/f;Nedd4-2f/f mice(B) and NEX-Cre;Nedd4-1f/f;Nedd4-2f/f mice (C). Neurons were transfected with an EGFP expression vector at DIV1, fixed at DIV7, and immunostained forendogenous PTEN. High-magnification images of the regions in the white boxes are shown (Lower Left). The soma and axonal growth cone of the EGFP-expressing neuron are indicated by arrowheads and arrows, respectively. Note that PTEN signals at the axonal growth cone were very weak compared withthose in the soma and proximal neurites. This finding is strikingly different from data obtained in Xenopus laevis retinal ganglion cells (1). [Scale bars, 50 μm(low-magnification images) and 10 μm (high-magnification images).] (D) Representative images for control (Nedd4-1f/f;Nedd4-2f/f) and Nedd4-1;Nedd4-2double KO neurons (NEX-Cre;Nedd4-1f/f;Nedd4-2f/f) immunostained for endogenous PTEN. (Scale bars, 50 μm.) (E) Quantification demonstrated no significantdifference between the two groups (D) in the ratio of nuclear PTEN vs. total PTEN in the soma. P = 0.991, unpaired t test with Welch’s correction. Nedd4-1f/f;Nedd4-2f/f, n = 83; NEX-Cre;Nedd4-1f/f;Nedd4-2f/f, n = 78. Data are expressed as mean ± SEM.

1. Drinjakovic J, et al. (2010) E3 ligase Nedd4 promotes axon branching by downregulating PTEN. Neuron 65(3):341–357.

Hsia et al. www.pnas.org/cgi/content/short/1400737111 6 of 9

Fig. S6. Up-regulation of Nedd4-1 in GFAP-Cre;PTENf/f brains. (A) Representative Western blotting results showing up-regulation of Nedd4-1 in GFAP-Cre;PTENf/f brains. GFAP-Cre;PTENf/f mice specifically express Cre recombinase in adult neural stem cells in the subventricular zone (SVZ) and in a subpopulation ofastrocytes. Mice at 8 wk of age were used to ensure a fully developed SVZ and to achieve a nearly complete PTEN deletion. Areas of Cre recombination (SVZ)were dissected from PTENf/f and GFAP-Cre;PTENf/f mouse brains. Protein lysates were prepared using RIPA buffer containing 1% SDS. (B) Quantification of theWestern blotting results. Signals for Nedd4-1 were normalized to signals for actin. Numbers on the y axis represent arbitrary units. *P = 0.0261, unpaired t test,n = 4 per group. Data are expressed as mean ± SEM.

Fig. S7. Role of Nedd4-1 as a downstream target of PTEN-dependent signaling in neurite morphogenesis is reflected by effects on miniature excitatorypostsynaptic currents (mEPSCs). (A) Representative mEPSC traces recorded from control (blue), PTEN KO (violet), and PTEN;Nedd4-1 double KO (green) autapticneurons. Autaptic neurons were prepared from PTENf/f and PTENf/f;Nedd4-1f/f mice, and infected with the EGFP- and Cre-coexpressing lentivirus (PTEN KO andPTEN;Nedd4-1 double KO, respectively). Control autapic neurons were prepared from littermates and infected with the only-EGFP–expressing lentivirus. (B andC) Quantification of mEPSC frequencies (B) and amplitudes (C). Note that the increase in mEPSC frequency in PTEN KO neurons is partially rescued by additionalloss of Nedd4-1, indicating a rescue of the increased cell size and synapse numbers seen in PTEN KO neurons (Fig. 4 A–E). The increased mEPSC amplitudes inPTEN;Nedd4-1 double KO neurons may be due to increased levels of postsynaptic AMPA receptors, as indicated by a previous study (1). Control, 1.99 ± 0.22 Hz,23.7 ± 0.86 pA; PTEN KO, 4.12 ± 0.43 Hz, 24.3 ± 0.87 pA; PTEN;Nedd4-1 double KO, 2.49 ± 0.33, 28.9 ± 1.6 pA. ***P < 0.001, **P < 0.01, *P < 0.05, one-wayANOVA and Tukey’s post hoc test, n > 50 cells each. Data are expressed as mean ± SEM.

1. Schwarz LA, Hall BJ, Patrick GN (2010) Activity-dependent ubiquitination of GluA1 mediates a distinct AMPA receptor endocytosis and sorting pathway. J Neurosci 30(49):16718–16729.

Hsia et al. www.pnas.org/cgi/content/short/1400737111 7 of 9

Fig. S8. Enhanced mTORC1 activity partially accounts for the hypertrophy of neurites in PTEN KO neurons. (A–D) Representative images of a control neurontreated with DMSO (A), a control neuron treated with rapamycin (B), a PTEN KO neuron treated with DMSO (C), and a PTEN KO neuron treated with rapamycin(D). Cultured hippocampal neurons were prepared from PTENf/f mice and transfected with an EGFP- and Cre-expressing vector (PTEN KO). Control neuronswere prepared from littermates and transfected with the EGFP-only–expressing vector. (Scale bars, 50 μm.) (E) Sholl analyses for the four groups in A–D. Notethe difference between the PTEN KO+rapa and control+rapa groups, indicating that pathways independent of mTORC1 activity also contribute to the en-hanced neurite growth in PTEN KO neurons. (F) Statistical analysis of the total number of crossing neurites obtained in the Sholl analysis shown in E. **P < 0.01,***P < 0.001. Data were analyzed by one-way ANOVA and Tukey’s post hoc test (n > 49 neurons per group). (G and H) Representative images of a PTEN;Nedd4-1double KO (DKO) neuron treated with DMSO (G) or rapamycin (H). (Scale bars, 50 μm.) Cultured hippocampal neurons were prepared from littermates with thegenotype PTENf/f;Nedd4-1f/f and transfected with an EGFP- and Cre-coexpressing vector. (I) Sholl analyses for the two groups in G and H. Note that rapamycin hadhardly any effect on dendrite growth in PTEN;Nedd4-1 double KO neurons. (J) Statistical analysis of the total number of crossing neurites obtained in the Shollanalysis shown in I. Data obtained in G–J are from a subset of experiments shown in A–F. P = 0.9333, unpaired t test. PTEN;Nedd4-1 DKO+DMSO, n = 49; PTEN;Nedd4-1 DKO+Rapa, n = 60. Data are expressed as mean ± SEM.

Hsia et al. www.pnas.org/cgi/content/short/1400737111 8 of 9

Fig. S9. Nedd4-1 and Nedd4-2 are dispensable for PTEN-mediated establishment of neuronal polarity. (A) Representative images of control (Top), Nedd4-1;Nedd4-2 double KO (Middle), and Nedd4-1;Nedd4-2;PTEN triple KO neurons (Bottom). Primary hippocampal neurons prepared from the indicated genotypeswere transfected with an EGFP-expressing vector or an EGFP- and Cre-coexpressing vector at DIV1 and fixed at DIV7. Staining for AnkyrinG (AnkG) was used tolabel axon initial segments. (Scale bars, 50 μm.) (B) Statistical analysis of the number of axons for the three groups in A. The percentage of neurons projectingmultiple axons was increased in Nedd4-1;Nedd4-2;PTEN triple KO neurons compared with Nedd4-1;Nedd4-2 double KO neurons (***P < 0.0001, χ2 test) andcontrol neurons (***P = 0.0005, χ2 test). Nedd4-1;Nedd4-2 double KO neurons, on the other hand, showed no significant difference compared with controlneurons. n > 120 neurons per group. (C) Representative images of control (Upper) and PTEN KO neurons (Lower). (Scale bars, 50 μm.) (D) Statistical analysis ofthe number of axons for the two groups in C. A significantly increased percentage of PTEN KO neurons projected multiple axons compared with controlneurons. *P = 0.0394, χ2 test. Control, n = 123; PTEN KO, n = 115. Data are expressed as mean ± SEM.

Hsia et al. www.pnas.org/cgi/content/short/1400737111 9 of 9