PSD-95 stabilizes NMDA receptors by inducing thedegradation of STEP61Sehoon Wona, Salvatore Incontrob, Roger A. Nicollb,c,1, and Katherine W. Rochea,1

aReceptor Biology Section, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, MD 20892; bDepartment ofCellular and Molecular Pharmacology, University of California, San Francisco, CA, 94158; and cDepartment of Physiology, University of California, SanFrancisco, CA, 94158

Contributed by Roger A. Nicoll, June 23, 2016 (sent for review March 31, 2016; reviewed by David S. Bredt and John Isaac)

Phosphorylation regulates surface and synaptic expression ofNMDA receptors (NMDARs). Both the tyrosine kinase Fyn and thetyrosine phosphatase striatal-enriched protein tyrosine phosphatase(STEP) are known to target the NMDA receptor subunit GluN2B ontyrosine 1472, which is a critical residue that mediates NMDARendocytosis. STEP reduces the surface expression of NMDARs bypromoting dephosphorylation of GluN2B Y1472, whereas the synapticscaffolding protein postsynaptic density protein 95 (PSD-95) stabilizesthe surface expression of NMDARs. However, nothing is known abouta potential functional interaction between STEP and PSD-95. We nowreport that STEP61 binds to PSD-95 but not to other PSD-95 familymembers. We find that PSD-95 expression destabilizes STEP61 viaubiquitination and degradation by the proteasome. Using subcellularfractionation, we detect low amounts of STEP61 in the PSD fraction.However, STEP61 expression in the PSD is increased upon knockdownof PSD-95 or in vivo as detected in PSD-95–KO mice, demonstratingthat PSD-95 excludes STEP61 from the PSD. Importantly, only extrasy-naptic NMDAR expression and currents were increased upon STEPknockdown, as is consistent with low STEP61 localization in the PSD.Our findings support a dual role for PSD-95 in stabilizing synapticNMDARs by binding directly to GluN2B but also by promoting synap-tic exclusion and degradation of the negative regulator STEP61.

PSD-95 | NMDA receptor | STEP | ubiquitination

NMDA receptors (NMDARs) are ionotropic glutamate re-ceptors that are expressed throughout the nervous system

and play crucial roles in neuronal development, synaptic plas-ticity, and learning and memory (1–4). Functional NMDARs areheterotetrameric complexes that are composed of homologoussubunits (GluN1, GluN2A-D, and GluN3A-B). Two GluN1 sub-units typically combine with two GluN2 subunits, which modulatechannel activity and receptor properties (5–8). GluN2A andGluN2B are the predominant GluN2 subunits in hippocampusand cortex, and the subunit composition varies during neuronaldevelopment (9–12). Therefore the precise regulation of NMDARsubunit expression, composition, trafficking, and localization iscritical for proper neuronal function. NMDAR activity is dynami-cally regulated by protein phosphorylation, e.g. NMDAR currentsare potentiated by tyrosine kinases and suppressed by tyrosinephosphatases (13). In fact, GluN2B is the most prominent tyrosinephosphorylated protein within postsynaptic densities (PSDs) (14),and phosphorylation is increased during long-term potentiation(LTP) in CA1 hippocampus (15). Moreover, tyrosine phosphory-lation of GluN2B has been shown to increase in several patho-logical conditions, including ischemia and seizures (16–19).Striatal-enriched protein tyrosine phosphatase (STEP, also

known as “PTPN5”) is a brain-specific protein phosphatase thatis expressed in the striatum, hippocampus, and cortex (20, 21).The STEP family of protein tyrosine phosphatases includes bothmembrane-associated [striatal-enriched protein tyrosine phospha-tase 61 (STEP61)] and cytosolic (STEP46) variants that are gener-ated by alternative splicing of a single gene (22–24). STEP61 ispresent in the PSD of glutamatergic synapses (25, 26) and has beenshown to influence NMDAR-mediated synaptic currents and LTP

in the hippocampus (17, 27–29). GluN2B is specifically phos-phorylated by the Src family tyrosine kinase Fyn on tyrosine1472, which is part of an endocytic motif (YEKL). Importantly,GluN2B Y1472 phosphorylation promotes surface expression ofGluN2B-containing NMDARs by disrupting binding to the AP-2clathrin-associated adaptor protein complex, which targets pro-teins for endocytosis (30–32). In contrast, STEP61 dephosphor-ylates GluN2B Y1472, leading to internalization of NMDARcomplexes (17, 33, 34). STEP61 also dephosphorylates Fyn at aregulatory tyrosine residue (Y420), thereby inhibiting its activity(35) and indirectly decreasing Y1472 phosphorylation. In con-trast, postsynaptic density protein 95 (PSD-95), a member of themembrane-associated guanylate kinase (MAGUK) family, sta-bilizes the surface expression of GluN2B-containing NMDARsby its direct interaction with the GluN2B PDZ (PSD-95/Discs-large/ZO-1) ligand, which is adjacent to the endocytic motif.However, there is no evidence for specific physical or functionalinteractions between STEP61 and PSD-95.Here we used a variety of approaches to determine if STEP61

and PSD-95 interact to regulate the trafficking of NMDARs. Wedemonstrate that STEP61 binds specifically to PSD-95, but not toother MAGUKs in heterologous cells, cultured neurons, andmouse brain. Interestingly, we observed that PSD-95 decreasesSTEP61 expression by promoting its ubiquitination and degra-dation via the ubiquitin–proteasome system (UPS). In particular,we found that PSD-95 knockdown in cultured neurons or itsabsence in PSD-95–KO mice resulted in increased STEP61 ex-pression with a concomitant reduction in the phosphorylationof GluN2B Y1472 and receptor expression in the PSD. Using

Significance

NMDA receptors (NMDARs) are principal regulators of synapticsignaling in the brain. Modulation of NMDARs’ function andtrafficking is important for the regulation of synaptic trans-mission and several forms of synaptic plasticity. Postsynapticdensity protein 95 (PSD-95) acts as a scaffolding protein andstabilizes the surface and synaptic expression of NMDARs,whereas striatal-enriched protein tyrosine phosphatase (STEP),a brain-specific protein tyrosine phosphatase, dephosphorylatesand destabilizes NMDARs via endocytosis. We now demon-strate that PSD-95 binds to STEP61 and promotes its degradationvia the proteasome, thereby stabilizing surface expression ofNMDARs. We have revealed a dynamic role for PSD-95 insculpting protein content at excitatory synapses that is distinctfrom its canonical role as a scaffolding protein.

Author contributions: S.W., S.I., R.A.N., and K.W.R. designed research; S.W. and S.I.performed research; R.A.N. and K.W.R. contributed new reagents/analytic tools; S.W.and S.I. analyzed data; and S.W., S.I., R.A.N., and K.W.R. wrote the paper.

Reviewers: D.S.B., Johnson and Johnson; and J.I., Eli Lilly & Co.

The authors declare no conflict of interest.1To whom correspondence may be addressed. Email: rochek@ninds.nih.gov or nicoll@cmp.ucsf.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1609702113/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1609702113 PNAS Early Edition | 1 of 9

NEU

ROSC

IENCE

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 1

5, 2

021

immunofluorescence microscopy, we determined that STEP61knockdown increases NMDAR surface expression and also NMDA-mediated currents in hippocampal neurons. However, the simul-taneous knockdown of STEP61 and PSD-95 has no effect onNMDAR surface expression, indicating a precise interplay betweenthese two proteins. Therefore, PSD-95 has a dual function inregulating NMDARs by stabilizing NMDARs via direct interactionwith the GluN2B PDZ-binding domain and also controlling syn-aptic content by triggering the degradation of STEP61 to diminishthe destabilizing effect of STEP61 on surface NMDARs.

ResultsSTEP61 Is a Membrane Protein That Binds Specifically to PSD-95.STEP has two isoforms: STEP61 and STEP46 (Fig. 1A), whichvary in their predicted membrane association (SI Results and Fig.S1). We first tested whether each isoform is expressed at the cellsurface. We expressed STEP61 or STEP46 in HEK293T cells andperformed cell-surface biotinylation assays. We found that STEP61is expressed at the cell surface in heterologous cells (Fig. 1B), butSTEP46 is not. Similarly, STEP61 is surface-expressed in cultured ratcortical neurons (Fig. 1B). We could not analyze STEP46, becauseonly STEP61 is expressed in cortex (Fig. S1D). STEP61 andMAGUKs are expressed at synapses and affect synaptic NMDARs,but it was not known if they interact directly. To assess which

members of the MAGUK family might bind to STEP61, we coex-pressed STEP61 and several GFP-tagged MAGUKs [PSD-95, syn-apse-associated protein 97 (SAP97), and SAP102] in HEK293Tcells and immunoprecipitated each MAGUK using GFP antibody.Immunoblotting for STEP61 revealed that PSD-95, but not SAP97or SAP102, binds to STEP61 (Fig. 1C). Similarly, when we immu-noprecipitated STEP61 (using HA antibody) and immunoblottedfor MAGUKs, we found that only PSD-95 binds to STEP61 (Fig.1D). To test whether endogenous STEP61 can bind to PSD-95, weperformed coimmunoprecipitation assays with STEP antibody fromlysates prepared from primary cultured rat cortical neurons (Fig.1E) at day in vitro (DIV) 21 or from crude synaptosomal fractionsfrom adult mouse cortex using STEP antibody (Fig. 1F). In primarycortical neurons or in the crude synaptosomal fraction, GluN2A,GluN2B, and Fyn kinase, which are known STEP61-binding pro-teins, coimmunoprecipitated with STEP61. We also observed robustbinding of PSD-95 to STEP61 but not to PSD-93 or SAP102. Wedid not include SAP97 because it did not bind to STEP61 inHEK293T cells. To exclude the possibility that STEP61 binding toPSD-95 is mediated by Fyn, which binds to PSD-95 and NMDARs(36), we tested whether STEP61 can bind to PSD-95 in the absenceof Fyn using adult Fyn-KO mouse brain (Fig. 1G). We found thatSTEP61 binds to PSD-95 and GluN2B but not to SAP102.

Fig. 1. STEP61 is expressed at the cell surface and specifically binds to PSD-95. (A) Domain structure of STEP61 and STEP46. The green boxes represent the PRD,the black boxes represent the transmembrane domain; the red box represents the kinase-interacting motif (KIM); and the yellow box represents the proteintyrosine phosphatase domain (PTP). (B) Cell-surface biotinylation assays were performed in HEK293T cells expressing HA-tagged STEP61 or STEP46 (Left) or incultured rat cortical neurons (Right) to monitor endogenous proteins. GAPDH is a negative control, and GluN1 is a positive control for surface expression inneurons. (C and D) HA-STEP61 and a GFP-tagged MAGUK (PSD-95, SAP97, or SAP102) were expressed in HEK293T cells, and coimmunoprecipitations wereperformed, followed by immunoblotting. (C) MAGUKs were immunoprecipitated with GFP antibody. (D) STEP61 was isolated with HA antibody. (E–G) Primarycultured rat cortical neurons at DIV 21 (E) or the adult mouse cortex or the P2 fraction (crude synaptosomes) of adult WT mouse cortex (F) or adult Fyn-KOmouse cortex (G) was solubilized with 1% DOC (sodium deoxycholate), immunoprecipitated with STEP antibody or with IgG as control, and probed with theindicated antibodies.

2 of 9 | www.pnas.org/cgi/doi/10.1073/pnas.1609702113 Won et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

5, 2

021

PSD-95 Regulates STEP61 Expression Levels. STEP61 can bind toPSD-95, whereas a cytosolic form of STEP, STEP46, cannot (SIResults and Fig. S2 C and D). It is known that palmitoylation ofPSD-95 regulates its localization to the membrane; therefore, wenext examined whether palmitoylation of PSD-95 can affect itsbinding to STEP61. We coexpressed GFP-tagged WT PSD-95 orthe palmitoylation-deficient PSD-95 mutant (PSD-95 C3,5S) withSTEP61 in HEK293T cells and performed a coimmunoprecipita-tion assay for PSD-95 with GFP antibody. Our assay showed thatblocking palmitoylation of PSD-95 significantly reduced orblocked its interaction with STEP61 (Fig. S3), as is consistent withmembrane association being critical for the interaction. We alsotreated cortical neurons with 2-bromopalmitate (2-BP) to blockpalmitoylation. This treatment dramatically reduced the bindingof STEP61 to PSD-95 (Fig. S3B), further demonstrating the im-portance of palmitoylation in the binding of the two proteins.Next, we tried to determine the consequence of PSD-95

binding to STEP61. We tested whether PSD-95 can directly affectthe expression or trafficking of STEP61. First, we coexpressedSTEP61 with WT PSD-95 or palmitoylation-deficient mutantPSD-95 C3,5S in primary cultured cortical neurons and isolatedproteins. Immunoblotting of STEP61 coexpressed with WTPSD-95 reveals a reduction to 61.7% of total protein comparedwith STEP61 expressed alone. In contrast, PSD-95 C3,5S doesnot significantly affect STEP61 expression compared withSTEP61 expressed alone (Fig. 2A). Endogenous STEP61 isexpressed mostly in the Triton X-100–soluble fraction in ratbrain, most likely in extrasynaptic regions, whereas PSD-95 ispresent mainly in the Triton X-100–insoluble fraction, whichincludes the PSD (Fig. 2B). Because PSD-95 is mostly insolublein Triton X-100, whereas STEP61 is mostly soluble in Triton

X-100, we performed fractionation into two parts: Triton X-100–soluble and Triton X-100–insoluble fractions. Interestingly, STEP61expression is greatly reduced, to 40.3%, in the Triton X-100–insoluble fraction, whereas it shows a more modest reduction, to83.9%, in the Triton X-100–soluble fraction (Fig. 2C). Taken to-gether, these data indicate that PSD-95 decreases STEP61 ex-pression levels in both fractions, but this effect is more profoundin the Triton X-100–insoluble fraction, which contains the PSD.Therefore we focused primarily on the Triton X-100–insolublefraction for our biochemical studies.

The Regulation of STEP61 Stability by PSD-95 Is Dependent onUbiquitination and Degradation via the Proteasome. To investigatethe PSD-95–dependent regulation of STEP61 expression levels, wetreated cultured cortical neurons expressing STEP61 and PSD-95with the proteasome inhibitor MG-132 (1 μM) or the lysosomaldegradation inhibitor chloroquine (50 μM) for 16 h. Interestingly,MG-132 treatment markedly increased STEP61 expression to8.2-fold higher than STEP61 expression seen when STEP61 andPSD-95 are coexpressed and 1.8-fold higher than STEP61 ex-pression seen when only STEP61 is expressed (Fig. 3A). However,chloroquine did not significantly change STEP61 levels whenSTEP61 and PSD-95 were coexpressed (Fig. 3A). MG-132 treat-ment also increased PSD-95 levels, as is consistent with previousfindings showing that PSD-95 is a ubiquitinated protein and isdegraded by the UPS. To investigate if the changes in STEP61protein by PSD-95 are regulated by the UPS, we immunopreci-pitated STEP61 from neurons expressing STEP61 and WT PSD-95or PSD-95 C3,5S. Immunoblotting for ubiquitin shows that thechanges in STEP61 expression mediated by PSD-95 WT orPSD-95 C3,5S are consistent with ubiquitinated STEP61 being

Fig. 2. STEP61 expression is reduced by PSD-95 in a palmitoylation-dependent manner. (A) In primary cultured rat cortical neurons, GFP-tagged WT or mutantPSD-95 was cotransfected with HA-STEP61, and cells were solubilized with 1% SDS and immunoblotted (STEP61 and WT PSD-95, P = 0.0045; STEP61 and mutantPSD-95, P = 0.0247). Error bars represent ± SEM; *P < 0.05, **P < 0.005; n = 3. (B) Adult rat brain was fractionated to isolate the SPM and PSD. Samples fromeach fraction (10 μg of protein per lane) were immunoblotted with STEP61, PSD-95, Fyn, β-actin, and synaptophysin antibodies. (C) GFP-tagged WT or theunpalmitoylated (C3,5S) mutant PSD-95 was transfected with HA-STEP61 in primary cortical neurons. Cells were solubilized with 1% Triton X-100 (STEP61 andWT PSD-95, P = 0.0304; STEP61 and mutant PSD-95, P = 0.0079). The Triton X-100–insoluble fraction (STEP61 and WT PSD-95, P = 0.0045; STEP61 and mutantPSD-95, P = 0.0043) was solubilized with 1% SDS, and the lysates were immunoblotted as indicated. Error bars represent ± SEM; *P < 0.05, **P < 0.005; n = 3.

Won et al. PNAS Early Edition | 3 of 9

NEU

ROSC

IENCE

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 1

5, 2

021

degraded via the UPS (Fig. 3B). To test this idea, we treatedneuronal cultures cotransfected with STEP61, PSD-95 WT, orthe PSD-95 C3,5S mutant with MG-132 (1 μM), a proteasomeinhibitor, for 16 h. MG-132 treatment leads to a decrease inSTEP61 degradation by PSD-95 and an increase in the ubiquiti-nation of STEP61. Thus, these data show that PSD-95 is associatedwith the degradation of STEP61 via the UPS but is not dependenton lysosomes.

STEP61 Expression Is Elevated by PSD-95 Knockdown in Neurons. Wehad shown that increasing PSD-95 expression reduced STEP61expression levels. Therefore, we next tested if PSD-95 knock-down influences STEP61 expression (Fig. 4 A, SI Results, and Fig.S4). We transduced cultured cortical neurons with lentiviruscontaining shRNA against PSD-95, which specifically decreasedthe expression of PSD-95 (37) but not of PSD-93, even thoughthey are homologous MAGUKs. We then carried out fraction-ation as shown in Fig. 2B. In the PSD fraction (Fig. 4A), GluN2Bis reduced to 59% by PSD-95 knockdown, but PSD-93 and Fynare not changed. Interestingly, STEP61 expression is significantlyincreased by 2.6-fold in the PSD fraction (Fig. 4A). This elevatedSTEP61 leads to the reduction of Y1472 phosphorylation ofGluN2B. The ratio of Y1472 to GluN2B is reduced to 74.6%,although GluN2B is also decreased (Fig. 4A).We also generated a shSTEP lentivirus to knock down STEP61

and tested its efficiency in cultured cortical neurons. STEP61expression was significantly decreased, to 6.3% (Fig. 4B), andSTEP knockdown increased total GluN2B by 48.2%, but PSD-95and Fyn were not changed. Also, STEP knockdown leads to

increased phosphorylation of GluN2B Y1472, and the ratio ofphosphorylated Y1472 to total GluN2B is increased by 52.3% eventhough GluN2B is also increased (Fig. 4B). Using shPSD-95,shSTEP, or both, we performed immunostaining for surface-expressed GluN2B to determine whether PSD-95 knockdown,STEP knockdown, or both would regulate the surface expressionof GluN2B-containing NMDARs in hippocampal neurons. Com-pared with control, shPSD-95 considerably decreased the surfaceexpression of GluN2B, whereas the level of intracellular GluN2Bexpression was increased, as is consistent with a previous reportthat PSD-95 stabilizes the surface expression of NMDARs (32).The ratio of surface GluN2B to intracellular GluN2B is decreasedto 56% upon PSD-95 knockdown. Interestingly, shSTEP increasesthe surface expression of GluN2B and decreases intracellularGluN2B, and the ratio of surface to intracellular GluN2B isincreased by 73%. Next, we tested whether the transduction ofboth shPSD-95 and shSTEP lentivirus in hippocampal neuronsincreases or decreases the surface expression of GluN2B. In-terestingly, the expression levels of surface and intracellularGluN2B were not significantly different from the levels in con-trol (Fig. 4C).Taken together, these data indicate that STEP61 expression is

increased in PSD-95 knockdown experiments, thereby decreasing theY1472 phosphorylation level of GluN2B. Also, these results indicatethat PSD-95 and STEP61 have opposite roles in regulating the surfaceexpression of GluN2B-containing NMDARs (SI Results and Fig. S4)and, importantly, that STEP effects on NMDARs are regulatedby PSD-95.

Fig. 3. Proteasomal, but not lysosomal, inhibitors regulate STEP61 degradation by PSD-95. (A) HA-STEP61 and GFP-tagged WT PSD-95 were coexpressed inprimary cultured cortical neurons, and cells were treated with DMSO (P = 0.5e-5), MG-132 (1 μM, 16 h, P = 0.0002), or chloroquine (50 μM, 16 h, P = 0.9e-5).The Triton X-100–insoluble fraction was solubilized with 1% SDS. Lysates then were resolved by SDS/PAGE and immunoblotted as indicated (DMSO, P = 0.5e-5;MG-132, P = 0.0002; chloroquine, P = 0.9e-5). Error bars represent ± SEM; ***P < 0.001; n = 3. (B) Cultured cortical neurons expressing HA-STEP61 and GFP-tagged PSD-95 (WT or the unpalmitoylated mutant) were treated with MG-132 (1 μM, 16 h) or DMSO (control). The Triton X-100–insoluble fraction wasresolubilized with 1% SDS lysis buffer and neutralized with 1% Triton X-100 lysis buffer, immunoprecipitated with HA antibody to isolate STEP61, andimmunoblotted for ubiquitin (P4D1 antibody).

4 of 9 | www.pnas.org/cgi/doi/10.1073/pnas.1609702113 Won et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

5, 2

021

STEP61 Knockdown Increases Extrasynaptic NMDARs but Not SynapticNMDARs in Neurons. To investigate the effect of STEP knockdownon NMDARs, we transduced shSTEP lentivirus in culturedcortical neurons and 7 d later performed fractionation using 1%Triton X-100, yielding Triton X-100–soluble and Triton X-100–insoluble fractions. Interestingly, STEP knockdown significantlyincreased GluN2B and GluN1 in the Triton X-100–solublefraction but not in the –insoluble fraction (Fig. 5 A and B).PSD-95, GluN2A, and Fyn were not changed in either TritonX-100–soluble or –insoluble fractions.To analyze the effect of STEP61 knockdown on the functional

properties of NMDARs on the surface, we performed electro-physiology in hippocampal organotypic slice cultures. We firstcoated gold nanoparticles with a STEP61 shRNA expressing GFPby an independent promoter in a plasmid. These particles werebiolistically delivered to hippocampal slice cultures, and recordingswere made 7 d later from a transfected cell and simultaneouslyfrom a neighboring control cell. To measure surface NMDARs,using a large-diameter pipette, NMDA+ glycine was puffed ontoneighboring, simultaneously recorded neurons, one expressing theshRNA and the other serving as a control (Fig. 5 C and D).NMDAR currents were significantly increased in transfected neu-rons compared with control (Fig. 5 E and F). Intriguingly, evokedNMDAR excitatory postsynaptic potentials (EPSCs) did notchange in transfected neurons (Fig. 5 G and F). These findings,along with the immunostaining and fractionation experiments,suggest that STEP61 knockdown selectively increases the extra-synaptic surface expression of NMDARs.

STEP61 Expression Is Elevated in PSD-95–KO Mouse Brain. To de-termine whether PSD-95 regulates STEP61 expression in vivo, weexamined PSD-95–KO mouse brain. We performed subcellularfractionation of PSD-95–KO mouse forebrain and observed thatPSD-95 KO significantly increases STEP61 expression. STEP61expression is 1.8-fold higher in the PSD fraction from PSD-95–KO mice than in WT mice, whereas GluN2B, PSD-93, and Fynare not significantly changed (Fig. 6A). The ratio of phosphor-ylated GluN2B Y1472 versus total GluN2B is reduced to 40% byelevated STEP61 expression (Fig. 6A). Next, we immunopreci-pitated STEP and immunoblotted for ubiquitin, revealing a re-duction of ubiquitinated STEP and an increase in STEP61expression in the PSD-95–KO mice (Fig. 6B). Taken together,these data show that STEP61 expression is increased in PSD-95knockdown experiments and in PSD-95–KO mouse brain; thereis less ubiquitination of STEP in the PSD-95–KO mouse brain,resulting in elevated STEP61 expression, which decreases theY1472 phosphorylation level of GluN2B.

DiscussionSTEP (encoded by the Ptpn5 gene) is a brain-specific proteintyrosine phosphatase, which has several splice variants (23, 24).STEP61 and STEP46 are expressed in the striatum, amygdala, andthe optic nerve, but only STEP61 is expressed in the hippocam-pus, neocortex, and spinal cord (25, 38, 39). STEP61 has twotransmembrane regions and is targeted to the membrane in-cluding the PSD, whereas STEP46 is restricted to the cytosol.STEP61 functions to regulate NMDAR surface expression di-rectly by dephosphorylating Y1472 in the GluN2B endocytic

Fig. 4. STEP61 expression is increased by PSD-95 knockdown and regulates the surface expression of NMDARs. (A) Primary cortical neurons were transducedwith shPSD-95 lentivirus at DIV 14, and 7 d later PSD fractions were isolated and immunoblotted with PSD-95 (P = 0.7e-8), STEP61 (P = 0.0275), GluN2B (P =0.0022), PSD-93 (P = 0.4473), Fyn (P = 0.4697), and pY1472 GluN2B (P = 0.0169) antibodies. Error bars represent ± SEM; *P < 0.05, **P < 0.005, ***P < 0.001;n = 3. (B) To test the efficiency and investigate the effects of STEP knockdown on NMDARs, primary cortical neurons were transduced with shSTEP lentivirus atDIV 14, and total proteins were isolated with 1% SDS lysis buffer at DIV 21 (STEP61, P = 0.3e-5; GluN2B, P = 0.0011; PSD-95, P = 0.0980; Fyn, P = 0.2404; andpY1472 GluN2B, P = 0.0155). Error bars represent ± SEM; *P < 0.05, **P < 0.005, ***P < 0.001; n = 3. (C) Primary hippocampal neurons were transduced withshRNAs as indicated at DIV 7, were transfected with Flag-GluN2B at DIV 14, and were immunostained 3 d later. Surface receptors were labeled with anti-Flagand Alexa 555-conjugated secondary antibody (red). After fixation and permeabilization, intracellular receptors were visualized with anti-Flag and Alexa647-conjugated secondary antibody (pseudocolored white) (shPSD-95, P = 0.0001; shSTEP, P = 0.0009; and shPSD-95 and shSTEP, P = 0.2793). Error barsrepresent ± SEM, ***P < 0.001; n = 3. (Scale bar, 5 μm.)

Won et al. PNAS Early Edition | 5 of 9

NEU

ROSC

IENCE

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 1

5, 2

021

motif (YEKL) or indirectly by dephosphorylating Y420 in thecatalytic domain of Fyn kinase, thereby inactivating Fyn kinase.Previous studies have demonstrated that dysregulation of NMDARtrafficking is associated with Alzheimer’s disease (19, 40), andNMDAR hypofunction is implicated in schizophrenia in humansand animal models (16, 19, 41). These findings suggest that theregulation of STEP is a critical factor underlying neuropsychiatricdisorders such as Alzheimer’s disease, schizophrenia, Huntington’sdisease (42), and stroke/ischemia. In contrast to STEP61, PSD-95binds to the PDZ ligand in the C-terminal tail of GluN2B andstabilizes the surface expression of NMDARs (43–45). More-over, PSD-95 is important for the regulation of NMDAR activityby Fyn (36). Although both PSD-95 and STEP61 bind to GluN2Band regulate NMDAR surface expression, until now there havebeen no studies on the functional interaction between PSD-95and STEP61.

In the current study, we reveal a profound interaction betweenPSD-95 and STEP61 in heterologous cells, cultured corticalneurons, and mouse brain synaptosomes. Even though PSD-95has high homology with PSD-93, we detected no STEP61 bindingwith PSD-93 or any other MAGUK. This MAGUK-specific in-teraction is unusual, although there are some reports of speci-ficity (46, 47). Next, because Fyn, PSD-95, and NMDARs make acomplex, we confirmed that the interaction between STEP61 andPSD-95 is independent of Fyn using Fyn-KO mouse brain.Furthermore, we delineated the binding regions, showing thatthe PDZ3 domain of PSD-95 is essential for binding to STEP61and that the N-terminal 65 amino acids of STEP61 containing thefirst proline-rich domain (PRD) are critical for direct binding toPSD-95. Interestingly, these results show that STEP61 can makea complex with PSD-95 and GluN2B-containing NMDARs in-dependent of Fyn.

Fig. 5. STEP knockdown increases extrasynaptic NMDARs and NMDAR-mediated currents but not NMDAR EPSCs. (A and B) Primary cultured cortical neuronswere transduced with shSTEP lentivirus at DIV 7–10. After 7 d neurons were solubilized with 1% Triton X-100, and the insoluble fraction was solubilized with1% SDS. Lysates were immunoblotted with the indicated antibodies: STEP61 (P = 0.8e-7), PSD-95 (P = 0.4334), GluN2B (P = 0.0038), GluN1 (P = 0.0165), GluN2A(P = 0.1275), and Fyn (P = 0.3729) in Triton X-100–soluble fractions and STEP61 (P = 0.1e-8), PSD-95 (P = 0.4710), GluN2B (P = 0.3768), GluN1 (P = 0.0527),GluN2A (P = 0.3429), and Fyn (P = 0.1210) in Triton X-100–insoluble fractions. Error bars represent ± SEM; *P < 0.05, **P < 0.005, ***P < 0.001; n = 3.(C) Timeline for the electrophysiology experiments in rat organotypic hippocampal slice cultures. (D) Diagram for the simultaneous whole-cell recordings fromneighboring WT and transfected pyramidal neurons. (E) Scatterplot showing whole-cell currents in response to fast application of NMDA. Data representpairs of simultaneously recorded neurons in slice cultures from shSTEP-transfected and neighboring control cells. (F) Paired average of single pairs fromcontrol and transfected cells. Mean ± SEM for control and shSTEP are 760.3 ± 89.15 pA, n = 12, and 1,141 ± 134.9, n = 12 pA, respectively. **P = 0.0098;Wilcoxon signed-rank test. (Scale bars, 100 pA and 5 s.) (G) Scatterplot shows amplitudes of NMDA EPSCs for single pairs (open circles) and mean ± SEM (filledcircle) for shSTEP-transfected vs. control cells. (Scale bars, 50 pA and 50 ms.) (H) Paired average of single pairs. Mean ± SEM for control and shSTEP are 59 ± 6.2(n = 14) and 70 ± 6.8 (n = 14), respectively. P = 0.54; Wilcoxon signed-rank test.

6 of 9 | www.pnas.org/cgi/doi/10.1073/pnas.1609702113 Won et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

5, 2

021

Most importantly, we show that PSD-95 binding to STEP61leads to its ubiquitination and degradation, which is dependenton the UPS (Fig. 3). This regulation of synaptic protein levels is apreviously unappreciated role for PSD-95. Previous models ofPSD-95 have focused on its well-documented role as a scaffoldingprotein. We also determined that the effects of PSD-95 on STEP61were dependent on membrane localization. PSD-95 has two criticalcysteine residues, Cys3 and Cys5, which are sites for palmitoylation,and this palmitoylation of PSD-95 is essential for clustering at thePSD (48, 49). By targeting PSD-95 to the plasma membrane, pal-mitoylation may facilitate the interaction of the protein with othermembrane-associated proteins such as ion channels and receptors(49–51). We demonstrated that PSD-95 C3,5S (a palmitoylation-deficient mutant) does not regulate STEP61 expression levels. In-triguingly, STEP61 expression is reduced more dramatically in theTriton X-100–insoluble fraction than in the –soluble fraction whencoexpressed with PSD-95. These results are consistent with PSD-95being mainly Triton X-100–insoluble, showing that PSD-95 activelycontrols protein content at the PSD.We also demonstrated that PSD-95 knockdown increases STEP61

expression levels. Using lentivirus containing shRNA against PSD-95,we performed PSD-95 knockdown experiments in cultured corticalneurons and in the purified PSD fraction and found increasedSTEP61 in the PSD fraction. Interestingly, the expression level ofGluN2B and the phosphorylation of GluN2B Y1472 were decreased.A reduction of GluN2B pY1472 may lead to diminished GluN2Blevels at the PSD, because previous studies (19, 27, 30) have shownthat phosphorylation of Y1472 in GluN2B increases the surfaceexpression of GluN2B-containing NMDARs at the synapse.Using immunofluorescence microscopy to investigate if PSD-

95 and STEP61 regulate the surface expression of GluN2B-con-taining NMDARs, we observed a decrease in the surface ex-pression of GluN2B upon knockdown of PSD-95, whereas STEPknockdown increased the surface expression of NMDARs. In-terestingly, simultaneous knockdown of both PSD-95 and STEPdid not significantly change the surface expression. This findingindicates that PSD-95 and STEP61 act as counterregulators forthe surface expression of GluN2B-containing NMDARs. Fur-thermore, using hippocampal slice cultures, we found that STEPknockdown increases NMDAR currents in hippocampal slices,indicating that PSD-95 and STEP61 are critical for the surfaceexpression of GluN2B-containing NMDARs. The observationthat synaptic currents are not affected by STEP knockdown islikely explained by synaptic PSD-95 already having strongly re-duced synaptic STEP. Thus, the STEP knockdown preferentiallyaffected extrasynaptic NMDAR currents. These data indicatethat PSD-95 actively excludes STEP from the PSD of neurons.

Consistent with our findings from PSD-95–knockdown ex-periments in cultured cortical neurons, STEP61 is increased invivo in PSD-95–KOmouse brain. An increase of STEP61 in PSD-95–KO mouse brain results from reduced STEP ubiquitination.However, we do not know which E3 ligase is involved, and this isan important topic for future investigation. Interestingly, we donot detect a bidirectional regulatory effect between STEP61 andPSD-95 because PSD-95 expression is not significantly changedin STEP-KO mouse brain (Fig. S5). Thus, PSD-95 is a dominantregulator of STEP61, but not vice versa. Taken together, our resultsshow that PSD-95 plays a dominant role in decreasing STEP61expression and minimizing its enrichment at the PSD. Importantlywe elucidate a dynamic role for PSD-95 as an active organizer ofprotein content at synapses, driving a reduction in STEP in addi-tion to stabilizing NMDARs directly via its PDZ interaction.

Materials and MethodscDNA Constructs. Human STEP61 cDNA (ptpn5; NM032781.3) was purchasedfrom Origene Technologies. STEP was subcloned with an N-terminal HA tag.HA-STEP46 starts with methionine197, and a HA-tagged truncation mutant ofSTEP61 (PRD1-C, ΔPRD1, or PRD2-C) starts with Pro57, Ser66, or Pro169, re-spectively. Rat PSD-95 cDNA (Dlg4; NM019621) was tagged with C-terminalGFP, and a truncation mutant of PSD-95 (ΔG, ΔS, ΔP3, or ΔP2) was constructedto delete 221, 325, 483, or 567 amino acids from the C terminus, respectively.Rat SAP97 and rat SAP102 cDNAs were tagged with N-terminal GFP. Allplasmids were confirmed by sequencing.

Reagents and Antibodies. The following reagents and antibodies were used:DMEMhigh glucose (Thermo Fisher Scientific, catalog no. 10313-021); FBS (GEHealthcare Life Sciences, catalog no. SH30071.03); Opti-MEM (Thermo FisherScientific, catalog no. 31985-062); Lipofectamine 2000 (Invitrogen, catalogno. 11668-019); L-glutamine (Sigma, catalog no. G7513); penicillin-streptomycin(Thermo Fisher Scientific, catalog no. 15140-122); trypsin (Thermo FisherScientific, catalog no. 25300-054); Neurobasal medium (Thermo Fisher Sci-entific, catalog no. 21103-049); B-27 supplement (Thermo Fisher Scientific,catalog no.17504-044); HBSS (Thermo Fisher Scientific, catalog no. 14170-112); DNase I (Worthington, catalog no. 3170); poly-D-lysine (Sigma, catalogno. P7886); MG-132 (Sigma, catalog no. C2211); chloroquine (Sigma, catalogno. C6628); 2-bromohexadecanoic acid (Sigma, catalog no. 21604); rabbitanti-GFP (Invitrogen, catalog no. A11122); mouse anti-GFP (NeuroMab, catalogclone N86/8); rabbit anti-HA (Cell Signaling, catalog no. 3724); mouse anti-HA(Cell Signaling, catalog no. 2367); mouse anti-Flag (Sigma, catalog no. F1804);rabbit anti-Flag (Sigma, catalog no. 7425); mouse anti-STEP (Novagen, catalogno. NB300-202); rabbit anti-GluN2A (Sigma, catalog no. M264); mouse anti-GluN2B (NeuroMab, catalog clone N59/36); rabbit anti-GluN2B pY1472 (Milli-pore, catalog no. AB5403); mouse anti-GluN1 (54.2), a gift of the late Robert J.Wenthold, National Institute of Deafness and Other Communication Disorders,Bethesda, MD; rabbit anti–PSD-93 (Cell Signaling, catalog no. 9445); rabbitanti–PSD-95 (Cell Signaling, catalog no. 2507); mouse anti–PSD-95 (NeuroMab,catalog clone K28/43); rabbit anti-SAP102 (Cell Signaling, catalog no. 3733);mouse anti-SAP102 (NeuroMab, catalog clone N19/2); rabbit anti-Fyn (Cell

Fig. 6. In PSD-95–KO mouse brain, the STEP61 expression level is increased at synapses. (A) WT or PSD-95–KO adult mouse forebrain was subjected tofractionation, and the PSD fraction was isolated and immunoblotted with STEP61 (P = 0.0234), GluN2B (P = 0.3894), PSD-93 (P = 0.3545), Fyn (P = 0.3029), andpY1472 GluN2B (P = 0.0035) antibodies. Error bars represent ± SEM; *P < 0.05, **P < 0.005; n = 3. (B) The SPM fraction of WT or PSD-95–KO mouse forebrainwas isolated and solubilized with 1% SDS lysis buffer. The lysate was neutralized with cold 1% Triton X-100 lysis buffer, immunoprecipitated with STEPantibody, and immunoblotted for ubiquitin (P4D1). The arrowhead indicates IgG.

Won et al. PNAS Early Edition | 7 of 9

NEU

ROSC

IENCE

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 1

5, 2

021

Signaling, catalog no. 4023); rabbit anti-Src (Cell Signaling, catalog no. 2110);mouse anti-CKII (Sigma, catalog no. 3617); mouse anti–beta-actin (ABM, cat-alog no. G043); and mouse anti-ubiquitin P4D1 (Santa Cruz Biotechnology,catalog no.sc-8017).

Rat Primary Neuronal Cultures. Primary rat hippocampal and cortical neuronswere prepared from embryonic day 18 Sprague–Dawley rats following theguidelines of the National Institutes of Health Guide for the Care and Use ofLaboratory Animals (52). Hippocampi and cortices were isolated, and thetissue was dissociated for 30 min at 37 °C by 0.05% trypsin in 10 mM HBSScontaining 1.37 mg/mL DNase I. Cells were triturated using fire-polishedglass Pasteur pipettes. Neurons were plated on poly-D-lysine–coated dishesand coverslips and were maintained in serum-free Neurobasal Mediumsupplemented with 2% (vol/vol) B-27 and 2 mM L-Glutamine. The neuronswere maintained in an incubator at 37 °C and 5% CO2. One-half of themedium was replaced with new medium every 7 d.

HEK293T Cell and Primary Neuron Transfection.HEK293T cells were maintainedwith 5% (vol/vol) FBS and 1%penicillin-streptomycin in DMEM containing highglucose. To express various cDNAs of STEP and PSD-95, each plasmid DNA wasdiluted in Opti-MEM I, to which Lipofectamine 2000 diluted in Opti-MEM I wasadded and mixed thoroughly. After 20-min incubation at room temperature,HEK293T cells were added to the mixture and were incubated for 24–48 hafter transfection. For biochemistry, cortical neurons were plated onto 100-mmdishes coated with poly-D-lysine at a density of 8 × 106 cells per dish. The cellswere transfected at 5–6 DIV using a modified calcium phosphate method. After2 d, neurons were collected and lysed for immunoblotting.

Lentiviral Particle Packaging. To knock down PSD-95 or STEP in culturedneurons, lentiviral particles were produced in 100-mm dishes of HEK293Tcells. Cells were cotransfected with the lentiviral vector FHUGW containingthe shRNA sequence (5′-tcacgatcatcgctcagta-3′) against PSD-95 or a scrambledsequence (5′-gacaccgtacatcatagat-3′) or the shRNA sequence (5′-gcgtggtaga-catcctaaag-3′) against STEP, the packaging vector Δ8.9, and the vesicular sto-matitis virus G envelope glycoprotein vector by using FuGENE HD (Promega,catalog no. E2311) in UltraCULTURE medium (Lonza, catalog no. 12-725F)containing 2 mM L-Glutamine, 1 mM sodium pyruvate, and 0.075% sodiumbicarbonate. Forty to forty-eight hours after transfection, culture mediacontaining lentiviral particles were collected and centrifuged at 100,000 × gfor 2 h at 4 °C. The lentiviral particle pellet was resuspended with 100 μL PBS,aliquoted, and frozen at −80 °C.

Cell-Surface Biotinylation Assay and Immunoprecipitation. STEP61 or STEP46was transfected into HEK293T cells. On the next day, transfected HEK cells orcultured cortical neurons at DIV21 were washed with PBS (including 2 mMCaCl2 and 1 mM MgCl2) and then were incubated with 0.5 mg/mL biotin (EZ-Link Sulfo-NHS-LC-Biotin, Thermo Fisher Scientific, catalog no. 21335) in PBS(CaCl2 and MgCl2) for 30 min at 4 °C. Then cells were washed with PBS (in-cluding 10 mM glycine) to quench free biotin and were lysed with RIPA buffer.Biotinylated proteins were bound with NeutrAvidin agarose resin (ThermoFisher Scientific, catalog no. 29202) for 30 min at 4 °C and washed with PBS.Proteins were eluted from agarose resin using SDS sample buffer, andimmunoblotted with the indicated antibodies. In HEK293T cells, coimmuno-precipitation was performed 24–48 h after transfection with the indicatedcDNA plasmids in 1% Triton lysis buffer [50 mM Tris·HCl (pH 7.4), 150 mMNaCl,1 mM EDTA]. For coimmunoprecipitation of STEP and synaptic proteins, thecrude membrane fraction (P2) from WT or Fyn-KO mouse cortex and culturedcortical neurons were solubilized in lysis buffer [50 mM Tris·HCl (pH 8.8), 1%sodium deoxycholate] at 37 °C for 30 min and centrifuged. The supernatantwas neutralized with a fourfold volume of 1% Triton X-100 lysis [50 mMTris·HCl (pH 8.0), 150 mM NaCl, 2 mM EDTA] at 4 °C for 30 min on ice and wasincubated with 10 μL STEP antibody. On the next day, extracts were incubatedwith protein G-agarose beads for 4 h at 4 °C. Beads were washed three timeswith 1% Triton X-100 lysis buffer, and immunoprecipitated proteins wereeluted with SDS sample buffer for subsequent immunoblotting.

Subcellular Fractionation of Rodent Brain Tissues and Cultured Neurons. Thefractionation experiment was performed following standard methods (53).Brain tissues dissected out from adult female rats, Fyn-KO mice, STEP-KOmice (purchased from The Jackson Laboratory), or PSD-95 KO mice or cul-tured neurons (DIV 20–22), were homogenized with ice-cold TEVP buffer[320 mM sucrose, 10 mM Tris·HCl (pH 7.5), 5 mM EDTA, 1× protease inhibitormixture (Roche, 11697498001), and phosphatase inhibitor mixture II (Sigma,catalog no. P5726) and III (Sigma, catalog no. P0044)]. Homogenates (H)were centrifuged at 1,000 × g for 10 min at 4 °C. The supernatant (S1) was

centrifuged at 10,000 × g for 20 min at 4 °C to obtain the crude synapto-somal membrane (P2 fraction). S2, the supernatant obtained from S1, wasrecentrifuged at 165,000 × g for 2 h at 4 °C to isolate the P3 pellet. The P2pellet was resuspended with ice-cold TEVP buffer [35.6 mM sucrose, 10 mMTris·HCl (pH 7.5), 5 mM EDTA, 1× protease inhibitor mixture, and phospha-tase inhibitor mixture II and III] and was incubated on ice for 30 min andcentrifuged at 25,000 × g for 20 min at 4 °C to obtain synaptic plasmamembrane (SPM or LP1). The supernatant (LS1) was centrifuged at 165,000 × g,for 2 h at 4 °C to isolate the LP2 pellet. The SPM pellet was added to 1% TritonX-100 lysis buffer [10 mM Tris·HCl (pH 7.5), 5 mM EDTA, 1× protease inhibitormixture, and phosphatase inhibitor mixture II and III] and was incubated withgentle agitation at 4 °C for 30 min. Lysates were centrifuged at 33,000 × g for30 min at 4 °C to obtain a soluble fraction (the Triton X-100–soluble fraction)and a pellet (the Triton X-100–insoluble fraction/PSD fraction), which was sol-ubilized by brief sonication and heating at 37 °C for 30 min in 1% SDS [10 mMTris·HCl (pH 7.5), 5 mM EDTA, 1× protease inhibitor mixture, and phosphataseinhibitor mixture II and III]. To obtain the PSD fraction, the lysates werecentrifuged at 100,000 × g for 30 min at 4 °C.

Immunocytochemistry and Image Acquisition. At DIV 7 hippocampal neurons in12-well plates were transduced with 1 μL of lentiviruses containing shPSD-95,shSTEP, or both. After 7 d, neurons were transfected with Flag-GluN2B usingLipofectamine 2000. Three days posttransfection, neurons were used forimmunostaining. To measure the surface expression of Flag-GluN2B, neuronswere labeled by incubating the live neurons at 37 °C for 15 min in culturemedium containing mouse anti-Flag antibody. After the neurons were washedwith cold PBS, the cells were fixed with 4% (vol/vol) paraformaldehyde/4%(vol/vol) sucrose in PBS for 15 min. Before permeabilization, cells were blockedin 10% (vol/vol) normal goat serum (NGS). Surface Flag-GluN2B was labeledusing Alexa Fluor 555-conjugated secondary antibody. Then cells were per-meabilized with 0.25% Triton X-100 in PBS for 10 min and blocked in 10%(vol/vol) NGS. The neurons were labeled with rabbit anti-Flag antibody andwere incubated with Alexa Fluor 647-conjugated secondary antibody. Immu-nocytochemistry images were acquired on a Zeiss LSM 510 confocal microscope.Serial optical sections collected at 0.35-μm intervals were used to create maxi-mum projection images. MetaMorph (Universal Imaging Corporation) was usedto measure the integrated intensity of three dendrite sections from each se-lected neuron and the three sections for one cell were averaged; data arereported as the average from three independent experiments. For every sec-tion measured, the ratio of surface GluN2B to intracellular GluN2B fluorescencewas calculated and then was normalized.

Electrophysiology in Slice Cultures. Dual whole-cell recordings in area CA1were done by simultaneously recording responses from fluorescent transfectedneurons and neighboring untransfected control neurons. Pyramidal neuronswere identified by morphology and location. Series resistance was monitoredconstantly, and recordings in which series increased by >30 MOhm or varied by>50% between neurons were discarded. Whole-cell NMDA recordings used anextracellular solution bubbled with 95% O2/5% CO2 consisting of 119 mM NaCl,2.5 mM KCl, 4 mM CaCl2, 4 mM MgSO4, 1 mM NaH2PO4, 26.2 mM NaHCO3,11 mM glucose. Picrotoxin (100 μM) was added to block inhibitory currents, and1 μM TTX was added to block evoked potentials. Whole-cell NMDA responseswere evoked at +40 mV by 200 μM NMDA/200 μM glycine delivered to theneuronal soma by a large-diameter (20- to 30-μM tip diameter) pipette. Perfu-sion was 1 s in duration and was controlled by a Picospritzer II (General ValveCorporation). For NMDA-evoked EPSCs, 2 μM 2-Chloroadenosine was used tocontrol epileptiform activity. The intracellular solution contained 135 mMCsMeSO4, 8 mM NaCl, 10 mM Hepes, 0.3 mM EGTA, 5 mM QX314-Cl, 4 mMMgATP, 0.3 mM Na3GTP, and 0.1 mM spermine. A bipolar stimulation electrode(FHC) was placed in the striatum radiatum, and responses were evoked at 0.2 Hz.NMDAR currents were measured at +40 mV and were temporally isolated bymeasuring amplitudes 100 ms after the stimulus.

Statistical Analysis. The statistical significance between conditions was cal-culated using Student’s t test (n = number of independent experiments) andwas considered significant at P < 0.05 (*P < 0.05, **P < 0.005, ***P < 0.001).All experiments were repeated at least three times independently. The sig-nificance of evoked dual whole-cell recordings compared with controls wasdetermined using the Wilcoxon signed-rank sum test.

ACKNOWLEDGMENTS. We thank John D. Badger II for technical assistanceand the National Institute of Neurological Disorders and Stroke (NINDS)imaging facility for their assistance. This research was supported by theNINDS Intramural Research Program (S.W. and K.W.R.) and by NationalInstitute of Mental Health Grant MH-38256 (S.I. and R.A.N.).

8 of 9 | www.pnas.org/cgi/doi/10.1073/pnas.1609702113 Won et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

5, 2

021

1. Bliss TV, Collingridge GL (1993) A synaptic model of memory: Long-term potentiationin the hippocampus. Nature 361(6407):31–39.

2. Choi DW (1994) Glutamate receptors and the induction of excitotoxic neuronal death.Prog Brain Res 100:47–51.

3. Collingridge GL, Isaac JT, Wang YT (2004) Receptor trafficking and synaptic plasticity.Nat Rev Neurosci 5(12):952–962.

4. Hollmann M, Heinemann S (1994) Cloned glutamate receptors. Annu Rev Neurosci 17:31–108.

5. Carroll RC, Zukin RS (2002) NMDA-receptor trafficking and targeting: Implications forsynaptic transmission and plasticity. Trends Neurosci 25(11):571–577.

6. Cull-Candy SG, Leszkiewicz DN (2004) Role of distinct NMDA receptor subtypes atcentral synapses. Sci STKE 2004(255):re16.

7. Dingledine R, Borges K, Bowie D, Traynelis SF (1999) The glutamate receptor ionchannels. Pharmacol Rev 51(1):7–61.

8. Furukawa H, Singh SK, Mancusso R, Gouaux E (2005) Subunit arrangement andfunction in NMDA receptors. Nature 438(7065):185–192.

9. Al-Hallaq RA, Conrads TP, Veenstra TD, Wenthold RJ (2007) NMDA di-heteromericreceptor populations and associated proteins in rat hippocampus. J Neurosci 27(31):8334–8343.

10. Barria A, Malinow R (2002) Subunit-specific NMDA receptor trafficking to synapses.Neuron 35(2):345–353.

11. Bellone C, Nicoll RA (2007) Rapid bidirectional switching of synaptic NMDA receptors.Neuron 55(5):779–785.

12. Sanz-Clemente A, Nicoll RA, Roche KW (2013) Diversity in NMDA receptor composi-tion: Many regulators, many consequences. Neuroscientist 19(1):62–75.

13. Wang YT, Salter MW (1994) Regulation of NMDA receptors by tyrosine kinases andphosphatases. Nature 369(6477):233–235.

14. Moon IS, Apperson ML, Kennedy MB (1994) The major tyrosine-phosphorylatedprotein in the postsynaptic density fraction is N-methyl-D-aspartate receptor subunit2B. Proc Natl Acad Sci USA 91(9):3954–3958.

15. Rostas JA, et al. (1996) Enhanced tyrosine phosphorylation of the 2B subunit of theN-methyl-D-aspartate receptor in long-term potentiation. Proc Natl Acad Sci USA93(19):10452–10456.

16. Lau CG, Zukin RS (2007) NMDA receptor trafficking in synaptic plasticity and neuro-psychiatric disorders. Nat Rev Neurosci 8(6):413–426.

17. Mohn AR, Gainetdinov RR, Caron MG, Koller BH (1999) Mice with reduced NMDAreceptor expression display behaviors related to schizophrenia. Cell 98(4):427–436.

18. Parsons MP, Raymond LA (2014) Extrasynaptic NMDA receptor involvement in centralnervous system disorders. Neuron 82(2):279–293.

19. Snyder EM, et al. (2005) Regulation of NMDA receptor trafficking by amyloid-beta.Nat Neurosci 8(8):1051–1058.

20. Lombroso PJ, Murdoch G, Lerner M (1991) Molecular characterization of a protein-tyrosine-phosphatase enriched in striatum. Proc Natl Acad Sci USA 88(16):7242–7246.

21. Lombroso PJ, Naegele JR, Sharma E, Lerner M (1993) A protein tyrosine phosphataseexpressed within dopaminoceptive neurons of the basal ganglia and related struc-tures. J Neurosci 13(7):3064–3074.

22. Bult A, et al. (1996) STEP61: A member of a family of brain-enriched PTPs is localizedto the endoplasmic reticulum. J Neurosci 16(24):7821–7831.

23. Bult A, et al. (1997) STEP: A family of brain-enriched PTPs. Alternative splicing pro-duces transmembrane, cytosolic and truncated isoforms. Eur J Cell Biol 72(4):337–344.

24. Sharma E, Zhao F, Bult A, Lombroso PJ (1995) Identification of two alternativelyspliced transcripts of STEP: A subfamily of brain-enriched protein tyrosine phospha-tases. Brain Res Mol Brain Res 32(1):87–93.

25. Boulanger LM, et al. (1995) Cellular and molecular characterization of a brain-enrichedprotein tyrosine phosphatase. J Neurosci 15(2):1532–1544.

26. Oyama T, et al. (1995) Immunocytochemical localization of the striatal enrichedprotein tyrosine phosphatase in the rat striatum: A light and electron microscopicstudy with a complementary DNA-generated polyclonal antibody. Neuroscience69(3):869–880.

27. Braithwaite SP, et al. (2006) Regulation of NMDA receptor trafficking and function bystriatal-enriched tyrosine phosphatase (STEP). Eur J Neurosci 23(11):2847–2856.

28. Pelkey KA, et al. (2002) Tyrosine phosphatase STEP is a tonic brake on induction oflong-term potentiation. Neuron 34(1):127–138.

29. Zhang Y, et al. (2010) Genetic reduction of striatal-enriched tyrosine phosphatase(STEP) reverses cognitive and cellular deficits in an Alzheimer’s disease mouse model.Proc Natl Acad Sci USA 107(44):19014–19019.

30. Lavezzari G, McCallum J, Lee R, Roche KW (2003) Differential binding of the AP-2adaptor complex and PSD-95 to the C-terminus of the NMDA receptor subunit NR2Bregulates surface expression. Neuropharmacology 45(6):729–737.

31. Lavezzari G, McCallum J, Dewey CM, Roche KW (2004) Subunit-specific regulation ofNMDA receptor endocytosis. J Neurosci 24(28):6383–6391.

32. Roche KW, et al. (2001) Molecular determinants of NMDA receptor internalization.Nat Neurosci 4(8):794–802.

33. Kurup P, et al. (2010) Abeta-mediated NMDA receptor endocytosis in Alzheimer’sdisease involves ubiquitination of the tyrosine phosphatase STEP61. J Neurosci 30(17):5948–5957.

34. Venkitaramani DV, Moura PJ, Picciotto MR, Lombroso PJ (2011) Striatal-enrichedprotein tyrosine phosphatase (STEP) knockout mice have enhanced hippocampalmemory. Eur J Neurosci 33(12):2288–2298.

35. Nguyen TH, Liu J, Lombroso PJ (2002) Striatal enriched phosphatase 61 dephos-phorylates Fyn at phosphotyrosine 420. J Biol Chem 277(27):24274–24279.

36. Tezuka T, Umemori H, Akiyama T, Nakanishi S, Yamamoto T (1999) PSD-95 promotesFyn-mediated tyrosine phosphorylation of the N-methyl-D-aspartate receptor subunitNR2A. Proc Natl Acad Sci USA 96(2):435–440.

37. Elias GM, et al. (2006) Synapse-specific and developmentally regulated targeting ofAMPA receptors by a family of MAGUK scaffolding proteins. Neuron 52(2):307–320.

38. Lorber B, et al. (2004) Stimulated regeneration of the crushed adult rat optic nervecorrelates with attenuated expression of the protein tyrosine phosphatases RPTPal-pha, STEP, and LAR. Mol Cell Neurosci 27(4):404–416.

39. Raghunathan A, Matthews GA, Lombroso PJ, Naegele JR (1996) Transient compart-mental expression of a family of protein tyrosine phosphatases in the developingstriatum. Brain Res Dev Brain Res 91(2):190–199.

40. Mattson MP (2004) Pathways towards and away from Alzheimer’s disease. Nature430(7000):631–639.

41. Sawa A, Snyder SH (2002) Schizophrenia: Diverse approaches to a complex disease.Science 296(5568):692–695.

42. Gladding CM, et al. (2012) Calpain and STriatal-Enriched protein tyrosine phos-phatase (STEP) activation contribute to extrasynaptic NMDA receptor localizationin a Huntington’s disease mouse model. Hum Mol Genet 21(17):3739–3752.

43. Chung HJ, Huang YH, Lau LF, Huganir RL (2004) Regulation of the NMDA receptorcomplex and trafficking by activity-dependent phosphorylation of the NR2B subunitPDZ ligand. J Neurosci 24(45):10248–10259.

44. Kornau HC, Schenker LT, Kennedy MB, Seeburg PH (1995) Domain interaction be-tween NMDA receptor subunits and the postsynaptic density protein PSD-95. Science269(5231):1737–1740.

45. Niethammer M, Kim E, Sheng M (1996) Interaction between the C terminus of NMDAreceptor subunits and multiple members of the PSD-95 family of membrane-associ-ated guanylate kinases. J Neurosci 16(7):2157–2163.

46. Kalia LV, Salter MW (2003) Interactions between Src family protein tyrosine kinasesand PSD-95. Neuropharmacology 45(6):720–728.

47. Zhu YC, et al. (2013) Palmitoylation-dependent CDKL5-PSD-95 interaction regulatessynaptic targeting of CDKL5 and dendritic spine development. Proc Natl Acad Sci USA110(22):9118–9123.

48. Craven SE, El-Husseini AE, Bredt DS (1999) Synaptic targeting of the postsynapticdensity protein PSD-95 mediated by lipid and protein motifs. Neuron 22(3):497–509.

49. Topinka JR, Bredt DS (1998) N-terminal palmitoylation of PSD-95 regulates associationwith cell membranes and interaction with K+ channel Kv1.4. Neuron 20(1):125–134.

50. Christopherson KS, et al. (2003) Lipid- and protein-mediated multimerization of PSD-95: Implications for receptor clustering and assembly of synaptic protein networks.J Cell Sci 116(Pt 15):3213–3219.

51. El-Husseini AE, et al. (2000) Dual palmitoylation of PSD-95 mediates its ves-iculotubular sorting, postsynaptic targeting, and ion channel clustering. J Cell Biol148(1):159–172.

52. Committee on Care and Use of Laboratory Animals (1996) Guide for the Care and Useof Laboratory Animals (Natl Inst Health, Bethesda), DHHS Publ No (NIH) 85-23.

53. Hallett PJ, Collins TL, Standaert DG, Dunah AW (2008) Biochemical fractionation ofbrain tissue for studies of receptor distribution and trafficking. Curr Protoc NeurosciChapter 1:Unit 1-16.

Won et al. PNAS Early Edition | 9 of 9

NEU

ROSC

IENCE

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 1

5, 2

021