Synaptic Targeting of AMPA Receptors Is Regulated by a Camkii Site in the ﬁrst Intracellular Loop of Glua1

Synaptic targeting of AMPA receptors is regulated by a CaMKII site in the ﬁrst intracellular loop of GluA1

Wei Lua,1, Kaname Isozakib,1, Katherine W. Rocheb,2, and Roger A. Nicolla,2

aDepartments of Cellular and Molecular Pharmacology and Physiology, University of California, San Francisco, CA 94143; and bNational Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, MD 20892

Contributed by Roger A. Nicoll, November 5, 2010 (sent for review October 10, 2010) The accumulation of AMPA receptors (AMPARs) at synapses is Results essential for excitatory synaptic transmission. However, the mech- Expression of the GluA1 Loop1 Impairs Synaptic Transmission. In anisms underlying synaptic targeting of AMPARs remain elusive. agreement with previous studies (12), expression of the GluA1 We have now used a molecular replacement approach on an C-tail in hippocampal slice cultures did not impair AMPAR- AMPAR-null background to investigate the targeting mechanisms mediated synaptic transmission (Fig. 1 C and H). We thus ex- necessary for regulating AMPAR trafficking in the hippocampus. plored the possibility that Loop1 (Fig. 1 A and B), an ∼30-aa Although there is an extensive literature on the role of the GluA1 cytoplasmic domain, plays a role in AMPAR trafficking. We first C-tail in AMPAR trafficking, there is no effect of overexpressing expressed GFP tagged GluA1 Loop1 in pyramidal neurons as the C-tail on basal transmission. Instead, we found that the first a potential dominant negative. Next, 2–4 d after biolistic trans- intracellular loop domain (Loop1) of GluA1, a previously over- fection in hippocampal slice cultures, we made simultaneous looked region within AMPARs, is critical for receptor targeting to dual whole-cell recordings from GFP positive pyramidal neurons synapses, but not for delivery of receptors to the plasma mem- and untransfected neighboring neurons, and compared the brane. We also identified a CaMKII phosphorylation site (S567) in amplitudes of AMPAR and NMDAR excitatory postsynaptic the GluA1 Loop1, which is phosphorylated in vitro and in vivo. currents (EPSCs) evoked by a common input. Expression of the Furthermore, we show that S567 is a key residue that regulates GluA1 Loop1 specifically reduced the amplitude of AMPAR Loop1-mediated AMPAR trafficking. Thus, our study reveals a EPSCs by ∼40%, without affecting either NMDAR EPSC am- unique mechanism for targeting AMPARs to synapses to mediate plitude (Fig. 1 D and H) or paired-pulse ratio (Fig. S1), a mea- synaptic transmission. sure of presynaptic release probability. Loop1 from GluA2, which is 72% identical to that of GluA1 (Fig. 1B), also impaired GluA2 | GluR1 | postsynaptic density | hippocampus | pyramidal neurons synaptic transmission (Fig. 1 E and H). In contrast, expression of Loop1 from the kainate receptor subunit GluK2, which is 36% MPA receptors (AMPARs) are tetramers composed of identical to GluA1 Loop1 (Fig. 1B), did not affect synaptic AGluA1-4 (GluR1-4 or GluRA-D) subunits, and mediate the transmission (Fig. 1 F and H). Furthermore, to control for po- majority of fast excitatory synaptic transmission in the brain (1, 2). tential nonspecific effects, we expressed GluA1 Loop1 in pyra- The dynamic regulation of AMPAR trafficking into and out of midal neurons from GluA1 KO mice but observed no effect on synapses is a major mechanism underlying synaptic plasticity (3–6). AMPAR EPSCs (Fig. 1 G and H). Therefore the effect of GluA1 However, the mechanisms underlying AMPAR trafficking to syn- Loop1 expression on synaptic transmission depends on the apses to mediate basal synaptic transmission remain largely elusive presence of GluA1-containing receptors. Taken together, these (3–6). In the hippocampus, the majority of AMPARs are either data show that GluA1 Loop1 functions in AMPAR trafficking. GluA1A2 or GluA2A3 heteromers (7), and associate with their accessory subunits, transmembrane AMPAR regulatory proteins GluA1 Loop1 Is Required for Synaptic Delivery of GluA1 Homomers. fi (TARPs), which are critical for receptor trafficking (8). Further- AMPAR traf cking involves two general steps: delivery of more, GluA1A2 heteromers are the major species at both syn- receptors to the surface and targeting of the receptors to the aptic and extrasynaptic membranes of hippocampal CA1 pyra- synapse. Which step might depend on GluA1 Loop1? To address fi midal neurons (9). Most studies to date have focused on the role of this question, we compared the traf cking of GluA1 to that of AMPAR C-termini in receptor trafficking (3–5). However, in- GluA1/K2, a mutant of GluA1 in which the Loop1 region has been A terference with GluA1 C-tail function has little effect on basal syn- swapped with the loop domain from GluK2 (Fig. 2 ). In – HEK293T cells, agonist-evoked currents from outside-out patches aptic transmission (10 12). For example, basal synaptic transmission fi is not impaired upon overexpression of the GluA1 C-tail (12), nor in showed that GluA1/K2, similar to GluA1, traf cs to the surface, forms homomeric receptors, as judged by the strong inward rec- knockin mice in which GluA1 lacks its C-tail PDZ ligand (10) or key fi A B phosphorylation sites (11). These data suggest that regions other ti cation (Fig. S2 and ), and binds TARPs (Fig. S3). Thus, than the GluA1 C-tail may function in GluA1-containing AMPAR swapping the Loop1 from GluK2 did not change basic biophysical and trafficking properties of GluA1 in heterologous cells. synaptic delivery to maintain synaptic transmission. fi To study AMPAR trafficking, we have generated a triple floxed To study GluA1/K2 traf cking in neurons, we used a molecu- fl/fl lar replacement approach. In hippocampal slice cultures from Gria1-3 mouse line in which expression of the genes encoding fl/fl GRIA1-3 mice, expression of mCherryCre (hereafter Cre) GluA1, A2 and A3 are conditionally regulated (9). Expression of fl/fl Cre recombinase in CA1 pyramidal neurons from Gria1-3 mice leads to the loss of AMPARs (9), creating a null background for Author contributions: W.L., K.W.R., and R.A.N. designed research; W.L. and K.I. the reintroduction of WT or mutant AMPAR subunits that can be performed research; W.L., K.I., K.W.R., and R.A.N. analyzed data; and W.L., K.W.R., and studied independent of native receptors. Using this molecular R.A.N. wrote the paper. replacement approach, we found that delivery of AMPARs to The authors declare no conflict of interest. fi synapses requires the rst cytoplasmic domain (Loop1) of the 1W.L. and K.I. contributed equally to this work. fi AMPAR GluA1 subunit. In addition, we identi ed S567 of GluA1 2To whom correspondence may be addressed. E-mail: [email protected] or nicoll@ Loop1 as a CaMKII phosphorylation substrate, and found that cmp.ucsf.edu. S567 is a critical residue for the regulation of the Loop1- This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. dependent AMPAR trafficking to synapses. 1073/pnas.1016289107/-/DCSupplemental.

22266–22271 | PNAS | December 21, 2010 | vol. 107 | no. 51 www.pnas.org/cgi/doi/10.1073/pnas.1016289107 Downloaded by guest on September 25, 2021 A B 558 585

SRFSPYEWHSEEFEEGRDQTTSDQSNEF GluA1 Loop1 SRFSPYEWHTEEFEDGRETQSSESTNEF GluA2 Loop1 ARFSPYEWYNPHPCNPDSDVV-E--NNF GluK2 Loop1 Loop1 C-tail

C GluA1 C-tail D GluA1 Loop1 A M P A E P S C NM DA E P S C A M P A E P S C NM DA E P S C

- 20 0 15 0 - 80 10 0 A p A ( p ( 1 l i ) p )

t a - 10 0 75 - 40 50 C L o o o L - - 1 1 u l u u l G A G A 0 0 0 0 0 - 10 0 - 20 0 0 75 15 0 0 - 40 - 80 0 50 10 0

E GluA2 Loop1 F GluK2 Loop1 - 18 0 12 0 - 80 10 0 A p ( 1 p 1 p ) o L o o o L

- - 90 60 - 40 50 L o L 2 A 2 K l u l u l u G G - 0 0 0 0 0 - 90 - 18 0 0 60 12 0 0 - 40 - 80 0 50 10 0 Co n t ro l (pA) Co n t ro l (pA) ) ) t 12 0 12 0 t n G GluA1 Loop1 in GluA1 KO H C n C f f

o % o * * % o - 10 0 15 0 60 60 ( A P M A M P A ( ( A ( ) A ) i 1 p o o L o o p 1 i D p M (

0 0 N O - 50 75 t il 1 1 1 O t ai l 1 O K Cn a op n 1 1 - t p op K Lo op o C t p p K

1 n o o Lo o 1 C Lo L o o 1 C 1 1 2 L A A 1 L A A 1 2 l u u u l u u l u A 1 2 2 l u lu A A K G lu A lu lu u lu A K G G G A G l u u G in G l l in G G p1 G G G p1

0 0 o NEUROSCIENCE o o 0 - 50 - 10 0 L Lo 0 75 15 0 1 1 A A Co n t ro l (pA) Co n t ro l (pA) lu lu G G

Fig. 1. Expression of GluA1 Loop1 impairs synaptic transmission. (A) Topology of the AMPAR subunit with Loop1 and C-tail indicated with arrows. (B)Se- quence alignment of Loop1 domains from GluA1, GluA2, and GluK2. (C–G) Scatter plots show amplitudes of EPSCs for single pairs (○) and mean ± SEM (●), respectively for expression of GluA1 C-tail (C, AMPA: n = 11, P = 0.71; NMDA: n =11,P = 0.90), GluA1 Loop1 (D, AMPA: n =28,*P < 0.001; NMDA: n =25,P = 0.77), GluA2 Loop1 (E, AMPA: n = 36, *P < 0.05; NMDA: n = 36, P = 0.20), GluK2 Loop1 (F, AMPA: n = 16, P = 0.62; NMDA: n = 15, P = 0.21) in WT pyramidal cells, and GluA1 Loop1 in GluA1 KO cells (G, AMPA: n =12,P = 0.89; NMDA: n = 12, P = 0.98). (Scale bar: 0.02 s, 20, 50, 20, 25 and 20 pA in C, D, E, F, and G, respectively.) (H) Summary bar graphs show the percentage of amplitudes (mean ± SEM) of AMPAR and NMDAR EPSCs of control cells presented in C–G.

eliminated AMPAR-mediated synaptic transmission (Fig. S4) GluA1A2 heteromers. In HEK293T cells, GluA1/K2, similar to and AMPAR currents from somatic outside-out patches in ∼2–3 GluA1, could assemble with GluA2 to form heteromeric recep- wk (9). We next biolistically transfected neurons with two ex- tors (Fig. S2C). To study GluA1A2 heteromer trafficking in pression constructs: Cre and GluA1-IRES-EGFP or Cre and neurons, we cotransfected pyramidal neurons in hippocampal fl/fl GluA1/K2-IRES-EGFP (identified by both red and green fluo- slice cultures from GRIA1-3 with Cre-IRES-GluA1 or Cre- rescence). Both GluA1 and GluA1/K2 fully rescued surface IRES-GluA1/K2 together with GluA2-IRES-EGFP. Outside-out AMPAR currents as revealed by recordings from somatic out- patch recordings from pyramidal neurons expressing either Cre, side-out patches (Fig. 2B). The glutamate-evoked currents were GluA1, and GluA2 or Cre, GluA1/K2, and GluA2 showed sim- strongly inwardly rectifying, indicating that native receptors have ilar glutamate-evoked currents from control cells, demonstrating been replaced with the expressed subunits (Fig. 2B). In addition, that we rescued surface expression of AMPARs (Fig. 3A). The I/ GluA1 rescued the majority of synaptic AMPARs (∼70%, Fig. V recordings in both conditions confirmed that the majority of 2C), and the synaptic currents were as rectifying as those from AMPARs on the surface in the rescued cells were heteromers the GluA2 KO (Fig. 2D), as expected for GluA1 homomers. In containing GluA2 (Fig. 3A). These data suggest that trafficking striking contrast, very little rescue occurred with GluA1/K2 (Fig. of GluA1/K2-GluA2 heteromers to the surface is not impaired. 2 E and F). Taken together, these findings indicate that, although We next tested the ability of the heteromers to rescue synaptic GluA1 Loop1 is not necessary for the delivery of GluA1 to the transmission. In contrast to the full rescue of synaptic transmission neuronal surface, it is essential for its targeting to the synapse. by wt GluA1 and GluA2 (Fig. 3B), expression of GluA1/K2 and GluA2 showed significantly smaller AMPAR EPSCs, compared GluA1 Loop1 Is Required for Synaptic Delivery of GluA1A2 with that in control cells (Fig. 3D). In addition, rectification indices Heteromers. As GluA1A2 heteromers are the major receptor in neurons expressing GluA1 and GluA2 (Fig. 3C) and GluA1/K2 complexes at synapses in CA1 pyramidal neurons (9), we ex- and GluA2 (Fig. 3E) are identical to that in control neurons, in- amined the role of GluA1 Loop1 in synaptic trafficking of dicating that receptors trafficked to synapses are heteromers. These

Lu et al. PNAS | December 21, 2010 | vol. 107 | no. 51 | 22267 Downloaded by guest on September 25, 2021 A B 800 1 600 0.8 0.6 400 pA Cnt Cre + GluA1 Cre + GluA1/K2 RI 0.4 * 200 0.2 * 0 0

Cnt Cnt GluA1 GluA1/K2 Cre + GluA1 Cre + GluA1 Cre + GluA1/K2 Cre + GluA1/K2

C AMPA EPSC NMDA EPSC -80 D Cre + GluA1 -40 1.4 -200 120

AMPA (pA) 0

50 RI 0.7 -100 60 * 25 * 0 NMDA (pA) 0 Cre + GluA1 (pA) Cre 0 0 Cnt Cnt 0 -100 -200 060120 GluA2 KO Control (pA) Control (pA) Cre + GluA1 Cre + GluA1 E F AMPA EPSC NMDA EPSC -120 Cre + GluA1/K2 -60 1.4 -200 180 *

AMPA (pA) 0 80

RI 0.7 -100 90 40 * *

NMDA (pA) 0 0

Cre + GluA1/K2 (pA) Cre 0 0 Cnt Cnt

0 -100 -200 0 90 180 GluA2 KO Control (pA) Control (pA) Cre + GluA1/K2 Cre + GluA1/K2

Fig. 2. GluA1 Loop1 is required for trafficking of GluA1 homomers to synapses. (A) Schematic of GluA1 and the GluA1/K2 chimera in which GluA1 Loop1 is replaced fl/fl by that of GluK2 (GluK2 Loop1 is in red). (B) Sample outside-out patch currents from pyramidal neurons in slice cultures from GRIA1-3 (Cnt: n = 12; Cre + GluA1: n = 10; Cre + GluA1/K2: n =9;P > 0.05 for all comparisons). (Scale bar: 1s, 200 pA.) RI of the I/V curves of outside-out patch currents shows that expressed receptors lacked GluA2 (Cnt: n = 10; Cre + GluA1: n =8;Cre+GluA1/K2:n =5;*P < 0.001 for comparing with Cnt; P = 0.47 for comparing Cre + GluA1 with Cre + GluA1/K2). (C–F) Scatter plots show amplitudes of EPSCs for single pairs (○)andmean± SEM (●). Bar graphs in C and E show amplitudes of AMPAR and NMDAR EPSCs (C, AMPA; n =13;P = 0.07; NMDA; n =13;P =0.26;andE, AMPA; n =13,*P < 0.001; NMDA; n =12;P =0.76).(D and F) Bar graphs show average RI (D, Cnt: n = 24; Cre + GluA1: n = 8; GluA2 KO: n =5;*P < 0.001 between Cnt and Cre + GluA1; P = 0.20 between Cre + GluA1 and GluA2 KO) and (F, Cnt: n = 24; Cre + GluA1/K2: n = 8; GluA2 KO: n =5; *P < 0.001 between Cnt and Cre + GluA1/K2; P = 0.58 between Cre + GluA1/K2 and GluA2 KO). Left are representative superimposed traces (Black, Cnt; green, Cre + GluA1 or GluA1/K2). (Scale bar: 0.02 s, 50 pA in C, 10 pA in D, 50 pA in E, and 20 pA in F.) Error bars represent mean ± SEM.

data indicate that GluA1A2 heteromers traffictosynapsesthrough (Fig. 5A). In addition, the phosphospecific antibody clearly rec- a GluA1 Loop1-dependent mechanism. ognized only the phosphorylated fusion protein and point mutations of S567 eliminated any immunoreactivity (Fig. 5B). We GluA1 Is Phosphorylated on S567 by CaMKII both in Vitro and in Vivo. next analyzed phosphorylation of immunoprecipitated full-length What is the molecular basis for the Loop1-dependent recep- GluA1 expressed in HEK293T cells and observed specific phos- tor targeting? We noted that GluA1 Loop1 contains several serine phorylation of GluA1, but not GluA2 (Fig. 5C). Importantly, in (S) /threonine (T) residues that could potentially be phosphory- detergent extracts of adult rat hippocampi, pS567-Ab recognized lated. We reasoned that, if GluA1 Loop1 was subject to phos- a single ∼100-kDa protein in GluA1 immunoprecipitates (Fig. 5D phorylation in neurons, this posttranslational modification could and Fig. S5), demonstrating that endogenous GluA1 is phos- be critical for the regulation of Loop1-dependent trafficking. Using phorylated at S567 in native brain tissue. in vitro phosphorylation assays, we found that GluA1 Loop1 was effectively phosphorylated by CaMKII (Fig. 4A), but not PKA or S567 of GluA1 Loop1 Negatively Regulates AMPAR Trafficking. Does PKC (Fig. 4B). In contrast, GluA2 Loop1, although sharing a high S567 in GluA1 Loop1 function in AMPAR trafficking? For instance, similarity with GluA1 Loop1, was not as efficiently phosphorylated phosphorylation of S567 could act as a triggering signal for forward by CaMKII (Fig. 4A), indicating substrate specificity for GluA1 trafficking of AMPARs to synapses. Alternatively, phosphorylation Loop1 phosphorylation. To identify the specific residue(s) phos- at S567 may hold receptors away from synapses, and thus function as phorylated by CaMKII in vitro, we generated point mutations of a negative regulator for synaptic delivery of AMPARs. To test these each S/T residue within GluA1 Loop1 and discovered that muta- possibilities, we used the molecular replacement approach to study tion of S567 to alanine (A) completely eliminated CaMKII phos- trafficking of GluA1 S567A, the nonphosphorylatable mutant, and phorylation of this region (Fig. 4C). Therefore, GluA1 S567 is the GluA1 S567D, the phospho-mimetic mutant in hippocampal slice CaMKII phosphorylation site in GluA1 Loop1. cultures. We transfected neurons with gold particles that were To evaluate the phosphorylation of GluA1 Loop1 in vivo, we coated with two expression constructs: Cre and GluA1 S567A-IRES- developed a phosphospecific antibody (pS567-Ab) that recog- EGFP or Cre and GluA1 S567D-IRES-EGFP. Both mutants fully nized GluA1 Loop1 phosphorylated at S567. Using an in vitro rescued surface AMPAR currents as revealed by recordings from phosphorylation assay of purified GST-GluA1Loop1, we found somatic outside-out patches, indicating that S567 is not involved in that CaMKII, but not PKA or PKC, phosphorylated GluA1 S567 AMPAR trafficking to neuronal surface (Fig. 6A). Interestingly,

22268 | www.pnas.org/cgi/doi/10.1073/pnas.1016289107 Lu et al. Downloaded by guest on September 25, 2021 A A 567 577 800 1.2 GluA1 Loop1 558-SRFSPYEWHSEEFEEGRDQTTSDQSNEF-585 600 GluA2 Loop1 565-SRFSPYEWHTEEFEDGRETQSSESTNEF-592

400 RI 0.6 pA 574 Cnt Cre + GluA1+A2Cre + GluA1/K2+A2 200 W T 0 0 1 a i l a i l p o o o Cnt Cnt - t - t L 2 L C C GST-GluA1 Loop1 1 2 A A A lu lu lu A D - G -G -G Cre + GluA1+A2 Cre + GluA1+A2 T 7 7 T T T T 5 7 7 Cre + GluA1/K2+A2 Cre + GluA1/K2+A2 CaMK II + S 5 S S S G W T T G G G Radiography 25 B AMPA EPSC NMDA EPSC -80 C Cre + GluA1+A2 -40 1.4 25 CBB staining

-160 100 AMPA (pA) 0 40

RI 0.7 PKA PKC GST-GluA1 Loop1 20 B Loop1 C-tail Loop1 C-tail C -80 50 1 2 1 2 1 2 1 2 NMDA (pA) 0 A A A A A A A A A A A A A A A 0 u l u u l lu lu l u lu l l u 5 8 1 7 7 8 9 2 G G T 6 6 7 7 5 7 8 - G G G - G -G G 5 5 5 5 5 5 Cnt Cnt - - - - - W T T T T T T T T T T S S S T T S S S S S S S S S S S S Cre + GluA1+A2 (pA) Cre 0 0 G G G G G G G G G G 37 0 -80 -160 0 50 100 25 Cre + GluA1+A2 Cre + GluA1+A2 Radiography Control (pA) Control (pA) 25 Radiography

AMPA EPSC NMDA EPSC 37 -120 CBB staining D 25 E 25 Cre + GluA1/K2+A2 * -60 1.4 CBB staining 100

-220 AMPA (pA) 0 Fig. 4. GluA1 Loop1 is phosphorylated on S567 by CaMKII. (A) GluA1 Loop1

50 RI 0.7 is phosphorylated by CaMKII. GST fusion proteins were incubated with -110 50 25 CaMKII and [γ-32P]ATP and analyzed by autoradiography. Sequence align- 0 NMDA (pA) 0 ment of Loop1 domains from GluA1 and GluA2 showed that among seven Cnt Cnt 0 0 Ser/Thr residues, T577 of GluA1 is a consensus CaMKII phosphorylation site. Cre + GluA1/K2+A2 (pA) Cre 0 -110 -220 050100 (B) GST fusion proteins were phosphorylated in vitro with purified PKA Control (pA) Control (pA) γ 32 Cre + GluA1/K2+A2 catalytic subunit or PKC together with [ - P]ATP, and analyzed by autora- Cre + GluA1/K2+A2 diography. (C) Identification of the CaMKII phosphorylation site, S567, in the Fig. 3. GluA1 Loop1 is required for trafficking of GluA1A2 heteromers to GluA1 Loop1 domain. Each of 7 Ser/Thr residues in GST-GluA1 Loop1 was 32 synapses. (A) Sample outside-out patch currents from pyramidal neurons in mutated to Ala, and phosphorylated by CaMKII in vitro using [γ- P]ATP and fl/fl slice cultures from GRIA1-3 transfected with mCherryCre-IRES-GluA1 and analyzed by autoradiography. Total GST fusion protein was visualized by GluA2 or mCherryCre-IRES-GluA1/K2 and GluA2 (Cnt: n = 12; Cre + GluA1 + protein staining in A–C. A2: n = 11; Cre + GluA1/K2 + A2: n =9;P > 0.2 for all comparisons. (Scale bar: 1 s, 200 pA). RI of I/V curves from outside-out patches pulled from indicated cells show that expressed receptors contained GluA2 (Cnt: n =10;Cre+ of AMPARs to synapses. Indeed, the enhanced synaptic locali- GluA1 + A2: n = 8; Cre + GluA1/K2 + A2: n =6;P > 0.20 for all comparisons). D zation of GluA1 S567A in Fig. 6 is reminiscent of the data from NEUROSCIENCE (B–E) Scatter plots show amplitudes of EPSCs for single pairs (○) and mean ± our molecular replacement assay in which synaptic rescue of the SEM (●). Bar graphs in B and D show that ∼98% AMPAR EPSCs were rescued ∼ B B, n P n GluA1 S567A mutant ( 82% of control, Fig. 6 ) is higher than by expression of GluA1 and GluA2 ( AMPA: = 14; = 0.93; NMDA: = 12; ∼ C P = 0.50), and ∼60% AMPAR EPSCs were rescued by expression of GluA1/K2 that of WT GluA1 ( 70% of control, Fig. 2 ). Furthermore, we and GluA2 (D, AMPA: n = 12, *P < 0.005; NMDA: n = 11; P =0.95).(C and E) reasoned that trafficking of the GluA1/A2-Loop1 chimera would Bar graphs show average RI (C, Cnt: n = 24; Cre + GluA1 + A2: n =8;P =0.70;E, be similar to GluA1 S567A, as GluA2 Loop1 was not an efficient Cnt: n = 24; Cre + GluA1/K2 + A2: n =5;P = 0.97). On left are representative phosphorylation substrate of CaMKII. Indeed, a molecular re- superimposed traces (Black, Cnt; green, Cre + GluA1 or GluA1/K2 + A2). (Scale bar: 0.02 s, 40 pA in B, 20 pA in C,50pAinD, and 50 pA in E.) Error bars represent mean ± SEM. GST-GluA1 Loop1 GST-GluA1 Loop1 A CaMK II PKA PKC B GST WT S567A S567D Kinase - + - + - + whereas GluA1 S567A rescued the majority of AMPAR-mediated WB CaMK II - + - + - + - + WB pS567-Ab synaptic transmission (∼82%), synaptic transmission in neurons 25 25 pS567-Ab

expressing GluA1 S567D was significantly smaller than that in GST 25 GST nearby controls (∼45%), showing that the phosphomimetic muta- 25 tion at S567 impaired receptor trafficking to synapses (Fig. 6B). FLAG-GluA1 FLAG-GluA2 C D Hippocampus (IP) WT S567A WT T574A Taken together, these data show that S567 is a critical residue that IgG GluA1 WB CaMK II - + - + - + - + WB regulates Loop1-dependent AMPAR trafficking. Furthermore, we 100 pS567-Ab identify S567 as a unique CaMKII site on GluA1, and phosphory- 100 pS567-Ab 100 GluA1 lation of GluA1 S567 in neurons inhibits receptor trafficking to 100 FLAG synapsesandactsasanegativeregulatorforsynaptictrafficking A B of AMPARs. Fig. 5. Phosphorylation of GluA1 S567 in vitro and in vivo. ( and )GST- GluA1 Loop1 was phosphorylated on S567 in vitro with CaMKII, but not We also evaluated synaptic targeting of GluA1 using immu- fl PKA or PKC. GST and GST-GluA1 Loop1 (WT, S567A, and S567D) were no uorescence microscopy and showed that the surface expres- phosphorylated in vitro as indicated, and analyzed by immunoblotting sion of GluA1 S567A and GluA1 S567D was indistinguishable with pS567-Ab. Equal loading of GST fusion protein was confirmed by from WT GluA1 when expressed in cultured hippocampal neu- immunoblotting with GST antibody (C) pS567-Ab recognizes GluA1, but rons (Fig. 6C), In contrast, the synaptic localization of GluA1 not GluA2, phosphorylation. FLAG-tagged AMPAR subunits were ex- fi S567A was significantly enhanced, whereas the synaptic expres- pressed in HEK293T cells, immunoprecipitated by FLAG M2 af nity gel, sion of GluA1 S567D was decreased (Fig. 6D), as determined by phosphorylated by CaMKII in vitro, and analyzed by immunoblotting with pS567-Ab. (D) GluA1 was phosphorylated at S567 in vivo. Adult rat hip- colocalization with endogenous PSD-95. Importantly, these im- pocampi were solubilized, and GluA1 was immunoprecipitated from ly- aging data support the physiology data in demonstrating that a sate. Immunoprecipitates were resolved by SDS/PAGE and immunoblotted phosphomimetic mutation at S567 specifically inhibits trafficking with indicated antibodies.

Lu et al. PNAS | December 21, 2010 | vol. 107 | no. 51 | 22269 Downloaded by guest on September 25, 2021 A B Cre + S567A -90 transmission onto CA1 pyramidal neurons (9, 21, 22), indicating AMPA EPSC NMDA EPSC -45 the importance of GluA1 in the maintenance of basal synaptic -200 200

AMPA (pA) 0 transmission. In addition, GluA1 plays a key role in synaptic Cnt Cre + S567ACre + S67D 80 plasticity, including both LTP (12, 18, 21, 23, 24) and LTD (11), -100 100 40 suggesting an important role for GluA1 in activity-dependent 0

NMDA (pA) ﬁ Cre + S567A (pA) Cre 0 0 AMPAR traf cking. Furthermore, the maintenance of extra- Cnt 0 -100 -200 0 100 200 S567A synaptic AMPARs is essentially dependent on the GluA1 sub- Cre + S567D 1000 -100 unit (9, 21, 22). Previous studies have shown that the GluA1 800 AMPA EPSC NMDA EPSC -50 * C-tail is critical for synaptic plasticity, but not for basal synaptic 600 -200 180

pA 400

AMPA (pA) 0 transmission (12). More recently, a membrane proximal region 200 100 0 -100 90 of the GluA1 C-tail that mediates the interaction with 4.1N 50 Cnt protein was found to function in surface delivery of GluA1, S567A S567D 0 NMDA (pA) but not in synaptic trafficking in the absence of activity (18). Cre + S567D (pA) Cre 0 0 0 -100 -200 090180 Cnt Therefore, the mechanisms regulating synaptic trafficking of Control (pA) Control (pA) S567D GluA1-containing AMPARs to mediate basal synaptic traffick-

WT S567A S567D ing are unclear. Our study now shows that the GluA1 Loop1 is C critical for synaptic trafficking, but not for surface delivery of 120 Surface GluA1, thus providing a linkage between extrasynaptic and synaptic trafficking of AMPARs. 60 level Our findings are reminiscent of the role of TARPs in regulating 0 AMPAR synaptic expression in a two-step process, with surface Intracellular surface Normalized WT S567AS567D expression being PDZ ligand independent, but synaptic expression requiring an intact PDZ ligand on TARPs (15, 25–27). However, our studies on Loop1-dependent trafficking also show that the fi D WT S567A S567D interaction of TARPs is not the only determining factor. Speci - * fi FLAG 0.5 cally, we nd that the GluA1/K2 chimera has decreased synaptic (GluA1) 0.4 * expression, yet retains TARP binding (Fig. S3), Therefore, further 0.3 studies will be necessary to understand the additional mechanisms PSD-95 0.2 0.1 underlying this unique synaptic targeting process. Colocalization with PSD-95 0 Previous studies have reported that GluA1 homomers overex- Merge WT S567AS567D pressed in rat hippocampal slice cultures failed to traffic to syn- fi apses (12, 24, 28). In contrast, our molecular replacement assay Fig. 6. Phosphorylation of GluA1 S567 regulates AMPAR traf cking to reveals that GluA1 homomers can effectively traffic to synapses. It synapses. (A) Sample outside-out patch currents from pyramidal neurons in fl/fl slice cultures from GRIA1-3 (Cnt: n = 12; Cre + S567A: n = 10; Cre + S567D: is unclear what accounts for this discrepancy, although different n = 10; P > 0.05 for all comparisons). (Scale bar: 1s, 200 pA.) (B) Scatter plots expression strategies (overexpression vs. molecular replacement), show amplitudes of EPSCs for single pairs (○) and mean ± SEM (●). Bar different expression constructs (GFP fusion to GluA1 vs. wt GluA1 graphs in right show amplitudes of AMPAR and NMDAR EPSCs (Upper, in our study) and different slice culture preparations (rat vs. mouse) AMPA: n = 14; P = 0.18; NMDA: n = 14; P = 0.46) and (Lower, AMPA: n = 12, may underlie the differences. Nevertheless, the robust rescue of *P < 0.001; NMDA: n = 12; P = 0.47). Scale bar: 0.02 s, 50 pA in Upper, 20 pA synaptic transmission by expression of GluA1 homomers on an in Lower. (C) FLAG-GluA1 (WT, S567A, or S567D) was expressed in cultured AMPAR-null background allowed the efficient quantitative anal- hippocampal neurons and analyzed at DIV 17. Surface receptors were shown fi P > ysis of traf cking of expressed receptors (Figs. 2 and 3), which in white and intracellular receptors were shown in red. 0.05 for all revealed Loop1-dependent AMPAR trafficking. comparisons (n =31–39 cells; n = 4 independent experiments). (Scale bar, μ D Phosphorylation of GluA1 by protein kinases is critical for 20 m.) ( ) Colocalization of endogenous PSD-95 and FLAG-GluA1 (WT, fi – S567A, or S567D) was analyzed at DIV 17 in hippocampal neurons. *P < 0.001 AMPAR traf cking (11, 24, 29 31). For example, genetic evidence (n =30–40 cells; n = 3 independent experiments). (Scale bar, 5 μm.) Error bars demonstrates that regulation of GluA1 S831, a CaMKII/PKC site, represent mean ± SEM. and S845, a PKA/PKG site, is important for activity-dependent AMPAR trafficking (11, 30, 32–34). In addition, PKC phosphorylation of S818 in the GluA1 C-tail regulates the interaction placement experiment showed that synaptic delivery of GluA1/ between GluA1 and 4.1N and promotes GluA1 trafficking to ex- A2-Loop1 was as efficient as GluA1 S567A (Fig. S6). trasynaptic membranes (18). Furthermore, phosphorylation of S818 works synergistically with S831 and S845 phosphorylation to Discussion facilitate synaptic trafficking of AMPARs in response to neuronal Using a molecular replacement approach, we now show that activity (24). However, until now, all of the characterized phos- a cytoplasmic domain, Loop1, of the GluA1 subunit is critical for phorylation sites have been located within the GluA1 C-tail and AMPAR trafficking. Specifically, Loop1 is critical for the tar- none of these sites are critical for the maintenance of basal synaptic geting of AMPARs to the synapse, but not for the delivery of transmission (11, 18, 24, 33). receptors to extrasynaptic membranes. In addition, we identified As a multifunctional protein kinase, CaMKII functions both a unique CaMKII phosphorylation site within Loop1 and showed positively and negatively in AMPAR trafficking as a consequence that a phospho-mimetic mutation of S567 limits unregulated of phosphorylating its many substrates. For example, CaMKII trafficking of AMPARs to synapses. Therefore, regulation of phosphorylation of TARPs is critical for activity-dependent traf- GluA1 Loop1 contributes to the maintenance of the stability of ficking of AMPARs to synapses during LTP (35). However, basal synaptic transmission. CaMKII phosphorylation of PSD-95 inhibits LTP-induced syn- Synaptic trafficking of AMPARs is thought to involve two aptic potentiation (36). Furthermore, CaMKII phosphorylation of steps: (i) an extrasynaptic insertion of receptors on the surface, synGAP negatively regulates synaptic strength (37). Interestingly, and (ii) a subsequent synaptic insertion of AMPARs from CaMKII can phosphorylate itself at two separate sites, one en- extrasynaptic pools (13–20). The GluA1 subunit has been shown hancing kinase activity and the other inhibiting activity (38). This is to play a critical role in both steps. Indeed, genetic deletion of analogous to CaMKII phosphorylation of GluA1 in which C-tail the GluA1 subunit leads to a profound loss of excitatory synaptic phosphorylation enhances receptor trafficking in response to

22270 | www.pnas.org/cgi/doi/10.1073/pnas.1016289107 Lu et al. Downloaded by guest on September 25, 2021 neuronal activity, whereas phosphorylation of S567 in Loop1 results suggest that the exchange of receptors between extra- inhibits delivery of AMPARs to synapses. synaptic and synaptic sites is a highly regulated step in AMPAR Although overexpression of the CaMKII catalytic domain synaptic trafficking and that the GluA1 Loop1 domain provides promotes the trafficking of recombinant GluA1 homomers to a unique mechanism for such regulation of AMPAR trafficking synapses (28), the underlying mechanism is unknown. Complex in neurons. intracellular calcium dynamics and differential subcellular localization of CaMKII isoforms may allow its substrates to be phos- Materials and Methods phorylated in a temporally and spatially controlled manner to Organotypical hippocampal slice cultures were prepared from P6-8 rat or P5–P8 regulate AMPAR trafficking. Alternatively, overexpression of the GRIA1-3fl/fl mice and transfected biolistically with various plasmids after 2–3 CaMKII catalytic domain may lead to simultaneous phosphory- d in culture. Phosphorylation of the Loop1 domain of AMPAR subunits was lation of many proteins that function differentially in AMPAR characterized using purified GST fusion proteins, recombinant receptors trafficking, resulting in phosphorylation of proteins (such as expressed in HEK293T cells and receptors in native brain tissues. Immunocy- TARPs or GluA1 C-tail) that promote synaptic trafficking that tochemistry was used to study trafficking of expressed AMPAR subunits in can override phosphorylation of other substrates that negatively dissociated hippocampal cultures. Detailed protocols for subcloning con- regulate synaptic trafficking of AMPARs (such as GluA1 S567). A structs, antibody production, electrophysiological recordings, biochemical relevant example is the activation of PKC by phorbol esters (39) experiments, and immunostaining are provided in SI Materials and Methods. and postsynaptic loading of catalytic fragment of PKC (40), which enhance AMPAR-mediated mEPSC amplitude in CA1 pyramidal ACKNOWLEDGMENTS. We thank K. Bjorgan and M. Cerpas for technical assistance, and all members from the R.A.N. laboratory and K.W.R. neurons. However, PKC phosphorylation of GluA2 S880 excludes laboratory for helpful discussions. We are also grateful for the generosity AMPARs from synapses, and thus negatively regulates AMPAR of Profs. Peter Seeburg and Rolf Sprengel at The Max Planck Institute for trafficking (41). Thus, it is important in the future to determine Medical Research (Heidelberg, Germany) for sharing individual gene tar- neuronal activity patterns in which individual substrates of protein geted conditional mice for GluA1, GluA2, or GluA3, from which we were fl Gria1-3fl/fl kinases are phosphorylated in a temporally and spatially con- able to make the triple oxed mice. We also thank the National fi fi fi Institute of Neurological Disorders and Stroke (NINDS) Light Imaging Facility, trolled manner to ful ll speci c traf cking events. particularly C. Smith, and the NINDS sequencing facility for DNA sequencing. In hippocampal CA1 pyramidal neurons, AMPARs at extra- W.L. is funded by a postdoctoral fellowship from the American Heart Asso- synaptic membranes are highly abundant (9, 22) and have been ciation; K.I. is funded by the Japan Society for the Promotion of Science Research Fellowship for Japanese Biomedical and Behavioral Researchers proposed to serve as a reserve pool of receptors, which not only at the National Institutes of Health; K.W.R. is funded by the NINDS Intra- supply synaptic pools to maintain synaptic transmission but are mural Research Program; and R.A.N. is funded by grants from the National also trafficked to synapses during synaptic plasticity (13–20). Our Institute of Mental Health.

1. Hollmann M, Heinemann S (1994) Cloned glutamate receptors. Annu Rev Neurosci 17: 23. Zamanillo D, et al. (1999) Importance of AMPA receptors for hippocampal synaptic 31–108. plasticity but not for spatial learning. Science 284:1805–1811. 2. Dingledine R, Borges K, Bowie D, Traynelis SF (1999) The glutamate receptor ion 24. Boehm J, et al. (2006) Synaptic incorporation of AMPA receptors during LTP is channels. Pharmacol Rev 51:7–61. controlled by a PKC phosphorylation site on GluR1. Neuron 51:213–225. 3. Malinow R, Malenka RC (2002) AMPA receptor trafficking and synaptic plasticity. 25. Chen L, et al. (2000) Stargazin regulates synaptic targeting of AMPA receptors by two Annu Rev Neurosci 25:103–126. distinct mechanisms. Nature 408:936–943. NEUROSCIENCE 4. Shepherd JD, Huganir RL (2007) The cell biology of synaptic plasticity: AMPA receptor 26. Schnell E, et al. (2002) Direct interactions between PSD-95 and stargazin control trafficking. Annu Rev Cell Dev Biol 23:613–643. synaptic AMPA receptor number. Proc Natl Acad Sci USA 99:13902–13907. 5. Bredt DS, Nicoll RA (2003) AMPA receptor trafficking at excitatory synapses. Neuron 27. Ziff EB (2007) TARPs and the AMPA receptor trafficking paradox. Neuron 53:627–633. 40:361–379. 28. Hayashi Y, et al. (2000) Driving AMPA receptors into synapses by LTP and CaMKII: 6. Newpher TM, Ehlers MD (2008) Glutamate receptor dynamics in dendritic micro- Requirement for GluR1 and PDZ domain interaction. Science 287:2262–2267. domains. Neuron 58:472–497. 29. Barria A, Muller D, Derkach V, Griffith LC, Soderling TR (1997) Regulatory phos- 7. Wenthold RJ, Petralia RS, Blahos J II, Niedzielski AS (1996) Evidence for multiple AMPA phorylation of AMPA-type glutamate receptors by CaM-KII during long-term receptor complexes in hippocampal CA1/CA2 neurons. J Neurosci 16:1982–1989. potentiation. Science 276:2042–2045. 8. Nicoll RA, Tomita S, Bredt DS (2006) Auxiliary subunits assist AMPA-type glutamate 30. Lee HK, Barbarosie M, Kameyama K, Bear MF, Huganir RL (2000) Regulation of receptors. Science 311:1253–1256. distinct AMPA receptor phosphorylation sites during bidirectional synaptic plasticity. 9. Lu W, et al. (2009) Subunit composition of synaptic AMPA receptors revealed by Nature 405:955–959. a single-cell genetic approach. Neuron 62:254–268. 31. Man HY, Sekine-Aizawa Y, Huganir RL (2007) Regulation of alpha-amino-3-hydroxy-5- 10. Kim CH, et al. (2005) Persistent hippocampal CA1 LTP in mice lacking the C-terminal methyl-4-isoxazolepropionic acid receptor trafficking through PKA phosphorylation PDZ ligand of GluR1. Nat Neurosci 8:985–987. of the Glu receptor 1 subunit. Proc Natl Acad Sci USA 104:3579–3584. 11. Lee HK, et al. (2003) Phosphorylation of the AMPA receptor GluR1 subunit is required 32. Barria A, Derkach V, Soderling T (1997) Identification of the Ca2+/calmodulin- for synaptic plasticity and retention of spatial memory. Cell 112:631–643. dependent protein kinase II regulatory phosphorylation site in the alpha-amino-3- 12. Shi S, Hayashi Y, Esteban JA, Malinow R (2001) Subunit-specific rules governing AMPA hydroxyl-5-methyl-4-isoxazole-propionate-type glutamate receptor. J Biol Chem 272: receptor trafficking to synapses in hippocampal pyramidal neurons. Cell 105:331–343. 32727–32730. 13. Heine M, et al. (2008) Surface mobility of postsynaptic AMPARs tunes synaptic 33. Roche KW, O’Brien RJ, Mammen AL, Bernhardt J, Huganir RL (1996) Characterization transmission. Science 320:201–205. of multiple phosphorylation sites on the AMPA receptor GluR1 subunit. Neuron 16: 14. Ehlers MD, Heine M, Groc L, Lee MC, Choquet D (2007) Diffusional trapping of GluR1 1179–1188. AMPA receptors by input-specific synaptic activity. Neuron 54:447–460. 34. Serulle Y, et al. (2007) A GluR1-cGKII interaction regulates AMPA receptor trafficking. 15. Bats C, Groc L, Choquet D (2007) The interaction between Stargazin and PSD-95 Neuron 56:670–688. regulates AMPA receptor surface trafficking. Neuron 53:719–734. 35. Tomita S, Stein V, Stocker TJ, Nicoll RA, Bredt DS (2005) Bidirectional synaptic 16. Tardin C, Cognet L, Bats C, Lounis B, Choquet D (2003) Direct imaging of lateral plasticity regulated by phosphorylation of stargazin-like TARPs. Neuron 45:269–277. movements of AMPA receptors inside synapses. EMBO J 22:4656–4665. 36. Steiner P, et al. (2008) Destabilization of the postsynaptic density by PSD-95 serine 73 17. Borgdorff AJ, Choquet D (2002) Regulation of AMPA receptor lateral movements. phosphorylation inhibits spine growth and synaptic plasticity. Neuron 60:788–802. Nature 417:649–653. 37. Krapivinsky G, Medina I, Krapivinsky L, Gapon S, Clapham DE (2004) SynGAP-MUPP1- 18. Lin DT, et al. (2009) Regulation of AMPA receptor extrasynaptic insertion by 4.1N, CaMKII synaptic complexes regulate p38 MAP kinase activity and NMDA receptor- phosphorylation and palmitoylation. Nat Neurosci 12:879–887. dependent synaptic AMPA receptor potentiation. Neuron 43:563–574. 19. Makino H, Malinow R (2009) AMPA receptor incorporation into synapses during LTP: 38. Elgersma Y, et al. (2002) Inhibitory autophosphorylation of CaMKII controls PSD The role of lateral movement and exocytosis. Neuron 64:381–390. association, plasticity, and learning. Neuron 36:493–505. 20. Groc L, et al. (2004) Differential activity-dependent regulation of the lateral mobilities 39. Carroll RC, Nicoll RA, Malenka RC (1998) Effects of PKA and PKC on miniature of AMPA and NMDA receptors. Nat Neurosci 7:695–696. excitatory postsynaptic currents in CA1 pyramidal cells. J Neurophysiol 80:2797–2800. 21. Jensen V, et al. (2003) A juvenile form of postsynaptic hippocampal long-term 40. Wang LY, Dudek EM, Browning MD, MacDonald JF (1994) Modulation of AMPA/ potentiation in mice deficient for the AMPA receptor subunit GluR-A. J Physiol 553: kainate receptors in cultured murine hippocampal neurones by protein kinase C. 843–856. J Physiol 475:431–437. 22. Andrásfalvy BK, Smith MA, Borchardt T, Sprengel R, Magee JC (2003) Impaired 41. Seidenman KJ, Steinberg JP, Huganir R, Malinow R (2003) Glutamate receptor subunit regulation of synaptic strength in hippocampal neurons from GluR1-deficient mice. 2 Serine 880 phosphorylation modulates synaptic transmission and mediates plasticity J Physiol 552:35–45. in CA1 pyramidal cells. J Neurosci 23:9220–9228.

Lu et al. PNAS | December 21, 2010 | vol. 107 | no. 51 | 22271 Downloaded by guest on September 25, 2021