To identify novel NR2C interacting proteins, we screened a rat brain yeast two-hybrid library with the NR2C C-terminus (a.a. 1076 – a.a. 1250; ), and isolated 14-3-3ε (Fig. S1
). The mammalian 14-3-3 proteins consist of seven highly homologous members that interact with a variety of proteins, regulating signal transduction and protein trafficking. The 14-3-3 family of proteins were originally isolated as abundant proteins in the brain and are highly conserved in a broad range of eukaryotic organisms.
Fig. 1 NR2C specifically interacts with 14-3-3ε, and binding is regulated by S1096. A, Schematic diagram of NR2C including an alignment of NR2A (amino acids 1281-1300), NR2B (amino acids 1293-1312) and NR2C (amino acids 1086-1105). NR2C S1096 is indicated (more ...)
To characterize the specificity of the NR2C-14-3-3ε interaction, we evaluated the binding of 14-3-3ε to the C-terminus of NR2A, NR2B or NR2C using yeast two-hybrid direct interaction assays. By evaluating growth on histidine-deficient plates (-HIS) as an assay for protein-protein interactions, we observed no specific binding when the C-terminus of NR2A or NR2B was co-expressed with 14-3-3ε (Fig. S1
). In contrast, NR2C interacted strongly with 14-3-3ε, demonstrating that the interaction between 14-3-3ε and NMDA receptors is specific for the NR2C subunit.
Upon sequence examination, we found two regions (1093-R-H-A-S-L-P-1098 and 1141-R-L-P-S-Y-P-1146) within the NR2C C-terminus that are similar to previously identified 14-3-3 binding motifs, R-S-X-SP
-X-P and R-X-X-X-SP
-X-P (X can be any amino acid and SP
indicates a phosphorylated serine) (Yaffe et al., 1997
). To determine whether these two regions of NR2C are required for 14-3-3ε binding, we tested the interaction of 14-3-3ε with NR2C containing a mutation in the critical serine residue within each putative binding site, S1096A () or S1144A. The NR2C S1096A mutation disrupted the 14-3-3ε binding, whereas NR2C S1144A mutant had no effect (), indicating that S1096 is part of the 14-3-3ε binding motif on NR2C.
We also examined 14-3-3ε binding to NR2C using a co-immunoprecipitation assay in HEK-293 cells expressing FLAG-NR2C or FLAG-NR2C S1096A and Myc-14-3-3ε. Surprisingly, Myc-14-3-3ε was not co-immunoprecipitated efficiently with NR2C (). Because NR2 subunits are ER-retained in the absence of NR1, we reasoned that perhaps the correct assembly of NR2C with NR1 is required for the 14-3-3 interactions. Indeed, there is evidence that 14-3-3 family members preferentially bind to assembled proteins in the ER-Golgi intermediate compartment thus facilitating export of properly assembled multimers from the ER (O'Kelly et al., 2002
; Yuan et al., 2003
). Consistent with this hypothesis, we observed robust binding of 14-3-3 to NR2C only upon co-transfection with NR1. Furthermore, the NR2C S1096A mutation disrupted the NR2C-14-3-3ε interaction (). These results not only provided evidence that the R-H-A-S-L-P (S1096) motif of NR2C is responsible for 14-3-3 binding but also show that 14-3-3ε specifically binds NR1/NR2C heteromers.
To investigate the interaction between NR2C and 14-3-3ε in neurons, 14-3-3ε was immunoprecipitated from cerebellum homogenate and immunoblotted with an NR2C antibody, which also cross-reacted with NR2A and NR2B. Consistent with the results from yeast two-hybrid interaction assay, we found that NR2C, but not NR2A and NR2B, co-immunoprecipitated with 14-3-3ε (), demonstrating that NR2C is a specific binding partner for 14-3-3ε in vivo.
NMDA receptors are dynamically regulated by protein phosphorylation, and many sites have been identified on the C-termini of NR2 subunits (Chen and Roche, 2007
). Because 14-3-3 proteins typically bind to phosphorylated serine residues (Yaffe et al., 1997
) and S1096 on NR2C is required for 14-3-3 binding, we investigated whether S1096 on NR2C is phosphorylated. Interestingly, sequence examination revealed a prototypical PKB consensus motif surrounding S1096 in the NR2C intracellular C-terminus, which coincides with the 14-3-3 binding motif (). Although the serine is conserved in other NR2 subunits, the entire PKB recognition motif is not (). We performed in vitro
phosphorylation assays using GST fusion proteins (GST, GST-NR2C wild-type (WT) or GST-NR2C S1096A) incubated with recombinant PKB. The equal loading of GST-NR2C fusion proteins was confirmed by Coomassie blue staining (data not shown). GST-NR2C WT and GST-NR2C S1244A were efficiently phosphorylated by PKB in vitro
), but reduced phosphorylation was observed for GST-NR2C S1096A (Fig. S2A
), demonstrating the direct phosphorylation of S1096 by PKB in vitro
. The residual phosphorylation of GST-NR2C S1096A indicates that there is a second site on NR2C that PKB phosphorylates in vitro
, although less efficiently than S1096.
We next raised a phosphorylation-state specific polyclonal antibody recognizing phosphorylated S1096 that specifically detected PKB phosphorylated GST-NR2C fusion protein (Fig. S2B
). In contrast, the antibody did not recognize GST-NR2C S1096A even when the fusion protein was incubated with PKB (Fig. S2B
), demonstrating the direct phosphorylation of S1096 by PKB in vitro
and confirming the specificity of the antibody for NR2C phosphorylated on S1096.
To test if NR2C S1096 is phosphorylated by PKB activation in situ, we co-expressed NR2C with a constitutively active form of PKB (caPKB)(Cong et al., 1997
) in HEK-293 cells. Co-expression of NR2C with either caPKB or a dominant negative form of PKB (dnPKB) in HEK-293 cells did not affect NR2C expression (). The phosphorylation of S1096 increased dramatically upon co-transfection with caPKB, but no increase was observed for NR2C S1096A or for NR2C WT or NR2C S1096A co-expressed with dnPKB (). Similar results were obtained when NR1 was co-transfected with NR2C and caPKB or dnPKB (Fig. S3A
Fig. 2 NR2C is phosphorylated on S1096 by PKB and phosphorylation is regulated by IGF-I and NMDA receptor activity. A, HEK-293 cells were co-transfected with NR2C WT or NR2C S1096A and vector, a constitutively active form of PKB (caPKB), or a dominant-negative (more ...)
Various growth factors stimulate the phosphoinositide 3-kinase (PI3K) signal transduction cascade, which results in the activation of PKB (Datta et al., 1999
). Because HEK-293 cells are insulin-sensitive, we investigated whether IGF-1 treatment, which activates PKB via PI3K, would stimulate NR2C phosphorylation on S1096. We observed a marked increase in NR2C S1096 phosphorylation in response to PKB activation that was blocked by incubation with the PI3K inhibitor wortmannin to prevent activation of the PI3K-PKB pathway (). Taken together, these results demonstrate that IGF-1, through the activation of PI3K/PKB pathway, is able to induce NR2C S1096 phosphorylation. Similar results were obtained when NR2C was co-transfected with NR1 (Fig. S3B
To evaluate the phosphorylation of endogenous NR2C, we analyzed NR2C isolated from mouse cerebellum slices (prepared from P12 animals). Immunoblots probed with the NR2C S1096 phosphorylation state-specific antibody revealed that endogenous NR2C was phosphorylated on S1096 (). Furthmore, treatment of the membrane with lambda-phosphatase prior to immunoblotting eliminated the NR2C band recognized with the S1096 phosphorylation state-specific antibody (), confirming that the immunoreactivity was specific for phosphorylated NR2C. NR2C is expressed at a very low level in hippocampus compared to cerebellum, accordingly in hippocampus we observed only a low level of NR2C expression ().
To investigate if S1096 phosphorylation on endogenous NR2C is dynamically regulated in neurons, we treated cultured cerebellar granule cells (12 days in vitro) with IGF-1 for various periods of time to stimulate PKB activity. Immunoblot analysis with the NR2C S1096 phosphorylation state-specific antibody revealed that NR2C S1096 is phosphorylated even in the absence of IGF-1, indicating that there is a modest basal level of phosphorylation in cerebellar granule cells. However, IGF-1 treatment resulted in a dramatic increase in phosphorylation of NR2C S1096 (), showing that endogenous NR2C S1096 is regulated by PKB in vivo
. NMDA receptor activation has been shown to stimulate PKB activity (Lafon-Cazal et al., 2002
); therefore, we examined if agonist exposure regulates phosphorylation of NR2C S1096 and found that brief NMDA treatment transiently increased phosphorylation of NR2C S1096 (). In contrast, application of the NMDA receptor antagonist APV decreased NR2C S1096 phosphorylation over the same time course (), revealing a tonic stimulation of NR2C S1096 phosphorylation by endogenous glutamate. These results demonstrate that both IGF-1 and NMDA receptor activity regulate phosphorylation of NR2C on S1096 in cerebellar granule cells.
Phosphorylation of ionotropic glutamate receptors is an important mechanism modulating receptor trafficking (Lee, 2006
). We therefore examined if PKB phosphorylation of NR2C affects surface expression of NR2C-containing NMDA receptors using fluorescence activated cell sorting (FACS) analysis. A bicistronic expression construct containing the coding sequence of FLAG-tagged NR2C and EGFP was used to monitor NR2C-transfected cells (GFP positive cells), and a second bicistronic expression construct containing the coding sequence of NR1-1a and DsRed was used to monitor NR1-1a-transfected cells (DsRed positive cells). Surface expression of FLAG-NR2C in live HEK-293 cells expressing both GFP and DsRed was measured. We found that caPKB dramatically increased wild-type NR2C surface expression, whereas expression of caPKB resulted in only a modest increase in the surface expression of NR2C S1096A (), demonstrating that S1096 primarily mediates the PKB-dependent increase in NMDA receptor surface expression.
Fig. 3 PKB phosphorylation of NR2C on S1096 increases NMDA receptor surface expression. A, HEK-293 cells were co-transfected with NR1-1a-IRES-DsRed, FLAG-NR2C-IRES-EGFP and vector, caPKB or dnPKB. Cells were collected and analyzed by FACS analysis (see Experimental (more ...)
We also measured the effects of PKB on surface expression of NR2C/NR1 heteromers in HEK-293 cells using a cell surface biotinylation assay. We found that IGF-1 treatment robustly increased surface expression of NR2C WT with only a modest effect on NR2C S1096A (). α-Tubulin, an intracellular protein, was used as a negative control for surface biotinylation.
To determine whether surface expression of endogenous NR2C is regulated by PKB phosphorylation in neurons, we treated cultured cerebellar granule cells with IGF-1 and performed a cell surface biotinylation assay. IGF-1 treatment dramatically increased surface expression of NR2C in cerebellar granule cells (), and the increase was blocked by pre-incubation with the PI3K inhibitor wortmannin or LY294002 prior to IGF-1 treatment (Fig. S4A and S4C
). We also treated cultured cerebellar granule cells with NMDA and evaluated receptor surface expression using the cell surface biotinylation assay. Interestingly, NMDA treatment also dramatically upregulated surface expression of endogenous NR2C (), which can be blocked by pre-incubation with wortmannin or LY294002 (Fig. S4B and S4C
). Taken together, these data demonstrate that PKB phosphorylation of NR2C S1096 driven by IGF-1 signaling or NMDA receptor activation facilitates NR2C trafficking to the plasma membrane.
The biotinylation results indicated that PKB phosphorylation of NR2C is required for receptor trafficking in neurons, but these experiments did not specifically address the role of 14-3-3ε in regulating surface expression of NR2C. To determine whether 14-3-3ε binding to NR2C is necessary for NR2C trafficking, we used a dominant negative form of 14-3-3ε (14-3-3ε dn) to functionally deplete 14-3-3ε in neurons (Czirjak et al., 2008
; Thorson et al., 1998
). We made a GFP-NR2C construct, which contains a GFP tag in the extracellular N-terminal domain of the receptor, to specifically label surface-expressed receptors. Cerebellar granule cells were co-transfected with GFP-NR2C WT or GFP-NR2C S1096A and Myc-14-3-3ε or Myc-14-3-3ε dn. We labeled the surface-expressed receptors with anti-GFP antibody. Surface expression of NR2C WT was dramatically decreased upon co-transfection with 14-3-3ε dn (), indicating that the interaction between NR2C and 14-3-3 is required for NR2C trafficking. In addition, surface expression of NR2C S1096A was much less than that of NR2C WT when 14-3-3ε was co-expressed and was further reduced upon co-expression with 14-3-3ε dn (), suggesting that S1096 is not the sole molecular determinant in 14-3-3ε-regulated NR2C trafficking.
Fig. 4 Disruption of 14-3-3ε binding to NR2C reduces NR2C surface expression. Cerebelluar granule cells were co-transfected with NR2C containing an extracellular GFP protein tag (GFP-NR2C) or GFP-NR2C S1096A and Myc-14-3-3ε WT or Myc-14-3-3ε (more ...)
Because IGF-1 has been shown to promote neuronal survival by activating PKB and increased surface expression of NR2C was driven by the same signaling pathway, we investigated whether NR2C is involved in neuronal survival. Cultured cerebellar granule cells were co-transfected with vector, GFP-NR2A, GFP-NR2B or GFP-NR2C and Discosoma sp. red fluorescent protein (DsRed). At 6 DIV, the neurons were incubated with NMDA to induce excitotoxicity (Bonfoco et al., 1995
). To closely monitor the effect of NMDA treatment, we used time-lapse microscopy of DsRed-labeled neurons. We found that neurons overexpressing NR2A or NR2B developed cell body swelling and dendritic varicosities within 5 minutes following NMDA treatment ( and Movie S1
). Neurons expressing DsRed alone also showed swelling of the cell body and localized dendritic varicosities, although not as rapidly as neurons expressing NR2A or NR2B. In sharp contrast, NR2C expressing neurons were protected from the NMDA-induced toxicity ( and Movie S1
Fig. 5 NR2C surface expression is neuroprotective. A, Cerebellar granule cells (DIV4) were co-transfected with DsRed and vector, GFP-NR2A, GFP-NR2B or GFP-NR2C. On DIV 6, time-lapse imaging was performed during exposure to NMDA (200 μM). DsRed was used (more ...)
To determine if PKB-dependent trafficking of NR2C played a role in cell survival, we evaluated toxicity in cerebellar granule cells expressing NR2C WT or NR2C S1096A and found no difference in cell survival upon acute NMDA treatment (data not shown). However, we analyzed older cultures (DIV8) and found that neurons expressing NR2C S1096A displayed increased dendritic fragmentation and varicosities compared to NR2C WT, indicating neuronal injury (Hasbani et al., 2001
). Quantitation revealed that cerebellar granule cells expressing NR2C S1096A showed almost double the number of neurons with dendritic varicosities compared to NR2C WT-expressing cells (). This is consistent with NR2C surface expression, and specifically S1096, playing an important role in regulating the cell survival effect of NR2C.