GluA2 palmitoylation by SELENOK modulates AMPAR assembly and synaptic plasticity in Alzheimer's disease

Abstract

Se is essential for central nervous system function, and its deficiency is strongly associated with cognitive decline, especially in neurodegenerative disorders such as Alzheimer's disease (AD). Although Se exerts its effects through selenoproteins, the molecular basis of its neuroprotective action remains unclear. Among selenoproteins, the endoplasmic reticulum (ER)-resident selenoprotein K (SELENOK) is closely linked to cognitive function and therapeutic potential in AD. Here, we examined how SELENOK regulates synaptic plasticity and contributes to Se-mediated neuroprotection in AD. Using age-gradient SELENOK knockout models and palmitoyl-proteomics, we identified GluA2 (formerly GluR2) as a key downstream target. Mechanistically, SELENOK enhanced the activity of DHHC6, an ER-localized palmitoyltransferase, to promote GluA2 palmitoylation, facilitating its ER retention and efficient assembly of AMPA-type glutamate receptors (AMPARs). Notably, GluA2 palmitoylation was reduced in both AD model mice and postmortem brains of patients with AD. Importantly, neuronal overexpression of SELENOK in the hippocampus restored synaptic plasticity and cognitive function in AD mice. Overall, this study uncovers a novel SELENOK-dependent mechanism regulating AMPAR assembly, offering experimental support for developing Se-based therapeutic strategies for AD.

Keywords: AMPAR assembly; Alzheimer's disease; GluA2 palmitoylation; Selenoprotein K; Synaptic plasticity.

Fig. 1

Cognitive and synaptic plasticity deficits in knockout mice at 6 months of age. (a–c) Morris water maze test evaluating spatial learning and memory in wild-type (WT) and knockout (KO) mice at 2, 6, and 10 months of age. (a) Escape latency; (b) time spent in the target quadrant during the 24-h and 72-h probe trials; (c) number of platform crossings during the 24-h and 72-h probe trials. (n = 13–19; sex-balanced) (d–e) Open-field test measuring spontaneous locomotor activity. (d) Grid crossings; (e) Total travel distance. (n = 13–19; sex-balanced) (f–g) Long-term potentiation (LTP) in the hippocampal Schaffer collateral pathway recorded using an MED64 multi-electrode array system. (f) LTP recordings over 80 min; (g) average fEPSP amplitude during the final 10 min after high-frequency stimulation. (n = 6–9; sex-balanced). (h–i) Transmission electron microscopy (TEM) analysis of synapse number and structure. (h) Representative TEM images of synapses and (i) quantification of synapse number. (Scale bar = 2 μm; n = 11–15; four mice per group; sex-balanced) Data are presented as mean ± SEM. (a) Two-way analysis of variance (ANOVA) with Bonferroni post-hoc analysis. Other data were analyzed with Student's t-test. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001.

Fig. 2

SELENOK Modulates Synaptic Plasticity through GluA2 palmitoylation. (a) Western blot (WB) analysis of PSD95, synaptophysin, GluA1, GluA2, GluA3, GluN1, GluN2A, and GluN2B in hippocampal lysates from wild-type (WT) and knockout (KO) mice at 2, 6, and 10 months (n = 5; 3 males and 2 females). (b) qPCR analysis of synapse-related gene expression in the hippocampus of 10-month-old WT and KO mice (n = 5; 3 males and 2 females). (c) Heatmap of differentially palmitoylated proteins related to synaptic plasticity in hippocampi of 10-month-old WT and KO mice (normalized to WT) (n = 3; 2 males and 1 female). (d) Acyl-biotin exchange (ABE) followed by western blotting (ABE-WB) analysis of total and palmitoylated forms of indicated synaptic proteins in 10-month-old WT and KO hippocampi (n = 6; sex-balanced). (e-f) ABE-WB analysis of total protein and palmitoylation levels of GluA2 in hippocampi of 2-month-old (e) and 6-month-old (f) WT and KO mice (n = 6; sex-balanced). Data are presented as mean ± SEM and analyzed using Student's t-test. ∗p < 0.05.

Fig. 3

SELENOK facilitates DHHC6-mediated palmitoylation of GluA2 at Cys836. (a) HEK293 cells were co-transfected with sh-DHHC6 and Flag-tagged GluA constructs (GluA1, GluA2, or GluA3). Total and palmitoylation protein levels were assessed by acyl-biotin exchange (ABE) followed by western blotting (ABE-WB) (n = 6). (b) Co-immunoprecipitation (Co-IP) followed by western blotting (Co-IP/WB) analysis in HEK293 cells co-expressing Flag-GluA2 and HA-DHHC6 to assess protein interaction. (c) In HEK293 cells with adenoviral-mediated SELENOK knockdown (SELENOK-KD) or overexpression (SELENOK-OE), Flag-GluA2 was co-transfected with HA-DHHC6, sh-DHHC6, or control vectors. ABE-WB was used to analyze GluA2 palmitoylation (n = 3–5). (d–e) HEK293 cells were transfected with either an empty vector or HA-DHHC6, together with wild-type Flag-GluA2 (CC), single-point mutants (C610S and C836S), or the double mutant (C610S/C836S, referred to as SS). ABE-WB was performed to assess GluA2 palmitoylation (n = 3). Data are presented as mean ± SEM and analyzed using Student's t-test. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001.

Fig. 4

SELENOK is involved in AMPAR complex assembly. (a) Co-immunoprecipitation (Co-IP)/western blotting (WB) analysis of GluA1–GluA2 and GluA2–GluA3 interactions in the hippocampi of 6-month-old WT and KO mice (n = 6; sex-balanced). (b) Immunofluorescence (IF) and structured illumination microscopy (SIM) imaging to visualize the co-localization of GluA1 (red) and GluA2 (green) in hippocampal sections, with colocalized spots appearing yellow (scale bar = 8 μm). Quantification was performed using the Imaris 3D reconstruction (scale bar = 10 μm; n = 3; 7 non-overlapping, equal-area regions per mouse; 2 males and 1 female per group) (c) Co-IP/WB analysis of Flag-GluA2 interactions with HA-GluA1 or HA-GluA3 in SELENOK-KO N2a cells (n = 5–6). (d) Proximity ligation assay (PLA) quantifying GluA1–GluA2 interactions in SELENOK-KO N2a cells, with PLA signals shown in red (scale bar = 10 μm; n = 40). (e) Co-IP/WB analysis of GluA1–GluA2 interactions in primary neurons with SELENOK knockdown (SELENOK-KO) or overexpression (SELENOK-OE) (n = 6). (f) IF and SIM imaging of GluA1–GluA2 colocalization in primary neurons under SELENOK-KO or SELENOK-OE conditions (scale bar = 10 μm; n = 7 fields from 3 slices) Data are presented as mean ± SEM and analyzed using Student's t-test. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001.

Fig. 5

SELENOK regulates ER-localized GluA2 via palmitoylation to regulate AMPAR assembly. (a) ER fractions were isolated from the hippocampi of 6-month-old wild-type (WT) and knockout (KO) mice using an ER isolation kit, followed by WB analysis of ER-localized GluA2 (n = 8; sex-balanced). (b) SELENOK-KO N2a cells exogenously expressing Flag-GluA2 were subjected to ER fractionation, and ER-localized GluA2 levels were analyzed by WB (n = 5). (c) N2a cells were co-transfected with HA-GluA1 and Flag-GluA2 (WT, SS, or ER-SS). GluA1–GluA2 interactions were assessed by co-immunoprecipitation and western blotting (Co-IP/WB), and ER-localized GluA2 levels analyzed by WB after ER fractionation (n = 6 or 3). (d–e) Immunofluorescence (IF) and structured illumination microscopy (SIM) imaging of N2a cells expressing Flag-GluA2-WT or Flag-GluA2-SS, showing colocalization with the ER marker calnexin (colocalized spots are shown in yellow; scale bar = 8 μm; n = 18). (f) N2a cells were co-transfected with HA-GluA1 and ER-targeted Flag-GluA2 (ER-WT or ER-SS), followed by Co-IP/WB to assess GluA1–GluA2 interactions and WB analysis of ER-localized GluA2 levels after ER fractionation (n = 6 or 3). (g–h) N2a cells were co-transfected with HA-GluA1 and either WT or SS Flag-GluA2 along with a SELENOK overexpression plasmid. GluA1–GluA2 interactions were analyzed by CO-IP/WB. (h) Quantification of interactions in WT and SS groups from SELENOK-overexpressing cells (g) and corresponding controls from (c). (n = 4). Data are presented as mean ± SEM and analyzed using Student's t-test. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001.

Fig. 6

The AD Brain exhibits reduced GluA2 palmitoylation and impaired AMPAR assembly. (a) Acyl-biotin exchange-western blotting (ABE-WB) analysis of GluA2 palmitoylation and total protein levels of GluA2 and DHHC6 in the hippocampi from 6-month-old wild-type (WT) and 5 × FAD transgenic mice (n = 6; sex-balanced). (b) Co-immunoprecipitation (Co-IP)/WB analysis of GluA1–GluA2 interactions in the hippocampi from 6-month-old WT and 5 × FAD transgenic mice (n = 6; sex-balanced). (c) ABE-WB analysis of GluA2 palmitoylation and total protein levels in postmortem cortical tissues from normal controls and patients with AD. (n = 4) (d) Immunofluorescence (IF) and structured illumination microscopy (SIM) imaging to assess GluA1–GluA2 colocalization in hippocampal sections. GluA1 (red), GluA2 (green), and co-localized signals (yellow) (scale bar = 8 μm). Quantification was performed using Imaris 3D reconstruction (scale bar = 10 μm; n = 4 controls, 3 patients with AD; 5 non-overlapping, equal-area regions per section). Data are presented as mean ± SEM and analyzed using Student's t-test. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001.

Fig. 7

Neuronal SELENOK overexpression restores AMPAR function and cognitive impairment in 5 × FAD transgenic mice. (a) Schematic of the AAV2/B10 vector encoding SELENOK and experimental timeline for virus injection and behavioral testing. (b) Representative fluorescence image of enhanced green fluorescent protein (EGFP)-labeled adeno-associated virus (AAV) expression in brain tissue (scale bar = 1000 μm). (c) Quantitative real-time PCR (qPCR) analysis of SELENOK expression in the hippocampi of AD and AD-OE mice (n = 6; sex-balanced). (d) Morris water maze performance in 5 × FAD and AD-OE mice. Left: Escape latency; Middle: Time spent in the target quadrant during 24 h and 72 h post-training; Right: Number of platform crossings during 24 h and 72 h probe trials (n = 11 or 19; sex-balanced). (e) Open-field test evaluating spontaneous locomotor activity. Left: Grid crossings; Right: Rearing events (n = 11 or 19; sex-balanced). (f) Long-term potentiation (LTP) recordings in the hippocampal Schaffer pathway using the MED64 system. Left: Field excitatory postsynaptic potential (fEPSP) amplitude over 80 min; Right: Average fEPSP amplitude during the final 10 min after high-frequency stimulation (n = 6; sex-balanced). (g) Acyl-biotin exchange-western blotting (ABE-WB) analysis of GluA2 palmitoylation and total protein levels in the hippocampus (n = 6; sex-balanced). (h) Co-IP/WB analysis of GluA1–GluA2 interactions in hippocampal lysates (n = 6; sex-balanced). (i) Immunofluorescence (IF) and structured illumination microscopy (SIM) imaging of GluA1–GluA2 colocalization in the CA1, CA3, and DG regions (scale bar = 8 μm). (j) Quantification of colocalization using Imaris 3D reconstruction. GluA1 (red), GluA2 (green), colocalized (yellow) (scale bar = 10 μm; n = 4; 5 non-overlapping, equal-area regions per section). (k) WB analysis of ER-localized GluA2 in hippocampal samples following ER fractionation (n = 6; sex-balanced). Data are presented as mean ± SEM. Data from (d, left) were analyzed using Two-way ANOVA and Bonferroni post-hoc analyses were used for data from panel (d, left). Student's t-test was used for all other comparisons. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001.

Fig. 8

Proposed mechanism by which SELENOK regulates AMPAR assembly through GluA2 palmitoylation. SELENOK interacts with the palmitoyl acyltransferase DHHC6 in the ER, promoting DHHC6-mediated palmitoylation of GluA2. Palmitoylated GluA2 is retained in the ER and sorted into a GluA2 pool, which efficiently assembles with GluA1 and GluA3 subunits to form AMPAR complexes. These complexes are subsequently trafficked to the postsynaptic membrane, where they support synaptic transmission and plasticity. SELENOK-dependent GluA2 palmitoylation thus constitutes a key regulatory step in AMPAR assembly, trafficking, and the maintenance of excitatory synaptic function.