ADAR2-mediated Q/R editing of GluA2 in homeostatic synaptic plasticity

Abstract

Homeostatic synaptic plasticity is a negative feedback mechanism through which neurons modify their synaptic strength to counteract chronic increases or decreases in activity. In response to activity deprivation, synaptic strength is enhanced by increasing the number of AMPA receptors (AMPARs), particularly Ca²⁺-permeable AMPARs, at the synapse. Here, we found that this increase in Ca²⁺-permeable AMPARs during homeostatic upscaling was mediated by decreased posttranscriptional editing of GRIA2 mRNA encoding the AMPAR subunit GluA2. In cultured neurons, activity deprivation resulted in increases in the amount of unedited GluA2, such that its ion channel pore contains a glutamine (Q) codon instead of arginine (R), and in the number of Ca²⁺-permeable AMPARs at the synapse. These effects were mediated by a splicing factor-dependent decrease in ADAR2 abundance and activity in the nucleus. Overexpression of ADAR2 or CRISPR-Cas13-directed editing of GluA2 transcripts blocked homeostatic upscaling in activity-deprived primary neurons. In mice, dark rearing resulted in decreased Q-to-R editing of GluA2-encoding transcripts in the primary visual cortex (V1), and viral overexpression of ADAR2 in the V1 blocked the induction of homeostatic synaptic plasticity. The findings indicate that activity-dependent regulation of GluA2 editing contributes to homeostatic synaptic plasticity.

Figure 1.. Activity deprivation increases synaptic GluA1 in a CP-AMPAR–dependent manner.

(A and B) DIV15 primary hippocampal neurons were treated with TTX/APV with or without PhTx for the indicated time (0.5 to 24 hours), then stained for N-terminal GluA1 and PSD-95 under non-permeant conditions. The amount of cell-surface GluA1 was quantified and normalized to the control. Data are means ± SEM from n = 21 to 25 cells pooled from 3 independent experiments. Scale bars, 5μm. (C and D) Synaptosomal fractionation from DIV15 primary cortical neurons treated with TTX/APV with or without PhTx for the indicated time (0.5 to 48 hours), then blotted for GluA1 and PSD-95. The amount of synaptosomal GluA1 was quantified by band intensity and normalized to both the loading control and experimental control. Data are means ± SEM from n = 3 to 4 independent experiments. Data in (B and D) were analyzed with two-way ANOVA with Tukey correction: ns = p > 0.05, and ****p < 0.0001.

Figure 2.. Activity deprivation increases expression of unedited GluA2(Q) and GluA2(Q)-containing CP-AMPARs.

(A and B) DIV15-16 cortical neurons were treated with TTX/APV for varied times (0.5 to 24 hours) and subjected to Sanger sequencing at the GluA2 mRNA Q/R editing site. Unedited GluA2 was quantified by the relative expression of adenosine at the editing site (CGG). Data are means ± SEM from n = 4 to 9 independent experiments. (C) DIV15-16 cortical neurons were treated with TTX/APV for 0.5 to 12 hours and GluA2Q and GluA2R mRNA levels were quantified with qPCR. Unedited GluA2 was quantified by the relative expression of GluA2Q and GluA2R. Data are means ± SEM from n = 3 to 5 independent experiments. (D and E) DIV10 cortical neurons were infected with LV-ADAR2 and treated with TTX/APV on DIV15 for 2 hrs before Sanger sequencing of GluA2. Unedited GluA2 was quantified by the relative expression of adenosine at the editing site (CGG). Data are means ± SEM from n = 3 independent experiments. (F) DIV10 hippocampal neurons were infected with empty vector or LV-ADAR2, then treated with TTX/APV on DIV15 and subjected to cobalt staining. Scale bar, 50μm. (G to I) The number of cobalt-positive cells was quantified by counting the number of cells in a single image that expressed signal above a set threshold. Data are means ± SEM from n = 15 to 22 images pooled from 3 independent experiments) and the intensity of the cobalt signal in both the soma and dendrite (Data are means ± SEM from n = 20 to 40 cells pooled from 3 independent experiments) were quantified. Analyzed with student’s t-test or one-way ANOVA with Dunnett correction. ns = p > 0.05, *p < 0.05, ** p < 0.01, ***p < 0.001, ****p < 0.0001.

Figure 3.. Increased GluA2Q expression is necessary for homeostatic increase in AMPAR synaptic accumulation and synaptic insertion after activity deprivation.

(A and B) Primary hippocampal neurons were infected with empty vector or LV-ADAR2 on DIV10, treated with 24 hrs TTX/APV on DIV15, and fixed for immunocytochemical staining of N-terminal GluA1. The amount of cell-surface GluA1 was quantified and normalized to the control. Data are means ± SEM from n = 15 to 30 cells pooled from 3 independent experiments. Scale bar, 50μm (full cell), 5μm (dendrite). (C and D) DIV7 primary hippocampal neurons were co-transfected with GFP and a control plasmid, ADAR2, or CRISPR-Cas13-ADAR2_DD+ GluA2-targeting gRNA. Cells were untreated (C) or treated with TTX/APV (T) on DIV15 and stained for N-terminal GluA1. The amount of cell-surface GluA1 was quantified and normalized to the control. Data are means ± SEM from n = 12 to 20 cells pooled from 4 independent experiments). Scale bar, 5μm. (E to F) Primary hippocampal neurons were infected with empty vector or LV-ADAR2 on DIV10 and treated with TTX/APV on DIV15 before undergoing an insertion assay to measure GluA1 insertion. Inserted GluA1 was measured by fluorescence intensity and normalized to the control. Data are means ± SEM from n = 26 to 33 cells pooled from 3 independent experiments) Scale bar, 50μm (full cell), 5μm (dendrite). Analyzed with one-way or two-way ANOVA with Tukey or Bonferroni correction. ns = p < 0.05, *p < 0.05, **p < 0.01.

Figure 4.. Activity deprivation results in decreased nuclear and cytosolic ADAR2 expression.

(A and B) DIV15 primary cortical neurons were treated with TTX/APV for varied times (0.5 to 24 hours), and the total lysate was collected for western blot analysis from 0.5 to 12 hours, and expression was rescued at 24 hours. Protein expression was quantified by band intensity and normalized to both the loading control and experimental control. Data are means ± SEM from n = 4 independent experiments. (C and D) DIV15 primary hippocampal neurons were treated with TTX/APV for varied times (2 to 12 hours), fixed and immunostained for ADAR2. ADAR2 protein expression was quantified by fluorescent intensity and normalized to the control. Data are means ± SEM from n = 17 to 31 cells pooled from 3 independent experiments). Scale bar, 50μm. (E to G) Primary cortical neurons were treated with TTX/APV for varied times (0.5 - 12 hours) at DIV15 then subjected to nuclear fractionation and western blot analysis of (F) nuclear and (G) cytosolic ADAR2 expression. Protein expression was quantified by band intensity and normalized to both the loading control and experimental control. Data are means ± SEM from n = 3 to 4 independent experiments). Analyzed with one-way ANOVA with Bonferroni correction. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Figure 5.. Neuronal inactivity-induced ADAR2 ubiquitination is not involved in GluA2 editing.

(A and B) DIV15 primary cortical neurons were treated with TTX/APV for varied times (0.5 to 12 hours). Cells were lysed in the presence of deubiquitinase inhibitors and ADAR2 was immunoprecipitated (IP) before being used for western blot analysis. ADAR2 ubiquitination was quantified by band intensity and normalized to both immunoprecipitated ADAR2 and to the experimental control. Data are means ± SEM from n = 6 to 7 independent experiments. (C and D) DIV15 primary cortical neurons were treated with TTX/APV for varied times (0.5 to 12 hours) and total lysate was collected for western blot analysis. Protein expression was quantified by band intensity and normalized to both the loading control and experimental control. Data are means ± SEM from n = 3 independent experiments. (E to G) DIV15 primary cortical neurons were treated with TTX/APV with or without MG132 for 12 hrs and subject to nuclear fractionation for western blot analysis. Protein expression was quantified by band intensity and normalized to both the loading control within each fraction and to the experimental control. Data are means ± SEM from n = 3 to 4 independent experiments. (H and I) DIV15 primary cortical neurons were treated with TTX/APV with or without MG132 for 12 hrs and collected for Sanger sequencing of GluA2. Unedited GluA2 was quantified by the relative expression of adenosine at the editing site (CGG). Data are means ± SEM from n = 4 independent experiments. Analyzed with one-way or two-way ANOVA with Dunnett correction. ns = p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 6.. Inactivity-caused up-regulation of SRSF9 is necessary for HSP induction.

(A to C) DIV15 primary cortical neurons were treated with TTX/APV for varied times (0.5 to 12 hours) then the nuclear fraction was collected for western blot analysis. Protein expression was quantified by band intensity and normalized to both the loading control and experimental control. Data are means ± SEM from n = 3 to 5 independent experiments. (D) HEK293T cells were infected with viral shSRSF9 and collected for western blot analysis. (E) DIV7 primary cortical neurons were infected with viral shSRSF9 and then treated with TTX/APV on DIV15 for 12 hours before being collected for Sanger sequencing of GluA2. Unedited GluA2 was quantified by the relative expression of adenosine at the editing site (CGG). Data are means ± SEM from n = 3 independent experiments. (F-G) DIV7 primary cortical neurons were infected with viral shSRSF9 and collected for nuclear fractionation for western blot analysis at DIV15 after a 12-hour TTX/APV treatment. Protein expression was quantified by band intensity and normalized to both the loading control and experimental control. Data are means ± SEM from n = 4 to 5 independent experiments. (H-I) Primary hippocampal neurons were infected with shSRSF9 virus at DIV 7 and incubated with TTX/APV at DIV 15 for 24 hours. Cells were fixed for immunostaining of surface GluA1. The amount of cell-surface GluA1 was quantified and normalized to the control. Data are means ± SEM from n = 23 to 27 cells pooled from 3 independent experiments. Scale bar, 50 μm. Analyzed with one-way or two-way ANOVA with Dunnett or Tukey correction. ns = p > 0.05, *p < 0.05, **p < 0.01, ****p < 0.0001.

Figure 7.. Dark rearing induces homeostatic accumulation of synaptic GluA1 accompanied by changes in ADAR2 and increased levels in GluA2Q in the primary visual cortex (V1).

(A and B) P45-60 mice were dark reared for 48 hours before their visual cortices were collected and subjected to synaptosomal fractionation for western blot analysis. Protein expression was quantified by band intensity and normalized to both the loading control and experimental control. Data are means ± SEM from n = 5 to 6 animals. (C and D) Visual cortices of dark reared P45-60 mice were collected for Sanger sequencing of GluA2. Unedited GluA2 was quantified by the relative expression of adenosine at the editing site (CGG). Data are means ± SEM from n = 4 to 5 animals. (E and F) Visual cortices were collected from P45-60 mice following 48 hrs dark rearing and the nuclear fractions were prepared for western analysis of ADAR2. Protein expression was quantified by band intensity and normalized to both the loading control and experimental control. Data are means ± SEM from n = 4 to 5 animals. (G and H) Nuclear fractions from visual cortices of dark reared mice were used for western analysis of SRFS9. Protein expression was quantified by band intensity and normalized to both the loading control and experimental control. Data are means ± SEM from n = 3 animals. Analyzed with student’s t-test. *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 8.. Increased GluA2Q expression is necessary for HSP induction in V1 after dark rearing.

(A and B) P2 mice were intraventricularly injected with viral ADAR2 and subjected to dark rearing between P45-60 before collection of visual cortices for nuclear fractionation. Protein expression was quantified by band intensity and normalized to both the loading control and experimental control. Data are means ± SEM from n = 4 animals. (C and D) Viral ADAR2-infected mice were subject to dark rearing at P45-60 before visual cortices were collected for Sanger sequencing of GluA2. Unedited GluA2 was quantified by the relative expression of adenosine at the editing site (CGG). Data are means ± SEM from n = 4 animals. (E and F) P45-60 mice infected with viral ADAR2 were dark reared before the visual cortices were collected for synaptosomal fractionation. Western blots were probed with different antibodies. Protein expression was quantified by band intensity and normalized to both the loading control and experimental control. Data are means ± SEM from n = 4 animals. (G) Timeline of dark rearing for P45-60 Fos^CreER mice. After dark rearing, mice had 1 hour of light exposure before i.p. injection of tamoxifen. Mice were returned to normal rearing conditions (12-hour light/dark cycle) before brains were collected to examine tdTomato expression. (H and I) Fos^CreER mice were infected with sham or ADAR2 virus and then subjected to dark rearing. Visual cortices were imaged for tdTomato expression. Data are means ± SEM from n = 4 to 5 animals. Scale bar, 200 μm. Analyzed with one-way or two-way ANOVA with Tukey or Dunnett correction. ns = p > 0.05, *p < 0.05, ***p < 0.001.