Nedd4-2-dependent regulation of astrocytic Kir4.1 and Connexin43 controls neuronal network activity

Abstract

Nedd4-2 is an E3 ubiquitin ligase in which missense mutation is related to familial epilepsy, indicating its critical role in regulating neuronal network activity. However, Nedd4-2 substrates involved in neuronal network function have yet to be identified. Using mouse lines lacking Nedd4-1 and Nedd4-2, we identified astrocytic channel proteins inwardly rectifying K+ channel 4.1 (Kir4.1) and Connexin43 as Nedd4-2 substrates. We found that the expression of Kir4.1 and Connexin43 is increased upon conditional deletion of Nedd4-2 in astrocytes, leading to an elevation of astrocytic membrane ion permeability and gap junction activity, with a consequent reduction of γ-oscillatory neuronal network activity. Interestingly, our biochemical data demonstrate that missense mutations found in familial epileptic patients produce gain-of-function of the Nedd4-2 gene product. Our data reveal a process of coordinated astrocytic ion channel proteostasis that controls astrocyte function and astrocyte-dependent neuronal network activity and elucidate a potential mechanism by which aberrant Nedd4-2 function leads to epilepsy.

Figure 1.

γ-oscillations in acute slices from Nedd4 brain-specific knockout mice. (A–D) Reduced power of γ-oscillations in Nedd4-1^f/f;Nedd4-2^f/f;EMX-Cre (N4-1/2 bDKO) mice. Representative recordings in CA3 hippocampal regions of acute brain slices from Nedd4-1^f/f;Nedd4-2^f/f (N4-1/2 CTL) (A) and N4-1/2 bDKO (B) mice before (baseline) and during (Kainate) induction of γ-oscillations with 100 nM kainate. Average powers (C) and frequencies (D) of γ-oscillations in N4-1/2 CTL (black dots) and N4-1/2 bDKO (green dots). (E–H) Reduced power of γ-oscillations in Nedd4-2^f/f;EMX-Cre (N4-2 bKO) mice. Representative recordings in the CA3 region of hippocampus slices from Nedd4-2^f/f (N4-2 CTL) (E) and N4-2 bKO (F) mice before (Baseline) and during (Kainate) the application of 100 nM kainate. Average powers (G) and frequencies (H) of γ-oscillations in N4-2 CTL (black dots) and N4-2 bKO (green dots). Different tones and shapes of dots in dot plots in C, D, G, and H represent data from different mice. Numbers of recorded slices (n) and the animal number (N); (C and D), n = 16 and N = 3 for N4-1/2 CTL, n = 21 and N = 3 for N4-1/2 bDKO; (G and H), n = 29 and N = 6 for N4-2 CTL, n = 25 and N = 5 for N4-2 bKO. Results are shown as mean ± SEM. **, 0.001 < P < 0.01; *, 0.01 < P < 0.05; ns, 0.05 < P (two-tailed nested t test). Data distribution was assumed to be normal but this was not formally tested. See also Fig. S1 and Table S2.

Figure S1.

Intact γ-oscillations in acute slices from Nedd4-1 and Nedd4-2 neuron-specific knockout (N4-1/2 nDKO) and Nedd4-1 brain-specific knockout (N4-1 bKO) mice. (A and B) Representative recordings in CA3 hippocampal regions of acute brain slices from N4-1/2 CTL (A) and N4-1/2 nDKO (B) mice before (baseline) and during (Kainate) induction of γ-oscillations with 100 nM kainate. (C and D) Average powers (C) and frequencies (D) of γ-oscillations in N4-1/2 CTL (black dots) and N4-1/2 nDKO (green dots). (E–H) Intact power of γ-oscillations in Nedd4-1^f/f;EMX-Cre (N4-1 bKO) mice. Representative recordings in CA3 region of hippocampus slices from Nedd4-1^f/f (N4-1 CTL) (E) and N4-1 bKO (F) mice before (Baseline) and during (Kainate) the application of 100 nM kainate. Average powers (G) and frequencies (H) of γ-oscillations in N4-1 CTL (black dots) and N4-1 bKO (green dots). Different tones and shapes of dots in dot plots in C, D, G, and H represents data from different mice. Numbers of recorded slices (n) and mice (N); (C and D), n = 19 and N = 4 for N4-1/2 CTL, n = 20 and N = 4 for N4-1/2 nDKO; (G and H), n = 16 and N = 4 for N4-1 CTL, n = 16 and N = 4 for N4-1 bKO. Results are shown as mean ± SEM. ns, 0.05 < P (two-tailed Student’s t test). Data distribution was assumed to be normal, but this was not formally tested. See also Table S2.

Figure 2.

Polyubiquitin chain formation activities of Nedd4-2 WT, S233L, and H515P mutants. (A) Time course of ubiquitination using recombinant wild-type Nedd4-2 (Nedd4-2 WT). Purified Ub was incubated with ATP, E1, E2, and Nedd4-2 WT for indicated durations. Samples were subjected to Western blotting using anti-Ub (top panel) and anti-Nedd4-2 (bottom panel) antibodies. Note the time-dependent polyUb chain formation at the expense of free Ub (arrowhead). (B) The average time course of the depletion of free Ub (left) and formation of polyUb chains (right) in in vitro ubiquitination assay in A (n = 3 replicates). (C) Scheme of human Nedd4-2 and epileptic missense mutants. (D–G) S233L (D and E) and H515P (F and G) missense point mutants of Nedd4-2 cause gain-of-function of the catalytic activity. (D and F) Representative images of Western blotting using anti-Ub (upper panel) and anti-Nedd4-2 (lower panel) antibodies for in vitro ubiquitination assay samples using Nedd4-2 WT and mutants (S233L in D and H515P in F). Arrowheads, free Ub. (E and G) Quantifications of relative free Ub (top dot plots) and formation of polyUb chains relative to N4-2 WT (bottom dot plots) after 150 s incubation. Nedd4-2 WT (black dots) and Nedd4-2 mutants (magenta dots). Results are shown as mean ± SEM. The number of experiments are four for each set of experiments. ***, P < 0.001; **, 0.001 < P < 0.01; *, 0.01 < P < 0.05 (two-tailed Student’s t test). Data distribution was assumed to be normal, but this was not formally tested. See also Table S2.

Figure 3.

Screening for proteins upregulated in Nedd4-1/2 brain-specific knockout mice. (A) Specificity of anti-GluA1 antibody verified with cortical homogenates from CTL and GluA1 KO. (B and C) Quantitative Western blots with the anti-GluA1 antibody using cortical homogenates from N4-1/2 CTL and N4-1/2 bDKO mice. In C, black dots, N4-1/2 CTL; green dots, N4-1/2 bDKO. (D) Relative Nedd4-2 protein levels in the cortex and hippocampus in wild-type mice. (E) Protein profiles of PSD95, RabGDI, Nedd4-2, EAAT2, Kir4.1, and Cx43 in subcellular fractionated samples from wild-type mice. H, homogenate; S, soluble; CSS, crude synaptosome; SC/CSV, synaptic cytoplasm/crude synaptic vesicle; CSM, crude synaptic membrane; and SM, pure synaptic membrane fractions. (F) Scatter plot of relative protein abundance as quantified by mass spectrometry. The log₂-transformed fold-change ratios between N4-1/2 bDKO and N4-1/2 CTL in the forward (y-axis) against reverse (x-axis) experiments were plotted. Black circles indicate proteins significantly changed in both experiments (significance B values <0.05). Color filling indicates proteins consistently upregulated (magenta) or downregulated (green) in the same direction in both experiments. (G) Representative Western blotting of cortical brain lysates from N4-1/2 CTL and N4-1/2 bDKO mice. Faint bands crossreacting with the anti-Nedd4-1 antibody in N4-1/2 bDKO samples in B and G are likely from cell types without Cre-expression (i.e., inhibitory neurons, blood cells, blood vessel cells, or microglia cells). (H and I) Levels of Kir4.1 (H) and Cx43 (I) in N4-1/2 CTL (black dots) and N4-1/2 bDKO (green dots). (J and K) Quantitative Western blotting using the anti-EAAT2 antibody showed no difference between cortical lysates from N4-1/2 CTL (black dots in K) and N4-1/2 bDKO (green dots in K). (L) Representative images of S100 β− and Draq5-stained CA3 regions of hippocampi from the control and N4-1/2 bDKO mice. Scale bars, 50 μm. (M) The number of S100β-positive cells (green channel in L) normalized to the number of Draq5-positive particles (magenta channel in L) showed no difference. (N and O) Kir4.1 (N) and Cx43 (O) mRNA levels were not significantly different in N4-1/2 CTL (black dots) and N4-1/2 bDKO (green dots). (P–U) Protein levels of Kir4.1 and Cx43 in Nedd4-1^f/f; EMX-Cre (N4-1 bKO) and Nedd4-1^f/f (N4-1 CTL) (P–R), and N4-2 bKO and N4-2 CTL (S–U) cortical lysates. Results are shown as mean ± SEM. Numbers of mice (n); (C), n = 5 for each genotype; (H and I), n = 3 for each genotype; (K), n = 5 for each genotype; (M), n = 5 for N4-1/2 CTL, n = 4 for N4-1/2 bDKO; (N), n = 9 for N4-1/2 CTL, n = 6 for N4-1/2 bDKO; (O), n = 3 for each genotype; (Q and R), n = 4 for each genotype; (T and U), n = 3 for each genotype. ***, P < 0.001; **, 0.001 < P < 0.01; *, 0.01 < P < 0.05; ns, 0.05 < P. Two-tailed Student’s t test. Data distribution was assumed to be normal, but this was not formally tested. See also Tables S1 and S2.

Figure 4.

Biochemical characterization of Kir4.1 and Cx43 as substrates of Nedd4-2. (A and B) Kir4.1-HA (A) or Cx43-HA (B) was expressed in HEK293FT cells in the presence or absence of EGFP-tagged Nedd4 E3s. Levels of Nedd4 E3s were studied by an anti-GFP antibody (top panels). HA-tagged substrates were immunoprecipitated (IP) with an anti-HA antibody and subjected to Western blotting using anti-HA (second panel) and anti-Ub (third panel) antibodies. Note that increased smear signals cross-reacting the anti-Ub antibody when EGFP-Nedd4-1 (first lane) or EGFP-Nedd4-2 (second lane) were coexpressed. Lanes in the second and third panels in A were run on the same gel but were noncontiguous. Patterns of anti-Ub Western blotting differ between (A and B) because different anti-Ub antibodies were used. The same samples were blotted with anti-K48-linked and anti-K63-linked polyUb chain antibodies (bottom two panels in A and B). Equal amounts of K48- and K63-linked tetra Ub chains were loaded in the right two lanes for SDS-PAGE together with ubiquitination assay samples. Note that signals from K48-linked and K63-linked tetra Ub chains in the bottom two blots are comparable, indicating that the anti-K48 and anti-K63 antibodies have almost the same titers. (C) EGFP-Nedd4-2 with one of the missense mutations found in the epileptic patients was used for the HEK293FT cell-based ubiquitination assays. H536P corresponds to human H515P. (D and E) N4-1/2 CTL (EMX-Cre −) and N4-1/2 bDKO (EMX-Cre +) astrocytes were infected with lentivirus expressing Kir4.1-HA (D) or Cx43-HA (E). Immunoprecipitated HA-tagged proteins were analyzed by Western blotting with anti-Ub (upper panel) and anti-HA (lower panel) antibodies. Images are representative of at least two independent experiments. (F) Domain structure of Nedd4-2 (accession no. NM_001114386). The amino acid sequences covered by truncated mutants of Nedd4-2 are indicated. (G) Affinity purification experiment using purified GST-tagged Nedd4 E3s with Kir4.1-HA (top panels) and Cx43-HA (middle panels) expressed in and extracted from HEK293FT cells. Immobilized GST-tagged proteins are stained with Ponceau (bottom panels). More Kir4.1-HA and Cx43-HA bound to GST-Nedd4-2 than to GST-Nedd4-1 (third and fourth lanes in the left top and the left middle Western blotting panels; bottom bar diagrams). Images are representative of at least two independent experiments. (H) Rescue of Kir4.1 (third Western blotting panel) and Cx43 (fourth Western blotting panel) levels in N4-1/2 bDKO astrocytes (EMX-Cre +) over the control (EMX-Cre −) by re-expressing recombinant wild-type Nedd4-2 (rec.N4-2 WT) but not by the inactive mutant of Nedd4-2 (rec.N4-2 C/S). In dot plots, black dots, control astrocytes; green dots, N4-1/2 bDKO astrocytes; gray dots, N4-1/2 bDKO astrocytes expressing recombinant rec.N4-2 WT; magenta dots, N4-1/2 bDKO astrocytes expressing rec.N4-2 C/S. (I) Upregulation of Kir4.1 (third Western blotting panel) and Cx43 (fourth Western blotting panel) by blocking endocytosis using dynasore in cultured N4-1/2 CTL astrocytes (EMX-Cre −) but not in N4-1/2 bDKO astrocytes (EMX-Cre +). In dot plots, black dots, control astrocytes treated with vehicle; green dots, control astrocytes treated with 100 μM dynasore; gray dots, N4-1/2 bDKO astrocytes treated with vehicle; magenta dots, N4-1/2 bDKO astrocytes treated with 100 μM dynasore. Results are shown as mean ± SEM. The bar diagrams in G, n = 4 for Kir4.1-HA binding assay and n = 5 for Cx43-HA binding assay; the bot plot for Kir4.1 in H, n = 5 for each assay point; other plots, n = 4 for each assay point. ***, P < 0.001; **, 0.001 < P < 0.01; *, 0.01 < P < 0.05; no asterisk, 0.05 < P (one-way ANOVA with Tukey’s post-hoc test). Data distribution was assumed to be normal, but this was not formally tested. See also Tables S2.

Figure 5.

Kir4.1-dependent increase in membrane conductance in N4-2 bKO astrocytes. (A–C) Example current traces in N4-2 CTL astrocytes (A), N4-2 bKO astrocytes (B), and N4-2 bKO astrocytes treated with fluoxetine (Fluo) (C). Cells were voltage clamped at −80 and +10 mV voltage steps were applied from −120 to +40 mV during recording currents. (D–F) Quantifications of rescues of defects in N4-2 bKO by fluoxetine. (D) Voltage–current plots from N4-2 CTL (black trace), N4-2 bKO (green trace), and N4-2 bKO treated with fluoxetine (gray trace). (E and F) The current (E) and membrane resistance (F) at the membrane potential of −120 mV in N4-2 bKO (green dots) astrocytes were decreased as compared with N4-2 CTL astrocytes (black dots). Decreased currents (E) and membrane resistance (F) were restored to control levels by the application of Fluo in N4-2 bKO astrocytes (gray dots). Results are shown as mean ± SEM. Numbers of recorded cells (n) in D–F; n = 20 for N4-2 CTL, n = 26 for N4-2 bKO, n = 18 for N4-2 bKO + Fluo. **, 0.001 < P < 0.01; *, 0.01 < P < 0.05; no asterisk, 0.05 < P. (One-way ANOVA with Tukey’s post-hoc test for E and F). Data distribution was assumed to be normal, but this was not formally tested. See also Fig. S2 and Table S2.

Figure S2.

The lack of impacts of fluoxetine on membrane currents and γ-oscillations in Nedd4-2^f/f (N4-2 CTL). (A and B) Example current traces in N4-2 CTL astrocytes without (A) and with fluoxetine (B). Cells were voltage clamped at −80 and +10 mV voltage steps were applied from −120 to +40 mV during recording currents. (C and D) Quantifications of membrane currents in N4-2 CTL without and with fluoxetine. (C) Voltage–current plots from N4-2 CTL (black trace) and N4-2 CTL with fluoxetine (red trace). (D) The current at the membrane potential of −120 mV showed no effects of fluoxetine in N4-2 CTL. (E and F) Representative recordings in CA3 hippocampal regions of acute brain slices from N4-2 CTL (E) and N4-2 CTL with fluoxetine (F) mice before (baseline) and during (Kainate) induction of γ-oscillations with 100 nM kainate. (G and H) Average powers (G) and frequencies (H) of γ-oscillations in N4-2 CTL (black dots) and N4-2 CTL with fluoxetine (red dots). Results are shown as mean ± SEM. Numbers of recorded cells (n) in D; n = 6 for N4-2 CTL, n = 6 for N4-2 CTL + Fluo. Numbers of recorded slices (n); (G and H), n = 6 for N4-2 CTL, n = 5 for N4-2 CTL + Fluo. Results are shown as mean ± SEM. ns, 0.05 < P (two-tailed Student’s t test). Data distribution was assumed to be normal, but this was not formally tested. See also Table S2.

Figure 6.

Augmented astrocyte coupling in the absence of Nedd4 E3s. (A) Representative images of FRAP live imaging experiments in N4-1/2 CTL (top panels) and N4-1/2 bDKO astrocytes (bottom panels). Primary astrocytes prepared from N4-1/2 CTL and N4-1/2 bDKO mice were loaded with calcein-AM. Calcein in the astrocytes indicated with arrows was bleached, and FRAP was recorded. Scale bars, 20 μm. (B and C) Time courses of FRAP at bleached cells in A for N4-1/2 CTL (B) and N4-1/2 bDKO (C) astrocytes. (D) Average FRAP time courses of N4-1/2 CTL (black trace) and N4-1/2 bDKO (green trace) astrocytes. (E) Average fluorescence recoveries at 300 s after bleaching. N4-1/2 bDKO showed a significant increase in the recovery of calcein fluorescence over N4-1/2 CTL. Results are shown as mean ± SEM. Numbers of imaged cells (n); n = 24 for N4-1/2 CTL, n = 23 for N4-1/2 bDKO. **, 0.001 < P < 0.01 (two-tailed Student’s t test). Normality of the distribution of was confirmed with the Kolmogorov–Smirnov test. See also Table S2.

Figure 7.

Characterization of astrocyte-specific N4-2 conditional KO. (A) Aldh1l1-CreERT2 mouse crossed with tdTomato-expressing Cre indicator mouse (Aldh1l1-CreERT2;ROSA26-tdTom) was injected with tamoxifen to induce Cre recombination with the same protocol used for Nedd4-2^f/f;Aldh1l1-CreERT2 (N4-2 AstKO) mice. The hippocampal CA3 region was immunostained for an astrocyte marker S100β. Scale bar, 200 μm. (B) The efficiency of Cre recombination in astrocytes. 88.5 ± 1.5 (SEM) % of S100β-positive cells are also positive for tdTomato. (C) The specificity of Cre recombination. 95.1 ± 0.77 (SEM) % of tdTomato-expressing Cre-recombined cells are positive for S100β. The total number of cells counted in B and C is 898 altogether. Six images taken from three mice were analyzed. (D) Representative Western blotting results using cortical lysates from N4-2 CTL and N4-2 AstKO with antibodies to N4-2, Kir4.1, Tubulin, and Cx43. (E–G) Quantification of relative N4-2, Kir4.1, and Cx43 protein levels in cortical lysates. (H–M) An increase in plasma membrane currents in N4-2 AstKO was restored by blocking Kir4.1 with barium chloride (Ba²⁺). Experiments were performed in a way similar to those in Fig. 5. All mutants were crossed with ROSA26-tdTom. (H–J) Example trances of N4-2 CTL crossed with ROSA26-tdTom (H), N4-2 AstKO crossed with ROSA26-tdTom (I), and N4-2 AstKO crossed with ROSA26-tdTom and treated with Ba²⁺ (J). (K–M) Voltage-current plots (K), currents at −100 mV (L), and membrane resistance (M) from three conditions. (N–P) Unchanged number of parvalbumin-positive cells in N4-2 AstKO. (N) Representative images of N4-2 CTL (top panels) and N4-2 AstKO (bottom panels) hippocampal sections stained with anti-Neuronal Nuclei (NeuN) and anti-Parvalbumin (PV) antibodies. Scale bars, 300 μm. (O and P) Absolute numbers of PV-positive cells in each entire imaged field (O) and the percentages of PV-positive cells with respect to total NeuN-positive cells (P). The number of mice for (E–G) is three for each genotype. Numbers of recorded cells (n) in K–M; n = 10 for N4-2 CTL, n = 5 for N4-2 AstKO, n = 5 for N4-2 AstKO + Ba²⁺. Numbers of brain sections in O and P (n); n = 13 for N4-2 CTL, n = 11 for N4-2 AstKO. ***, P < 0.001; **, 0.001 < P < 0.01; *, 0.01 < P < 0.05; ns, 0.05 < P (two-tailed Student’s t test for E–G, O, and P; one-way ANOVA with Newman–Keuls test for L and M). The normality of the distribution was confirmed with the Kolmogorov–Smirnov or Shapiro–Wilk test. See also Table S2.

Figure 8.

Rescue of reduced γ-oscillation in N4-2 AstKO by pharmacological inhibition of Kir4.1 or gap junction. (A and B) Representative recordings in the CA3 region of acute hippocampal slices from N4-2 CTL (A) and N4-2 AstKO (B) mice before (baseline) and after (Kainate) induction of γ-oscillations with 100 nM kainate application. (C) Representative recordings in the CA3 region of acute hippocampal slices from N4-2 AstKO mice before (baseline) and after (Kainate + Fluo) induction of γ-oscillations with 100 nM kainate application in the presence of fluoxetine. (D and E) Average powers (D) and frequencies (E) of γ-oscillations in N4-2 CTL (black dots) and N4-2 AstKO without (green dots) and with (gray dots) fluoxetine. (F–H) Representative recordings from the same set of experiments as (A–C), except for the usage of GAP26 as a pharmacological blocker in H. (I and J) Average powers (I) and frequencies (J) of γ-oscillations in N4-2 CTL (black dots) and N4-2 AstKO hippocampal slices without (green dots) and with (gray dots) GAP26. Results are shown as mean ± SEM. Numbers of recorded slices (n); (D and E), n = 33 for N4-2 CTL, n = 32 for N4-2 AstKO, n = 40 for N4-2 AstKO + Fluo; (I and J), n = 19 for N4-2 CTL, n = 27 for N4-2 AstKO, n = 19 for N4-2 AstKO + GAP26. ***, P < 0.001; **, 0.001 < P < 0.01; *, 0.01 < P < 0.05; no asterisk, 0.05 < P (one-way ANOVA with Newman–Keuls test for D, E, I, and J). Data distribution was assumed to be normal, but this was not formally tested. See also Fig. S2 and Table S2.

Figure 9.

Decreased γ-oscillations in N4-2 AstKO mice in vivo. (A and B) Representative local field potentials (LFP) were recorded in the CA3 region of the hippocampus in vivo from anesthetized N4-2 CTL (A) and from N4-2 AstKO (B) mice before (Before Kainate) and after (Kainate) intraperitoneal injection of kainate. Third traces in A and B are representative slow-γ filtered LFPs. Bottom heat maps are representative normalized spectrograms. Kainate was injected at time point 0 min. (C) The normalized average power of slow-γ-band oscillation in N4-2 CTL (black trace) and N4-2 AstKO (green trace) mice. Slow γ-band oscillation in N4-2 AstKO showed a siginficantly reduced power in comparison to the N4-2 CTL. Results are shown as mean ± SEM. N = 4 animals for N4-2 CTL and N = 4 animals for N4-2 AstKO. *, P < 0.05 (repeated measures two-way ANOVA with LSD post-hoc test). Data distribution was assumed to be normal, but this was not formally tested. See also Table S2.

Figure 10.

Physiological and pathophysiological homeostasis of astroglial channel proteins by Nedd4-2. Top: Kir4.1 and Cx43 are ubiquitinated by Nedd4-2 and thus degraded by lysosome in astrocytes. This process limits the uptake of extracellular potassium and the subsequent dissipation of intracellular potassium via gap junctions. Sustained extracellular potassium is critical for the maintenance of the neuronal network activity. Left bottom: In astrocyte-specific Nedd4-2 KO mouse, the neuronal network activity is depressed because of reduced extracellular potassium caused by the augmentation of Kir4.1 and Cx43 levels in astrocytes. Right bottom: Nedd4-2 missense mutation (S233L, E271A, or H515P) causes the gain-of-function of E3 ligase activity, increasing ubiquitination levels of Kir4.1 and Cx43, and thus reducing their protein expression. This is the potential cause of an enhanced neuronal network activity and epilepsy.