Astroglial Kir4.1 potassium channel deficit drives neuronal hyperexcitability and behavioral defects in Fragile X syndrome mouse model

Abstract

Fragile X syndrome (FXS) is an inherited form of intellectual disability caused by the loss of the mRNA-binding fragile X mental retardation protein (FMRP). FXS is characterized by neuronal hyperexcitability and behavioral defects, however the mechanisms underlying these critical dysfunctions remain unclear. Here, using male Fmr1 knockout mouse model of FXS, we identify abnormal extracellular potassium homeostasis, along with impaired potassium channel Kir4.1 expression and function in astrocytes. Further, we reveal that Kir4.1 mRNA is a binding target of FMRP. Finally, we show that the deficit in astroglial Kir4.1 underlies neuronal hyperexcitability and several behavioral defects in Fmr1 knockout mice. Viral delivery of Kir4.1 channels specifically to hippocampal astrocytes from Fmr1 knockout mice indeed rescues normal astrocyte potassium uptake, neuronal excitability, and cognitive and social performance. Our findings uncover an important role for astrocyte dysfunction in the pathophysiology of FXS, and identify Kir4.1 channel as a potential therapeutic target for FXS.

Fig. 1. Neuronal hyperexcitability and activity-dependent increase of extracellular potassium levels in Fmr1 knockout hippocampus.

a Whole-cell recording from CA1 pyramidal neuron (light blue) surrounded by stratum radiatum astrocytes (gray). b Representative traces of neuronal responses to 60 pA current pulse in WT (dark blue) and Fmr1 KO (magenta) in the presence of synaptic blockers (picrotoxin, CPP and NBQX). Scale bar: 20 mV, 100 ms. Fmr1 KO pyramidal neurons (n = 14 neurons from 14 slices in 6 mice) show (c), reduced rheobase (P = 0.008, t = 2.898, df = 22) and (d) increased number of evoked action potentials (APs) as a function of the injected current in comparison to WT neurons (n = 10 neurons from 10 slices in 6 mice; P < 0.001, F(14, 308) = 18.66). e Scheme of simultaneous recordings of extracellular potassium levels ([K⁺]_o, dark blue) and of field excitatory postsynaptic potentials (fEPSP, light blue) in response to Schaffer collateral (SC) stimulation. f [K⁺]_o under basal conditions is unchanged in Fmr1 KO (n = 8 slices in 7 mice) as compared to WT hippocampus (n = 9 slices in 9 mice; P = 0.786, t = 0.276, df = 15). g Representative traces of simultaneous recording of fEPSP (light blue) and [K⁺]_o (dark blue) in response to 10 Hz, 30 s stimulation of SC. Scale bars, upper panel for fEPSP: 0.2 mV, 5 s; lower panel for [K⁺]_o: 0.2 mM, 5 s. h Representative recordings of [K⁺]_o in response to 10 Hz, 30 s stimulation (horizontal bar) in WT (dark blue) and Fmr1 KO (magenta) hippocampal slices. Scale bar for [K⁺]_o: 0.1 mM, 10 s. Stimulation of SC induces rise in [K⁺]_o showing (i), increased area (P = 0.006, t = −3.236, df = 15) and (j) peak amplitude (P = 0.017, t = −2.679, df = 15) in Fmr1 KO (n = 8 slices in 7 mice) as compared to WT mice (n = 9 slices in 9 mice). [K⁺]_o undershoot (gray) has enlarged (k), area (P = 0.004, t = 3.449, df = 15), (l) peak amplitude (P = 0.007, t = −3.112, df = 15) as well as the time of return (P = 0.003, t = −3.550, df = 15) in Fmr1 KO (n = 8 slices in 7 mice) in comparison to WT mice (n = 9 slices in 9 mice). Data are presented as mean values ± SEM (c, d, f, i–m). *P < 0.05, **P < 0.01, ***P < 0.001. Two-sided unpaired Student’s t test (c, f, i–m), Two-way ANOVA repeated measures, post hoc Fisher LSD (d). CPP: (RS)-3-(2-carboxypiperazin-4-yl)-propyl-1-phosphonic acid; NBQX: 2,3-Dioxo-6-nitro-1,2,3,4-tetrahydrobenzo[f]quinoxaline-7-sulfonamide. Source data are provided as a Source Data file.

Fig. 2. Synaptically-evoked K⁺ uptake through Kir4.1 channels is reduced in Fmr1 knockout astrocytes.

a Hippocampus scheme illustrating simultaneous recording of neuronal activity as field excitatory postsynaptic potentials (fEPSPs, light blue) and astrocyte current (dark blue) with extracellular or patch-clamp electrodes, respectively, in response to Schaffer collateral (SC) stimulation. b Simultaneous recordings of fEPSP (light blue, trace 1) and synaptically-evoked astrocyte current (dark blue, trace 1) in response to SC stimulation (0.05 Hz) in the presence of picrotoxin (control). Addition of glutamate receptor antagonists CPP and NBQX inhibits fEPSP (light blue, trace 2) and potassium component of astrocyte current (dark blue, trace 2). Subtraction of CPP + NBQX-insensitive component (2) from the total astrocyte current (1) reveals synaptically-evoked astroglial K⁺ current (I_K). This current is carried by Kir4.1 channels as confirmed by the absence of I_K in Kir4.1^−/− mice (purple). Scale bars, upper: 0.2 mV, 10 ms; lower left: 10 pA, 10 ms; lower right: 20 pA, 0.2 s. c Representative traces of pharmacologically isolated astrocyte I_K in WT (dark blue) and Fmr1 KO mice (magenta). Arrows indicate stimulation artifact. Scale bar: 10 pA, 1 s. Quantification of astrocyte I_K properties reveals: (d) decrease in I_K peak amplitude (P < 0.001, t = −5.429, df = 28), (e) decrease in I_K peak normalized to fEPSP slope (P = 0.005, U = 45), (f) decrease in charge (P < 0.001, U = 24), (g) increase in time of peak (P = 0.018, t = −2.510, df = 28), (h) increase in rise time (P = 0.014, t = −2.614, df = 28) and i decrease in decay time (P = 0.005. U = 44) in Fmr1 KO (n = 15 astrocytes from 15 slices in 11 mice) as compared to WT (n = 15 astrocytes from 15 slices in 14 mice). Data are presented as mean values ± SEM (d–i). *P < 0.05, **P < 0.01, ***P < 0.001. Statistical significance was calculated using two-sided unpaired Student’s t test (d, g, h) or two-sided Mann–Whitney test (e, f, i). Source data are provided as a Source Data file.

Fig. 3. Kir4.1 expression is diminished in Fmr1 knockout hippocampus.

a Representative examples of immunofluorescent labeling of Kir4.1 (red) and astrocyte marker glial fibrillary acidic protein (GFAP, green) in WT (dark blue) and Fmr1 KO (magenta) hippocampus. Scale bar: 50 µm. b Quantification of Kir4.1 integrated density (P = 0.018, U = 171) and c Kir4.1/ GFAP integrated density ratio (P = 0.010, U = 161) reveals decreased expression in Fmr1 KO stratum radiatum (n = 26 images from 6 mice) as compared to WT (n = 22 images from 5 mice). d Representative high magnification confocal images of a single astrocyte in CA1 stratum radiatum labeled by GFAP (green) and Kir4.1 (red) in WT and Fmr1 KO mice. Distribution of Kir4.1 puncta was determined at any radial position within 25 µm diameter (white arrow) starting from soma center (white point). Scale bar: 10 µm. e Kir4.1 radial intensity profile is similar in Fmr1 KO (n = 29 astrocytes from 5 mice) and WT (n = 23 astrocytes from 6 mice) astrocytes, but displays significant shift (P < 0.001, F(1, 3892) = 147.1) toward lower Kir4.1 intensity in Fmr1 KO when compared to WT astrocytes. f Examples of western blots showing surface expression of Kir4.1 in WT and Fmr1 KO hippocampi. Actin was used as a loading control. g Decreased level of surface Kir4.1 amount in Fmr1 KO (n = 3 mice; P = 0.030, t = 3.314, df = 4) as compared to WT hippocampus (n = 3 mice). Data are presented as mean values ± SEM (b, c, e, g). *P < 0.05, ***P < 0.001. Statistical significance was assessed by performing two-sided Mann–Whitney test (b, c), two-way ANOVA, post hoc Fisher LSD test (e) or two-sided unpaired Student’s t test (g). Arb. units: arbitrary units. Source data are provided as a Source Data file.

Fig. 4. FMRP interacts with Kir4.1 mRNA.

a Schematic illustration of fragile X mental retardation protein (FMRP, cyan) immunoprecipitation from hippocampus lysates using specific anti-FMRP antibody (green) coupled to magnetic beads (gray). Only mRNAs bound to FMRP are recognized by the antibody-bead complex and are further isolated for RT-PCR analysis. Mouse IgG was used as a negative control. b Upper panel: Kir4.1 mRNAs (red) were detected in the FMRP immunoprecipitate (IP, cyan) from WT (dark blue), but not from Fmr1 KO (magenta) hippocampus. The bands corresponding to Kir4.1 mRNA were identified in the Input (black), and were absent from the immunoglobulin (IgG) immunoprecipitate (purple) of both WT and Fmr1 KO hippocampus. Middle panel: previously identified FMRP-associated mRNA encoding postsynaptic density protein 95 (PSD-95, gray) was co-immunoprecipitated with FMRP in WT, but not in Fmr1 KO mice. PSD-95 mRNA was detected in the Input fractions, but was absent from IgG immunoprecipitation complex. Lower panel: glutamate transporter 1 (GLT-1) mRNA was identified in Input, but was not observed in FMRP and IgG immunoprecipitates from both WT and Fmr1 KO hippocampus. c Confocal images of fluorescent in situ hybridization (FISH)-detected Kir4.1 mRNA (red) and immunofluorescence (IF) labeling of FMRP (cyan) in WT hippocampal sections showing their colocalization. Scale bar: 1 µm. d Line profiles of individual fluorescent signals along white dashed lines in (c) depict colocalization. Experiments were performed in WT mice (n = 4) and Fmr1 KO mice (n = 2) (b, c). Source data are provided as a Source Data file.

Fig. 5. Viral delivery of Kir4.1-GFP into FRMP-deficient astrocytes fully rescues K⁺ uptake.

a Diagram of adeno-associated vectors (AAV) designed to express Kir4.1 tagged with green fluorescent protein (GFP) or GFP control under the glial fibrillary acidic protein (GFAP) gene promoter (gfaABC₁D), and scheme of unilateral AAV2/5 microinjection into the mouse hippocampus. b Mice were injected at P15-17 and electrophysiology and immunohistochemistry (IHC) were performed 2 weeks post-injection (p.i.). c Prominent expression of Kir4.1-GFP transgene (green) in CA1 stratum radiatum (sr) after Kir4.1-GFP virus delivery into Fmr1 KO astrocytes; pyr: pyramidal layer. Experiments were performed in triplicate. Scale bar: 50 µm. d Confocal images representing co-immunostaining of GFP (green), Kir4.1 (magenta) and GFAP (yellow) following Kir4.1-GFP virus delivery; white solid line outlines GFP-positive (GFP⁺) stratum radiatum astrocytes, whereas white dashed line marks GFP-negative (GFP⁻) astrocyte. Experiments were performed in 5 mice. Scale bar: 20 µm. e Representative traces of synaptically-evoked astroglial potassium currents (I_K) after injection of Kir4.1-GFP (GFP⁺ astrocyte: green; GFP⁻ astrocyte: black) or GFP (GFP⁺ astrocyte: gray) into Fmr1 KO hippocampus. Stimulation artifacts are indicated by arrows. Scale bar: 10 pA, 1 s. f GFP⁺ and GFP⁻ astrocytes were loaded with Alexa 594 dye (red) during I_K recording. Scale bar: 20 µm. Delivery of Kir4.1-GFP into Fmr1 KO astrocytes significantly restores (g) I_K peak amplitude (P < 0.0001, F(2, 13) = 20.50), (h) I_K peak amplitude normalized to field excitatory postsynaptic potential (fEPSP) slope (P = 0.0003, F(2, 13) = 16.12) and (i) I_K charge (P = 0.0008, F(2, 13) = 13.11) to the level observed in WT astrocytes (blue dashed line). Number of recorded and Alexa 594 loaded cells (f–i): Kir4.1-GFP-injected GFP⁺ astrocytes, n = 6 astrocytes from 6 slices in 5 mice (green); Kir4.1-GFP-injected GFP^- astrocytes, n = 4 astrocytes from 4 slices in 3 mice (black); GFP-injected GFP⁺ astrocytes, n = 6 astrocytes from 6 slices in 4 mice (gray). Data are presented as mean values ± SEM (g–i). **P < 0.01, ***P < 0.001. Statistical significance was calculated using one-way ANOVA, post hoc Fisher LSD test (g–i). Source data are provided as a Source Data file.

Fig. 6. Neuronal excitability is rescued by viral delivery of Kir4.1-GFP in astrocytes of Fmr1 knockout mice.

a Whole-cell recording from CA1 pyramidal neuron (black) surrounded by green fluorescent protein (GFP)-positive astrocytes (green). b Representative traces of neuronal voltage responses evoked by 60 pA current pulse in WT mice injected with GFP (WT + GFP, light blue), Fmr1 KO mice injected with GFP (Fmr1 KO + GFP, gray), Fmr1^fl/fl mice injected with Cre-GFP (Fmr1^fl/fl + Cre-GFP, red), Fmr1 KO mice injected with Kir4.1-GFP (Fmr1 KO + Kir4.1-GFP, green) and Kir4.1^fl/fl mice injected with Cre-GFP (Kir4.1^fl/fl + Cre-GFP, black). Scale bar: 20 mV, 100 ms. c Rheobases of pyramidal neurons in Fmr1 KO + GFP and Fmr1^fl/fl + Cre-GFP mice are significantly decreased in comparison to WT + GFP mice (light blue dashed line, P = 0.005, F(2, 25) = 6.739). d Rheobase of neurons in Fmr1 KO + Kir4.1-GFP mice is increased in comparison to Fmr1 KO + GFP mice and reached the values found in WT + GFP mice (P = 0.024, F(2, 26) = 4.343). e Neurons from Fmr1^fl/fl + Cre-GFP and Kir4.1^fl/fl + Cre-GFP mice display similar rheobases, which are decreased in comparison to WT + GFP mice (P = 0.004, F(2, 28) = 6.648). f The number of evoked action potentials (APs) is significantly higher in neurons from Fmr1 KO + GFP and Fmr1^fl/fl+Cre-GFP mice in comparison to WT + GFP (P = 0.0008, F(2, 25) = 9.715). g Delivery of Kir4.1-GFP into Fmr1 KO astrocytes fully restores the number of evoked APs to the level observed in WT + GFP mice (P = 0.0023, F(2, 26) = 7.754). h Neurons from Fmr1^fl/fl + Cre-GFP and Kir4.1^fl/fl + Cre-GFP mice display similar number of evoked APs, which are increased in comparison to neurons from WT + GFP (P = 0.0014, F(2, 28) = 8.434). Number of recorded cells (c–h): WT + GFP (n = 10 neurons from 10 slices in 4 mice), Fmr1 KO + GFP (n = 9 neurons from 9 slices in 6 mice), Fmr1^fl/fl + Cre-GFP (n = 9 neurons from 8 slices in 4 mice), Fmr1 KO + Kir4.1-GFP (n = 10 neurons from 9 slices in 6 mice), Kir4.1^fl/fl + Cre-GFP (n = 12 neurons from 10 slices in 7 mice). Data are presented as mean values ± SEM (c–h). *P < 0.05, **P < 0.01. Statistical significance was assessed by performing one-way ANOVA, post hoc Fisher LSD test (c–e) and repeated measures two-way ANOVA, post hoc Fisher LSD test (f–h). Source data are provided as a Source Data file.

Fig. 7. Viral delivery of Kir4.1-GFP in Fmr1 knockout astrocytes rescues hippocampus-dependent cognitive and social defects.

a Adeno associated virus (AAV) bilateral injection into the hippocampus of 3-month-old mice. Behavioral testing was performed 1−2 months post injection (p.i.). b Novel object recognition test composed of training phase when mice were exposed to two identical objects (F1 and F2) and testing phase whereas one familiar object (F) was replaced by a novel object (N). c In the testing phase, WT mice injected with green fluorescent protein (GFP; WT + GFP, n = 11 mice; light blue) exhibited a preference for the N (P = 0.007, t = 3.356, df=10), while GFP-injected Fmr1 KO mice (Fmr1 KO + GFP, n = 7 mice; gray) had a similar preference (P = 0.991, t = 0.01219, df = 6) for the F and the N. Delivery of Kir4.1-GFP into Fmr1 KO (Fmr1 KO + Kir4.1-GFP, n = 9 mice; green) hippocampal astrocytes restored (P = 0.030, t = 2.636, df = 8) the preference for the N. d Cre-GFP-injected WT mice (WT + Cre-GFP, n = 10 mice; blue) show higher recognition index (P = 0.002, t = 4.193, df = 9) for the N, whereas Kir4.1^fl/fl mice injected with Cre-GFP (Kir4.1^fl/fl + Cre-GFP, n = 8 mice; black) do not display preference for N (P = 0.565, t = 0.603, df = 7). e, h Test mouse was introduced to stranger mouse (S1) and object (O) during the sociability phase and then the object was replaced by a novel stranger mouse (S2) during the social recognition phase. f, g Mice in each experimental group spend more time sniffing the S1 than the O (f: WT + GFP, n = 11 mice; Fmr1 KO + GFP, n = 8 mice; Fmr1 KO + Kir4.1-GFP, n = 11 mice; P < 0.001, F(1, 54) = 43.57; g: WT + Cre-GFP, n = 9 mice; Kir4.1^fl/fl + Cre-GFP mice, n = 9 mice; P < 0.001, F(1, 32) = 89.13). (i) WT + GFP mice (n = 11 mice) displayed preference for S2 mouse whereas Fmr1 KO + GFP mice (n = 8 mice) failed to distinguish the S1 and S2 mice. This social interaction impairment was corrected after Kir4.1-GFP injection into the hippocampus of Fmr1 KO mice (n = 11 mice; P < 0.001, F(1, 54) = 19.02). WT + Cre-GFP mice (n = 9 mice) show preference for S2, whereas Kir4.1^fl/fl + Cre-GFP mice (n = 9 mice) failed to distinguish the S1 and S2 mice (P = 0.005, F(1, 32) = 9.093). Data are presented as mean values ± SEM (c, d, f, g, i, j). *P < 0.05, **P < 0.01, ***P < 0.001. Two-sided one sample t-tests to 50% (c, d) and Two-way ANOVA followed by a Sidak post hoc test (f, g, i, j). Source data are provided as a Source Data file.