Inflammasome signaling in astrocytes modulates hippocampal plasticity

Abstract

Emerging evidence indicates that a baseline level of controlled innate immune signaling is required to support proper brain function. However, little is known about the function of most innate immune pathways in homeostatic neurobiology. Here, we report a role for astrocyte-dependent inflammasome signaling in regulating hippocampal plasticity. Inflammasomes are multiprotein complexes that promote caspase-1-mediated interleukin (IL)-1 and IL-18 production in response to pathogens and tissue damage. We observed that inflammasome complex formation was regularly detected under homeostasis in hippocampal astrocytes and that its assembly is dynamically regulated in response to learning and regional activity. Conditional ablation of caspase-1 in astrocytes limited hyperexcitability in an acute seizure model and impacted hippocampal plasticity via modulation of synaptic protein density, neuronal activity, and perineuronal net coverage. Caspase-1 and IL-18 regulated hippocampal IL-33 production and related plasticity. These findings reveal a homeostatic function for astrocyte inflammasome activity in regulating hippocampal physiology in health and disease.

Keywords: IL-18; IL-33; astrocytes; caspase-1; hippocampal plasticity; inflammasome; memory; neuroimmunology; seizure.

Figure 1.. Inflammasome activation is dynamically regulated in the adult brain

(A) Representative brain section from an adult ASC^Citrine mouse. ASC specks are represented by green dots generated using Imaris software. (B) Quantification of ASC speck density by brain region in adult ASC^Citrine mice. (C) Representative images of inflammasome activation in the hippocampus (top), cerebellum (middle), and cortex (bottom) from an adult ASC^Citrine mouse. Insets show magnification. (D and E) Engagement of inflammasome signaling in the adult brain assessed by western blot on microdissected hippocampus, cerebellum, and cortex WT lysates. (D) Representative blots. FL, full length. (E) Quantification of relative protein abundance. Signal mean fluorescence intensity (MFI) was measured across all bands (in arbitrary units [a.u.]) then percent of total signal was calculated for each band (n = 4). (F–H) Citrine fluorescence in primary ASC^Citrine CNS cells in response to CaCl₂ (2 mM) applied for 6 h. Treatment was washed out overnight and fluorescence read again at 18 h. (F) Experimental design. (G) ASC^Citrine fluorescence over time normalized to background signal. The gray box indicates the treatment window. (H) Area under the curve (AUC) of ASC^Citrine fluorescence from 0 to 6 h. Data combined from 4 to 6 wells per group. (I–L) ASC^Citrine mice (8–12 weeks old) were exposed to an EE for 48 h (I and J) or left to age (K and L). Control mice were left in their home cages. (I and K) Quantification of the number of ASC specks over the entire hippocampus. (J and L) Representative CA1 images. (M–O) Adult (8–12 weeks old) ASC^Citrine mice were trained on the MWM and brains were harvested immediately following 2 days of training (learn group) or after the probe trial (recall group). Mice that underwent the same treatment but without a hidden platform were used as controls and harvested following the probe trial (swim group). (M) Experimental design. (N) Quantification of the number of ASC specks over the entire hippocampus. (O) Representative CA1 images. Violin plots represent quantification per image and dots represent average data per an individual mouse. Error bars represent mean ± SEM. Statistical significance calculated by mixed effects analysis (G), one-way ANOVA with Tukey’s multiple comparisons test (H), or using linear mixed effects modeling (I, K, and N). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. See also Figure S1.

Figure 2.. Caspase-1 inhibition alters neuronal plasticity

(A–D) Adult TRAP2;tdTomato mice were injected intraperitoneal (i.p.) with 50 mg/kg VX765 or vehicle control, left to rest for 10 min, and then injected with 150 mg/kg tamoxifen. After 30 min, mice were placed into an EE for 48 h, followed by brain harvesting for immunofluorescence. (A) Experimental design. (B–D) Hippocampal trapped tdTomato analysis. (B) Representative images. Quantification of the percent area of total hippocampal coverage of tdTomato+ area (C) and total count of tdTomato+ neurons (D). (E–N) Adult Thy1^YFP mice were treated with 30 mg/kg VX765 or vehicle control by i.p. injection 3×/week until harvesting. (E–H) Neuronal spines in the hippocampus (Hp; E and F) and cortex (Cx; G and H) of vehicle- or VX765-treated mice were 3D reconstructed using Imaris software. (E and G) Quantification of average spine density (number of spines per 10 μm). (F and H) Representative images and 3D renderings. (I, J, M, and N) Synaptophysin, VGLUT1, and GABA_ARα1 protein assessed by western blot on hippocampal lysates from vehicle- or VX765-treated mice. (I) Representative blots. Quantification of MFI of synaptophysin (J), VGLUT1 (M), and GABA_ARα1 (N) normalized to percent of the control samples for each blot. (K and L) Immunohistochemistry of synaptophysin in hippocampal CA1. (K) Representative images. (L) Quantification of the number of synaptophysin puncta per field of view (FOV). Violin plots represent quantification per image and dots represent average data per an individual mouse. Error bars represent mean ± SEM. Statistical significance calculated using linear mixed effects modeling (C–E, G, and L) or by unpaired Student’s t test (J, M, and N). *p < 0.05, **p < 0.01. See also Figure S2.

Figure 3.. Astrocyte-specific inflammasome activity influences neuronal plasticity

(A and B) Inflammasome cell-type localization in the hippocampus of ASC^Citrine mice. (A) Representative images. Images to the right show distinct colocalized channels within the area of the dotted box to the left. White arrows point to ASC specks. (B) Pie chart depicting the relative presence of ASC specks within astrocytes (GFAP+ cells) compared with microglia (IBA1+ cells) divided by the total number of ASC specks in a FOV. 3–5 FOVs analyzed per mouse (n = 3 mice). (C–L) Brains were harvested from Casp1^ΔAst and Casp1^cont mice at 4–6 months of age for western blot on hippocampal lysates (C–F) and hippocampal CA1 immunohistochemistry (G–L). Quantification of MFI of synaptophysin (C), VGLUT1 (D), and GABA_ARα1 (E) normalized to percent of the control samples for each blot. (F) Representative blots. (G) Representative images and (H) quantification of the percent area of synaptophysin coverage per FOV. (I) Representative images and (J) quantification of the number of VGLUT1 and Homer-1 puncta colocalized per FOV. (K) Representative images from hippocampal DG and (L) quantification of the total number of c-Fos+ puncta per hippocampus. (M and N) Whole-cell electrophysiology recordings were conducted in slices from male mice (8–12 weeks) in Casp1^cont (n = 11, 5) and Casp1^ΔAst (n = 9, 3) CA1 neurons (n = cells, animals). (M) Average number of action potentials elicited relative to current injection steps. (N) Representative traces of neuronal firing at 400 and 600 pA current injections for Casp1^cont (top) and Casp1^ΔAst (bottom) CA1 neurons. Violin plots represent quantification per image and dots represent average data per an individual mouse (C–E, H, J, and L) or the average of all mice per group (M). Error bars represent mean ± SEM. Statistical significance calculated by unpaired Student’s t test (C–E), linear mixed effects modeling (H, J, and L), or Kolmogorov-Smirnov test (M). *p < 0.05, **p < 0.01, ****p < 0.001. See also Figures S2 and S3.

Figure 4.. Caspase-1 signaling in astrocytes regulates hippocampal function, neurogenesis, and PNNs

(A–C) Hippocampi from Casp1^ΔAst and Casp1^cont mice (n = 3 per group) were harvested at 14 weeks of age for scRNA-seq. (A) tSNE plot representing the cell types retrieved from Casp1^cont and Casp1^ΔAst hippocampi. BAMs, border-associated macrophages; OPCs, oligodendrocyte progenitor cells; CP cells, choroid plexus cells. (B and C) Bar charts representing select GO terms collected from upregulated and downregulated genes between all hippocampal cells. (B) Cellular component and (C) biologic process GO terms. (D–G) Memory function assessed in adult Casp1^ΔAst and Casp1^cont mice using a fear conditioning memory persistence paradigm. (D) Experimental design. (E) Percent time spent immobile (freezing) during the context test assessed on consecutive test sessions. (F and G) At the 11-week test session, mice were also exposed to a novel, untrained context. (F) The percent time spent immobile (freezing) during the exposure to the trained (“fearful”) context compared with the untrained (“novel”) context. (G) Memory discrimination between trained and novel contexts. (H–K) Brains of Casp1^ΔAst and Casp1^cont mice were harvested at 4–6 months of age for immunohistochemistry. (H) Representative images and (I) quantification of the percent area of DCX coverage in hippocampal DG. (J) Representative images and (K) quantification of the percent area of WFA coverage in hippocampal CA1. Violin plots represent quantification per image and dots represent average data per an individual mouse (F, G, I, and K) or the average of all mice per group (E). Error bars represent mean ± SEM. Statistical significance calculated by mixed effects analysis with Šídák’s multiple comparisons test (E), two-way ANOVA with Šídák’s multiple comparisons test (F), one-sample t test with Wilcoxon test against zero (G), or linear mixed effects modeling (I and K). *p < 0.05, **p < 0.01, ***p < 0.001, ns = not significant. See also Figures S3–S5.

Figure 5.. Astrocyte inflammasome signaling modulates IL-33-related hippocampal plasticity

Brains were harvested from Casp1^ΔAst, Il33^ΔAst, Il33^ΔNeur, and respective Cre-negative littermate controls (Casp1^cont and Il33^cont) at 4–6 months of age. (A–E) Relative IL-33 protein in astrocytes (SOX9+ cells) and neurons (MAP2+ cells) in hippocampal CA1 of Casp1^cont and Casp1^ΔAst mice by immunohistochemistry. (A) Representative images. Quantification of (B) percent area coverage and (C) MFI of IL-33 signal per FOV. Quantification of IL-33 within thresholded SOX9+ (D) or MAP2+ (E) area plotted as a percent of total IL-33 signal per FOV. (F and G) Il33 transcriptional expression in Casp1^ΔAst and Casp1^cont mice relative to Gapdh. (F) Expression of Il33a in sorted astrocytes. (G) Expression of Il33b in sorted neurons. (H–M) Immunohistochemistry in hippocampal CA1 of Il33^ΔAst and Il33^cont mice. (H) Representative images and (I) quantification of the number of colocalized VGLUT1 and Homer-1 puncta per FOV. (J) Representative images and (K) quantification of the total number of c-Fos+ puncta per FOV. (L) Representative images and (M) quantification of WFA percent area coverage per FOV. (N) Relative IL-18 concentration by ELISA in microdissected hippocampus, cerebellum, and cortex lysates from adult WT mice. (O and P) Relative expression of Il18 (O) and Il18r1 (P) in sorted neurons and astrocytes from Casp1^cont mice (O: astrocyte n = 7, neuron n = 7; P: astrocyte n = 7, neuron n = 3). (Q–U) Primary halved hippocampi isolated from Casp1^cont mice (R, T, and U) or Il33^ΔNeur and Il33^cont mice (S) exposed to treatments then homogenized for IL-33 (R and S) or IL-18 (T and U) ELISA. (Q) Experimental design. (R) Hippocampal lysate IL-33 concentration under increasing IL-18 (1, 100, or 500 ng/mL) treatments applied for 30 min. (S) Supernatant IL-33 concentration after 2 h treatment with 100 ng/mL IL-18. (T and U) Relative IL-18 concentration normalized to total protein per hippocampus slice under increasing forskolin (FSK; 30 or 60 μM; T) or glutamate (50, 100, or 1,000 μM; U) concentrations. Violin plots represent quantification per image and dots represent average data per an individual mouse. Error bars represent mean ± SEM. Statistical significance calculated using linear mixed effects modeling (B–E, I, K, and M), unpaired Student’s t test (F, G, O, and P), one-way ANOVA with Tukey’s multiple comparisons test (N), one-way ANOVA with Dunnett’s multiple comparisons test (R, T, and U), or mixed effects analysis with uncorrected Fisher’s least significant difference test (S). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. See also Figures S5 and S6.

Figure 6.. Inflammasome activation propagates seizure progression

Mice (6 weeks old) were injected i.p. with KA and then scored for 2 h using a modified Racine seizure scale. (A–D) Male and female WT (n = 65) and Casp1^−/− (n = 18) mice were injected with 24 mg/kg KA. (E–H) Casp1^cont (n = 17) and Casp1^ΔAst (n =17) mice were injected with 16 mg/kg KA. (A and E) Seizure activity scored every 5 min. (B and F) Time to seizure onset (Racine score 3) for mice that actively seized during the 2 h scoring time. (C and G) Time spent seizing represented by sum of time from reaching Racine score of 4 until end of scoring period. (D and H) Survival curve. Data combined from at least three independent experiments. Each data point represents the average of every mouse within the group (A and E). Error bars indicate mean ± SEM. Statistical significance calculated by two-way ANOVA (A and E), unpaired Student’s t test (B, C, F, and G), or Gehan-Breslow-Wilcoxon test (D and H). *p < 0.05, ***p < 0.001, ****p < 0.0001. See also Figure S6.