Caspase-1 inhibition alleviates cognitive impairment and neuropathology in an Alzheimer's disease mouse model

Abstract

Alzheimer's disease (AD) is an intractable progressive neurodegenerative disease characterized by cognitive decline and dementia. An inflammatory neurodegenerative pathway, involving Caspase-1 activation, is associated with human age-dependent cognitive impairment and several classical AD brain pathologies. Here, we show that the nontoxic and blood-brain barrier permeable small molecule Caspase-1 inhibitor VX-765 dose-dependently reverses episodic and spatial memory impairment, and hyperactivity in the J20 mouse model of AD. Cessation of VX-765 results in the reappearance of memory deficits in the mice after 1 month and recommencement of treatment re-establishes normal cognition. VX-765 prevents progressive amyloid beta peptide deposition, reverses brain inflammation, and normalizes synaptophysin protein levels in mouse hippocampus. Consistent with these findings, Caspase-1 null J20 mice are protected from episodic and spatial memory deficits, neuroinflammation and Aβ accumulation. These results provide in vivo proof of concept for Caspase-1 inhibition against AD cognitive deficits and pathologies.

Fig. 1

VX-765 treatment restores J20 mice cognitive function. a Experimental treatment paradigm. b NOR Discrimination index from vehicle-treated WT (grey squares), vehicle-treated J20 (black circles) and VX-765-treated J20 (blue triangles). Each mouse tested is represented by one symbol. Data represent mean and s.e.m. (Treatment, F(2,20) = 85.8, p < 0.0001; Time, F(4,80) = 4.188, p = 0.0039; Treatment × Time, F(8,80) = 3.599, p = 0013). c Distance travelled during open field task (Treatment, F(2,19) = 11.47, p = 0.0005; Treatment × Time, F(8,76) = 5.69, p < 0.0001). b, c Two-way repeated-measures ANOVA and Dunnett’s post-hoc versus J20 + vehicle, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. d–i Barnes maze: learning acquisition # of primary errors during d T2 (Treatment, F(2,212) = 19.93, p < 0.0001) and g WO (Treatment, F(2,56) = 10.86, p = 0.001; Training day, F(3,56) = 3.392, p < 0.0241). d, g Two-way repeated-measures ANOVA and Dunnett’s post-hoc versus J20 + vehicle *p < 0.05, ***p < 0.001, ****p < 0.0001. Probe primary latency and errors during e T2 and h WO. Target preference: # of pokes of each hole labelled +1 to +9 to the right or −1 to −9 to the left of the target (T) during the probe after f T2 and i WO (T2 primary latency, F(2,52) = 5.879, p = 0.0050; T2 primary errors, F(2,52) = 9.998, p = 0.0002; WO primary latency, F(2,22) = 4.076, p = 0.0312; WO primary errors, F(2,22) = 10.84, p = 0.0005). e, h ANOVA, Tukey’s post-hoc, *p < 0.05, **p < 0.01, ***p < 0.001

Fig. 2

VX-765 dose-dependently restores cognitive function in J20 mice. a NOR discrimination index at T1 (F(3,24) = 20.42, p < 0.0001), T2 (F(3,23) = 12.83, p < 0.0001), WO, and T3 (F(3,23) = 8.828, p = 0.0004). Vehicle-treated (black circles), and 50 mg kg⁻¹ (blue triangles), 25 mg kg⁻¹ (pink triangles), or 10 mg kg⁻¹ (purple hexagon) VX-765-treated J20 mice. Each mouse tested is represented by one symbol. Data represent mean and s.e.m. b−g Barnes maze during b−d T2 and e−g WO. Learning acquisition at b T2 and e WO; probe and target preference at c, d T2 and f, g WO. Learning acquisition # of primary errors during b T2 (Treatment, F(2,56) = 3.377, p = 0.0412; Training day, F(3,56) = 7.102, p = 0.0004), and e WO (Treatment, F(2,56) = 5.434, p = 0.007; Training day, F(3,56) = 11.52, p < 0.0001), probe primary errors at c T2 (F(2,14) = 5.69, p = 0.0155). a, c, f ANOVA and b, e Two-way repeated-measures ANOVA, Dunnett’s post-hoc versus J20 + vehicle, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Fig. 3

Casp1 KO restores cognitive function in J20 mice. a−e Genotypes: J20^−/−/Casp1^−/− WT/WT (grey squares), J20^−/−/Casp1^+/− WT/Het (pink diamonds), J20^−/−/Casp1^−/− WT/KO (blue symbol), J20^−/+/Casp1^+/+ J20/WT (black circles), J20^−/+/Casp1^−/+ J20/Het (purple triangles), J20^−/+/Casp1^−/− J20/KO (blue triangles). Each mouse tested is represented by one symbol. Data represent mean and s.e.m. a NOR discrimination index (F(5,67) = 16.22, p < 0.0001) and b distance travelled during open field task (F(5,67) = 3.717, p = 0.005). c−e Barnes maze: c # of primary errors during learning acquisition (Genotype, F(5,263) = 6.469, p < 0.0001; Training day, F(3,263) = 29.44, p < 0.0001), d probe primary latency and errors (Primary errors, F(5,66) = 4.8, p = 0008) and e target preference. a, b, d ANOVA and c two-way ANOVA, Dunnett’s post-hoc versus WT/WT. *p < 0.05, **p < 0.01, ****p < 0.0001

Fig. 4

VX-765 reverses neuroinflammation in J20 mice. a Iba1-immunopositive microglia in hippocampal SLM and cortical S1 regions. b, c, e Vehicle-treated J20^−/−/Casp1^−/− WT (grey squares), vehicle-treated J20^−/−/Casp1^−/− WT/KO (blue hexagons), J20^−/+ at 5 months baseline (J20-BL; dark grey hexagons), vehicle-treated J20^−/+ (black circles), and VX-765-treated J20^−/+ (purple triangles). b Stereological quantification of Iba1-positive microglia from hippocampal pyramidal cell layer to the SLM (F(2,11) = 58, p < 0.0001) and cortex (F(2,11) = 39.95, p < 0.0001) (top panels), average % distribution of morphological microglial subtypes I, II, III and IV (middle panels), and Il-1-β levels (bottom panels) in 5-month-old baseline J20 and in 8-month-old WT + vehicle, J20 + vehicle, and J20 + VX-765 WT mice. ANOVA, Dunnett’s post-hoc versus J20 + vehicle, ***p < 0.001. c Stereological quantification of Iba1-positive microglia in the hippocampal CA1 (F(3,21) = 50.53, p < 0.0001) and cortex (F(3,21) = 96.21, p < 0.0001) (top panels), average % distribution of morphological microglial subtypes I, II, III and IV (middle panels), and Il-1-β levels (bottom panels) in WT/WT, WT/Casp1^−/− (WT/KO), J20/WT, and J20/Casp1^−/− (J20/KO) mice. ANOVA, Dunnett’s post-hoc versus J20/WT, ****p < 0.0001. d Micrographs of GFAP immunopositive astrocytes. e GFAP immunostaining density in vehicle-treated WT and J20 and in VX-765-treated J20 brain hippocampi (F(2,12) = 5.234, p = 0.0232) and cortex (F(2,12) = 8.582, p = 0.0049). f GFAP immunostaining density in J20/KO and littermate control brain hippocampi (F(3,21) = 41.15, p < 0.0001) and cortex (F(3,21) = 4.746, p = 0.0111). ANOVA, Dunnett’s post-hoc versus J20 + vehicle (e) or J20/WT (f), *p < 0.05, ***p < 0.001, ****p < 0.0001. Scale bar in a, d = 50 µm

Fig. 5

VX-765 prevents progressive Aβ accumulation in J20 mice. a Aβ micrographs of hippocampal SLM and S1 cortex (scale bar = 50 µm). b Quantitative analysis comparing Aβ immunostaining density between J20 + vehicle (black circles) and J20 + VX-765 (blue triangles) mice in the CA1 hippocampal pyramidal cell layer to the SLM (p = 0.0029) and cortex (p = 0.0314, unpaired t test). c Quantitative analysis comparing Aβ immunostaining density between J20^−/+/Casp1^+/+ J20/WT (black circles) and J20^−/+/Casp1^−/− J20/KO (blue triangles) mice in the CA1 hippocampal cell layer (p = 0.0040) and cortex (0.0276, unpaired t test). d−g RIPA- (d, e) and formic acid- (f, g) soluble Aβ levels in the hippocampus and cortex of 5-month baseline J20^−/+ (black diamonds), J20 + vehicle (black circles), and J20 + VX-765 (blue triangles) mice. d, f Aβ₄₂/Aβ₃₈ + Aβ₄₀ + Aβ₄₂ levels (RIPA soluble, hippo: F(2,17) = 27.85, p < 0.0001) and e, g total Aβ ANOVA, Dunnett’s post-hoc versus J20 + vehicle, ***p < 0.001, ****p < 0.0001. h−k J20^−/+/Casp1^+/+ J20/WT (black circles), J20^−/+/Casp1^−/+ J20/Het (purple triangles), and J20^−/+/Casp1^−/− J20/KO (blue triangles) RIPA- (h, i) and formic acid- (j, k) soluble Aβ levels in the hippocampus and cortex of J20/WT, J20/Casp1^+/− (J20/Het), and J20/Casp1^−/− J20/KO mice. h, j Aβ₄₂/Aβ₃₈ + Aβ₄₀ + Aβ₄₂ levels (RIPA-soluble, hippo: F(2,26) = 16.03, p < 0.0001; RIPA-soluble, cortex: F(2,26) = 6.602, p < 0.0048) and i, k total Aβ ANOVA, Dunnett’s post-hoc versus J20/WT, **p < 0.01, ***p < 0.001, ****p < 0.0001. l−o Human APP protein levels (6E10 immunostaining) and quantification in the hippocampus and cortex of l, m J20 + vehicle (black circles) and J20 + VX-765 mice (blue triangles), and n−o J20/WT (black circles), J20/Het (purple triangles), and J20/KO (blue triangles) mice

Fig. 6

VX-765 reverses loss of synaptophysin in J20 mice. a Synaptophysin immunopositive micrographs. Scale bar = 200 µm for hippocampus and 50 µm for cortex. b Synaptophysin immunopositive density in WT + vehicle (grey squares, n = 4), J20 + vehicle (black circles, n = 4), and J20 + VX-765 (blue triangles, n = 4) hippocampus (F(2,9) = 7.974, p = 0.0102, ANOVA, Tukey’s post-hoc, **p < 0.01) and cortex. c Significantly altered (Kruskall−Wallis) synaptic protein mRNA levels in vehicle or VX-765-treated J20 mice hippocampi (n = 3 per group)

Fig. 7

VX-765 protects human neurons against stress-mediated neuritic beading. a MTT assay. b EGFP fluorescent micrographs of EGFP- or APP+ EGFP-transfected neurons. c % beading in 25 or 50 µM VX-765 or 5 µM Z-YVAD-fmk Casp1 peptide inhibitor pretreated (1 h) or post-treated (48 h) APP-transfected or serum-deprived neurons. DMSO vehicle treatment of EGFP-transfected neurons (grey squares), DMSO- (black circles), 5 µM YVAD- (purple hexagons), 25 µM VX-765- (blue triangles) or 50 µM VX-765- (purple triangles)-treated stressed (APP transfection in upper panels or serum deprived in bottom panel) neurons. Two-way repeated-measures ANOVA, Dunnett’s post hoc versus APP^WT or serum-deprived, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Pretreatment: APP^WT (Treatment F(4,10) = 10.17, p = 0.0015; Time F(2,20) = 93.32, p < 0.0001; Treatment × Time F(8,20) = 6.736, p = 0.0003), serum-deprived (Treatment F(4,10) = 10.2, p = 0.0015; Time F(2,20) = 37.65, p < 0.0001). Postbeading treatment: APP^WT (Time F(4,40) = 102.7, p < 0.0001), serum-deprived (Treatment F(4,23) = 14.12, p < 0.0001; Time F(4,23) = 24.35, p < 0.0001). d, e Serum-treated normal (grey square), serum-deprived (black circles), serum-deprived and treated with 25 µM VX-765 (blue triangles), and serum-deprived and treated with 50 µM VX-765 (pink triangles) neurons. d Secreted or cellular Aβ₄₂/ Aβ₃₈ + Aβ₄₀ + Aβ₄₂. e Secreted Il-1β (F(3,8) = 6.636, p = 0.0146), IFN-γ, TNF-α (F(3,8) = 3.931, p = 0.054), and IL-6 (F(3,8) = 10.24, p = 0.0041) relative to untreated (+Serum) HPN. ANOVA, Dunnett’s post-hoc versus +Serum, *p<0.05. a, c, d, e. Each individual neuron preparation tested is represented by one symbol. Data represent mean and s.e.m.