Inactivation of retinogeniculate synapses induces caspase-3 activity.

(A) Schematics of experimental setup. AAVs expressing tetanus toxin light chain (TeTxLC) and/or mTurquiose2 (Turq) were injected into the right eye of E15 mice (left). By P5, retinogeniculate synapses in dLGN were inactivated to varying extents depending on injection and side (right). (B-D) Confocal images of Turq (left panels) and activated caspase-3 (right panels) in left dLGN (B) and right dLGN (C) of a TeTxLC-injected P5 animal and in left dLGN of a control P5 animal (D). Dotted lines delineate dLGN boundaries. Only signals within dLGNs were analyzed. Images from the same fluorescent channel were adjusted to the same contrast. The compass in B marks tissue orientation. Scale-bars: 100 μm. D, dorsal; V, ventral; M, medial; L, lateral. (E) Quantification of caspase-3 activity in indicated dLGNs. Activated caspase-3 signals in each dLGN (highlighted areas in B-D) were summed and normalized to dLGN area. Each point represents the result from one dLGN. Data from two dLGNs of the same animal were paired for analysis (grey lines). n=10 for TeTxLC-injected animals and n=9 for control animals. Mean and standard deviation (S.D.) are shown. P-values were calculated from two-tailed t-tests (paired when applicable). (F) Example images showing punctate caspase-3 activities in ventral-medial regions of indicated dLGNs. Images were adjusted to the same contrast. Scale-bar: 20 μm. (G) High-resolution images of dLGN showing TeTxLC-expressing RGC axons (yellow) and activated caspase-3 (magenta). Two regions of interest (dotted squares) are magnified to illustrate that caspase-3 activity was found juxtaposing TeTxLC-expressing axon terminals but not within them. Scale-bar: 5 μm.

Caspase-3 activation at weak synapses requires the presence of strong synapses.

(A) Schematics illustrating experimental conditions. No synapses are inactivated in the control (left); only synapses from right eyes are inactivated in single inactivation condition (middle); synapses from both right and left eyes are inactivated in dual inactivation condition (right). (B) Confocal images of P5 left side dLGNs in the three conditions showing caspase-3 activity in the dLGN. Dotted lines mark dLGN boundaries. Scale-bars: 100 μm. (C) Quantification of dLGN caspase-3 activity in the indicated conditions. Caspase-3 activity in each dLGN were summed and normalized to the dLGN area. For the single inactivation condition, values were from left dLGNs only. For the other two conditions, values from both dLGNs were averaged. n=8 animals for control, n=11 animals for single inactivation, and n=10 animals for dual inactivation. Mean and S.D. are shown. P-values were calculated from Tukey’s multiple comparison tests.

Caspase-3 is required for segregation of eye-specific territories.

(A-B) Representative confocal images of retinogeniculate inputs in the dLGN of P10 wild-type (A) and Casp3−/− (B) mice. Contralateral inputs are labeled with AlexaFlour488 (AF488) conjugated cholera toxin subunit B (CTB) and ipsilateral inputs with AF594-CTB. Original images were thresholded into 0-or-1 images using the Otsu method (34), and the overlap between thresholded contralateral and ipsilateral inputs is shown. (C) Percentage overlap between eye-specific territories in wildtype and Casp3−/− mice under a series of increasing signal cutoff thresholds. Note that the percentage overlap is plotted on a log scale. Each circle represents one animal. Mean and S.D. are shown. n=9 for wildtype mice and n=6 for Casp3−/− mice. (D) Mean percentage overlap values in wildtype and Casp3−/− mice and p-values of two-tailed t-tests between the two genotypes are listed for each cutoff threshold.

Caspase 3 is required for retinogeniculate circuit refinement.

(A-B) Example recordings of dLGN relay neuron responses in P30 wildtype (A) and Casp3−/− (B) mice. Excitatory postsynaptic currents (EPSCs) were evoked by increasing stimulation currents in the optic tract. Both AMPAR-mediated inward current at -70 mV membrane potential and AMPAR and NMDAR-mediated outward current at +40 mV membrane potential are shown. Peak response amplitudes at each stimulation intensity are plotted to the right of recording traces. Scale bars represent 0.5 nA and 10 ms. (C) Distribution of RGC input numbers on individual dLGN relay neurons in wildtype and Casp3−/− mice. The number of RGC inputs was inferred by manually counting the number of steps in AMPAR-mediated EPSC response curves (lower right in A and B) while blind to the genotypes. P-value was calculated from two-tailed t-test. n=37 cells from 22 wildtype mice and n=32 cells from 16 Casp3−/− mice. (D) Example recordings of mEPSC measurements from wildtype and Casp3−/− mice. Scale-bars represent 0.5 s (horizontal) and 5 pA (vertical). (E-F) Cumulative distribution curves of inter-mEPSC intervals (E) and mEPSC amplitudes (F) in wildtype and Casp3−/− mice. P-values were calculated from Kolmogorov-Smirnov tests. n=16 cells from 4 wildtype mice and n=17 cells from 4 Casp3−/− mice. (G) Example recordings from paired pulse measurements at -70 mV membrane potential in wildtype and Casp3−/− mice. Traces from experiments with 50, 150, 250, 500, and 1000 ms inter-stimulus intervals are overlayed. Stimulus artifacts were removed from the traces for clarity. Scale-bars represent 100 ms (horizontal) and 0.3 nA (vertical). (H) Paired-pulse ratio (calculated as amplitude of the second response over that of the first response) in wildtype and Casp3−/− mice at various inter-stimulus intervals. Mean and standard error of the mean (SEM) are shown. P-values were calculated from Bonferroni’s multiple comparison test. p=0.0067 for 50 ms interval, p=0.0369 for 150 ms interval, and p=0.0097 for 250 ms interval. n=16 cells from 7 wildtype mice and n=13 cells from 4 Casp3−/− mice.

Microglia-mediated synapse elimination depends on caspase-3.

(A) Representative 3D-reconstructed images of a P5 Casp3+/−; Cx3cr1-Gfp+/− mouse dLGN with microglia displayed in green, contralateral RGC axon terminals in red, and ipsilateral RGC terminals in blue. In the merged image, the region from which microglia are selected for analysis is indicated with the dashed line. The scale-bar represents 100 μm. (B) Representative surface rendering of microglia (green) from P5 dLGNs of Casp3+/−; Cx3cr1-Gfp+/− and Casp3−/−; Cx3cr1-Gfp+/− mice. Intracellular contralateral (red) and ipsilateral (blue) RGC axon terminals are shown. Microglia from caspase-3 deficient mice engulf visibly less synaptic material. Scale-bars represent 10 μm. (C) Total volume of engulfed synaptic material in individual microglia from Casp3+/−; Cx3cr1-Gfp+/− and Casp3−/−; Cx3cr1-Gfp+/− mice. (D) Total volume of engulfed synaptic material in each microglia (from C) is normalized to the volume of that microglia. In C-D, each point represents one microglia. Mean and S.D. are shown. p-values were calculated from unpaired two-tailed t-tests. n=61 microglia from 8 Casp3+/−; Cx3cr1-Gfp+/− mice and n=54 microglia from 5 Casp3−/−; Cx3cr1-Gfp+/− mice.

Removal of weak synapses by microglia requires caspase-3 activity.

(A) Schematics illustrating the experimental rationale. In wildtype mice (upper panel), inactivating retinogeniculate synapses from the right eye activates caspase-3 (magenta) and recruits microglia to preferentially engulf right eye-originated synapses (red) over left eye-originated synapses (green). If caspase-3 activation is blocked (lower panel), engulfment of inactive synapses should be attenuated. (B-E) Surface rendering of representative microglia from P5 left dLGN of Casp3+/+ (B-C) or Casp3−/− (D-E) mice injected with AAV that provided mTurquoise2 (B and D) or TeTxLC (C and E) in the right eye at E15. Microglia were labeled by immunostaining against Iba1. RGC axon terminals from the left eye are shown in yellow and terminals from the right eye in red. Scale bars represent 15 μm. (F-G) Ratio between volumes of right-eye and left-eye-originated synaptic material engulfed by microglia from Casp3+/+ (F) or Casp3−/− (G) mice injected with AAV that carried the gene for mTurquoise2 (blue) or TeTxLC (red). Each dot represents one microglia. Engulfment ratios are displayed on a log scale. 0, 25, 50, 75, and 100 percentiles are shown. p-values were calculated from unpaired two-tailed Mann-Whitney tests. n=52 microglia from 7 Turq-injected Casp3+/+ mice, n=50 microglia from 8 TeTxLC-injected Casp3+/+ mice, n=64 microglia from 5 Turq-injected Casp3−/− mice, and n=51 microglia from 6 TeTxLC-injected Casp3−/− mice.

Caspase-3 deficiency protects against Aβ induced synapse loss.

(A and C) Representative 3D-reconstructed images (3 μm in z) showing presynaptic (SV2, in green) and postsynaptic (Homer1, in red) signals in dentate gyrus of female 6 month-old APP/PS1 mice on caspase-3 wildtype (A) and deficient (C) backgrounds. Ellipsoids were fitted to the original images (upper panels) to isolate pre-(green) and post-synaptic (red) puncta (lower panels). Homer1 ellipsoids found with 300 nm of a SV2 ellipsoid are highlighted in white in the fitted images. Original images are adjusted to the same contrast. For the fitted images, only ellipsoids from the upper half of the z-stack are shown. Scale-bar represents 4 μm. (B and D) Quantification of synapse density in APP/PS1 mice on caspase-3 wildtype (B) and deficient (D) backgrounds. Mean and S.D. are shown. p-values were calculated from unpaired two-tailed t-tests. n=5 for App/Ps1−/− mice, n=6 for App/Ps1+/− mice, n=6 for Casp3−/−; App/Ps1−/− mice, and n=4 for Casp3−/−; App/Ps1+/− mice.