DKK3 mRNA and protein levels are increased in the human AD brain.

(A) Temporal cortex RNAseq dataset logistic regression shows that DKK3 mRNA levels are increased in AD cases relative to controls. Ordinal regression shows that DKK3 is differentially expressed for Braak scores IV-VI but not for CERAD scores 2-4.

(B, C) Representative immunoblots of DKK3 and loading control actin in (B) soluble and (C) insoluble protein fractions from the hippocampus of control (CT; n = 15-16), Braak stages I-III (n = 16), and Braak stage IV-VI (n = 16) individuals (One-Way ANOVA test followed by Tukey’s multiple comparisons). See also Table S1.

(D) Abundance of DKK3 protein in dorsolateral prefrontal cortex from control (CT, n = 106, asymptomatic (Asym) AD (n = 200), and AD (n = 182) individuals was evaluated using a tandem mass tag mass spectrometry (TMT-MS) proteomic dataset study (Johnson et al., 2022) (ANOVA with two-sided Holm correction, AD vs CT p-value = 0.00000168, Asym AD vs. CT p-value = 0.042492481, AD vs. Asym AD = 0.000402907).

DKK3 localizes to dystrophic neurites around Aβ plaques, and DKK3 extracellular levels are increased in the brain of AD mouse models.

(A) Diagram of the components of an Aβ plaque (blue), astrocytes (orange), microglia (green), and dystrophic neurites (red).

(B) Confocal images of DKK3 protein (red) and amyloid plaques stained with the 6E10 antibody (blue) in the hippocampus of 18-month-old J20 and 8-months NLGF mice. Scale bar = 10 μm.

(C) Confocal images of DKK3 (red) and Aβ plaques labeled by Thioflavin S (ThioS; green) in the hippocampus of 18-month-old WT and J20 mice. ThioS+ plaques not containing DKK3 (-DKK3; arrowheads), ThioS+ plaques containing DKK3 (+ DKK3; asterisks). Scale bar = 150 μm and 100 μm in zoom-in pictures. Graph depicts quantification of the percentage of ThioS+ plaques containing or not DKK3 (Student’s T-test, n = 3 animals per genotype).

(D) Z-stack confocal images show that DKK3 (red) accumulates at Aβ plaques (ThioS; blue) and colocalizes with atrophic axons (Neurofilament-H; green and LAMP1; green) but not with dendrites (MAP2; green). XY views of one plane are shown in the last panel. For LAMP1, a zoom-in picture showing colocalization between DKK3 and LAMP1 puncta is shown. Scale bar = 6 μm. Graphs show Pearson’s correlation coefficient between DKK3 and Neurofilament-H, LAMP1, or MAP2, n = 3-4 animals.

(E, F) Immunoblot images show DKK3 levels in the cell lysate and secreted fraction of acute hippocampal slices of (E) 3-4-month-old WT and J20 mice or (F) 2-3-months old WT and NLGF mice. Slices were incubated with vehicle (Ctrl) or APV for 3 hours. Actin or Vinculin was used as a loading control in the homogenate. Graphs show densitometric quantifications of lysate and extracellular (extracell) DKK3 levels relative to control and the ratio of extracellular/lysate DKK3 levels (Two-Way ANOVA followed by Tukey’s post-hoc test; n = 4-5 animals).

Gain-of-function of DKK3 leads to opposing effects on the number of excitatory and inhibitory synapses in the hippocampus.

(A) Diagram depicting the treatment of hippocampal brain slices obtained from 3-month-old adult WT mice with vehicle (Ctrl) or recombinant DKK3 protein. Synapses were evaluated by confocal microscopy and electrophysiological recordings.

(B) Confocal images of the CA3 SR region labeled with the presynaptic excitatory marker vGLUT1 (green) and the postsynaptic marker PSD-95 (red). Arrows indicate excitatory synapses as colocalized pre- and postsynaptic puncta. Scale bar = 5 μm and 2.5 μm in zoomed-in pictures. Quantification is shown on the right-hand side (Mann-Whitney test, n = 5 animals per condition).

(C) Representative mEPSC traces recorded at -60mV from CA3 cells. Stars indicate mEPSC events. Quantification of mEPSC frequency and amplitude is shown on the right-hand side (Mann-Whitney test, n = 10-13 cells from 5 animals).

(D) Confocal images of the CA3 SR region labeled with the presynaptic inhibitory marker vGAT (green) and the postsynaptic marker gephyrin (red). Arrows indicate inhibitory synapses as colocalized pre- and postsynaptic puncta. Scale bar = 5 μm and 2.5 μm in zoomed-in pictures. Quantification is shown on the right-hand side (Mann-Whitney test, n = 4 animals per condition).

(E) Representative mIPSC traces recorded at 0mV from CA3 cells. Stars indicate mIPSC events. Quantification of mIPSC frequency and amplitude is shown on the right-hand side (Student’s T-test for mIPSC frequency and Mann-Whitney test for mIPSC amplitude, n = 11-12 cells from 5-7 animals).

DKK3 regulates excitatory and inhibitory synapse number through the Wnt/GSK3β and Wnt/JNK pathways respectively.

(A) Diagram of the canonical Wnt pathway through inhibition of GSK3β (Wnt/GSK3β pathway), resulting in elevation of β-catenin and transcriptional activation via TCF/LEF.

(B) Confocal images show excitatory synapses, visualized by colocalization of vGLUT1 (green) and Homer1 (red), as well as β-catenin puncta (grey) in the CA3 SR after treatment with vehicle (Ctrl) or DKK3 in the absence or presence of BIO. Arrows indicate extra-synaptic β-catenin puncta. Scale bar = 5 μm. Quantification of extrasynaptic β-catenin puncta density as a percentage of control is shown on the right-hand side (Two-Way ANOVA followed by Tukey’s multiple comparisons, n = 2-3 brain slices/animal from 5 animals).

(C) Confocal images show excitatory synapses (co-localized vGLUT1 puncta in green and PSD-95 puncta in red) in the CA3 SR after treatment with vehicle (Ctrl) or DKK3 in the absence or presence of BIO. Scale bar = 5 μm and 2.5 μm. Graph shows the quantification of puncta density of pre- and postsynaptic markers and excitatory synapses as a percentage of control (Kruskal-Wallis followed by Dunn’s multiple comparisons, n = 5 animals).

(D) Diagram of the Wnt pathway through activation of JNK (Wnt/JNK pathway), resulting in increased levels of phospho-JNK and transcriptional changes.

(E) Representative immunoblots of phospho-JNK Thr183/Tyr185 (P-JNK) and total JNK of brain slices treated with DKK3 and/or the JNK inhibitor CC-930. Actin was used as a loading control. Graph shows densitometric quantification of P-JNK vs. total JNK relative to the control condition (Kruskal-Wallis followed by Dunn’s multiple comparisons, n = 2 brain slices/animal from 4-5 animals).

(F) Confocal images showing inhibitory synapses defined by the colocalization of vGAT (green) and gephyrin (red) puncta in the CA3 SR after treatment with vehicle (Ctrl) or DKK3 in the absence or presence of CC-930. Scale bar = 5 μm and 2.5 μm. Graph shows the quantification of puncta density of pre and postsynaptic markers and inhibitory synapses as a percentage of control (Kruskal-Wallis followed by Dunn’s multiple comparisons, n = 5 animals).

In vivo loss-of-function of DKK3 decreases inhibitory synapses but does not affect excitatory synapses in the wild-type hippocampus.

(A) Diagram showing the experimental design. 3-month-old WT mice were injected with AAV9 scrambled (Scr) or DKK3 shRNA in the CA3 region. Confocal images showing GFP (green) and DKK3 (red) in Scr- and DKK3-shRNA injected hippocampus. Scale bar = 145 μm. Graph shows quantification of DKK3 intensity in the area injected with the viruses.

(B) Confocal images from CA3 SR show excitatory synapses (colocalized vGLUT1 puncta in green and PSD-95 puncta in red). Arrows indicate excitatory synapses. Scale bar = 5 μm and 2.5 μm in zoomed-in images. Quantification is shown on the right-hand side (Student’s T-test, n = 5 animals per condition).

(C) Representative mEPSC traces recorded at -60mV from CA3 cells. Stars indicate mEPSC events. Quantification of mEPSC frequency and amplitude is shown on the right-hand side (Student’s T-test, n = 8-9 cells from 4 animals).

(D) Confocal images from CA3 SR show inhibitory synapses (colocalized vGAT in green and gephyrin in red). Arrows point to inhibitory synapses. Scale bar = 5 μm and 2.5 μm in zoomed-in pictures. Quantification is shown on the right-hand side (Student’s T-test, n = 5 animals).

(E) Representative mIPSC traces recorded at 0mV from CA3 cells. Stars indicate mIPSC events. Quantification of mIPSC frequency and amplitude is shown on the right-hand side (Mann-Whitney test, n = 10-12 cells from 6 animals).

In vivo loss-of-function of DKK3 ameliorates synaptic changes in the hippocampus of J20 mice before and after Aβ plaque formation.

(A) Diagram depicting the experimental design. In green, 3-month-old WT and J20 mice were injected bilaterally with AAV9-Scr shRNA or AAV9-DKK3 shRNA in the CA3 region. The density of synapses was evaluated at 4-month-old before plaque deposition starts. In blue, 7-month-old J20 mice were injected bilaterally with AAV9-Scr shRNA or AAV9-DKK3 shRNA in the CA3 region. The density of synapses around plaques was evaluated at 9-month-old.

(B, C) Representative confocal images from the CA3 SR region of 4-month-old WT and J20 mice. Images show (B) excitatory synapses (Bassoon in green and Homer1 in red) and (C) inhibitory synapses (vGAT in green and Gephyrin in red). Arrows point to synapses. Scale bar = 2.5 μm. Quantification of synapse number as a percentage relative to WT-Scr shRNA animals is shown on the right-hand side (Two-Way ANOVA followed by Tukey’s post-hoc test, n = 9-11 animals per condition and 2-3 brain slices per animal).

(D, E) Representative confocal images from the CA3 SR region of 9-month-old J20 mice. Images show an Aβ plaque (6E10; blue) and (D) excitatory synapses or (C) inhibitory synapses at different distances relative to the core of the plaque. Scale bar = 2.5 μm. Graphs show synapse number per 200 μm3 at each distance (Two-Way ANOVA followed by Tukey’s post-hoc test, n = 6-7 animals per condition and 2-3 brain slices per animal).

In vivo loss-of-function of DKK3 improves spatial memory in J20 mice.

(A) Diagram depicting that 3-month-old WT and J20 mice were injected bilaterally with AAV9-Scr shRNA or AAV9-DKK3 shRNA in the CA3 area of the hippocampus. One month later, the behavior of animals was assessed using the Open-field, Novel Object Location (NOL) test, Elevated-Plus Maze (EPM), and the Morris water maze (MWM).

(B) Novel Object Location Test. The percentage of time exploring the new object location versus the total time was evaluated (Two-Way ANOVA with Tukey’s post-hoc test, n = 12 WT Scr shRNA, 14 WT DKK3 shRNA, 17 J20 Scr shRNA, 16 J20 DKK3 shRNA).

(C-E) Morris Water Maze.

(C) Representative traces for the MWM Trials 1 and 6 are shown. Graph on the right shows the escape latency. Two-way ANOVA with repeated measures showed a significant effect over trials (animal group F(3,55) = 16.97, p-value<0.0001; trial F(5,259) = 42.94, p-value = 0.457; animal group and trial interaction F(15,275) = 2.753, p-value = 0.0006). For all analyses (n=12 WT Scr shRNA, 14 WT DKK3 shRNA, 17 J20 Scr shRNA, 16 J20 DKK3 shRNA). Graph show comparison between groups (Two-way ANOVA followed by Tukey’s multiple comparisons).

(D, E) Representative traces for the (D) Early and (E) Late probes. Graphs on the right show the time (sec) to first reach the target location (Kruskal Wallis followed by Dunns’ multiple comparisons) and the distance (cm) traveled in the target quadrant (Two-way ANOVA followed by Tukey’s post-hoc test for the early trial or Kruskal Wallis followed by Dunns’ multiple comparisons).