Protective effect of PDE4B subtype-specific inhibition in an App knock-in mouse model for Alzheimer's disease

Abstract

Meta-analysis of genome-wide association study data has implicated PDE4B in the pathogenesis of Alzheimer's disease (AD), the leading cause of senile dementia. PDE4B encodes one of four subtypes of cyclic adenosine monophosphate (cAMP)-specific phosphodiesterase-4 (PDE4A-D). To interrogate the involvement of PDE4B in the manifestation of AD-related phenotypes, the effects of a hypomorphic mutation (Pde4b^Y358C) that decreases PDE4B's cAMP hydrolytic activity were evaluated in the App^NL-G-F knock-in mouse model of AD using the Barnes maze test of spatial memory, ¹⁴C-2-deoxyglucose autoradiography, thioflavin-S staining of β-amyloid (Aβ) plaques, and inflammatory marker assay and transcriptomic analysis (RNA sequencing) of cerebral cortical tissue. At 12 months of age, App^NL-G-F mice exhibited spatial memory and brain metabolism deficits, which were prevented by the hypomorphic PDE4B in App^NL-G-F/Pde4b^Y358C mice, without a decrease in Aβ plaque burden. RNA sequencing revealed that, among the 531 transcripts differentially expressed in App^NL-G-F versus wild-type mice, only 13 transcripts from four genes - Ide, Btaf1, Padi2, and C1qb - were differentially expressed in App^NL-G-F/Pde4b^Y358C versus App^NL-G-F mice, identifying their potential involvement in the protective effect of hypomorphic PDE4B. Our data demonstrate that spatial memory and cerebral glucose metabolism deficits exhibited by 12-month-old App^NL-G-F mice are prevented by targeted inhibition of PDE4B. To our knowledge, this is the first demonstration of a protective effect of PDE4B subtype-specific inhibition in a preclinical model of AD. It thus identifies PDE4B as a key regulator of disease manifestation in the App^NL-G-F model and a promising therapeutic target for AD.

Fig. 1. Protective effect of hypomorphic PDE4B on spatial memory in App^NL-G-F mice assessed by the Barnes maze.

A Primary latency (s). All three genotypes showed a significant decrease over the five training days (Friedman’s ANOVA, WT: χ²₄ = 37.819, p < 0.001; App^NL-G-F: χ²₄ = 39.847, p < 0.001; App^NL-G-F/Pde4b^Y358C: χ²₄ = 43.518, p < 0.001). Genotypic differences were observed from day 2 (Kruskal-Wallis test, day 2: χ²₂ = 8.939, p = 0.011; day 3: χ²² = 10.142, p = 0.006; day 4: χ²₂ = 15.830, p < 0.001; day 5: χ²₂ = 10.712, p = 0.005). B Primary path length (m). All three genotypes showed a significant decrease over the five training days (Friedman’s ANOVA, WT: χ²₄ = 34.880, p < 0.001; App^NL-G-F: χ²₄ = 27.6²₄, p < 0.001; App^NL-G-F/Pde4b^Y358C: χ²₄ = 37.867, p < 0.001). Genotypic differences were observed on days 4 and 5 (Kruskal-Wallis test, day 4: χ²₂ = 10.078, p = 0.006; day 5: χ²₂ = 9.441, p = 0.009). C Number of errors. All three genotypes showed a significant decrease over the five training days (Friedman’s ANOVA, WT: χ²₄ = 26.726, p < 0.001; App^NL-G-F: χ²₄ = 15.075, p = 0.005; App^NL-G-F/Pde4b^Y358C: χ²₄ = 34.562, p < 0.001). Genotypic differences were observed on days 4 and 5 (Kruskal-Wallis test, day 4: χ²₂ = 11.461, p = 0.003; day 5: χ²₂ = 7.209, p = 0.027). D Time (s) spent in target quadrant. Genotypic differences were observed (ANOVA, F_2,41 = 4.304, p = 0.02). E Time (s) spent in target sector. Genotypic differences were observed (ANOVA, F_2,41 = 6.111, p = 0.005). F Number of head entries into the target hole annulus. Genotypic differences were observed (ANOVA, F_2,41 = 5.547, p = 0.007). G Heat maps showing the cumulative time spent in localities of the arena. 12-month-old WT (n = 15), App^NL-G-F (n = 17) and App^NL-G-F/Pde4b^Y358C (n = 15) mice. Data are plotted as mean ± SEM. *p < 0.05; **p < 0.01 vs. WT. ^#p < 0.05; ^##p < 0.01 vs. App^NL-G-F. Open circles, females; closed circles, males; broken line, chance level.

Fig. 2. Protective effect of hypomorphic PDE4B on brain metabolism in App^NL-G-F mice assessed by ¹⁴C-2-DG brain imaging.

App^NL-G-F mice show hypometabolism that is corrected in App^NL-G-F/Pde4b^Y358C mice in multiple subfields of the PFC (A), cerebral cortex (B), septum (C), thalamus (D), basal ganglia (E), and mesolimbic pathway (F). App^NL-G-F also show hypometabolism selectively in the DHSLM and SLM that was corrected in App^NL-G-F/Pde4b^Y358C mice (G, H). In other hippocampal subfields, App^NL-G-F mice did not show hypometabolism, whereas metabolism was enhanced in these subfields in App^NL-G-F/Pde4b^Y358C mice. 13-month-old WT (n = 15), App^NL-G-F (n = 17) and App^NL-G-F/Pde4b^Y358C (n = 15) mice. Data are plotted as mean ± SEM. Student’s t-test with Bonferonni-Holm post hoc correction for multiple comparisons. *p < 0.05; **p < 0.01 vs. WT. ^#p < 0.05; ^##p < 0.01 vs. App^NL-G-F. Brain region abbreviations are defined in Table S1.

Fig. 3. Exemplar ¹⁴C-2-DG color-coded autoradiographs obtained from coronal brain sections of 13-month-old WT, App^NL-G-F, and App^NL-G-F/Pde4b^Y358C mice.

A Orbitofrontal cortex (+2.46 mm from bregma). B Medial PFC ( + 1.98 mm from bregma). C Hippocampus (−3.16 mm from bregma). Higher rates of metabolism are indicated by red/orange and lower rates indicated by yellow/white. Scale bar indicates tissue ¹⁴C concentration (nCi/g).

Fig. 4. Partial attenuation of neuroinflammation in App^NL-G-F mice by hypomorphic PDE4B.

A Heatmap analysis of Z scores of 32 inflammatory markers, each column indicating a different mouse. App^NL-G-F mice show elevated levels of inflammatory markers, which are partly decreased in App^NL-G-F/Pde4b^Y358C mice. Composite Z scores for cytokines (B), chemokines (C), growth factors (D), and all 32 markers (total; E) are elevated in App^NL-G-F mice but modulated by the PDE4B inhibition in App^NL-G-F/Pde4b^Y358C mice. 12-month-old WT (n = 7), App^NL-G-F (n = 6) and App^NL-G-F/Pde4b^Y358C (n = 5) mice. Data are plotted as mean ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001 vs. WT. ^#p < 0.05; ^##p < 0.01; ^###p < 0.001 vs. App^NL-G-F. Open circles, females; closed circles, males; ♀, significant difference in females only.

Fig. 5. Modulation of differentially expressed cerebral cortical genes in App^NL-G-F mice by hypomorphic PDE4B.

A Ide (mean of 7 transcripts). B Btaf1 (mean of 3 transcripts). C Padi2 (mean of 2 transcripts). D C1qb (1 transcript). E Immunoreactivity of IDE protein normalized to β-Actin in brains. Genotypic differences were observed (Kruskal-Wallis test, χ²₁₈ = 14.259, p = 0.001). F Typical blot of 50 µg protein from brains probed with anti-IDE and anti-β-Actin antibodies. 12-month-old WT (n = 7), App^NL-G-F (n = 6) and App^NL-G-F/Pde4b^Y358C (n = 5) mice. Data are plotted as mean ± SEM. **adj. p < 0.01, ****adj. p < 0.0001 vs. WT; ^####adj. p < 0.0001 vs. App^NL-G-F for A–D. ***p < 0.001 vs. WT; ^##p < 0.01 vs. App^NL-G-F for (E).