Alzheimer's disease pathology is attenuated in a CD38-deficient mouse model

Abstract

Objective: Alzheimer's disease (AD)-associated dementia is due to tissue damage caused by amyloid β (Aβ) deposition within the brain and by accompanying neuroinflammation. The nicotinamide adenine dinucleotide (NAD) glycohydrolase CD38, which is expressed by neurons, astrocytes, and microglial cells, regulates inflammatory and repair processes in the brain and other tissues by degrading NAD and repressing the activity of other NAD-consuming enzymes and by producing NAD-derived metabolites that regulate calcium signaling and migration of inflammatory cells. Given the role of CD38 in neuroinflammation and repair, we examined the effect of CD38 deletion on AD pathology.

Methods: We crossed APPswePS1ΔE9 (APP.PS) mice with Cd38(-) (/) (-) mice to generate AD-prone CD38-deficient animals (APP.PS.Cd38(-) (/) (-) ) and examined AD-related phenotypes in both groups.

Results: APP.PS.Cd38(-) (/) (-) mice exhibited significant reductions in Aβ plaque load and soluble Aβ levels compared to APP.PS mice, and this correlated with improved spatial learning. Although CD38 deficiency resulted in decreased microglia/macrophage (MM) accumulation, the transcription profile of the Cd38(-) (/) (-) and Cd38(+/) (+) MM was similar, suggesting that the decreased Aβ burden in APP.PS.Cd38(-) (/) (-) mice was not due to alterations in MM activation/function. Instead, APP.PS.Cd38(-) (/) (-) neuronal cultures secreted less Aβ and this reduction was mimicked when APP.PS neuronal cultures were treated with inhibitors that blocked CD38 enzyme activity or the signaling pathways controlled by CD38-derived metabolites. Furthermore, β- and γ-secretase activity was decreased in APP.PS.Cd38(-) (/) (-) mice, which correlated with decreased Aβ production.

Interpretation: CD38 regulates AD pathology in the APP.PS model of AD, suggesting that CD38 may be a novel target for AD treatment.

Figure 1. Loss of CD38 reduces Aβ plaque load in APP.PS mice

Brain sections of APP.PS and APP.PS.Cd38^−/− mice, 8 and 14 months of age were stained with 4G8 anti-Aβ mAb. (A) Representative images of APP.PS and APP.PS.Cd38^−/− mice at 8 and 14 months of age. Lower panels represent magnifications of the selected areas in the upper panels marked by rectangles. Scale bar = 1 mm and 200 μm for higher and lower panels, respectively. (B) Representative high magnification (20×) images of 8 and 14 month old APP.PS and APP.PS.Cd38^−/− mice. Scale bar = 200 μm. (C, D) Quantification of the stained area representing Aβ plaques (C) or the number of Aβ plaques (D) (**p < 0.005, ***p < 0.0005, Student’s t test). (E) Analysis of the average plaque area (*p < 0.05, **p < 0.005, Student’s t test). (F) Plaques were categorized according to their size to three groups: small (250–500 μm²), medium (500–1000 μm²) and large (> 1000 μm²) plaques. Similar distribution of the plaques in APP.PS and APP.PS.Cd38^−/− mice was observed, both at 8 and 14 months of age. The quantified values shown are presented as mean ± SEM (bars) (n = 8 and 6 for APP.PS and APP.PS.Cd38^−/− aged 8 months, respectively and n = 8 for APP.PS and APP.PS.Cd38^−/− aged 14 months).

Figure 2. Loss of CD38 reduces soluble and insoluble Aβ peptide but not APP levels in APP.PS mice

(A,B) Immunoblot analysis of Aβ peptide levels. Soluble (A) and insoluble (B) protein extracts were prepared as described in Materials and Methods. Proteins (175 and 2 μg for soluble and insoluble respectively) were separated by SDS-PAGE and membranes were probed with anti-Aβ Abs. Aβ levels in the soluble fraction were normalized to a ~12 KDa band in the Ponceau staining (lower panels). (C, D) Quantification of immunoblot results of soluble (C) and insoluble (D) Aβ levels (*p < 0.05, **p < 0.005, ***p < 0.0005, Student’s t test). (n = 8). (E, G) Immunoblot analysis of APP levels. Cortical protein extracts were prepared from brains of APP.PS and APP.PS.Cd38^−/− mice at 8 (E) and 14 (G) months of age. Proteins (50 μg) were separated on SDS-PAGE and membranes were probed with anti-APP Abs as described in Materials and Methods. The membranes were cut into two parts and each part was stained separately for either APP or for cytochrome c for normalization. APP levels were normalized to cytochrome c (Cyt c). (F, H) Quantification of APP immunoblot results at 8 (F) and 14 (H) months of age. (NS, not significant; Student’s t test) The quantified values shown are presented as the mean ± SEM (bars). (n = 8 for each genotype).

Figure 3. Reduction in α-, β- and γ-secretase activity in APP.PS.Cd38^−/− mice

Brain hemispheres were isolated from 10 week old APP.PS and APP.PS.Cd38^−/− mice, membranes where purified and used for secretase activity measurements as described in Materials and Methods. α-secretase (A) β-secretase (B) and γ-secretase (C) activities were measured over time using fluorometric assays and the enzymatic activities were calculated as described in Materials and Methods. (D–F) The slope of the linear range of the reaction representing the α-secretase (D), β-secretase (E) or γ-secretase (F) enzyme activity rates was calculated for each mouse. Displayed are the calculated slopes, normalized to APP.PS mice. (*p < 0.05, **p < 0.005, ***p < 0.0005, student’s t-test). The values are presented as the mean ± SEM (bars). (n = 8).

Figure 4. mRNA and protein levels’ measurements of the different secretases

(A) Total RNA samples, prepared from brain hemispheres, were analyzed by RT-qPCR as described in Materials and Methods to determine relative mRNAs levels of the γ-secretase components; human PS1 transgene (PSEN1), mouse PS1 (Psen1), Psen2, Ncstn, Psenen, Aph1a, Aph1b and Aph1c. The results are expressed as fold induction for each animal in both groups relative to one APP.PS mouse. The data are presented as individual value plots displaying the relative expression level of the indicated gene for each mouse. (n = 4–5). (*p < 0.05, Student’s t test). (B–G) Protein levels of ADAM10, β and γ-secretase components. Brain hemispheres from 10 week old APP.PS and APP.PS.Cd38^−/− mice were dissected, and protein extracts were prepared as described in Materials and Methods. (B) Proteins (50 μg) were separated on SDS-PAGE and probed with antibodies against the indicated proteins. Cytochrome c (cyt c) was used as loading control. The membranes were cut into two parts and each part was stained separately for either the corresponding protein or for cytochrome c for normalization. (C–G) Densitometry was used to normalize protein levels in the samples to cytochrome c. Quantification of the normalized results did not reveal a difference in ADAM10 (C), nicastrin (D), PS1 (E), PS2 (F) and BACE1 (G) protein levels between the groups. The values are presented as the mean ± SEM (bars). (n = 8).

Figure 5. Regulation of neuronal Aβ production by CD38, NAD and the CD38 generated metabolite, cADPR

(A) Aβ levels in primary neuronal cultures. Primary neuronal cultures were prepared from brains isolated from APP.PS, APP.PS.Cd38^−/−, WT and Cd38^−/− neonatal mice and the amount of secreted Aβ present in the cultures was measured as described in Materials and Methods. (p = 0.01 for transgene and p = 0.005 for CD38, two-way ANOVA).(n = 16, 15, 22 and 17 cultures for WT, Cd38^−/−, APP.PS and APP.PS.Cd38^−/−, respectively). (B, C) Determination of NAD levels. Brains from 2, 4, and 8 month old APP.PS, APP.PS.Cd38^−/−, WT and Cd38^−/− mice were homogenized and intracellular NAD levels were measured as described in Materials and Methods. (B) NAD levels in WT and Cd38^−/− mice (p = 3×10⁻⁶ for CD38, Two-way ANOVA, *p = 0.006 and ***p = 8×10⁻⁶, Fisher’s LSD post hoc test; n = 5). (C) NAD levels in APP.PS and APP.PS.Cd38^−/− mice (p = 4×10⁻⁸ for CD38, Two-way ANOVA, ***p < 5×10⁻⁴, Fisher’s LSD post hoc test; n = 4). (D) Determination of NAD levels in primary neuronal cultures. Intracellular NAD levels were measured, as described in Materials and Methods, in cells isolated from the primary neuronal cultures. (p = 5×10⁻⁵ for CD38, Two-way ANOVA, *p < 0.05 and ***p < 0.0005, Fisher’s LSD post hoc test) (n = 15 for APP.PS and APP.PS.Cd38^−/− mice; n = 16 for WT and Cd38^−/− mice. 1–2 samples from each mouse). (E) The effect of NAD on Aβ levels. Primary neuronal cultures were treated with 100 μM NAD and Aβ levels were measured as described in Materials and Methods. (***p = 0.0001; p = 0.07 in WT cultures, Student’s t test). (n = 8 and 7 for APP.PS control and NAD, respectively, n = 8 for WT cultures). Results are expressed relative to the control (untreated) of the relevant genotype. (F) Effect of CD38 enzymatic inhibitors and cADPR antagonists on Aβ levels. Primary neuronal cultures were treated with the cADPR antagonist, 8Br-cADPR (50 μM), or the CD38 inhibitor, luteolin (20 μM), for 96 hours and Aβ levels were measured as described in Materials and Methods. (in APP.PS cultures, p = 0.002, one-way ANOVA, *p = 0.007 and **p = 0.0007 for 8Br-cADPR and luteolin, respectively, Fisher’s LSD post hoc test). In APP.PS.Cd38^−/− cultures, no significant effect was observed for any of the treatments (n = 8). Results are shown relative to the control (untreated) of the relevant genotype.

Figure 6. CD38 regulates MM accumulation in the brains of APP.PS AD-prone mice

(A) Analysis of IL-1β and TNFα mRNA levels in MM isolated from APP.PS and APP.PS.Cd38^−/−. MM were isolated from the brains of the two groups at 1.5, 3, 8 and 14 months of age, RNA was extracted and gene expression was examined by qRT-PCR as described in Materials and Methods. (For IL-1β, p = 0.04 for genotype and p = 0.03 for age, two-way ANOVA, *p < 0.05, Fisher’s LSD post-hoc test. For TNFα, p = 0.0005 for age and p = 0.02 for the age × genotype interaction, two-way ANOVA, **p < 0.005, Fisher’s LSD post-hoc test). The values are presented as the mean ± SEM (bars) (n = 2 pools of 2–3 mice for each group). (B) Representative images of Iba1 staining in hippocampi of APP.PS and APP.PS.Cd38^−/− mice at 1.5 and 8 months of age. Images were captured at a 4× (upper panel) and 20× magnification (lower panel). Rectangles indicate areas taken for 20× magnification. Scale bar = 1000 μm or 200 μm for upper and lower panels, respectively. Higher resolution representative (60×) images in the in hippocampi of 8 month old APP.PS and APP.PS.Cd38^−/− mice are shown in (C). Scale bar = 50 μm. (D) Quantification of the total area stained by Iba1. The results are expressed as the ratio between APP.PS.Cd38^−/− and APP.PS. (*p < 0.05 and **p < 0.005, student’s t-test). (E) Quantification of the area occupied by the MM cell bodies in the hippocampus. Differentiation between cell bodies and extension was based on higher Iba1 staining intensity in cell bodies than in extensions. Thus, implementing differential range of intensity of the Iba1 staining, cell bodies were identified as high intensity staining areas and extensions were identified as low intensity staining areas. The results are expressed as the ratio between APP.PS.Cd38^−/− and APP.PS. (*p < 0.05 and **p < 0.005, student’s t-test), (F) Quantification of the area stained by the MM extensions in the hippocampus. The results are expressed as the ratio between APP.PS.Cd38^−/− and APP.PS. (**p < 0.005 and ***p < 0.0005, student’s t-test). (G) Quantification of the ratio between the area of the MM extensions and the MM cell bodies. The results are expressed as the ratio between APP.PS.Cd38^−/− and APP.PS. (*p < 0.05, student’s t-test).The values are presented as the mean ± SEM (bars). (n = 8 for all groups except for APP.PS.Cd38^−/− 3 months of age, n = 6). (H) Correlation between the extent of Aβ and Iba-1 staining at 8 months of age. For each mouse, the average total area stained by Aβ was plotted against the average total area stained by Iba1 in the hippocampus. A significant correlation was found only in APP.PS mice (pearson’s r = 0.89, **p < 0.005). (n = 8 for APP.PS and n = 6 for APP.PS.Cd38^−/− mice).

Figure 7. Loss of CD38 improves spatial learning in APP.PS mice

(A) Morris Water Maze Test. 10 month old APP.PS, APP.PS.Cd38^−/−, WT and Cd38^−/− mice were tested. The results presented are the average latency of 4 daily trials to reach the hidden platform for each of the 8 days of the learning phase of the task. (p < 10⁻¹⁷ for day and p = 2×10⁻¹² for transgene, repeated measures ANOVA, p = 5×10⁻⁵ between APP.PS and APP.PS.Cd38^−/− mice, p = 0.66 between WT and Cd38^−/− mice, p = 10⁻⁵ between APP.PS.Cd38^−/− and WT mice, p = 2×10⁻⁶ between APP.PS.Cd38^−/− and Cd38^−/− mice, Fisher’s LSD post hoc test). (n = 8, 11, 9, 10 for APP.PS, APP.PS.Cd38^−/−, WT and Cd38^−/− mice respectively). (B) APP.PS and APP.PS.Cd38^−/− were tested at 14 months of age. The results presented are the average latency of 4 daily trials to reach the hidden platform for each of the 8 days of the learning phase of the task. (p < 10⁻¹⁷ for day and p = 0.0017 for CD38, repeated measures ANOVA). (n = 20). (C, E) Visible platform test. The results shown are latencies to climb a visible platform. No significant difference (student’s t-test) was observed between the latencies of APP.PS mice and APP.PS.Cd38^−/− mice at 10 (C) or at 14 months of age (E). (D, F) Single probe trial test. The test was conducted 24 h after the last learning trial. The results shown are the percentage of the time that the mice spent in the quadrant in which the platform was previously located. No significant difference (student’s t-test) was observed between APP.PS and APP.PS.Cd38^−/− mice at 10 (D) or 14 months of age (F). The values are presented as the mean ± SEM (bars). (G–J) Expression of synaptophysin. The cortex of APP.PS and APP.PS.Cd38^−/− mice at 8 (G) and 14 (H) months of age was dissected, and protein extracts prepared as described in Materials and Methods. Proteins (50 μg) were separated on SDS-PAGE and probed with anti-synaptophysin (SYP) or with anti-cytochrome c (Cyt c) Abs as loading control. (I, J) Quantification of the normalized results at 8 (I) and 14 (J) months of age. (*p = 0.02, Student’s t test). The values are presented as the mean ± SEM (bars). (n = 8).