Early involvement of D-serine in β-amyloid-dependent pathophysiology

Abstract

The N-methyl-D-aspartate subtype of glutamate receptors (NMDAR) is a key regulator of brain plasticity encoding learning and memory. In addition to glutamate, NMDAR activation requires the binding of the co-agonist D-serine. The beta-amyloid (Aß) peptide which accumulates in Alzheimer's disease (AD), affects the D-serine-dependent NMDAR activation in vitro, but whether this alteration would significantly contribute to AD-related pathophysiology and memory deficits remains unclear. Herein, we report a decrease in the maximal pool of recruitable NMDAR and in the expression of NMDAR-dependent long-term potentiation together with impaired basal neurotransmission at CA3/CA1 synapses from hippocampal slices of 5xFAD mouse, an AD-related model with elevated Aß levels. The NMDAR synaptic impairments develop from 1.5 to 2 months of age with the initial rise of Aß and is correlated to a transient increase in D-serine levels. Deficits in working and spatial memories as well as cognitive flexibility then occurred in 10-12 months-old animals. Importantly, the NMDA-related synaptic deregulations (but not the altered basal neurotransmission) and behavioral impairments (working and cognitive flexibility) are prevented or reduced (spatial memory) in 5xFAD mice devoid of D-serine after genetic deletion of its synthesis enzyme serine racemase. Altogether, these results therefore provide in vivo evidence for the implication of D-serine at least in the early pathogenic signatures of AD driven by the increase in amyloid load suggesting that the recent proposal of preventive therapy of AD by administration of the precursor L-serine remains questionable.

Keywords: Alzheimer’s disease; Long-term potentiation; Memory deficits; NMDA receptors; Serine racemase.

Fig. 1

SR deletion does not impact the Aβ load and amyloid deposits in the hippocampus of 5xFAD mice. A. Box plots with the median ± mini/max of total Aβ (including Aβ₄₀ and Aβ₄₂ isoforms) concentrations (in ng/mg of protein) determined in hippocampal tissues of 1.5-2 months-old; 3–4 months-old and 10–12 months-old WT, 5xFAD and 5xFAD/SR-KO mice (* P < 0.05, ** P < 0.01 and *** P < 0.0001 compared to WT mice; Kruskal-Wallis test with Dunn’s multiple comparisons). B. Examples of CA1 area from hippocampal sections stained with Congo red solution in a 10–12 months-old WT, 5xFAD and 5xFAD/SR-KO mouse to visualize amyloid deposits (Bar = 50 μm). C. Box plots with median ± mini/max of the number of amyloid plaques/mm²) in the hippocampus of 5xFAD and 5xFAD/SR-KO mice compared to WT (* P < 0.05 compared to WT mice; Kruskal-Wallis test with Dunn’s multiple comparisons. D. Bar graphs with mean ± SEM comparing the quantification of immunoreactivity for SR normalized to β-actin protein levels in hippocampal tissues of WT, SR-KO, 5xFAD and 5xFAD/SR-KO mice. In the insert are representative immunoblots for β-actin (upper band) and SR (lower band) in in each experimental group

Fig. 2

Transient increase in hippocampal D-serine levels induced by the Aβ load in 5xFAD mice A. Representative amino acid chromatogram obtained showing the peaks relative to the elution of L-serine, glutamine and D-serine in a WT (black line) vs. an 1.5-2 months-old (turquoise line) mouse. B. Bar graphs with mean ± SEM comparing the D-serine levels in the hippocampus of 1.5-2 months-old; 3–4 months-old and 10–12 months-old WT (white column) and 5xFAD (turquoise column) mice (* P < 0.05 with two-way ANOVA and Dunnett T3 multiple comparisons test). C. Bar graphs with mean ± SEM comparing the L-serine levels in the hippocampus of 1.5-2 months-old; 3–4 months-old and 10–12 months-old WT and 5xFAD mice

Fig. 3

SR deletion prevents age-dependent changes in CA1 synaptic plasticity of 5xFAD mice. A. Bar graphs with mean ± SEM comparing the basal synaptic transmission calculated as the fEPSP/PFV ratio at 400µA stimulation intensity in slices from 1.5-2 months-old; 3–4 months-old and 10–12 months-old WT (white column), 5xFAD (turquoise column) and 5xFAD/SR-KO (purple column) mice (** P < 0.01, one-way ANOVA with Dunnett T3 multiple comparisons). In the insert are superimposed representative traces obtained in respective 10–12 months-old experimental groups (Bars: 25ms and 0.1mV). B. Bar graphs with mean ± SEM comparing the paired-pulse facilitation ratio in slices of respective aged-related experimental groups. C. Time-course of LTP induced by 1 × 100 Hz tetanus in 10–12 months-old mice ($ P < 0.05 with one sample t test compared to baseline of 100 for the 15 last minutes). In the inserts are superimposed representative traces obtained before and 60 min after the conditioning stimulation for each age group (Bars: 25ms and 0.1mV). D. Bar graphs with mean ± SEM comparing the LTP magnitude determined for the last 15 min of recording in slices from 1.5-2 months-old; 3–4 months-old and 10–12 months-old WT, 5xFAD and 5xFAD/SR-KO mice (# P = 0.06 and ** P < 0.01 with two-way ANOVA in 15 last minutes and Dunnett T3 multiple comparisons test)

Fig. 4

SR deletion prevents age-dependent changes in NMDAR activation of 5xFAD mice. Box plots with the median ± mini/max illustrating the fEPSP/PFV ratio calculated for 400µA stimulation intensity of isolated NMDAR fEPSPs from slices of 1.5-2 months-old (A₁), 3–4 months-old (B₁) and 10–12 months-old (C₁) WT 5xFAD and 5xFAD/SR-KO mice. A₂, B₂, C₂. Box plots with the median ± mini/max illustrating the percentage increase in fEPSP/PFV ratio by the addition of a saturating dose of D-serine (100µM) ($ P < 0.05 one sample Wilcoxon test vs. 0 for each group and * P < 0.05 compared to WT mice Kruskal-Wallis with Dunn’s multiple comparisons test). In right are superimposed respective traces of isolated NMDAR fEPSPs obtained in the three groups before and in the presence of saturating D-serine (100µM) (Bars: 25ms and 0.1mV)

Fig. 5

SR deletion alleviates memory deficits in 10–12-month-old 5xFAD mice. A. Bar graphs with mean ± SEM illustrating the working memory assessed in WT, 5xFAD and 5xFAD/SR-KO mice ($ P < 0.05 univariate t-test compared to 50% chance value). B. Time-course of the swimming distance during the 5 days of spatial learning in WT, 5xFAD and 5xFAD/SR-KO mice (# for P < 0.05 between the first and last day with ANOVA repeated measures and * P < 0.5; *** P < 0.001 for the distance in the last day of training ANOVA one-way with Dunnett T3 multiple comparisons compared to WT mice). C. Bar graphs with mean ± SEM illustrating the percentage of time in the target quadrant assessed during the first probe-test ($ P < 0.05 vs. 25% with univariate t-test). D. Bar graphs with mean ± SEM illustrating the number of crossing the platform zone in the first probe-test (* P < 0.05, *** P < 0.001, ANOVA one-way with Dunnett T3 multiple comparisons). E. Time-course of the spatial learning flexibility after changing the location of the platform (# P < 0.05 between the first and last day with ANOVA repeated measures and *** for P < 0.001 compared to WT mice ANOVA one-way with Dunnett T3 multiple comparisons for the distance in the last day of relearning). F. Bar graphs with mean ± SEM illustrating the percentage of time in the target quadrant assessed during the second probe test ($ P < 0.05 vs. 25% with univariate t-test and * P < 0.05 ANOVA one-way with Dunnett T3 multiple comparisons). G. Bar graphs with mean ± SEM illustrating the number of crossings the platform zone in the second probe test (** P < 0.01 compared to WT mice ANOVA one-way with Dunnett T3 multiple comparisons)