Vitamin C deficiency in the brain impairs cognition, increases amyloid accumulation and deposition, and oxidative stress in APP/PSEN1 and normally aging mice

Abstract

Subclinical vitamin C deficiency is widespread in many populations, but its role in both Alzheimer's disease and normal aging is understudied. In the present study, we decreased brain vitamin C in the APPSWE/PSEN1deltaE9 mouse model of Alzheimer's disease by crossing APP/PSEN1(+) bigenic mice with SVCT2(+/-) heterozygous knockout mice, which have lower numbers of the sodium-dependent vitamin C transporter required for neuronal vitamin C transport. SVCT2(+/-) mice performed less well on the rotarod task at both 5 and 12 months of age compared to littermates. SVCT2(+/-) and APP/PSEN1(+) mice and the combination genotype SVCT2(+/-)APP/PSEN1(+) were also impaired on multiple tests of cognitive ability (olfactory memory task, Y-maze alternation, conditioned fear, Morris water maze). In younger mice, both low vitamin C (SVCT2(+/-)) and APP/PSEN1 mutations increased brain cortex oxidative stress (malondialdehyde, protein carbonyls, F2-isoprostanes) and decreased total glutathione compared to wild-type controls. SVCT2(+/-) mice also had increased amounts of both soluble and insoluble Aβ1-42 and a higher Aβ1-42/1-40 ratio. By 14 months of age, oxidative stress levels were similar among groups, but there were more amyloid-β plaque deposits in both hippocampus and cortex of SVCT2(+/-)APP/PSEN1(+) mice compared to APP/PSEN1(+) mice with normal brain vitamin C. These data suggest that even moderate intracellular vitamin C deficiency plays an important role in accelerating amyloid pathogenesis, particularly during early stages of disease development, and that these effects are likely modulated by oxidative stress pathways.

Keywords: Alzheimer’s disease; Vitamin C; amyloid; cognition; mouse models; oxidative stress.

Figure 1. Learning and memory tasks

At 5 months of age (5M) SVCT2^+/−APP/PSEN1⁺ mice were the only group to show no decrease in interest of the familiar odor on the second day of testing (24 hour recall) (A, left). Only wild-type mice showed evidence of a significant preference for the novel over the familiar odor on day 2. In contrast, SVCT2^+/−APP/PSEN1⁺ mice spent significantly less time exploring the novel odor than the familiar odor (A, right). At 5 months of age, mice carrying APP/PSEN1 mutations were impaired on the Y-maze alternation task regardless of vitamin C status (B), whereas at 12 months of age (12M) impairments were seen in SVCT2^+/− and SVCT2^+/−APP/PSEN1⁺ mice compared to mice with normal vitamin C levels (C). The older cohort of mice was also tested in the conditioned fear paradigm. All groups spent similar time freezing during the 4-minute trial in the original training context (D). However, mice carrying APP/PSEN1 mutations spent less time freezing in a new context, when they heard the tone that had been the conditioned stimulus, indicating impaired cue recall in these mice (E). Data shown are mean +S.E.M. p<0.05 t-test between exploration of odors (time in seconds, log₁₀ transformed) on two separate trials (left: familiar on day 1 versus day 2, right: novel versus familiar on day 2); * p<0.05, main effect of SVCT2^+/− compared to SVCT2^+/+, or main effect of APP/PSEN1⁺ compared to APP/PSEN1⁻.

Figure 2. Morris water maze learning

Mice were trained on the hidden version of the water maze for 8 days to ensure that all mice had been given sufficient trials to learn the location of the hidden platform (A, E). Swimming locations during the no-platform probe trial were recorded and analyzed by group according to quadrant (target versus non-target), time spent within 20 cm of the platform edge, and crosses of the platform location, as depicted in (I). At both 5 months (5M) and 12 months (12M), each group showed some degree of preference for swimming in the target quadrant, although these preferences were clearer in mice that did not carry APP/PSEN1 mutations (B, F). Time spent in the 50 cm-diameter zone around the platform (20 cm from platform edge) was lower in APP/PSEN1 mutant mice at 5 months (C), but did not differ according to group at 12 months (G). Number of crosses of the platform location was similar across groups at 5 months (D), but was significantly lower in all mice with low vitamin C (SVCT2^+/−) at 12 months of age (H). Data shown are mean +S.E.M. p<0.05 non-target quadrants compared to target quadrant for each group; * p<0.05, Main effect of SVCT2^+/− compared to SVCT2^+/+, or main effect of APP/PSEN1⁺ compared to APP/PSEN1⁻.

Figure 3. Activity, anxiety and neuromuscular co-ordination

Locomotor activity was tested in two 15-minute sessions on two consecutive days (D1, D2). At 5 months of age (5M) there were no differences according to genotype (A), but at 12 months (12M) SVCT2^+/− mice were slightly hyperactive compared to SVCT2^+/+ mice with normal vitamin C levels (B). Performance on the elevated zero maze was similar among groups at 5 months (C), but at 12 months SVCT2^+/− mice traveled further than SVCT2^+/+ mice (D). Performance on the accelerating rotarod, tested across 2 days (D1, D2), was significantly poorer in SVCT2^+/− mice than SVCT2^+/+ at both 5 months (E) and 12 months of age (F). Data shown are mean +S.E.M. * p<0.05, **p<0.01, main effect of SVCT2^+/− compared to SVCT2^+/+.

Figure 4. Measures of oxidative stress, antioxidant status and neuroinflammation

Mice that had undergone behavioral testing were sacrificed at 6 months (6M: “5-month” group) and 14 months of age (14M: “12-month” group). Lacking one copy of the SVCT2 successfully lowered vitamin C in the brains of SVCT2^+/− mice (A). Low vitamin C significantly increased MDA at both ages, and at 5 months, MDA was also higher in APP/PSEN1⁺ mice (B). Protein carbonyls were also higher in all groups compared to wild-type mice at 6 months, but by 14 months of age the levels of carbonyls were similar regardless of genotype (C). F₂-isoprostanes were increased in mice carrying APP/PSEN1 mutations, but only in 6-month old mice (D). Total glutathione levels were higher in wild-type mice than all other groups at 6 months, but no further differences were seen at 14 months (E). APP/PSEN1 mutations increased GFAP expression as detected by Western blot in 14-month old mice, although no differences were apparent in the younger cohort (F). Data shown are mean +S.E.M. *p<0.05, **p<0.01, ***p<0.001, Main effect of SVCT2^+/− compared to SVCT2^{+/+; +}p<0.05, main effect of APP/PSEN1⁺ compared to APP/PSEN1⁻; p<0.05 different from wild-type controls according to pairwise comparisons following significant interaction between genotypes in the Univariate ANOVA (comparisons between APP/PSEN1⁺ and wild-type at each level of SVCT2, and between SVCT2^+/− and SVCT2^+/+ within APP/PSEN1⁺ or wild-type mice).

Figure 5. Measurement of amyloid-β

Amyloid-β levels are only reported in mice that carry APP/PSEN1 mutations. At 6 months of age (6M), soluble and insoluble amyloid-β_1-40 and amyloid-β_1-42 were very low, and insoluble amyloid proteins of both lengths were more numerous in mice with low vitamin C (A, C). This relationship was confirmed by the higher ratio of total amyloid-β_1-42 /amyloid- β_1-40 in SVCT2^+/−APP/PSEN1⁺ mice with low brain vitamin C (E). In the older cohort that was sacrificed at 14 month of age (14M), amyloid-β_1-40 and amyloid-β_1-42 levels were much higher than in the younger mice, but did not differ according to group (B, D). Neither did the ratio of total amyloid- β1-42 /amyloid-β1-40 differ according to vitamin C level at that age (F). Thioflavin-S-positive plaques were infrequent and small in 6-month old mice and coverage did not vary according to vitamin C level (G). By 14 months of age plaque deposits were far more numerous and were also significantly greater in SVCT2^+/−APP/PSEN1⁺ mice compared to APP/PSEN1⁺ mice with normal vitamin C transport (H). Data shown in panels A to F are mean +S.E.M. Data shown in panels G and H are medians, +quartiles, with maximum and minimum group values represented by whiskers. *p<0.05, SVCT2^+/−APP/PSEN1⁺ compared to SVCT2^+/+APP/PSEN1⁺ mice. Example images of thioflavin-S stained sections are given in (I). Panels Ii and Iii show representative sections of the average percent coverage values for SVCT2^+/−APP/PSEN1⁺ and SVCT2^+/+APP/PSEN1⁺ mice at 6 months of age. Panels Iiii and Iiv depict representative sections of the average coverage for 14-month-old mice.

Figure 6. Experimental design

Behavioral testing was begun at 5 months, or 12 months of age. Mice were sacrificed for biochemical measures following behavioral testing, at 6 months, or 14 months of age.