Cognitive Effects of Simulated Galactic Cosmic Radiation Are Mediated by ApoE Status, Sex, and Environment in APP Knock-In Mice

Abstract

Cosmic radiation experienced during space travel may increase the risk of cognitive impairment. While simulated galactic cosmic radiation (GCRsim) has led to memory deficits in wildtype (WT) mice, it has not been investigated whether GCRsim in combination with genetic risk factors for Alzheimer's disease (AD) worsens memory further in aging mice. Here, we investigated the central nervous system (CNS) effects of 0 Gy (sham) or 0.75 Gy five-ion GCRsim or 2 Gy gamma radiation (IRR) in 14-month-old female and male APP^NL-F/NL-F knock-in (KI) mice bearing humanized ApoE3 or ApoE4 (APP;E3F and APP;E4F). As travel to a specialized facility was required for irradiation, both traveled sham-irradiated C57BL/6J WT and KI mice and non-traveled (NT) KI mice acted as controls for potential effects of travel. Mice underwent four behavioral tests at 20 months of age and were euthanized for pathological and biochemical analyses 1 month later. Fecal samples were collected pre- and post-irradiation at four different time points. GCRsim seemed to impair memory in male APP;E3F mice compared to their sham counterparts. Travel tended to improve cognition in male APP;E3F mice and lowered total Aβ in female and male APP;E3F mice compared to their non-traveled counterparts. Sham-irradiated male APP;E4F mice accumulated more fibrillar amyloid than their APP;E3F counterparts. Radiation exposure had only modest effects on behavior and brain changes, but travel-, sex-, and genotype-specific effects were seen. Irradiated mice had immediate and long-term differences in their gut bacterial composition that correlated to Alzheimer's disease phenotypes.

Keywords: APP mouse models; Alzheimer’s disease; apolipoprotein E; galactic cosmic radiation; gamma radiation; gut microbiome; sex differences.

Figure 1

Performance of APP;E3F, APP;E4F, and wildtype (WT) mice in open field (OF) test was affected by travel, sex, and genotype, while in rotarod test, it was influenced by irradiation, travel, and genotype. Fourteen-month-old female and male APP;E3F and APP;E4F were irradiated with 0 Gy, 0.75 Gy simulated galactic cosmic radiation (GCRsim), or 2 Gy gamma at Brookhaven National Laboratory (BNL). WT mice were only sham-irradiated and traveled to BNL. All mice underwent behavioral testing 6 months post-irradiation, together with age-matched transgenic, non-traveled controls. (A–C) Effects of radiation, travel, sex, or genotype on locomotor activity, measured as total ambulatory distance in the OF, were analyzed. (D–F) Anxiety-like behavior was quantified as percentage time spent in the center of the OF for the full duration from 0–30 min. (G–I) Anxiety-like behavior as percentage center duration between 0–5 min. (J–L) Motor coordination was measured as latency to fall on the rotarod. (M–O) Percentage improvement between two rotarod trials was used to assess motor learning. #: p < 0.1, *: p < 0.05, **: p < 0.01. Data (A,B,D,E,G,H,J,K,M,N) were analyzed using three-way ANOVAs with Fisher’s least significant difference (LSD). Data (C,F,I,L,O) were analyzed using two-way ANOVAs with Fisher’s LSD to determine significance within sex and unpaired t-tests between sex. p values were not adjusted for multiple comparisons.

Figure 2

Performance of APP;E3F, APP;E4F, and WT mice in spatial novelty Y-maze (SNYM) test was influenced by radiation, travel, and genotype, while no effects on cognition in novel object recognition (NOR) test were detected. Fourteen-month-old female and male APP;E3F and APP;E4F were irradiated with 0 Gy, 0.75 Gy GCRsim, or 2 Gy gamma at BNL. WT mice were only sham-irradiated and traveled to BNL. All mice underwent behavioral testing 6 months post-irradiation, together with age-matched transgenic, non-traveled controls. (A–C) Long-term memory after 24h was measured with NOR (% time spent with novel object). (D–F) Locomotion was quantified as % distance in the novel arm of the SNYM. (G–H) Short-term memory after 2 min was analyzed as % time spent in the novel arm of the SNYM. #: p < 0.1, *: p < 0.05, **: p < 0.01. Data (A,B,D,E,G,H) were analyzed using three-way ANOVAs with Fisher’s LSD. Data (C,F,I) were analyzed using two-way ANOVAs with Fisher’s LSD to determine significance within sex and unpaired t-tests between sex.

Figure 3

GCRsim and gamma radiation had no effects on microhemorrhages and Aβ levels in APP;E3F and APP;E4F mice 7 months post-irradiation. (A–C) Microhemorrhages were quantified based on Prussian blue staining for hemosiderin deposits across the whole-hemibrain section. (D–F) Total amyloid beta load was assessed by S97 Aβ immunohistochemical staining and quantified by the % region of interest (ROI) in the hippocampus (HC) using ImageJ (G–I). Fibrillar amyloid load was examined by Thioflavin S staining and quantified as the % ROI in the HC via ImageJ. #: p < 0.1, *: p < 0.05, **: p < 0.01. Data (A,B,D,E,G,H) were analyzed using three-way ANOVAs with Fisher’s LSD. Data (C,F,I) were analyzed using two-way ANOVAs with Fisher’s LSD to determine significance within sex and unpaired t-tests between sex. p values were not adjusted for multiple comparisons.

Figure 4

GCRsim and gamma radiation had effects on cholesterol and triglyceride levels in the plasma of APP;E3F and APP;E4F mice 7 months post-irradiation. (A–C) Total cholesterol concentration in the plasma was quantified using a Fujifilm Cholesterol E assay kit. (D–F) High-density lipoprotein (HDL) cholesterol concentration in the plasma was assessed via a Fujifilm HDL-Cholesterol E assay kit. (G–I) Triglyceride concentration in the plasma was determined by the use of a Fujifilm LabAssay Triglyceride assay kit. #: p < 0.1, *: p < 0.05, **: p < 0.01. Data (A,B,D,E,G,H) were analyzed using three-way ANOVAs. Data (C,F,I) were analyzed using two-way ANOVAs for within-sex comparisons and multiple t-tests within genotypes. p values were not adjusted for multiple comparisons.

Figure 5

GCRsim and gamma irradiation decreased plasma cytokine levels in APP;E3F and APP;E4F mice 7 months post-irradiation. (A–X) Cytokine levels in the plasma were quantified using an MSD V-PLEX Proinflammatory Panel 1 ELISA. #: p < 0.1, *: p < 0.05, **: p < 0.01. Data (A,B,D,E,G,H,J,K,M,N,P,Q,S,T,V,W) were analyzed using three-way ANOVAs. Data (C,F,I,L,O,R,U,X) were analyzed using two-way ANOVAs for within-sex comparisons and multiple t-tests within genotypes. Data with highly different variances were compared with either one-way ANOVA or multiple t-test. p values were not adjusted for multiple comparisons.

Figure 5

GCRsim and gamma irradiation decreased plasma cytokine levels in APP;E3F and APP;E4F mice 7 months post-irradiation. (A–X) Cytokine levels in the plasma were quantified using an MSD V-PLEX Proinflammatory Panel 1 ELISA. #: p < 0.1, *: p < 0.05, **: p < 0.01. Data (A,B,D,E,G,H,J,K,M,N,P,Q,S,T,V,W) were analyzed using three-way ANOVAs. Data (C,F,I,L,O,R,U,X) were analyzed using two-way ANOVAs for within-sex comparisons and multiple t-tests within genotypes. Data with highly different variances were compared with either one-way ANOVA or multiple t-test. p values were not adjusted for multiple comparisons.

Figure 6

Effects of radiation, travel, sex, and genotype on ApoE and Aβx-42 levels in brain homogenates of GCRsim- and gamma-irradiated APP;E3F and APP;E4F mice 7 months post-irradiation. (A–C) ApoE concentration in the brain was quantified using an in-house sandwich ELISA. (D–F) Levels of Aβx-42 in the brain were quantified using an ELISA. #: p < 0.1, *: p < 0.05, ***: p < 0.001, ****: p < 0.0001. Data (A,B,D,E) were analyzed using a three-way ANOVA. Data (C,F) were analyzed using a two-way ANOVA for comparisons between sex and multiple t-tests within genotypes. Data with highly different variances were compared with either one-way ANOVA or multiple t-test. p values were not adjusted for multiple comparisons.

Figure 7

Gamma- and GCRsim-irradiated mice had immediate and long-term differences in their gut bacterial composition that correlated to Alzheimer’s disease phenotypes. Male (M) and female (F) APP-ApoE3 knock-in (E3) and APP-ApoE4 knock-in (E4) mice were transported (Tr) to the Brookhaven National Laboratory facility at 14 months of age and were exposed to either 0.75 GCRsim, 2 Gy gamma, or 0 Gy (Sham) radiation. Fecal pellets from the mice were collected before (0 days), the day after, 145 days after, and 190 days after irradiation and analyzed with 16S rRNA sequencing. Differences in the relative abundance of bacteria were analyzed with linear discriminant analysis (LDA) effect size (LEfSe). Several amplicon sequence variants (ASVs) were differently abundant one day after irradiation compared to sham controls. The graph shows each bacterial taxa that were significantly different as their LDA score (higher than 0 is more abundant, lower is less abundant) plotted against the logarithm of their average relative abundance (A–H). Heatmaps of bacteria that were significantly different from baseline at both 145 days and 190 days post-irradiation. All colored boxes had a p < 0.05 (I,J). Heatmap representing the Spearman R of relative abundance correlated with S97 and Thioflavin S (ThioS) staining of amyloid plaques and % of time spent in the novel arm of the forced-alternation Y-maze. All colored boxes had a p < 0.05 (K). The relative abundance over time of ASV a8b0bb4465364446253d4efd0ec16a89 (Muribaculaceae), which had a significantly higher abundance in male E4 mice and correlated with higher plaque burden in the same group (L,M). The relative abundance over time of ASV 598b197a5e45205c84efc7cb64258bbb (Candidatus stoquefichus), which had a significantly higher abundance in male E3 mice and correlated with higher plaque burden in the same group (N,O).

Figure 8

Genotype influenced the gut microbial composition in control and irradiated mice. Male (M) and female (F) APP-ApoE3 knock-in (E3) and APP-ApoE4 knock-in (E4) mice travelled (Tr) to the Brookhaven National Laboratory facility at 14 months of age and were exposed to either 0.75 GCRsim, 2 Gy gamma, or 0 Gy (Sham) radiation. Fecal pellets from the mice were collected before (0 days), the day after, 145 days after, and 190 days after irradiation and analyzed with 16S rRNA sequencing. Bacterial taxa that were detected in at least two-thirds of the samples were included in the analysis. Differences in the relative abundance of bacteria were analyzed with linear discriminant analysis (LDA) effect size (LEfSe) at the amplicon sequence variant (ASV) level. Unweighted permutational analysis of variance (Adonis test) of sex, genotype, treatment, travel status, and timepoint of the microbiome samples (A). Unweighted UniFrac distance colored by genotype and split by type of radiation (B). Differently abundant ASVs one day after irradiation compared to sham controls. The graph shows each bacterial taxa that were significantly different as their LDA score (higher than 0 is more abundant, lower is less abundant) plotted against the logarithm of their average relative abundance (C). Heatmaps of bacteria that were significantly different from baseline at both 145 days and 190 days post-irradiation. All colored boxes had a p < 0.05 (D).