3.1. ES affects GFAP expression in 10-month-old in WT mice.
ES led to physiological signs of stress in the pups, including decreased body weight gain between P2-P9 (t(7)=4.596, p=0.002), increased adrenal gland weight (t(9)=-3.591, p=0.006), and decreased DG volume at P9 (t(11)=2.368, p=0.037) (data not shown).
The effect of ES on astrocyte GFAP coverage, GFAP+ cell density and GFAP+ cell complexity in WT animals was analyzed across different ages. At P9, there was significantly increased GFAP coverage in the SLM, which was unaffected in the other hippocampal subregions (fig. 1A (CTL) and fig. 1B (ES); HPC: t(7.095)=-2.029, p=0.082; DG: t(11)=-1.876, p=0.087; CA: t(7.372)=-2.190, p=0.063 fig. 1C; SLM t(11)=-3.289, p=0.007 fig. 1D). No differences in GFAP+ cell numbers were found (hilus: t(11)=-1.158, p=0.272; CA1: t(11)=-0.458, p=0.656; SLM: t(11)=-0.869, p=0.403 fig. 1E), nor could differences in cell complexity be demonstrated (number of intersections: F(1,286)=0.001, p=0.971 fig. 1F; number of primary processes: t(286)=0.607, p=544) in the SLM of P9 mice. Gene expression analyses revealed no changes in mRNA expression of any of the astrocyte markers between CTL and ES at P9 (see table 2).
At P30, no differences in GFAP coverage were found in the whole HPC or hippocampal subregions (HPC: t(8)=-0.995, p=0.349; DG: t(8)=-0.721, p=0.492; CA: t(8)=-1.070, p=0.316).
At 6mo, there were no differences in the HPC and its subregions in GFAP coverage (fig. 1G (CTL) and fig. 1H (ES) HPC: t(10)=2.054, p=0.067; DG: t(10)=1.909, p=0.085; CA: t(10)=2.127, p=0.059 fig. 1I; SLM t(10)=2.128, p=0.059 fig. 1J). A decrease in GFAP+ cell density after ES was found in the hilus (t(10)=3.169, p=0.010), but not in other hippocampal subregions (CA1: t(10)=-0.956, p=0.362; SLM: t(10)=1.472, p=0.172 fig. 1K; DG: t(10)=0.773, p=0.457). Furthermore, no alteration in cell complexity was found at 6 mo (number of intersections: F(1,261)=0.7064, p=0.4014 fig. 1L; primary processes: Mann-Whitney U: p=0.144).
3.2. ES and amyloid pathology do not affect expression of GFAP and astrocyte-related genes at 4 months.
In figure 2, example images of GFAP coverage of 4-month-old transgenic mice are shown for the HPC: APP/PS1-CTL (fig. 2A), APP/PS1-ES (fig. 2B), and the EC: APP/PS1-CTL (fig. 2D), APP/PS1-ES (fig. 2E). At 4 mo, there was no difference in GFAP coverage induced by either ES or APP1/PS1 overexpression in hippocampal subregions or the EC (HPC: Fcondition(1,31)=0.000, p=0.986, Fgenotype(1,31)=1.650, p=0.208, Fcondition*genotype(1,31)=1.700, p=0.202 fig. 2C; DG: Fcondition(1,31)=0.094, p=0.761, Fgenotype(1,31)=0.967, p=0.333, Fcondition*genotype(1,31)=1.425, p=0.242; CA: Fcondition(1,31)=0.059, p=0.810, Fgenotype(1,31)=1.690, p=0.203, Fcondition*genotype(1,31)=1.879, p=0.180; EC: Fcondition(1,28)=0.807, p=0.377, Fgenotype(1,28)=0.982, p=0.330, Fcondition*genotype(1,28)=0.604, p=0.444 fig. 2F).
Gene expression analyses show that ES increased Fasn gene expression levels at 4 months without affecting the other genes that were analyzed (see table 2).
3.3. The effects of amyloid pathology and ES in the EC and HPC of 10-month-old mice
3.3.1 Amyloid pathology increases GFAP expression in the EC at 10 months, which is not further affected by ES.
In figure 3, example images of GFAP coverage of 10-month-old mice are shown for the EC of WT-CTL (fig. 3A), WT-ES (fig. 3B), APP/PS1-CTL (fig. 3C), and APP/PS1-ES (fig. 3D). Clustering of GFAP is observed in APP/PS1 but not WT mice. In the EC, APP/PS1 causes a global increase in GFAP coverage that is not further affected by ES (EC: Fcondition(1,20)=0.488, p=0.493, Fgenotype(1,20)=12.002, p=0.002, Fcondition*genotype(1,20)=3.447, p=0.078 fig. 3E). Furthermore, APP/PS1 increased cytoskeletal complexity of GFAP+ cells in the EC (EC; AUC: Fcondition(1,46.637)=0.586, p=0.448, Fgenotype(1,69.271)=9.653, p=0.003, Fcondition*genotype(1,66.065)=0.847, p=0.361 fig. 3F, 3G; primary processes: Fcondition(1,24)=0.890, p=0.355, Fgenotype(1,24)=12.240, p=0.002, Fcondition*genotype(1,24)=0.882, p=0.357 fig. 3H).
3.3.2. Effects of amyloid pathology and ES on global and clustered GFAP expression and astrocyte-related gene expression in the HPC at 10 months.
In figure 4, example images of GFAP coverage of 10-month-old mice are shown for the HPC of WT-CTL (fig. 4A), WT-ES (fig. 4B), APP/PS1-CTL (fig. 4C), and APP/PS1-ES (fig. 4D). Clustering of GFAP is observed in APP/PS1 but not WT mice. At 10mo, global GFAP expression in the HPC was affected by both APP/PS1 overexpression and ES (HPC: Fcondition(1,20)=11.914, p=0.003, Fgenotype(1,20)=6.419, p=0.020, Fcondition*genotype(1,20)=4.458, p=0.048 fig. 4E). Post hoc analyses revealed a significant difference between CTL and ES in WT animals, with GFAP coverage being decreased after ES exposure (WT-CTL–WT-ES: p=0.001 fig. 4E), but not in APP/PS1 animals (APP/PS1-CTL–APP/PS1-ES: p=832 fig. 4E). APP/PS1-CTL mice had significantly decreased GFAP coverage compared to WT-CTL mice (WT-CTL–APP/PS1-CTL: p=0.015 fig. 4E) but this was not the case in ES mice (WT-ES–APP/PS1-ES: p=0.991 fig. 4E). As for the specific HPC subregions, in the DG, expression of global GFAP was differentially affected by APP/PS1 in ES animals as compared to CTL animals (DG: Fcondition(1,20)=12.830, p=0.002, Fgenotype(1,20)=2.499, p=0.130, Fcondition*genotype(1,20)=10.874, p=0.004). While GFAP expression was unaltered by APP/PS1 in CTL animals (WT-CTL–APP/PS1-CTL: p=0.608), it was increased after ES in transgenic mice (WT-ES–APP/PS1-ES: p=0.015). Also here, ES decreased GFAP coverage in WT (WT-CTL–WT-ES: p<0.001), but not APP/PS1 mice (APP/PS1-CTL–APP/PS1-ES: p=0.998).
In the CA region of the HPC, both ES exposure and APP/PS1 overexpression decreased global GFAP coverage (CA: Fcondition(1,20)=10.030, p=0.005, Fgenotype(1,20)=21.457, p<0.001, Fcondition*genotype(1,20)=1.893, p=0.184 fig. 4F). Interestingly, when GFAP coverage was separately analyzed in the SLM region of the CA, APP/PS1 increased while ES decreased global GFAP expression (SLM: Fcondition(1,20)=20.190, p<0.001, Fgenotype(1,20)=13.812, p=0.001, Fcondition*genotype(1,20)=0.786, p=0.386 fig. 4G). Cell complexity analyses showed no differences in either the number of intersections (HPC AUC: Fcondition(1,24)=2.109, p=0.159, Fgenotype(1,24)=0.314, p=0.580, Fcondition*genotype(1,24)=1.195, p=0.285 fig. 4H) or primary processes (HPC: Fcondition(1,24)=1.506, p=0.232, Fgenotype(1,24)=1.643, p=0.212, Fcondition*genotype(1,24)=1.058, p=0.314).
Although global GFAP coverage was decreased by APP/PS1, clustering of GFAP immunoreactive signal was observed in the transgenic animals (fig. 4A, B, C, D). For this reason, we performed an additional analysis aimed at unraveling the alterations in clustering of GFAP (see methods). Example pictures of the masking are shown in figure 4 I, J, K, L. We confirm the earlier found decrease in GFAP as induced by ES in the HPC (fig. 4M). However, clustered GFAP signal was increased in APP/PS1 mice (HPC: Fcondition(1,20)=19.972, p<0.001, Fgenotype(1,20)=33.749, p<0.001, Fcondition*genotype(1,20)=4.288, p=0.052 fig. 4M). Similar effects were found in the CA (CA: Fcondition(1,20)=13.734, p=0.001, Fgenotype(1,20)=5.970, p=0.024, Fcondition*genotype(1,20)=2.451, p=0.133 fig. 4N) and SLM (SLM: Fcondition(1,20)=19.974, p<0.001, Fgenotype(1,20)=31.041, p<0.001, Fcondition*genotype(1,20)=0.127, p=0.725 fig. 4O).
In the DG, expression of GFAP was differentially affected by APP/PS1 in ES animals as compared to CTL animals (DG: Fcondition(1,20)=12.375, p=0.002, Fgenotype(1,20)=77.601, p<0.001, Fcondition*genotype(1,20)=6.695, p=0.018). Further post hoc testing revealed a significant difference between CTL and ES in WT animals, with GFAP coverage being decreased after ES exposure (WT-CTL–WT-ES: p<0.001), but not in the APP/PS1 animals (APP/PS1-CTL–APP/PS1-ES: p=0.935). APP/PS1 significantly increased GFAP coverage in both CTL animals (WT-CTL–APP/PS1-CTL: p=0.001) and ES animals (WT-ES–APP/PS1-ES: p<0.001).
Gene expression levels at 10mo were affected by APP/PS1 overexpression, but not by ES (see table 2). In 10-month-old animals, APP/PS1 overexpression increased mRNA levels of Aqp4, Gfap, and Vimentin, and decreased Aldh1l1.
3.4. The interaction of astrocytes and microglia in ES and amyloid pathology
We have previously described microglial profile and amyloid load in the same cohort of mice used in this study (summarized in fig. 5A). This allowed us to investigate the link between GFAP coverage and this current set of data. As Aβ protein in APP/PS1 mice is present as cell-associated amyloid at early stages of pathology, and as extracellular plaques at later stages of pathology, we used cell-associated amyloid for 4mo data and plaque load for 10mo data.
Normalized GFAP coverage to cell-associated amyloid is increased in the DG but not CA region in 4mo APP/PS1 mice (DG: Mann-Whitney U=5.0, p=0.0221, fig. 5B, CA: t(11)=1.532, p=0.1538, not shown). In the 10mo APP/PS1 mice clustered, but not global, GFAP signal in the DG when normalized to extracellular amyloid plaque-load is decreased, (masked GFAP DG: t(6)=4.691, p=0.0034; masked GFAP CA: t(6)=0.4, p=0.703; global GFAP CA: t(7)=0.603, p=0.566; global GFAP DG: t(7)=2.096, p=0.0743). To further explore interactions between the stainings for astrocytes (GFAP), microglia (Iba1, CD68) and Amyloid pathology (6E10), we created correlation matrices of the data derived from the stainings, shown in figure 5D (4-month-old APP/PS1 mice) and figure 5E (10-month-old APP/PS1 mice). Correlation coefficients and matrices were also calculated and created for WT mice (Supplementary Fig. 1).
3.4.1. Interaction of GFAP, amyloid pathology, and microglia at 4 months
Amyloid pathology did not correlate with GFAP (CA: r=0.211, p=0.469; DG: r=0.422, p=0.133) or Iba1 (CA: r=0.328, p=0.215; DG: r=-0.055, p=0.841) expression in the HPC. In contrast, the number of Aβ-positive cells at 4mo was negatively correlated to expression of the microglial phagocytic marker CD68 (DG, r=-0.561, p=0.046, fig. 5F; CA, r=-0.4594, p=0.099). When exploring the relationship between astrocyte and microglial markers in the HPC in WT and APP/PS1 mice, while we do not find significant correlations between astrocytes and microglial markers in the HPC, the magnitude of the correlation coefficients in WT mice (Supplementary Fig. 1) were much stronger than in APP/PS1 mice. This was the case for both the CA (WT GFAP vs Iba1: r=0.376, p=0.124; WT GFAP vs CD68: r=0.504, p=0.056; APP/PS1 GFAP vs Iba1: r=0.151, p=0.606; APP/PS1 GFAP vs CD68: r=0.022, p=0.947) and the DG (WT GFAP vs Iba1: r=0.3910, p=0.109; WT GFAP vs CD68: r=0.414, p=0.125; APP/PS1 GFAP vs Iba1: r=-0.116, p=0.694; APP/PS1 GFAP vs CD68: r=0.153, p=0.655).
3.4.2. Interaction of GFAP, amyloid pathology and microglia at 10 months.
Plaque load did not significantly correlate with Iba1 (CA: r=0.043, p=0.906; DG: -0.283, p=0.428), CD68 (CA: r=0.170, p=0.687; DG: -0.398, p=0.329), or GFAP (CA: r=-0.599, p=0.089; DG: -0.064, p=0.870). Similarly, there was no correlation at this age between total Aβ and either Iba1 (CA: r=0.073, p=0.841; DG: -0.119, p=0.743), CD68 (CA: r=0.084, p=0.843; DG: r=-0.166, p=0.694) or GFAP (CA: r=-0.473, p=0.198; DG: -0.085, p=0.827). Regarding the relationship between astrocytes and microglia, hippocampal GFAP expression was not significantly correlated with Iba1 expression in 10-month-old WT (CA: r=0.140, p=0.620; DG: 0.311, p=0.260) or APP/PS1 animals (CA: r=-0.184, p=0.636; DG: 0.080, p=0.839). There was similarly no correlation between GFAP expression and CD68 coverage in WT (CA: r=0.331, p=0.248; DG: 0.230, p=0.428) or APP/PS1 (CA: r=0.640, p=0.122; DG: 0.235, p=0.612) animals, although there was a significant correlation in the CA region of APP/PS1 animals between CD68 and masked GFAP coverage (CA: r=0.779, p=0.039, fig. 5G; DG: 0.216, p=0.642). Similar to the 4mo data, there were several strong positive correlations between these proteins between hippocampal subregions in both WT (not shown) and APP/PS1 (fig. 5E) mice, although plaque load (r=0.218, p=0.545) and GFAP (GFAP: r=0.393, p=0.296; masked GFAP: r=0.320, p=0.401) of APP/PS1 mice were not correlated between the CA and DG. GFAP was correlated in WT animals between both subregions (GFAP: r=0.885, p<0.001; masked GFAP: r=0.860, p<0.001).