Peroxiredoxin 6 mediates protective function of astrocytes in Aβ proteostasis

Background Disruption of β-amyloid (Aβ) homeostasis is the initial culprit in Alzheimer’s disease (AD) pathogenesis. Astrocytes respond to emerging Aβ plaques by altering their phenotype and function, yet molecular mechanisms governing astrocytic response and their precise role in countering Aβ deposition remain ill-defined. Peroxiredoxin (PRDX) 6 is an enzymatic protein with independent glutathione peroxidase (Gpx) and phospholipase A2 (PLA2) activities involved in repair of oxidatively damaged cell membrane lipids and cellular signaling. In the CNS, PRDX6 is uniquely expressed by astrocytes and its exact function remains unexplored. Methods APPswe/PS1dE9 AD transgenic mice were once crossed to mice overexpressing wild-type Prdx6 allele or to Prdx6 knock out mice. Aβ pathology and associated neuritic degeneration were assessed in mice aged 10 months. Laser scanning confocal microscopy was used to characterize Aβ plaque morphology and activation of plaque-associated astrocytes and microglia. Effect of Prdx6 gene dose on plaque seeding was assessed in mice aged six months. Results We show that hemizygous knock in of the overexpressing Prdx6 transgene in APPswe/PS1dE9 AD transgenic mice promotes selective enticement of astrocytes to Aβ plaques and penetration of plaques by astrocytic processes along with increased number and phagocytic activation of periplaque microglia. This effects suppression of nascent plaque seeding and remodeling of mature plaques consequently curtailing brain Aβ load and Aβ-associated neuritic degeneration. Conversely, Prdx6 haplodeficiency attenuates astro- and microglia activation around Aβ plaques promoting Aβ deposition and neuritic degeneration. Conclusions We identify here PRDX6 as an important factor regulating response of astrocytes toward Aβ plaques. Demonstration that phagocytic activation of periplaque microglia vary directly with astrocytic PRDX6 expression level implies previously unappreciated astrocyte-guided microglia effect in Aβ proteostasis. Our showing that upregulation of PRDX6 attenuates Aβ pathology may be of therapeutic relevance for AD.

There is a significant reduction in the fibrillar and the total A plaque load in the brain cortex and in the hippocampus both in female and male APP/Prdx6 Tg mice when compared to sex-matched APP/Prdx6 +/+ controls. Conversely, APP/Prdx6 +/-mice of both sexes show increase in the fibrillar and the total A plaque load compared to sex-matched APP/Prdx6 +/+ controls. When matched for Prdx6 genotype, female mice have significantly higher fibrillar and total A plaque load in the brain cortex and in the hippocampus. Two-way ANOVA results summarized in Table S1, demonstrate significance of the main effect of sex and the main effect of Prdx6 genotype for all measures. 1 Significant interaction between the main effects is noted for the fibrillar plaque load in the brain cortex and the total A plaque load in the hippocampus. Overall, these findings demonstrate that Prdx6 gene dose inversely modulates A deposition both in female and male mice, but when matched for Prdx6 genotype the load of A plaque is invariably higher in females irrespective of plaque type and brain structure analyzed.

Fig. S2
Segmentation analysis of LSCM images of A plaques. a A representative A plaque from APP/Prdx6 +/+ mice (also used in Fig. 4) triple labeled with X-34 (fibrillar core), anti-A (clone HJ3.4), and anti-GFAP antibodies. A collapsed stack of LSCM tomograms taken at the level of plaque core is presented. b Shown is separation of individual LSCM acquisition channels. Edges of A + label are manually contoured to create the A plaque mask. The mask is then transferred into the other channels to segment remaining components of the plaque. c Shown are black and white images of segmented plaque components as used for ensuing quantitative analysis. A plaque cross-section area is defined as the area within the outline of A plaque mask. "Plaque / core ratio" is calculated by dividing the A plaque cross-section area by the area of X-34 label within the mask, while "GFAP + % plaque area" is the ratio of segmented GFAP + label to A plaque cross-section area. Scale bar: 10 m in a, b, and c

Fig. S3
Overexpression of Prdx6 is associated with increased accumulation of apoE within A plaques, while Prdx6 haplodeficiency conversely reduces apoE plaque content. a Shown are epifluorescent microscopy images of representative A plaques co-labeled with anti-A and anti-apoE antibodies, and with X-34 (fibrillar core) from 10-month-old female mice of indicated genotypes. b Shown is quantitative analysis of the apoE plaque content. Ratio between the area of apoE + label to the area of A + label (ApoE + % plaque area) was quantified in 30 plaques per genotype, and presented as the mean value + SEM. In APP/Prdx6 Tg mice the ApoE + plaque coverage increases to 71.2% (p < 0.0001) from 50.6% in APP/Prdx6 +/+ controls, while in APP/Prdx6 +/-mice it is reduced to 37.6% (p < 0.0001). b p < 0.0001 (ANOVA); ****p < 0.0001 (Holm-Sidak's post hoc test). Scale bar: 20 m in a Also shown are representative, low power, epifluorescent microscopy images of astrocytic clusters associated with A plaques (arrows) and activated astrocytes associated with brain vessels (arrowheads) from the brain cortex of 10-month-old mice of indicated genotypes. All sections are triple-labeled with X-34 (fibrillar Aβ), and with anti-PRDX6, and anti-GFAP antibodies. The photographs evidence the following gradient of PRDX6 expression across the genotypes: APP/Prdx6 Tg > APP/Prdx6 +/+ > APP/Prdx6 +/-. This gradient is apparent both in mice aged 3 months in a and those aged 10 months in b and c, which represent animals before the onset of Aβ deposition and those in which Aβ plaques are present, respectively. There is marked upregulation of PRDX6 expression in astrocytes associated with Aβ plaques in 10month-old mice compared to 3-month-old animals for matching genotypes. LSCM images in b also reveal that upregulation of PRDX6 expression is associated with the spread of PRDX6 immunoreactivity from the soma to astrocytic processes. This effect is enhanced in astrocytes from APP/Prdx6 Tg mice, while in APP/Prdx6 +/-mice it is nearly absent. The photographs in c also evidence expression of PRDX6 by activated, perivascular astrocytes. The expression level of PRDX6 protein by perivascular astrocytes is commensurate to the Prdx6 gene dose across mouse lines. Perivascular astrocytic activation prior to the onset of Aβ deposition was not observed in any of the mouse lines (not shown). Although Prdx6 RNA can be detected in microglia cultured in vitro, its level represents merely 1% of that in astrocytes and does not increase with LPS stimulation. Results of this experiment remain consistent with negative anti-PRDX6 immunostaining of microglia in the brain parenchyma including microglia, which undergoes phagocytic activation around A plaques. Values represent mean + SEM from 6 female mice per genotype. b and c p < 0.0001 in (ANOVA); *p < 0.05, **p < 0.01, and ****p < 0.0001 (Holm-Sidak's post hoc test). Abbreviations: Crtxcortex, Hip -hippocampus. Scale bar: 500 m in a Fibrillar plaque load in APP/Prdx6 Tg mice in the brain cortex is reduced by 17.3% (p < 0.01) and by 20.3% in the hippocampus (p < 0.05) compared to APP/Prdx6 +/+ controls, while in APP/Prdx6 +/mice it is increased in respective structures by 22.7% (p < 0.0001) and 40.4% (p < 0.0001).
APP/Prdx6 Tg mice also show significant reduction in the numerical density of fibrillar plaques by 12.2% in the brain cortex (p < 0.05) and by 22.0% in the hippocampus (p < 0.05) compared to APP/Prdx6 +/+ controls, while in APP/Prdx6 +/-mice plaque density is increased by 30.0% (p < 0.0001) and by 25.5% (p < 0.01) in the brain cortex and in the hippocampus, respectively. In direct comparison with APP/Prdx6 Tg line, the fibrillar plaque load in 6-month-old female APP/Prdx6 +/-mice is 1.5-fold higher in the brain cortex and 1.8-fold higher in the hippocampus (p < 0.0001), while the numerical density of fibrillar plaques is 1.5-fold higher (p < 0.0001) and 1.6fold higher (p < 0.0001) in respective structures.

Quantitative analysis of the apoE plaque content
Free -floating coronal brain sections were pretreated with 10 mM sodium citrate and 0.05% Tween 20 (pH 6.0) for 20 min at 85°C [1], and then triple labeled with X-34, anti-A 1-16 (rabbit polyclonal anti-mouse (1:500; Jackson ImmunoResearch), respectively. Microphotographs of X-34/A/apoE triple-labeled plaques were captured in the brain cortex under x40 objective using 80i Nikon fluorescent microscope equipped in monochrome DSQi1Mc camera and NIS Elements Imaging Software v. 4.13 (Nikon Corp., Tokyo, Japan). Areas of apoE and A labels for each plaque were determined using NIH ImageJ v1.52 and divided by one another to obtain ApoE + % plaque area.
Ten plaques in the brain cortex per animal, from three females per genotype were randomly selected for analysis.

Analysis of C3 expression in plaque-associated astrocytes
Sodium citrate pretreated, free-floating brain sections were triple labeled with X-34, and anti-C3 (rat monoclonal 1:100; Hycult Biotech Inc., Germantown, PA) and anti-GFAP (rabbit polyclonal 1:2000; DAKO) primary antibodies followed by secondary antibodies: Alexa Fluor 594 conjugated goat anti-rat (1:500; Jackson ImmunoResearch) and Alexa Fluor 488 conjugated goat anti-rabbit (1:500; Jackson ImmunoResearch). C3 / GFAP co-labeled astrocytes surrounding the X-34 + core of A plaques were photographed under x40 objective using 80i Nikon fluorescent microscope as described above. Captured images were analyzed using NIH ImageJ v1.52. Individual clusters of plaque-associated astrocytes were manually outlined on digitized images and the integrated densities for C3 and GFAP immunofluorescence were measured within the outlined area of interest as previously described [3]. C3 + / GFAP + Index was calculated by dividing C3 integrated density by that of GFAP for the same plaque. Ten plaques in the brain cortex per animal, from three females per genotype were randomly selected for analysis.

Analysis of Prdx6 RNA level in astrocytes and microglial cells in vitro
Mixed glia cultures were established from the brain cortex of P0-P3 pups as previously described [4] and maintained in Dulbecco's Modified Eagle Medium with F12 Nutrient Mixture (DMEM/F12) 7 PRDX6 Regulates A Proteostasis Pankiewicz, JE et al. (Thermo Fisher Scientific) supplemented with 10% fetal bovine serum (FBS), 5% horse serum, 1 mM sodium pyruvate, 100g/mL streptomycin, and 100U/mL penicillin. At 18 day in vitro (DIV) the culture was treated with 60 mM Lidocaine in Ca 2+ /Mg 2+ -free Hanks' Balanced Salt Solution (Thermo Fisher Scientific) for 15 min to detach microglia cells [5]. The suspension of free-floating cells was collected and immediately centrifuged at 1,000 x g for 5 minutes in the presence of 50 µM EDTA. Resulting pellet was resuspended in a buffer containing 2.6 mM KCl, 137.9 mM NaCl,

month-old animals
The load of fibrillar A plaques was analyzed in female mice aged 6 months on sections stained with Th-S for consistency with the results obtained in 10-month-old animals. The same wholesection approach to plaque load analysis was used as described in the Material and Methods section of the main manuscript. In addition, numerical density of fibrillar plaques on crosssectional profiles of the brain cortex and the hippocampus was determined using NIH ImageJ v1.52.
Phagocytic activity of microglia was assessed by determining the CD68 + / X-34 + Index of nascent plaques. Free-floating brain sections were first pretreated with sodium citrate, as described above, and then stained with X-34 (10 mM in 40% ethanol [pH 10]), and subsequently immunostained with anti-CD68 rat monoclonal antibody(1:200; Abcam Inc.), which was followed by Alexa Fluor 594 conjugated goat anti-rat secondary antibody (1:500; Jackson ImmunoResearch). Three coronal sections from three brains randomly selected per genotype were used for analysis. The sections were taken at the approximated levels of the anterior commissure, the rostral portion of the hippocampus, and the mammillary bodies. The entire brain cortex on each section was photographed under x20 objective using 80i Nikon fluorescent microscope as described above which produced 5-6 non-overlapping, 0.44 mm 2 test areas per cortical profile. X-34 + and CD68 + loads were thresholded and quantified on each test area using NIH ImageJ v1.52. Since the analysis focused on the nascent plaques, infrequent mature plaques, arbitrary defined as having X-34 + core > 40um 2 were edited out from digitized images. CD68 + / X-34 + Index was calculated by dividing the CD68 + load by the X-34 + load for each test area.
9 Table S1. Two-way ANOVA analysis of fibrillar and A plaque load in the brain cortex and in the hippocampus in 10-month-old female and male littermates from APP/Prdx6 Tg , APP/Prdx6 +/+ , and APP/Prdx6 +/-transgenic mouse lines.

Parameter
Main