Tau reduction in aged mice does not impact Microangiopathy

Abstract

Microangiopathy, including proliferation of small diameter capillaries, increasing vessel tortuosity, and increased capillary blockage by leukocytes, was previously observed in the aged rTg4510 mouse model. Similar gene expression changes related to angiogenesis were observed in both rTg4510 and Alzheimer's disease (AD). It is uncertain if tau is directly responsible for these vascular changes by interacting directly with microvessels, and/or if it contributes indirectly via neurodegeneration and concurrent neuronal loss and inflammation. To better understand the nature of tau-related microangiopathy in human AD and in tau mice, we isolated capillaries and observed that bioactive soluble tau protein could be readily detected in association with vasculature. To examine whether this soluble tau is directly responsible for the microangiopathic changes, we made use of the tetracycline-repressible gene expression cassette in the rTg4510 mouse model and measured vascular pathology following tau reduction. These data suggest that reduction of tau is insufficient to alter established microvascular complications including morphological alterations, enhanced expression of inflammatory genes involved in leukocyte adherence, and blood brain barrier compromise. These data imply that 1) soluble bioactive tau surprisingly accumulates at the blood brain barrier in human brain and in mouse models, and 2) the morphological and molecular phenotype of microvascular disturbance does not resolve with reduction of whole brain soluble tau. Additional consideration of vascular-directed therapies and strategies that target tau in the vascular space may be required to restore normal function in neurodegenerative disease.

Keywords: Alzheimer’s disease; Blood vessels; Brain microvessels; Microangiopathy; Neurodegeneration; Tau.

Fig. 1

Tau protein in human vasculature. a Blood vessel isolations containing capillaries were validated by immunofluorescent labeling. DAPI-positive nuclei were also positive for endothelial cell markers such as von Willebrand factor (vWF) and zona occludins (ZO-1). b Isolates also contained basement membrane (collagen IV, ColIV) and some vessels were arteriolar in origin, indicated by the presence of smooth muscle actin (SMA). c Glut1 labeled endothelial cells were also occasionally observed to be surrounded by tau (HT7 antibody). d Western blotting of blood vessel isolations from human temporal cortex samples with varying degrees of tau pathology (Braak stage) indicates that vessels are frequently positive for tau (DAKO), but not other neuronal components (NeuN). Vessels are also enriched in Glut1 compared to total brain extracts. e When applied to a tau biosensor cell assay, vessels appear to retain their bioactivity with AD vessels exhibiting elevated seeding potential compared to controls (Mann-Whitney U test, p = 0.007, A.U. = arbitrary units). f Total frontal cortex protein from subjects with frontotemporal lobar dementia (FTLD) with tau exhibited greater bioactivity than control brains (Student’s t test, p = 0.038). g Total tau protein in total brain versus blood vessel protein preparations as measured by ELISA. All graphs are plotted with means +/− standard deviations. * indicates p < 0.05, ** p < 0.01

Fig. 2

Tau protein in vasculature from transgenic mice. a Tissue sections from rTg4510 or wild-type controls at 15 months of age were labeled for tau (HT7) and endothelial cells (Glut1). b An enlarged view of the box from (a) shows the extent of tau pathology at this age in transgenic (TG) and not wild-type (WT) mice. c Numerous blood vessels were observed to be closely associated with tau. d Western blotting of total cortex and isolated vessels indicates tau is present in vessels, but not other neuronal components such as NeuN. Glut1 is highly enriched. Vessels and cortex protein were run on the same blot, but separated here for clarity. e Tau bioactivity in a biosensor cell assay shows increased tau bioactivity in transgenic brain versus wild-type (Repeated Measures ANOVA, Sidak’s post hoc, p = 0.003) and in transgenic blood vessels versus wild-type (Sidak’s post hoc, p = 0.01). f Tau protein ELISA measures in total brain versus blood vessel protein. g Expression of the human Mapt transgene was observed by in situ hybridization in neurons but did not co-localize with Glut1-positive endothelial cells. Image is of a single plane captured using a confocal microscope. All graphs are plotted with means +/− standard deviations. * indicates p < 0.05, ** p < 0.01

Fig. 3

Doxycycline treatment reduces soluble tau but does not alter vascular tau. a Sections from rTg4510 mice (TG) with and without doxycycline treatment were labeled for tau protein (HT7) showing a significant reduction in tau outside of neurofibrillary tangles. Insets show higher magnification images of cortical pathology. Scale bar = 10 μm. b Western blotting of human tau (HT7) present in Triton X (TX), sarkosyl soluble (SS) and sarkosyl insoluble (SI) fractions from a protein insolubility assay. c Quantification of western blots confirms reduced soluble tau present in TX (Student’s t test, p = 0.001) and (d) SS fractions (p = 0.0002) and (e) no change in SI fractions (p = 0.31). Tau measures were normalized to a total protein stain to control for loading differences, which can be found along with uncropped blots in the supplement. f Western blotting of human tau in brain and isolated brain vessels from doxycycline treated (+) and untreated (−) mice. Glut1 is included to show enrichment of endothelial protein in vessels preparations. g Quantification of total human tau in vessels normalized to Glut1. h A biosensor cell assay shows retained tau bioactivity in transgenic blood vessels from Dox + mice (Repeated Measures ANOVA, Sidak’s post hoc p = 0.04) and no difference in Dox- mice (p = 0.89). All graphs are plotted with means +/− standard deviations. * indicates p < 0.05, ***p < 0.001

Fig. 4

Vascular remodeling is not altered in mice with reduced tau burden. a 100 μm thick z-projection images of fluorescein-labeled vasculature from wild-type (WT) and transgenic (TG) mice captured by two-photon microscopy. b The volume of blood vessels in each mouse was calculated and normalized to the cortical thickness to account for atrophy. Blood vessel volume was increased in transgenic mice and was not altered by treatment (Two-way ANOVA, genotype p = 0.04, treatment p = 0.4). c Total blood vessel length was also increased in transgenics (Two-way ANOVA, genotype p = 0.007, treatment p = 0.74). d The total percentage of all vessels imaged without red blood cell flow was also increased (Two-way ANOVA, genotype p < 0.0001, treatment p = 0.19). All graphs are plotted with means +/− standard deviations. * indicates p < 0.05, ** p < 0.01, ***p < 0.001

Fig. 5

Increased cell adhesion protein expression in tau expressing mice. a Quantitative PCR revealed an increase in Vcam1 (Two-way ANOVA, age p = 0.0117, genotype p = 0.0002, interaction p = 0.01), Icam1 (age p < 0.0001, genotype p < 0.0001, interaction p < 0.0001), and Icam2 (age p < 0.04, genotype p < 0.004, interaction p = 0.05). Increased expression was observed in transgenic 12-month mice and 15-month mice; Sidak’s multiple comparisons p-values are indicated. b Doxycycline treated and untreated transgenic mice had elevated Vcam1 (Two-way ANOVA, genotype p < 0.0001, treatment p = 0.06, interaction p = 0.56), Icam1 (genotype p = 0.002, treatment p = 0.90, interaction p = 0.94) and Serpine1 (genotype p = 0.02, treatment p = 0.16, interaction p = 0.19), but not Icam2 (genotype p = 0.13, treatment p = 0.50, interaction p = 0.74). Untreated 15-month mice are the same in panels (a) and (b). RQ = relative quatification using comparative Ct method. All graphs are plotted with means +/− standard deviations. * indicates p < 0.05, ** p < 0.01

Fig. 6

Compromised blood brain barrier function is not restored in tau expressing mice. a Sections from wild-type (WT) and transgenic (TG) mice were labeled to detect mouse IgG, which is normally excluded from the brain or found only within vascular lumens. b Enlarged views of boxes from (a) show that mouse IgG was detected in both wild-type and transgenic cortex, but to a greater extent in transgenic mice, including within vascular walls and in cells with glial morphology (inset). c A threshold-based quantification of cortical IgG labeling shows an increased in transgenic mice (Two-way ANOVA, genotype p < 0.0001, treatment p = 0.06). d A Western blot of tight junction proteins ZO-1 and Occludin (Ocln) as well as Glut1 (endothelial cell marker) and tubulin (loading control). e Quantification of the western blot from (d) showing ZO-1 protein normalized to Glut1 amount. Results are shown as the percentage versus wild-type (Dox-). No difference was seen in ZO-1 (Two-way ANOVA, genotype p = 0.59, treatment p = 0.50). f Quantification of Occludin normalized to Glut1 revealed a decrease in transgenic mice (Two-way ANOVA, genotype p = 0.003, treatment p = 0.71) including a significant difference between WT DOX+ and TG DOX+ (Sidak’s multiple comparison, p = 0.02). g Total vascular protein loaded on the Western blot was not significantly different between groups and is plotted normalized to tubulin (Two-way ANOVA, genotype p = 0.09, treatment p = 0.89)