Hypoxia promotes tau hyperphosphorylation with associated neuropathology in vascular dysfunction

Abstract

Background: Hypertension-induced microvascular brain injury is a major vascular contributor to cognitive impairment and dementia. We hypothesized that chronic hypoxia promotes the hyperphosphorylation of tau and cell death in an accelerated spontaneously hypertensive stroke prone rat model of vascular cognitive impairment.

Methods: Hypertensive male rats (n = 13) were fed a high salt, low protein Japanese permissive diet and were compared to Wistar Kyoto control rats (n = 5).

Results: Using electron paramagnetic resonance oximetry to measure in vivo tissue oxygen levels and magnetic resonance imaging to assess structural brain damage, we found compromised gray (dorsolateral cortex: p = .018) and white matter (corpus callosum: p = .016; external capsule: p = .049) structural integrity, reduced cerebral blood flow (dorsolateral cortex: p = .005; hippocampus: p < .001; corpus callosum: p = .001; external capsule: p < .001) and a significant drop in cortical oxygen levels (p < .05). Consistently, we found reduced oxygen carrying neuronal neuroglobin (p = .008), suggestive of chronic cerebral hypoperfusion in high salt-fed rats. We also observed a corresponding increase in free radicals (NADPH oxidase: p = .013), p-Tau (pThr231) in dorsolateral cortex (p = .011) and hippocampus (p = .003), active interleukin-1β (p < .001) and neurodegeneration (dorsolateral cortex: p = .043, hippocampus: p = .044). Human patients with subcortical ischemic vascular disease, a type of vascular dementia (n = 38; mean age = 68; male/female ratio = 23/15) showed reduced hippocampal volumes and cortical shrinking (p < .05) consistent with the neuronal cell death observed in our hypertensive rat model as compared to healthy controls (n = 47; mean age = 63; male/female ratio = 18/29).

Conclusions: Our data support an association between hypertension-induced vascular dysfunction and the sporadic occurrence of phosphorylated tau and cell death in the rat model, correlating with patient brain atrophy, which is relevant to vascular disease.

Keywords: Hypertension; Hypoxia; Neurodegeneration; SHRSP; Tau.

Fig. 1.

Structural alterations to the GM and impaired WM tracts in the SHRSP/JPD model. (A). Representative slices of T2-MRI maps from WKY and SHRSP/JPD and corresponding quantification in the right hemisphere. (B). Regions of interest (ROI), DLCTX and HIPP, were normalized to corresponding head muscle. A significantly higher T2-intensity ratio is observed in DLCTX of SHRSP/JPD (DLCTX: p=.018*) compared with WKY controls. (C). Lower FA in the HIPP of SHRSP/JPD model (p=.013*) compared to WKY controls reflects axonal damage to WM tracts. No changes were observed between groups in the DLCTX. (D). Quantification of ASL maps shows diminished CBF in the DLCTX (p=.005*) and HIPP (p < .001**) of SHRSP/JPD model compared to controls. (E). DLCTX and HIPP ROI of total brain VOL. Hippocampal region quantification suggests elevated brain edema (p=.040*), which is absent from WKY controls. (F). T2-MRI intensity ratios in the DLCTX and HIPP of the SHRSP/JPD model suggest that most of the neuropathological damage occurs in the right hemisphere compared to the left in both groups (*p < 0. 001). Data represent WKY (n=3) and SHRSP/JPD (n=13). (G). Quantification of CC and EC WM tracts in the right hemisphere of SHRSP/JPD model compared to controls. Increased pixel intensity T2-MRI ratios normalized to corresponding head muscle in SHRSP/JPD versus WKY controls (CC: p=.016*, EC: p=.049**). (H). No significant changes were observed in CC and EC FA measurements indicative of lack of axonal damage in these regions. (I). ASL shows a significant reduction in CBF in CC (p=.001*) and EC (p < .001**) brain ROI. Data is representative of WKY (n=3) and SHRSP/JPD (n=13).

Fig. 2.

Hgb leakage from the peripheral circulation occurs through an opened BBB in the SHRSP/JPD model. Red blood cell-derived Hgb from the peripheral circulation may enter the brain through a damaged BBB at 15W timepoint. (A). Representative images of IHC triple staining for IgG (BBB leakage, red), RECA1 (endothelial cell, green) markers. Microbleeds suggest increased BBB permeability in SHRSP/JPD compared to WKY controls (merged images, arrows). Bar=20 μm; 20× magnification. (B). Higher DAB immunoreactivity of Hgbα and Hgbβ is found in the DLCTX and HIPP of the experimental group while low expression is detected in controls. Bar=100 μm, 20×. (C). Stereology showing a higher Hgbα+ cell count in the DLCTX (*p=.047) and HIPP (**p=.043) of the SHRSP/JPD model versus WKY controls. (D). Stereology showing a higher Hgbβ⁺ cell count in the DLCTX (*p < .001) and HIPP (**p=.043) of the SHRSP/JPD model. n=4–5 rats/group. Hgb=Hemoglobin; BBB=blood brain barrier. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 3.

Cortical and hippocampal hypoxic hypoperfusion in the SHRSP/JPD model. (A). Brain section from SHRSP/JPD confirming the Lithium Phthalocyanine crystal implantation in the right DLCTX (arrow) at 10W of life. (B). Timecourse of EPR oximetry measurements of interstitial pO₂ readings obtained weekly from 10W–15W (WKY, n=5, squares; SHRSP/JPD, n=8, circles) demonstrates a significant drop in pO₂ levels at 12W (*p < .05) which remain low until 15W (**p < .01, ***p < .001). (C). Representative WB bands are presented for HIF1α and normalized to actin loading control. Quantification reveals elevated protein expression for HIF1α/actin ratio in the DLCTX (*p=.05; n=4–5/group) and HIPP (**p=.046; n=3/group) brain regions of SHRSP/JPD compared to controls. WB lanes were run on the same gel but are noncontiguous.

Fig. 4.

Hypoxia increases free radical production and reduces Ngb levels. (A). Increased NADPH oxidase activity after 4W of JPD diet in the DLCTX (p=.072, nonsignificant trend) and HIPP (*p=.013) of SHRSP/JPD model compared to age-matched controls. (B). Increased HIPP 3-NT protein expression normalized to actin loading control in total WB brain lysates of SHRSP/JPD versus WKY (*p=.003, n=4/group). (C). Double staining for lipid peroxidation marker (4-HNE, red) with DAPI nuclear counterstain (blue) with quantification. Results show increased free-radical mediated tissue injury in the DLCTX (*p < .001) and HIPP (**p=.006) of the SHRSP/JPD model (arrows) which is absent from controls. Quantification of 4-HNE pixel intensity was measured in n=5 animals/group. Bar=55 mm, 40×. (D). Representative images of IHC triple staining in the DLCTX and HIPP. Increased colocalization (arrows) of Ngb (O₂ carrying molecule, red) with NeuN (neuronal marker, green) and DAPI (nuclear marker, blue) in WKY versus SHRSP/JPD. Bar=10 μm, 40×. (E). ELISA quantification of total brain Ngb protein expression in the DLCTX and HIPP of experimental versus control rats. Results show diminished Ngb levels in the HIPP of SHRSP/JPD (*p=.008, n=6–11) compared with WKY controls (n=4). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 5.

Increased p-Tau but not α-SYNUC and Aβ neuropathology in the SHRSP/JPD model. (A). Representative images showing high AT180⁺ cells (at phosphorylated epitope site pThr231 of tau) DAB staining in the DLCTX and HIPP of SHRSP/JPD absent from WKY controls with corresponding stereology quantification. Bar=20 μm, 20×; n=4–5/group. (B). DLCTX: *p=.011; HIPP: **p=.003. (C). WB analysis of AT180 protein expression normalized to total tau (Tau5) and actin loading control. Representative bands demonstrate higher p-Tau levels in SHRSP/JPD versus WKY age-matched controls (DLCTX: *p=.010; HIPP: **p=.038; n=3–5/group). WB lanes were run on the same gel but are noncontiguous. (D). Representative DAB staining for α-SYNUC indicate no differences between SHRSP/ JPD and WKY groups in studied ROI. Bar=20 μm; 20× magnification, n=4–6/group. (E). Corresponding α-SYNUC WB bands normalized to actin reveal no significant differences between experimental and control groups in the DLCTX and HIPP brain regions. α-SYNUC=alpha Synuclein, n=3–5/group. (F). Representative DAB images for Aβ_1–42 (corresponds to amino acids 33–42 of human Aβ_1–42) show unchanged immunoreactivity in experimental and control groups. Aβ=Beta-Amyloid. Bar=10 μm; 40× magnification, n=3–5/group. (G). ELISA quantification of total Aβ_1–42 levels in DLCTX and HIPP brain lysates did not show significant differences between groups. n=4–7/group. (H). Triple IHC staining for Aβ (red), RECA1 (endothelial cell marker, green) and DAPI (nuclear stain, blue) demonstrates a lack of CAA in SHRSP/JPD and WKY groups. CAA=Cerebral Amyloid Angiopathy; Bar=10 μm; 40× magnification, n=4/group. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 6.

Pro-inflammatory cytokine release and increased neurodegeneration in the GM of the SHRSP/JPD model. (A). IHC double staining showing robust CX3CR1⁺ microglial activation (red) in the HIPP region colocalized with DAPI nuclear staining (blue) of SHRSP/JPD but not in age-matched controls (arrow); Bar=20 μm, 40×. (B). WB quantification showing elevated active- IL-1β levels in hippocampal lysates of SHRSP/JPD compared with controls (p < .001, WKY: n=5; SHRSP/ JPD: n=7). (C). Presence of NeuN⁺ DAB immunoreactivity in the DLCTX and HIPP regions is significantly diminished in SHRSP/JPD compared to age-matched WKY controls. Bar=20 μm representative of 20× and 40× magnification (black boxes represent 40× insert); HIPP 4× view, bar=100 μm and 40× view, bar=10 μm; n=5/group. (D). Stereology quantification of NeuN⁺ DAB staining shows reduced NeuN⁺ cell counts in SHRSP/JPD DLCTX (*p=.043) and HIPP (**p=.044) compared to WKY; n=4–5/group. (E). Presence of cortical apoptotic neurons (TUNEL⁺ (green) and NeuN⁺ (red)) in SHRSP/JPD (n=1) absent from WKY controls (n=1). Bar=10 μm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 7.

Decreased hippocampal volume and cortical thickness in SIVD-BD patients. The normalized hippocampal volume (A) and the cortical thickness (B) were significantly lower in subcortical ischemic vascular disease of the Binswanger type (SIVD-BD) as compared to healthy controls (*p < .05) after correcting for age, gender and scanner magnetic field strength effects. Outliers are represented by a (+) symbol.

Fig. 8.

Proposed working model for neuropathological change in SHRSP/JPD. We postulate that hypertension- induced hypoxia caused by changes in blood vessel morphology (1) promotes Ngb downregulation and free radical generation in neurons (2). Microglia activation and the release of pro-inflammatory cytokines ensues, thus damaging neuronal axons and triggering p-Tau (3) which may lead to Hgb leakage from the periphery through damaged microvasculature of the BBB (4) and apoptotic neuronal cell death (5). Neurotoxic cytokine release from active microglia may also directly promote BBB permeability. Ngb=neuroglobin, p=phosphorylation, BBB=blood brain barrier, Hgb=hemoglobin.