Amyloid-β and APP deficiencies cause severe cerebrovascular defects: important work for an old villain

Abstract

Alzheimer's disease (AD) is marked by neuritic plaques that contain insoluble deposits of amyloid-β (Aβ), yet the physiological function of this peptide has remained unclear for more than two decades. Using genetics and pharmacology we have established that Aβ plays an important role in regulating capillary bed density within the brain, a function that is distinct from other cleavage products of amyloid precursor protein (APP). APP-deficient zebrafish had fewer cerebrovascular branches and shorter vessels in the hindbrain than wild-type embryos; this phenotype was rescued by treatment with human Aβ peptide, but not a smaller APP fragment called p3. Similar vascular defects were seen in zebrafish treated with a β-secretase inhibitor (BSI) that blocked endogenous Aβ production. BSI-induced vascular defects were also improved by treatment with human Aβ, but not p3. Our results demonstrate a direct correlation between extracellular levels of Aβ and cerebrovascular density in the developing hindbrain. These findings may be relevant to AD etiology where high levels of Aβ in the brain parenchyma precede the development of neuritic plaques and dense aberrantly-branched blood vessel networks that appear between them. The ability of Aβ to modify blood vessels may coordinate capillary density with local metabolic activity, which could explain the evolutionary conservation of this peptide from lobe-finned fish to man.

Figure 1. Schematic of APP processing that produces Aβ and p3 peptides.

APP is initially cleaved by either α-secretase or β-secretases to yield a C87 in the alpha pathway, or C99 in the beta pathway, respectively. These cleavage events also produce an extracellular soluble APP (sAPP) fragment, from the amino terminus, that is slightly longer with α-secretase cleavage. The C87 and C99 fragments are subsequently cleaved within the transmembrane domains (tm) by γ-secretase to produce p3 and Aβ peptides, respectively. Both γ-secretase events produce an Aβ-intracellular-domain (AICD) fragment that is entirely cytosolic. Variability of γ-secretase cleavage on C99 produces Aβ fragments from 39–43 amino acids - only Aβ(1–42) is shown – and similar variability with C87 cleavage.

Figure 2. Cerebrovascular defects in APP-deficient zebrafish embryos.

(A) Dark field (top) and fluorescence (bottom) images of a control transgenic embryo at 3 dpf shows vascular structures dues to EGFP expression in endothelial cells. (B) Confocal image (projected stack) of cerebrovascular structures in the head of the fish in A. (C) Dark field (top) and fluorescence (bottom) images of a zAPP-MO embryo at 3 dpf. (D) Confocal image (projected stack) of cerebrovascular structures in the head of the fish in C. (E) Dark field (top) and fluorescence (bottom) images of a ctrl-MO embryo at 3 dpf. (F) Confocal image (projected stack) of cerebrovascular structures in the head of the fish in E. (G) Graph showing the number of CtA branches in control (N = 30), zAPP-MO (N = 15), and ctrl-MO (N = 15) zebrafish at 3 dpf (***, P < 8.9e-16). (H) Mean CtA branch lengths in control (N = 28), zAPP-MO (N = 14), and ctrl-MO (N = 8) embryos at 3 dpf (***, P < 9.8e-23); scale bars = 100 μm.

Figure 3. Aβ rescued vascular defects in APP-deficient (zAPP-MO) zebrafish embryos at 3 dpf.

(A) Confocal image (projected stack) of a control zebrafish embryo at 3 dpf. (B) Comparable image of a zAPP-MO embryo at 3 dpf. (C) Cerebrovascular structures of an Aβ-treated zAPP-MO were similar to non-injected controls. (D) p3 treatment did not rescue vascular defects in zAPP-MO embryos. (E) Graph of CtA branch numbers in embryos in the control (N = 30), zAPP-MO (N = 15), Aβ-treated zAPP-MO (N = 15), and p3-treated zAPP-MO (N = 15) embryos at 3 dpf. Differences between control and zAPP-MO were significant (P < 8.9e-16), but there were no significant differences between Aβ-treated zAPP-MO and control or ctrl-MO. p3-treated zAPP-MO had significantly fewer branches than control embryos (P < 8.7e-14). (F) Graph of mean CtA branch lengths in control (N = 28), zAPP-MO (N = 14), Aβ-treated zAPP-MO (N = 10), and p3-treated zAPP-MO (N = 10) embryos at 3 dpf. Differences between control and zAPP-MO were significant (P < 9.8e-23), but there were no significant differences between Aβ-treated zAPP-MO and control or ctrl-MO. p3-treated zAPP-MO had significantly shorter vessel lengths than control embryos (P < 1.3e-15).

Figure 4. Aβ-deficiency induced by BSI-treatment caused vascular defects that were rescued by Aβ, but not p3.

(A) Confocal image of cerebrovascular structures in an untreated control zebrafish at 3 dpf (control). (B) Vascular structures in a BSI-treated embryo at 3 dpf (BSI). (C) Cerebrovascular structures in a BSI-treated embryo that was treated with Aβ (BSI+Aβ) showed rescue of the vascular defects. (D) Vascular defects in a BSI-treated embryo were not rescued by p3 (BSI+p3). (E) Graph of CtA branch numbers in control (N = 30), BSI (N = 29), BSI+Aβ (N = 8), and Aβ+p3(N = 10) embryos at 3 dpf. Differences between control and BSI were significant (P < 0.0006), but there was no significant difference between BSI+Aβ and control embryos. BSI+p3 embryos had significantly fewer branches than control embryos (P < 3.0e-7). (F) Graph of mean CtA branch lengths in control (N = 28), BSI (N = 21), BSI+Aβ (N = 8), and BSI+p3(N = 10) embryos at 3 dpf. Differences between control and BSI were significant (P < 3.0e-13), but there was no significant difference between BSI+Aβ and control embryos. BSI+p3 embryos had significantly shorter vessel lengths than control embryos (P < 3.6e-13).