Neurochemical Research

, Volume 32, Issue 4, pp 833–835

Does Aβ 42 Have a Function Related to Blood Homeostasis?


Original Paper

DOI: 10.1007/s11064-006-9221-9

Cite this article as:
Hardy, J. Neurochem Res (2007) 32: 833. doi:10.1007/s11064-006-9221-9


In this review, I discuss the possibility that Aβ42 has a physiologic function in blood vessel homeostasis and the consequences that this might have for theories concerning the pathogenesis of Alzheimer’s disease and for treatment.


GeneticsAmyloidAlzheimer's disease


The amyloid hypothesis of Alzheimer’s disease has held sway for the last 15 years based upon the fact that all the autosomal dominant causes of the disease increase the production of Aβ42 (Fig. 1) [1, 2]. Although strictly, it is not important for the theory whether or not Aβ has a function or not, there has always been a general presumption that it did not and indeed, there has been comparatively little research on the function of the whole APP molecule. Most researchers have held the view, usually unwritten, that it was just a waste product of APP metabolism. In contrast, many of those who have been critical of the amyloid hypothesis have suggested Aβ may be “protective” and have not acknowledged the strength of the genetic data relating to Alzheimer causality.
Fig. 1

The amyloid hypothesis has the benefit of being able to accommodate all the known and defined causes of Alzheimer’s disease but, until it leads to therapies, it remains unproved. If it does not lead to therapies, alternatives will have to be considered

Over the last 5 years, I have become increasingly worried that research on Alzheimer’s disease was missing an important clue in not knowing the function of APP and in this article I air these continuing concerns and struggle towards a suggestion which may explain them.

The relationship between plaque and vascular amyloid

Occam’s razor would suggest that plaque and vascular amyloid should be related to each other as Miyakawa and colleagues suggested many years ago [3]. We suggested that those nerve cells which innervated cortical amyloidopathic blood vessels would be among those which showed the earliest damage [4]. The work showing the close pathogenic relationship between the amyloid cerebral hemorrhages and Alzheimer’s disease underscores this view of an association [57]. However, most convincing is the work showing that, in transgenic mouse models, there is a clear, close and intimate relationship between plaques and cerebral vessels [8].

Iron in every plaque

It seems almost certain that most neuritic plaques contain some iron [9, 10]. While there are possibly many ways in which iron could get into plaques, it seems that a hemorrhage is a particularly simple explanation and others have taken this view [9].

APP in the clotting cascade

Although we have little detailed notion of the function of APP, we do know that it is involved (as proteases nexin II) in the blood clotting cascade [11]: we also know that it is a gene whose expression is upregulated in response to injury [12].

Aβ as a synaptic depressant

One of the most remarkable features of the testing of amyloid immunizations in transgenic mice has been the remarkably quick partial recovery of memory functions in the immunized mice [13]. It has remained unclear whether the memory decrement in the mice was, in any way, related to the profound decrement in Alzheimer patients because the former had essentially no cell loss and the latter have profound atrophy of (for example) the hippocampus. It seems most likely that rather, this rapid improvement in function relates to an acute effect on synaptic activity [14] rather to any toxic effect on neurons.

Presenilin mutations: gain, loss or change of function?

It is clear that presenilin mutations increase the proportion of Aβ42 [15] and decrease the proportion of Aβ40 produced from APP metabolism and whether this corresponds to a gain or loss of function has been debated [16]. However, it is clearly a subtle change in function which suggests that presenilins have essentially (at least) two stable structures just as hemoglobin has and that the pathogenic mutations (and possibly ligands) alter the equilibrium between those structures.

Why have we not found a connection between Aβ and tau?

While the absence of evidence is not evidence of absence, it is remarkable that, despite 15 years of study, we have no convincing understanding of Aβ toxicity in vivo or of any direct relationship between Aβ and tau deposition. There have been only two direct pieces of evidence suggesting a relationship: the first is data showing Aβ toxicity was dependent on tau [17] and the second is that crossing mutant tau transgenic mice with APP transgenic mice potentiated tangle pathology [18].

Synthesis and suggestion (see also ref. [19])

Let us suppose that Aβ 42 has a role in the rapid sealing of microhemorrhages and that, during such an event, a number of things happen. First, Aβ in the perivascular space [20] (largely produced by neurons) comes into contact with blood and is deposited. Second, either because of anoxia or because of the other reactions to tissue damage, APP is upregulated and presenilins switch conformations towards the Aβ 42 producing conformation. This local increase in production has two effects: first, it promulgates perivascular deposition as amyloid angiopathy along from the site of damage and second, because it is a synaptic depressant, it causes a reduction in oxygen demand from around the damaged region. The upregulation of the N-terminal fragment of APP is, under this scheme, part of the attempt by the tissue to prevent clotting in the hemorrhaged region caused by blood contact with tissue.

This scheme has several strengths. First it suggests that we need not suppose that amyloid is directly neurotoxic nor that there is any direct effect on tau and tangles. These might more relate to the anoxic insult. Second, it fits with the observation that the complication of the amyloid immunization trial was only found in Alzheimer cases and related to a break down of the blood brain barrier. Third, it is consistent with the observation that, while apoe4 shows a clear association with Alzheimer’s disease (and with amyloid deposition), apoe2 shows a remarkable association with cerebral hemorrhage, suggesting this allele may be somehow less effective in dealing with the effect of such a hemorrhage [21].

Does this scheme fit with the genetic data on presenilins and APP mutations? What it would suggest in these individuals is that they are “overprimed”. As soon as they have an age related hemorrhage, they would have a catastrophic amyloidogenic response. However, for the disease to be truly progressive in such individuals, perhaps we should expect that excessive amyloid deposition may have a detrimental effect on the vascular bed [22] and cause further problems in this compartment rather than in the brain parenchyma. Such an effect on the vascular bed has in fact been reported in transgenic mice [23].

If treatments based on the amyloid hypothesis do not work (and the next few years should provide time to test this), then we need to consider, in a clear minded way, why this is the case. The vascular complications of the vaccine trial may have simply been one-off, or they may presage a more serious and general problem with amyloid therapies relating to a role of Aβ in vascular homeostasis [24]. Many amyloid-based trials are now in progress: their conclusions are anxiously awaited.


Work in the author’s lab was supported by the NIA/NIH intramural program.

Copyright information

© Springer Science+Business Media, LLC 2006