Problems with the amyloid hypothesis notwithstanding, knowing where amyloid-β (Aβ) plaques form might be key to developing strategies for clearing them from the brains of those with Alzheimer disease (AD). In this month’s American Journal of Pathology, two papers report that in three different mouse models of Alzheimer disease, plaques with dense cores, which have been associated with neuritic pathology and neuronal loss, are predominantly associated with blood vessel walls, and that this association can lead to vascular damage. The findings raise questions about just where plaque formation might occur most frequently during AD.

Christine van Broeckhoven and colleagues at the Flanders Interuniversity Institute for Biotechnology and the University of Antwerp in Belgium, together with collaborators in the US, carried out a detailed and systematic study of dense core plaques in the brains of both Tg2576 (see Hsiao et al., 1996) and PSAPP (see Duff et al., 1996 /pap/annotation.asp?powID=12907) mice. Both the former, expressing human Aβ precursor protein (AβPP) with the Swedish mutations (K670N and M671L), and the latter, a double transgenic expressing mutant human presenilin 1 (M146L) and AβPP with the Swedish mutations, generate generous quantities of plaques. Both sets of transgenic animals have been used as investigational models for AD.

First author Samir Kumar-Singh and colleagues found that the vast majority of dense core plaques are found adjacent to blood vessels. Using stereological microscopy of 4 μM thin slices, the authors found that of over 2,000 plaques sampled from the neocortices and hippocampuses, 94 and 85 percent of them were associated with blood vessels in Tg2576 (n = 9) and PSAPP (n = 5) mice, respectively.

To ensure that these numbers were not arrived at by chance, Kumar-Singh studied the entire brain of a 2-year-old Tg2576 and a 5-month-old PSAPP animal—at these ages the plaque burden in the different transgenics is about the same. The findings confirmed the data from the plaque sampling experiment. Ninety-five percent of 210 plaques from the Tg2576 animal and 85 percent of 258 plaques from the PSAPP mouse were positively associated with the microvasculature. Statistical analysis based on the size and area of the blood vessels and plaques also revealed that these associations could not be incidental. In fact, even in thinner (1 μM) sections from a PSAPP animal, the authors failed to identify even one plaque that did not associate with a vessel at some level. Moreover, the plaque/vessel relationship was further strengthened by the finding that plaques that had more than one dense core were more likely to be associated with more than one blood vessel. In fact, the total number of vessels associated with a plaque turned out to be the best predictor of how many cores might be present within it.

Tackling this question from a slightly different perspective, William van Nostrand and colleagues at Stony Brook University, New York, and the University of Nijmegen, Holland, report similar findings using the Tg-SwDI transgenic mice. These animals produce human AβPP with the same Swedish mutants as the Tg2576 mice, but mutations normally found in certain Dutch and Iowan populations (E693Q and D694N) are also included in the transgene for good measure. Previously, it was shown that the amyloid generated in these animals is predominantly vascular (see Davis et al., 2004). Now, first author Jianting Miao and colleagues show that the number of blood vessels that stained positive for Aβ increases dramatically as these animals age. In the thalamus and subiculum, areas where fibrillar Aβ accumulates—but not the frontal cortex where the amyloid is mostly diffuse—the percentage of vessels staining for Aβ reached 50 by 12 months and over 80 by 2 years of age.

When Miao and colleagues examined the brain ultrastructure, they found that the accumulation of microvascular amyloid was accompanied by anatomical changes. For example, the number of microvessels fell by 16 and 27 percent in the hippocampus of 1- and 2-year-old animals, respectively. Aβ in the meningeal vessels was also associated with apoptotic cells and loss of smooth muscle cells. Miao also found a dramatic age-related increase in the numbers of reactive astrocytes and microglia in the thalamus and subiculum. Both cell types were about five times more abundant in 2-year-old animals as compared to young, 6-month-old mice.

In fact, Kumar-Singh and colleagues also found obvious morphological changes in the Tg2576 and PSAPP mice. These included loss or thinning of the endothelium, basement membrane thickening or splitting to accommodate Aβ, and degeneration of smooth muscles. They also found that the transgenic animals had more microhemorrhages than did control mice.

How these findings relate to AD is unclear. The Dutch and Iowa mutations are known to cause primarily cerebral amyloid angiopathy, or CAA (see ARF related news story), and as Kumar-Singh and colleagues point out, their mouse data is more reminiscent of Flemish familial AD, characterized by many large, dense-core plaques and CAA, than it is of sporadic AD or Down syndrome, in which plaques are more diffuse and CAA minimal.

Nonetheless, these results would seem to cement the relationship between vessel walls and Aβ deposits. And as Kumar-Singh and colleagues write, “the present study, for the first time, demonstrates that dense amyloid plaques in Tg2576 and PSAPP mice are centered on vessel walls. If a similar mechanism is also operative in AD, therapeutics targeting Aβ clearance from the vascular compartment may be most beneficial.”—Tom Fagan


  1. I believe the bystander effect is heavily implicated here and anyone familiar with comments on alzforum know that although autoimmunity and BBB leakage are important aspects in AD I still hold firmly to the belief that amyloid beta is an important catalyst (or cocatalyst) for immune response gone awry and for BBB leakage and TAU, CDK5 (amongst others) involved in intracellular hyperphosporylation.

  2. Two independent studies provided morphological evidence suggesting that accumulations of amyloid in mouse cerebral blood vessels are associated with amyloid plaques, which are typically detected in the CNS of AD patients. There is a wealth of evidence confirming vascular pathology in AD, and it was suggested years ago that amyloid plaques might originate from vessels.

    The characterization of the different morphological types of amyloid plaques has been an important research focus in our laboratory. It is becoming increasingly clear that each distinct type of plaque may arise from separate mechanisms and that the concept that diffuse plaques gradually evolve into dense-core plaques or vice versa, may not be valid.

    The Kumar paper suggested that all plaques are of vascular origin in the transgenic mouse brains; hence, in these models, there is no randomness to the distribution of the amyloid plaques. The Miao paper implied that diffuse and nonfibrillar amyloid in the cortex of the Tg-SwD1 mice remained diffuse and nonfibrillar, and that fibrillar amyloid in the thalamus was tightly associated with vessels, suggesting a vasculature origin.

    Although the primary focus of these papers was on "extracellular" amyloid (plaques), I found it surprising that there was little mention of intracellular amyloid in nearby neurons, or in smooth muscle cells as observed in AD. Perhaps the intent was to keep the paper focused on plaques, or maybe there were technical issues that did not show intracellular amyloid, or were not observed in the mouse model.

    For example, there was a loss of smooth muscle cells in the amyloid-laden vessels (Miao paper), but no discussion of the presence of amyloid in smooth muscle cells, which is frequently observed. If one was to extrapolate observations that the intracellular accumulation of amyloid in neurons is linked directly to plaque formation, it is not unreasonable to suspect a similar scenario here. Hence, amyloid could accumulate in vascular smooth muscle cells over time before they die, leaving amyloid casts (= vascular plaques). Furthermore, any cell damage, especially smooth muscle cell death, would evoke the observed inflammatory response (gliosis).

    Apparently, only diffuse, thioflavin S-negative plaques were detected in the frontotemporal cortex. Were microglia associated with these diffuse plaques detected in the prefrontal cortex, or were they only associated with "affected" vessels of the Tg-SwD1 mice? I would speculate that neither microglia nor activated astrocytes were associated with these diffuse plaques of the frontocortex, akin to the diffuse plaques observed in the human AD cerebellum and cerebrum. Therefore, if our earlier hypothesis is true—that dense-core (classical, thioflavin S-positive, etc.) plaques, but not diffuse plaques, originate from amyloid-laden neurons—then the lack of those plaques in regions of the CNS could account for the lack of consistent AD-like behaviors in such “AD” models.

    On another aspect, it was noted that “Ig occasionally stained neuronal surfaces,” but no follow-up or comment was printed. One of our recent studies reported neurodegenerative characteristics on these Ig-positive neurons.

    I still have trouble with the concept that AD is well-characterized by the steady deposition of amyloid from neurons in the brain and entering into the vasculature (Kumar, Fig. 10). Why couldn’t it be the other way around—that the brain is slowly and steadily accumulating vascular-derived amyloid, perhaps via BBB dysfunction? This issue needs to be resolved.

  3. Regarding Dr. D’Andrea’s remarks, we studied only ThS-positive “dense” plaques and not diffuse plaques, as also suggested by the title “Dense-core plaques in Tg2576 and PSAPP mouse models of Alzheimer disease are centered on vessel walls.” Within the text, however, we mostly refrained from using the term dense-core plaques (calling them dense-plaques instead). That's because the plaques observed in the studied mouse models differ from the classical dense-core plaques observed in AD, especially those observed in the Flemish APP pathology where we had earlier shown their proximity to vessels (Kumar-Singh et al. Am J Pathol, 2002).

    Secondly, as Dr. D’Andrea suggested, we indeed came across intracellular amyloid in nearby neurons and sometimes amyloid related to smooth muscle cells. However, the primary focus of our paper were dense, extracellular amyloid deposits. Similarly, our observation that “Ig occasionally stained neuronal surfaces” was there to support our observation that there are at least subtle BBB disturbances in these mouse models, as has also been observed in humans, and that clearly has other important implications, especially towards therapeutics.

    Lastly, despite a strong neuronal promoter used in these mouse models, there are some reports showing Aβ secretion from non-neuronal cells (i.e., ref. 57). Nevertheless, we have to agree that the majority of Aβ still comes from the neurons. With saturating parenchymal Aβ-degrading pathways, Aβ (especially Aβ40) is trafficked towards vessels for clearance. Although the model sums up more probable theories and published work for mouse AD models, we kept open the possibility that Aβ might also come back from vessels into brain either by receptor-mediated influx or by free diffusion taken that BBB is dysfunctional at these sites. While more studies have to be performed (especially with models that have predominantly Aβ42), one of the strongest evidence for a neurovascular Aβ flow for PSAPP and Tg2576 was that the earliest vascular deposits observed ultrastructurally occurred on the abluminal surface of capillaries and small parenchymal vessels.


    . Dense-core plaques in Tg2576 and PSAPP mouse models of Alzheimer's disease are centered on vessel walls. Am J Pathol. 2005 Aug;167(2):527-43. PubMed.

    . Cerebral microvascular amyloid beta protein deposition induces vascular degeneration and neuroinflammation in transgenic mice expressing human vasculotropic mutant amyloid beta precursor protein. Am J Pathol. 2005 Aug;167(2):505-15. PubMed.

    . Dense-core senile plaques in the Flemish variant of Alzheimer's disease are vasocentric. Am J Pathol. 2002 Aug;161(2):507-20. PubMed.

  4. The source of Aβ is still an important question. We recently could show that the blood-brain barrier (BBB) endothelial cells could be important players for generating Aβ. We showed significant levels of BACE-1 in brain ECs and further show an upregulation of EC BACE-1 in a mouse AD model (hAPP-SL). These data bring back the vascular focus in AD, particularly with respect to generation of Aβ, and not just its clearance across the BBB.


    . BACE-1 is expressed in the blood-brain barrier endothelium and is upregulated in a murine model of Alzheimer's disease. J Cereb Blood Flow Metab. 2015 Oct 13; PubMed.

    View all comments by Kavi Devraj

Make a Comment

To make a comment you must login or register.


News Citations

  1. Study Poses Cautionary Question: Will Knocking Down Aβ42 Promote Amyloid Angiopathy?

Paper Citations

  1. . Correlative memory deficits, Abeta elevation, and amyloid plaques in transgenic mice. Science. 1996 Oct 4;274(5284):99-102. PubMed.
  2. . Increased amyloid-beta42(43) in brains of mice expressing mutant presenilin 1. Nature. 1996 Oct 24;383(6602):710-3. PubMed.
  3. . Early-onset and robust cerebral microvascular accumulation of amyloid beta-protein in transgenic mice expressing low levels of a vasculotropic Dutch/Iowa mutant form of amyloid beta-protein precursor. J Biol Chem. 2004 May 7;279(19):20296-306. Epub 2004 Feb 25 PubMed.

Further Reading

Primary Papers

  1. . Dense-core plaques in Tg2576 and PSAPP mouse models of Alzheimer's disease are centered on vessel walls. Am J Pathol. 2005 Aug;167(2):527-43. PubMed.
  2. . Cerebral microvascular amyloid beta protein deposition induces vascular degeneration and neuroinflammation in transgenic mice expressing human vasculotropic mutant amyloid beta precursor protein. Am J Pathol. 2005 Aug;167(2):505-15. PubMed.