One of the paradoxes of Alzheimer’s disease is that while the brain is stuffed with amyloid, levels of Aβ42 in the cerebrospinal fluid (CSF) drop off. Many scientists have proposed that this is because soluble brain Aβ gets mopped up by plaques, but direct evidence for this idea has been scarce. In the November 2 Journal of Neuroscience, researchers led by Dennis Selkoe at Brigham and Women’s Hospital, Boston, in collaboration with David Holtzman and John Cirrito at Washington University in St. Louis, Missouri, strengthen the case for this hypothesis. They used microdialysis to measure brain interstitial fluid (ISF) Aβ in young, middle-aged, and old mouse models of AD. In the old mice, Aβ42 levels were low in ISF and high in plaques, and newly generated Aβ quickly absorbed onto existing deposits, suggesting the peptide gets sequestered there. In addition, the authors report that most of the free-floating Aβ42 in old, plaque-ridden mice appears to come from plaques, not neurons. Their data provide perhaps the most direct evidence to date for processes that many researchers have speculated are occurring in AD brains, to which scientists have no direct access for study.

On the other hand, a paper in the October 26 Journal of Neuroscience puts forth a more novel explanation for falling extracellular Aβ in aging AD brains. Researchers led by Gunnar Gouras, previously at Weill Cornell Medical College in New York City and now at Lund University, Sweden, report that neurons from transgenic AD mice secrete less Aβ as they age in culture. Over time, these neurons also lose the ability to spit out Aβ in response to synaptic stimulation. The results imply that AD brains might accumulate more intraneuronal Aβ as they age than healthy brains do, Gouras said, suggesting that the failure to secrete Aβ may be central to pathology. “This turns the traditional amyloid hypothesis on its head,” he noted. If the findings are confirmed by other labs, and are found to apply to aging mice in vivo, they could have implications for therapeutic strategies. “Both papers are interesting and support findings of decreased CSF amyloid-β dynamics in aging and Alzheimer's disease,” Randall Bateman at WashU wrote to ARF. Bateman was not involved in either study.

To examine Aβ dynamics in situ, first author Soyon Hong in Selkoe’s group performed microdialysis on more than 40 awake, active APP (J20) mice at three, 12, and 24 months of age. Placing the dialysis probe in the mice hippocampus, she found that ISF Aβ, particularly Aβ42, the most toxic and aggregation-prone species, fell steadily with age. Meanwhile, levels of less soluble forms of Aβ climbed, as seen in brain extracts obtained with saline, detergent, or formic acid. Scientists have always assumed that saline extracts represented soluble, free-floating Aβ, Selkoe noted. However, they also contained large aggregates of more than 500 kD, which may actually represent Aβ globs that were loosely bound to membranes or plaques, or were intracellular before homogenization, Hong said. Dave Morgan at the University of South Florida, Tampa, who was not involved in the research, suggested the data may cause scientists to rethink what various chemically extracted fractions actually represent.

Although the microdialysis probe cutoff size allowed Aβ monomers and, to a lesser extent, dimers to pass, Hong and colleagues recovered only monomers in ISF. The complete lack of dimers suggests that this species is scarce in the ISF, Selkoe told ARF. Dimers and other oligomeric Aβ forms have more exposed hydrophobic portions than monomeric Aβ does, he pointed out, meaning they may tend to bind to nearby hydrophobic surfaces such as cell membranes and plaques. In ongoing work, Hong said, they are injecting wild-type mice with human Aβ monomers and dimers, and seeing where the peptides go.

To find out what happens to newly produced Aβ in older mouse brains, the authors injected radiolabeled, synthetic Aβ40 monomers into the hippocampal ISF through a cannula attached to the microdialysis probe. In the oldest, plaque-ridden mice, microdialysis recovered less than half as much radiolabeled Aβ as the procedure did in younger mice within 90 minutes after injection. The missing Aβ turned up in saline brain extracts. This provides direct evidence that plaques act as a sink, rapidly pulling Aβ out of the ISF and into less soluble fractions, Hong said. Some previous ISF studies in mice (see ARF related news story on Cirrito et al., 2003), and amyloid imaging studies in people (see ARF related news story on Fagan et al., 2006), also support this idea.

Finally, to get at the question of where endogenous ISF Aβ comes from at different ages, Hong and colleagues inhibited γ-secretase, the enzyme that produces Aβ. In young AD mice, this treatment resulted in a rapid fall in levels of all Aβ monomers tested. In the older mice, by contrast, Aβ38 and Aβ40 levels fell, but Aβ42 levels barely budged. The authors suggest that most of the Aβ42 in the ISF in older AD mice diffuses off deposits, rather than being freshly synthesized. Some previous work has turned up support for this idea as well (see, e.g., ARF related news story on Koffie et al., 2009).

Oligomers are widely believed to be the most toxic form of Aβ. If oligomers are drifting off plaques, then, “You’re not going to be able to do something that selectively affects oligomers without also getting rid of the fibrillar deposits, because the fibrillar deposits may be the source of the oligomers,” Morgan noted.

Gouras’ group took a different approach to the mystery of falling Aβ levels, focusing on Aβ secretion from neurons. Numerous studies have shown that when synapses are stimulated, they disgorge more of the peptide (see ARF related news story on Kamenetz et al., 2003; ARF related news story on Cirrito et al., 2005; and ARF related news story on Tampellini et al., 2009). This raised the question of whether brain activity accelerates AD pathology by contributing to extracellular plaque formation.

Gouras and colleagues suggested, however, that synaptic activity is beneficial because it removes intraneuronal Aβ that would otherwise harm synapses (see Gouras et al., 2010; Capetillo-Zarate et al., 2011). Some researchers have questioned studies on intraneuronal Aβ, claiming that existing technology is not good enough to distinguish it from its precursor (see ARF Webinar). That controversy aside, when Gouras’ group inhibited synaptic activity in AD transgenic mice, the animals made fewer extracellular plaques, but had more synapse damage and worse memory (see Tampellini et al., 2010).

Since age is the primary risk factor for AD, Gouras wondered how it might affect Aβ secretion. To investigate, first author Davide Tampellini prepared hippocampal and cortical cultures from embryonic Tg2576 mice, which express human APP with the Swedish mutation, and compared the cells at 12 and 19 days after plating. Previous work has suggested that this culture model mirrors in-vivo aging on an accelerated timeframe, with one week in culture corresponding to changes seen in one year in animals, Gouras said (see, e.g., Takahashi et al., 2004; Almeida et al., 2005; ARF related news story on Almeida et al., 2006; ARF related news story on Snyder et al., 2005; Martin et al., 2008; and Sodero et al., 2011). Some researchers expressed doubts about this in-vitro model, however. “I'm just not sure what 12-day versus 19-day neuronal cultures equate to in an aging disease,” Cirrito wrote to ARF.

At 19 days in vitro, Tampellini and colleagues saw no changes in wild-type neurons, but found that the transgenic neurons secreted about one-third less Aβ, and had about 50 percent more intracellular Aβ42, than they did at 12 days. Moreover, when older neurons were stimulated with glycine, the wild-type neurons spat out more than twice as much Aβ as before, but transgenic neurons failed to respond. These findings were unexpected, Gouras told ARF. “People have always assumed secretion went up [with age] because you see plaques outside cells.” It is not clear if secretion of other proteins also changes as the cells age.

If neurons release less Aβ with age, then why does the peptide accumulate in AD? The authors looked at Aβ clearance. They report that when the 12-day-old transgenic neurons were stimulated with glycine, the Aβ-degrading protease neprilysin migrated to the cell surface and co-localized with Aβ. The 19-day-old transgenic neurons expressed about 20 percent less neprilysin than the younger cultures. This dovetails with work showing that neprilysin expression falls with age in wild-type mouse brain (see Iwata et al., 2002). The drop in neprilysin could “play a critical role in synaptic accumulation of Aβ with aging and AD,” the authors write.

These data may shed some light on why AD develops in older brains, but not young ones, Gouras said. He pointed out that people at risk for AD show reduced brain activity decades before symptoms develop (see Reiman et al., 2004). Therefore, “AD patients should be secreting less Aβ than controls,” Gouras said, and may accumulate more intraneuronal Aβ and more synapse damage. He emphasized that he does not discount a role for extracellular Aβ and plaques in promoting pathology as well. These cell culture data would suggest that trying to lower Aβ secretion may be the wrong approach to the disease, Gouras said. Would promoting Aβ secretion be good, bad, or neutral? That question remains to be answered.—Madolyn Bowman Rogers


  1. The paper by Hong and colleagues is a valuable contribution to our understanding of interstitial β amyloid in the brain with AD pathogenesis. We had considered examining interstitial fluid (ISF) Aβ, but turned to our cultured neuron system because we felt that even if there was reduced ISF Aβ prior to plaques, microdialysis would be unable to fully tease apart changes in secretion from sequestration to extracellular aggregates.

    Our cultured neuron model has served us well in the past. With it we showed for the first time Aβ-dependent reductions in surface glutamate receptor subunits (Almeida et al., 2005) and Aβ-dependent alterations in the multivesicular body sorting, but not recycling endocytic pathways (Almeida et al., 2006). Moreover, our AD transgenic compared to wild-type neuron system was really put to the test when we saw that synaptic activity protected synapses of AD transgenic neurons while inducing Aβ secretion but reducing intraneuronal Aβ (Tampellini et al., 2009). This was borne out in vivo when we examined synapses and behavior in APP mice using two different models to inhibit cerebral activity (including the barrel cortex): Transiently inhibiting activity reduced plaques, but increased intraneuronal Aβ and synaptic damage, and worsened behavior in AD transgenic mice. Thus, synapse damage and behavioral dysfunction correlated with intraneuronal Aβ but not plaques (Tampellini et al., 2010).

    For our current paper, we returned to cultured neurons to more carefully define the intra- versus extracellular pools of Aβ40 and Aβ42 over time in culture, during which AD transgenic neurons develop progressive AD-like synapse alterations. In fact, our data of increasing intraneuronal but declining extracellular Aβ in AD transgenic neurons fit well with those on ISF Aβ by Hong et al. In contrast, our conclusions differ. We consider both the intra- and extracellular Aβ pools, while Hong and colleagues—like so many in the field—only consider the extracellular pools. A membrane-associated pool is also mentioned in their discussion, as an explanation for why they could not detect Aβ dimers in ISF. Our explanation would be that Aβ oligomers are clogging up within neurons in AD, and become extracellular following destruction of neurites. The latter is not just a hypothesis; there are plenty of electron microscopy data in AD transgenic mice and also human AD brains to back this up (Takahashi et al., 2002; Takahashi et al., 2004). But for those who are not familiar with immuno-EM and might find it difficult to interpret, please take a look at our most recent pathological study in the current issue American Journal of Pathology (Capetillo-Zarate et al., 2011). This paper clearly shows intraneuronal and even isolated intra-synaptic Aβ42 accumulation, oligomerization, and fibrillization using easily viewable 3D movies.

    Hong and colleagues provide valuable data that we could not using our system. They show that plaques appear to be a source of releasable Aβ, since the decrease of ISF Aβ42 in the setting of γ-secretase inhibition was less pronounced in the presence compared to absence of plaques. Their new data on native forms of Aβ in brain are also of considerable interest. Together, our publications are complementary and not mutually exclusive, and move us a bit further in our understanding of this complex illness.


    . Beta-amyloid accumulation in APP mutant neurons reduces PSD-95 and GluR1 in synapses. Neurobiol Dis. 2005 Nov;20(2):187-98. PubMed.

    . Beta-amyloid accumulation impairs multivesicular body sorting by inhibiting the ubiquitin-proteasome system. J Neurosci. 2006 Apr 19;26(16):4277-88. PubMed.

    . Synaptic activity reduces intraneuronal Abeta, promotes APP transport to synapses, and protects against Abeta-related synaptic alterations. J Neurosci. 2009 Aug 5;29(31):9704-13. PubMed.

    . Effects of synaptic modulation on beta-amyloid, synaptophysin, and memory performance in Alzheimer's disease transgenic mice. J Neurosci. 2010 Oct 27;30(43):14299-304. PubMed.

    . Intraneuronal Alzheimer abeta42 accumulates in multivesicular bodies and is associated with synaptic pathology. Am J Pathol. 2002 Nov;161(5):1869-79. PubMed.

    . Oligomerization of Alzheimer's beta-amyloid within processes and synapses of cultured neurons and brain. J Neurosci. 2004 Apr 7;24(14):3592-9. PubMed.

    . High-resolution 3D reconstruction reveals intra-synaptic amyloid fibrils. Am J Pathol. 2011 Nov;179(5):2551-8. PubMed.

    View all comments by Gunnar Gouras
  2. These are two very interesting papers discussing the production of Aβ with age. Soyon Hong and Dennis Selkoe's work in awake, behaving mice is particularly interesting as it elegantly shows that dense plaques are in equilibrium with soluble Aβ in the parenchyma, both sequestering exogenously added Aβ and acting as a source of Aβ when γ-secretase is inhibited. This supports the body of evidence showing that plaques are toxic to the nearby neurites and synapses because they are a local source of soluble Aβ species.

  3. This paper by Selkoe’s group is an important contribution to our understanding of amyloid-β (Aβ), an Alzheimer’s disease (AD) neurotoxin metabolism in plaque-free and plaque-rich brain. The authors used a well-established microdialysis technique to study the dynamics of soluble Aβ clearance by sampling and analyzing soluble Aβ species in the interstitial fluid (ISF) of young and old transgenic mice carrying an hAPP mini-gene with familial AD mutations. The study not only provides strong evidence for the hypothesis that cerebral amyloid deposits act as a sink for soluble Aβ in the ISF, but also suggests that amyloid plaques act as a large reservoir and a major source of soluble Aβ42 in the ISF.

    Mounting evidence suggests that diminished Aβ clearance from the brain, but not alteration in Aβ production (also confirmed in the present study), leads to Aβ accumulation in the brains of AD patients. Targeting Aβ clearance pathways (reviewed recently by Zlokovic, 2011) may help in correcting the faulty clearance from AD brain and across the blood-brain barrier.


    . Neurovascular pathways to neurodegeneration in Alzheimer's disease and other disorders. Nat Rev Neurosci. 2011 Dec;12(12):723-38. PubMed.

    View all comments by Abhay Sagare
  4. Aside from findings related to the intraneuronal accumulation of Aβ in the Alzheimer’s disease (AD) brain, this interesting and important paper from the Gouras lab brings to the forefront a neglected, but critical, issue in AD research: that of the age of the neurons that are investigated in cell culture. Most studies are done with embryonic or neonatal neurons, although the processes related to the aging brain and age-related diseases, such as AD, occur in very old neurons that may be very different from the embryonic ones. For example, the intracellular accumulation of oligomeric Aβ found at old age in some mouse models of AD (1), and in the human brain (2), is quite infrequent in primary neurons from embryonic mouse brains. The neurons in the old brain have gone through a history that marked them in very specific ways, and this history cannot be reproduced in culture, especially if these neurons are used without allowing them to age. The Gouras study allows for such aging, and this aging in the culture dish enabled the observations communicated in this study. Although the months-long aging required to generate the typical pathological lesions (e.g., plaques) in the brains of mouse models of AD differs from the weeks-long (at best) aging of neurons in culture, such experimental aging can be a good in-vitro model for studying the incipient stages of the disease process, as this study nicely shows.

    Experts in the biology of cultured neurons, such as Gary Banker, consider that the embryonic neurons mature pretty well, so they eventually could correspond to cells from a two- to three-week-old animal. Also, these cultured neurons might respond to environmental challenges similarly to an adult neuron, and they should be subjected to such stress whenever possible. Culturing neurons derived from the adult brain—which would be ideal—appears to pose insurmountable problems. Reprogramming cells with iPS technology, or direct conversion of adult skin fibroblasts into neurons in culture (3)—as now is becoming the norm—has its own problems. The “history” of cells converted into neurons is not that of a neuron; also, do they remember this history? Researchers will continue to use neurons obtained from the embryonic or neonate brain. However, to be more relevant to diseases of the old age, these neurons should be allowed to age in culture for extended periods of time, as the Gouras lab did.

    We also think that, in some cases, neuronal cell lines could provide insights into the biology of the aging neuron (see, e.g., 4). Although some researchers dislike this idea—often for good reasons—such cells can live in culture for quite a long time until senescence kicks in. But could a senescent neuronal cell not tell us something about neurons in the aging brain?


    . Intraneuronal Abeta causes the onset of early Alzheimer's disease-related cognitive deficits in transgenic mice. Neuron. 2005 Mar 3;45(5):675-88. PubMed.

    . Intraneuronal Abeta42 accumulation in human brain. Am J Pathol. 2000 Jan;156(1):15-20. PubMed.

    . Directed conversion of Alzheimer's disease patient skin fibroblasts into functional neurons. Cell. 2011 Aug 5;146(3):359-71. PubMed. RETRACTED

    . Neuritic deposits of amyloid-beta peptide in a subpopulation of central nervous system-derived neuronal cells. Mol Cell Biol. 2006 Jul;26(13):4982-97. PubMed.

    View all comments by Virgil Muresan

Make a Comment

To make a comment you must login or register.


News Citations

  1. Soluble Aβ: Getting a Grip on Its Fate
  2. Brain Imaging Speaks Volumes about AD and the Aβ Sink
  3. Spine Shrinkers: Aβ Oligomers Caught in the Act
  4. Does Aβ Normally Rein in Excited Synapses?
  5. Paper Alert: Synaptic Activity Increases Aβ Release
  6. The Ups and Downs of Aβ: Synaptic Activity Yields Mixed Results
  7. Paper Alert: Intraneuronal Aβ Impairs Multivesicular Body Sorting
  8. Amyloid-β Zaps Synapses by Downregulating Glutamate Receptors

Webinar Citations

  1. Intraneuronal Aβ: Was It APP All Along?

Paper Citations

  1. . In vivo assessment of brain interstitial fluid with microdialysis reveals plaque-associated changes in amyloid-beta metabolism and half-life. J Neurosci. 2003 Oct 1;23(26):8844-53. PubMed.
  2. . Inverse relation between in vivo amyloid imaging load and cerebrospinal fluid Abeta42 in humans. Ann Neurol. 2006 Mar;59(3):512-9. PubMed.
  3. . Oligomeric amyloid beta associates with postsynaptic densities and correlates with excitatory synapse loss near senile plaques. Proc Natl Acad Sci U S A. 2009 Mar 10;106(10):4012-7. PubMed.
  4. . APP processing and synaptic function. Neuron. 2003 Mar 27;37(6):925-37. PubMed.
  5. . Synaptic activity regulates interstitial fluid amyloid-beta levels in vivo. Neuron. 2005 Dec 22;48(6):913-22. PubMed.
  6. . Synaptic activity reduces intraneuronal Abeta, promotes APP transport to synapses, and protects against Abeta-related synaptic alterations. J Neurosci. 2009 Aug 5;29(31):9704-13. PubMed.
  7. . Intraneuronal beta-amyloid accumulation and synapse pathology in Alzheimer's disease. Acta Neuropathol. 2010 May;119(5):523-41. PubMed.
  8. . High-resolution 3D reconstruction reveals intra-synaptic amyloid fibrils. Am J Pathol. 2011 Nov;179(5):2551-8. PubMed.
  9. . Effects of synaptic modulation on beta-amyloid, synaptophysin, and memory performance in Alzheimer's disease transgenic mice. J Neurosci. 2010 Oct 27;30(43):14299-304. PubMed.
  10. . Oligomerization of Alzheimer's beta-amyloid within processes and synapses of cultured neurons and brain. J Neurosci. 2004 Apr 7;24(14):3592-9. PubMed.
  11. . Beta-amyloid accumulation in APP mutant neurons reduces PSD-95 and GluR1 in synapses. Neurobiol Dis. 2005 Nov;20(2):187-98. PubMed.
  12. . Beta-amyloid accumulation impairs multivesicular body sorting by inhibiting the ubiquitin-proteasome system. J Neurosci. 2006 Apr 19;26(16):4277-88. PubMed.
  13. . Regulation of NMDA receptor trafficking by amyloid-beta. Nat Neurosci. 2005 Aug;8(8):1051-8. PubMed.
  14. . Cholesterol loss enhances TrkB signaling in hippocampal neurons aging in vitro. Mol Biol Cell. 2008 May;19(5):2101-12. PubMed.
  15. . Cellular stress from excitatory neurotransmission contributes to cholesterol loss in hippocampal neurons aging in vitro. Neurobiol Aging. 2011 Jun;32(6):1043-53. PubMed.
  16. . Region-specific reduction of A beta-degrading endopeptidase, neprilysin, in mouse hippocampus upon aging. J Neurosci Res. 2002 Nov 1;70(3):493-500. PubMed.
  17. . Functional brain abnormalities in young adults at genetic risk for late-onset Alzheimer's dementia. Proc Natl Acad Sci U S A. 2004 Jan 6;101(1):284-9. PubMed.

Other Citations

  1. APP (J20) mice

Further Reading

Primary Papers

  1. . Dynamic analysis of amyloid β-protein in behaving mice reveals opposing changes in ISF versus parenchymal Aβ during age-related plaque formation. J Neurosci. 2011 Nov 2;31(44):15861-9. PubMed.
  2. . Impaired β-amyloid secretion in Alzheimer's disease pathogenesis. J Neurosci. 2011 Oct 26;31(43):15384-90. PubMed.