Scientists have long wondered why certain brain regions bear the brunt of Alzheimer’s disease. Plaques preferentially clog up the default-mode network, a set of interconnected brain regions that are most active when people are letting their minds wander. This network has a higher basal metabolism than other brain regions, and makes heavy use of glycolysis, or the incomplete burning of glucose, for energy. Now, a paper in the May 1 online Nature Neuroscience ties these observations together, suggesting that brain activity, as measured by glycolysis, influences Aβ deposition. Researchers led by David Holtzman at Washington University, St. Louis, Missouri, found that levels of extracellular, soluble Aβ in mouse brain correspond to local activity levels, with higher activity linked to more Aβ. Regions of the mouse brain that are analogous to the default-mode network in humans had the highest levels of glycolysis—and Aβ. Significantly, when the researchers raised or lowered neuronal activity, they saw corresponding changes in Aβ levels and plaque development later in life. “This may be one of the first explanations for why you get [AD pathology] in these areas,” Holtzman told ARF. The results also suggest that modulating brain activity in these networks, either pharmacologically or through lifestyle changes, could help prevent or delay AD.

The finding “argues that synaptic activity may be the dominant factor controlling Aβ levels and plaque formation,” Roberto Malinow at the University of California in San Diego wrote to ARF (see full comment below). He was not involved in the study. Sam Gandy at the Mount Sinai Medical Center in New York City noted in an e-mail, “The data here are beautiful and compelling.”

The default-mode network has caught the attention of scientists in the last decade. Early on, Randy Buckner at Washington University and Bill Klunk of University of Pittsburgh Medical School realized that its regions are the ones that deposit amyloid plaques early on in the course of AD (see ARF related news story on Buckner et al., 2005), and Buckner then worked with Keith Johnson’s group at Massachusetts General Hospital, Boston, to extend the analysis to cortical hub regions (see ARF related news story on Buckner et al., 2009). Other researchers have found that activity changes in these networks, as seen by functional imaging, foreshadow the development of AD (see ARF related news story on Greicius et al., 2004; Hedden et al., 2009; ARF related news story on Sperling et al., 2009). Most recently, research groups led by Marcus Raichle and Mark Mintun at WashU demonstrated increased glycolysis in these same regions using imaging techniques (see ARF related news story on Vaishnavi et al., 2010 and Vlassenko et al., 2010), which hinted at a connection between energy metabolism and Aβ pathology.

Holtzman’s group brings a different approach to the problem. These scientists developed in vivo microdialysis to measure the levels of soluble Aβ and other molecules in the interstitial fluid between cells. In previous work, they showed that jacking up synaptic activity with electrical probes increased the levels of extracellular, soluble Aβ in mice, while blocking activity with drugs lowered Aβ levels, demonstrating a direct relationship between activity and Aβ (see ARF related news story on Cirrito et al., 2005; ARF related news story on Cirrito et al., 2008). The group saw a similar link between activity and Aβ levels in people who received microdialysis as part of their neurological status monitoring following acute brain injury. Immediately after injury, levels of Aβ in the interstitial fluid were extremely low, but rose as the brain recovered and resumed synaptic activity (see ARF related news story on Brody et al., 2008).

Holtzman and colleagues wanted to take this further by examining the effects of endogenous synaptic activity on different brain regions, as well as looking at the long-term effects of varying Aβ levels. First author Adam Bero verified that aged Tg2576 AD mice had regional patterns of Aβ plaque deposition very similar to those seen in humans with AD, indicating that this mouse would be a useful model. Bero then showed that the concentrations of Aβ40 and Aβ42 in interstitial fluid in four-month-old mice were highest in areas that would have the most plaques at 18 months. Wild-type mice showed lower absolute levels but the same regional pattern in their ISF Aβ, suggesting this represents normal physiology.

To tie Aβ to brain activity, Bero and colleagues measured levels of lactate, which is produced by glycolysis. Lactate levels have been shown to correlate with neurotransmitter-mediated synaptic activity (see, e.g., Uehara et al., 2008) and indeed are frequently measured in humans as part of neurologic status monitoring after acute trauma. They found that the amount of lactate in the interstitial fluid varied in tandem with Aβ in both Tg2576 and wild-type mice.

The authors then employed several approaches to modulate neuronal activity. One was pharmacological: They administered picrotoxin, a GABA antagonist, to increase synaptic activity, and tetrodotoxin to decrease activity. Levels of lactate and Aβ in the interstitial space went up or down in parallel with activity. Bero and colleagues also manipulated physiological activity, by stimulating or trimming the mouse’s whiskers on one side and looking at corresponding changes in the barrel cortex connected to those whiskers. Aβ levels rose after whiskers were stimulated, while both lactate and Aβ dropped in barrel cortex after they were trimmed, again linking Aβ release to synaptic activity. In contrast to the clear relationship between Aβ levels and activity, Aβ levels did not correlate well with regional differences in clearance rates, or with differences in APP processing.

Bero and colleagues switched to seven-month-old APP/PS1 mice to examine the long-term effects of whisker manipulation, because this strain develops aggressive fibrillar plaques at a young age. After 28 days of whisker trimming, plaques on the deprived side grew only about one-quarter as much as those on the control side. Fewer new plaques formed, again suggesting that activity promotes pathology.

This presents a puzzle, as numerous epidemiological studies have suggested that cognitive stimulation and greater education, which presumably depend on greater neuronal activity, delay the onset of AD, not promote it. One possible answer, the authors suggest, is that activating task-oriented brain areas may reduce metabolism in the vulnerable default-mode network and therefore slow down Aβ deposition there. Holtzman told ARF, however, that it is equally plausible that education merely helps mask the early effects of AD, resulting in a later diagnosis, and does not affect Aβ pathways. If that is true, what causes some people to get AD while others do not?

Although the answer will require more research, there are some clues. For one thing, people who carry the ApoE4 risk allele have higher resting-state metabolism in the default-mode network (see ARF related news story on Filippini et al., 2009). They also have more brain amyloid. Holtzman noted that this network is the most used throughout the day. “I wonder if the explanation for what we’re finding in this paper has to do with how efficiently the brain utilizes its prominent networks,” he suggested. For example, perhaps specific genes regulate efficiency of brain energy metabolism, with some people having less efficient energy metabolism in the brain than others, resulting in more Aβ release, suggested Holtzman. Lifestyle factors such as stress and lack of sleep may also increase network metabolism, he said.

Holtzman is particularly interested in the sleep connection. His group previously showed that Aβ levels in the interstitial fluid fall during sleep (see ARF related news story on Kang et al., 2009). In the current paper, Holtzman and colleagues demonstrated that levels of lactate and Aβ are highest in young Tg2576 mice when they are awake, and lowest when they are asleep, suggesting that sleep is beneficial. In ongoing work, Holtzman is using animal models to study the mechanisms behind this effect. He is also measuring sleep patterns and AD biomarkers in healthy middle-aged people. His hypothesis is that people who regularly get more or better sleep will have less AD biomarkers as they age. “If you have very efficient sleep-wake cycles every day for most of your life, maybe that allows these networks to stay in balance better,” Holtzman speculated. Recently, researchers found that lack of sleep can cause localized “napping” among groups of neurons, which may explain why sleep-deprivation affects cognition. And other groups have linked dramatic changes in sleep patterns to cognitive decline, as well (see ARF related news story).

Holtzman believes the correlation between default-mode network activity and Aβ levels has significant implications for AD prevention. If people could modulate their brain network activity with lifestyle factors or medications, he said, it might delay disease onset.—Madolyn Bowman Rogers

Comments

Make a Comment

To make a comment you must login or register.

Comments on News and Primary Papers

  1. Both the paper by Bero and colleagues and the Alzforum news story make a tacit assumption concerning the relationship between synaptic activity and β amyloid-related synapse dysfunction: that reducing plaque by reducing activity-driven secretion of Aβ is good for the brain. But is this assumption true?

    As Bero and colleagues are aware, we reported last year in the Journal of Neuroscience (Tampellini et al., 2010) that deafferented barrel cortex causes reduced plaques in AD transgenic mice, findings now confirmed by Bero and colleagues. We then asked whether this plaque reduction in the setting of decreased synaptic activity was good or bad for synapses. Decreased plaques suggested it may be good, as Holtzman and colleagues posit. But there was reason to consider that reduced synaptic activity might actually be harmful to synapses, since in 2009 we published also in the Journal of Neuroscience that synaptic activation protected cultured neurons of Tg2576 mice against synaptic damage, even though Aβ secretion was increased, most likely because synaptic activity caused intracellular Aβ to decrease. Thus, active synapses were happy with extracellular Aβ up and intracellular Aβ down! Therefore, it was not surprising when we found that, even though plaques were decreased, decreasing synaptic activity by removal of whiskers actually increased intraneuronal Aβ and damaged synapses, as seen both by loss of synaptophysin and electron microscopy (Tampellini et al., 2010).

    We confirmed this finding using a second model of decreased synaptic activation—putting the mice to sleep. As reported by Holtzman and colleagues (Kang et al., 2009), we also found that sleep reduced plaque burden (Tampellini et al., 2010). However, again we looked at intraneuronal Aβ and synapses, and despite reduced plaques, intraneuronal Aβ was increased and synaptophysin was reduced in the mice made to sleep.

    Finally, while loss of synaptophysin and frank loss of synapses, as seen by electron microscopy, seemed not to be a good thing, we wanted to be even more certain and did behavioral testing. Consistent with the deterioration in the synapses, the Alzheimer’s transgenic mice that had been sedated did worse on memory testing despite having reduced plaques!

    So what is the role of synaptic activity effects on Aβ and synapses in AD? It does not seem to be as simple as Bero and colleagues suggest. Yes, areas of high synaptic activity appear prone to plaque formation, but decreasing (normal) synaptic activity increases intraneuronal Aβ, worsens synaptic degeneration, and impairs memory. These are important data that should not be ignored, particularly if one is considering modulation of synaptic activity as a potential therapeutic or prophylactic intervention. On the other hand, our work is not the whole story either—synaptic hyperexcitability and seizures also occur in AD and may be detrimental (another story), and blocking such hyperactivity may be beneficial.

    The work begun by Malinow and Holtzman relating synaptic activity and Aβ is crucial, and clearly the relationships are complex. We believe that some of the complexity is explained by considering that intraneuronal Aβ also plays a pathogenic role in disease and is modulated by activity. However, whether or not one thinks about intraneuronal Aβ, the negative effects of decreasing synaptic activity on synaptophysin levels, synaptic density counts, and cognitive performance are real.

    References:

    . 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.

    . 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.

  2. This is a very nice study. It provides further support that synaptic activity controls neuronal release of Aβ, and argues that synaptic activity may be the dominant factor controlling Aβ levels and plaque formation (rather than the recently proposed changes in Aβ removal). It will be very interesting in the future to correlate synaptic activity levels with Aβ release and plaque formation at a shorter distance scale (tens of microns).

    View all comments by Roberto Malinow
  3. The data here are beautiful and compelling, especially the demonstration of the temporal relationship between the accumulation of oligomers and the changes in interstitial fluid (ISF) lactate. Once again, the chicken-and-egg test shows that oligomerization is upstream of neurotoxicity, and not vice versa (although that relationship may become less black and white as the disease progresses).

    I think that there are some details that may have been glossed over for the sake of the narrative:

    1. “Neuronal activity” is described as though this were some homogenous, unitary phenomenon, when, in fact, the moment to moment, spatial summation of dozens of chemical signals arriving per second at the soma or at the synapse leads to highly diverse signaling and, in turn, a wide spectrum of Aβ40 and Aβ42 generation. I completely agree that this is not the whole story, but I would rank this as the determinant of some range of Aβ levels that will vary from one microenvironment to another.

    2. Different signals are differentially amyloidogenic. The soma versus the nerve terminal may be differentially competent for generating Aβ. Some signals regulate the relative utilization of α- versus β-secretase pathways, while other signals differentially regulate generation of Abeta42 versus 40. A slightly different but relevant point is that there are amyloidogenic pathways that may be more constitutive and others that may be more highly regulated (e.g., autophagy, nerve terminal generation of Aβ).

    3. The fraction of catabolized APP holoprotein or CTFs that gives rise to discrete fragments is tiny and cannot be predicted by steady levels of either holoAPP or APP CTFs. This is true in cell lines as well as tissue, and I would never try to predict Aβ levels from levels of APP or any APP fragment.

    4. Rarely mentioned in an “Aβ clinicopathological spatiotemporal specification” discussion, such as arises here, is an amazing fact that dates back to 1995 when the first authentic APP transgenic mouse was analyzed histologically: Regardless of whether the prion promoter or the Thy1 promoter was used to drive totally unnatural patterns of overexpression of human APP transgenes, the regional and laminar accumulation of plaque pathology recapitulated, virtually perfectly, the pattern observed in the human AD brain. This is pretty amazing when you think about it. How did the mouse’s brain “know” what to do with human APP regionally and according to cortical levels following overexpression on pan-neuronal or pan-cellular promoters? There are clearly pre-existing specifiers of the deposition pattern. The regional levels of secretases, insulin degrading enzyme, and neprilysin, and the respective states of activation, all play their respective roles, but what has been less clearly articulated is something that seems obvious to me—there is almost certainly one or more extracellular matrix pro-aggregation or pro-oligomerization factors that play important roles in specifying the pattern of regional and laminar conversion of soluble Aβ to oligomeric and fibrillar Aβ. These local factors (metals, glycosaminoglycans, etc.) may well represent good amyloid-lowering targets, and have been picked up with variable success (by Ashley Bush and Prana Biotechnology with CHQ and PBT2 on the upside and by Neurochem with Alzhemed on the downside).

    View all comments by Samuel Gandy
  4. We would like to reply to comments by Gouras and colleagues regarding our manuscript. Gouras, Lin, and Tampellini state, “Both the paper by Bero and colleagues and the Alzforum news story make a tacit assumption concerning the relationship between synaptic activity and β amyloid-related synapse dysfunction: that reducing plaque by reducing activity-driven secretion of Aβ is good for the brain. But is this assumption true?” We must point out that we did not make the tacit assumption being stated in any way.

    We would like to clarify the principal focus of our study: As deposition of amyloid plaques in specific brain regions is a fundamental feature of AD, we sought to elucidate the mechanisms that regulate brain region-specific amyloid deposition in AD. Using APP transgenic mice (Tg2576), we found that the steady-state level of neuronal activity in each brain region predicted interstitial fluid (ISF) Aβ levels and plaque deposition in a region-specific manner. We next found that physiological neuronal activity was sufficient to dynamically regulate ISF Aβ levels by acutely trimming or stimulating the whiskers on one side of the mouse facial pad while performing in vivo microdialysis in contralateral barrel cortex. Finally, we utilized longitudinal in vivo multiphoton microscopy to demonstrate that longer-term (28-day) unilateral whisker trimming was sufficient to prevent amyloid plaque formation and growth in contralateral barrel cortex, suggesting that physiological neuronal activity regulates amyloid plaque growth dynamics in living brain. Together, these data suggest that physiological neuronal activity regulates ISF Aβ levels and plaque deposition, and that regional differences in steady-state neuronal activity likely represent a key determinant of region-specific amyloid deposition in AD.

    The experiments described in the present paper did not aim to address whether intra- or extracellular Aβ assemblies represent the primary toxic Aβ species in AD. The pathological consequences of Aβ aggregation and extracellular Aβ deposition are well documented. However, intraneuronal Aβ accumulation may represent an additional mechanism of Aβ toxicity. This was not addressed in our study.

    Finally, if chronically elevated neuronal activity in specific brain regions was protective against AD neuropathology and its consequences, one might expect brain regions that exhibit greater neuronal activity throughout life to be less vulnerable to AD neuropathology. However, brain areas that are hypothesized to exhibit elevated neuronal activity throughout life (collectively termed the “default-mode network”) are precisely those that are most vulnerable to AD neuropathology. Further, these areas show dysfunction in cognitively normal people with amyloid deposition (Sperling et al., 2009; Hedden et al., 2009).

    Therefore, though neuronal activity forms the basis of brain function, chronic elevation of activity-dependent production and secretion of Aβ in specific brain regions appear to represent key determinants of region-specific amyloid deposition in AD. Of course, one would not want to globally suppress neuronal activity as any kind of therapy, or prevention of AD, with drugs such as sedatives or similar agents. However, we believe that further study of neuronal network modulation by environmental or even pharmacological means is warranted, not only to better understand network vulnerability to disease, but also potential therapeutic avenues.

    References:

    . Amyloid deposition is associated with impaired default network function in older persons without dementia. Neuron. 2009 Jul 30;63(2):178-88. PubMed.

    . Disruption of functional connectivity in clinically normal older adults harboring amyloid burden. J Neurosci. 2009 Oct 7;29(40):12686-94. PubMed.

  5. The Alzforum recently hosted a vivid Webinar that—among others—challenged the absolute supremacy of amyloid-β (Aβ) as the leading cause of Alzheimer’s disease (AD) neuropathology. Indeed, it is now accepted by an increasing number of investigators that AD-specific neuronal dysfunction and death could occasionally occur without being accompanied by the typical accumulation and aggregation of Aβ in specific brain regions. Yet, this hallmark of AD is still the one that unequivocally signals the presence of the disease for clinicians and investigators alike. Therefore, those who attempt to explain the AD process necessarily have to explain the accumulation and aggregation of Aβ in the AD brain. They also have to explain why these processes affect some—the so-called “vulnerable”—but not other, regions of the brain.

    The present study elegantly explains—first of all—why Aβ pathology preferentially develops in the vulnerable brain regions, by showing that these vulnerable regions are those where basal neural network activity is highest. The study also shows that this is so because increased network activity leads to increased levels of extracellular, soluble Aβ—a facilitating condition for the accumulation of soluble and insoluble Aβ aggregates, including neuritic plaques. The study is a tour de force: The many and difficult experiments are well controlled, and the conclusions—while intriguing—are logical, persuasive, and thought provoking.

    In spite of these remarkable achievements, the study does not explain the intimate mechanisms that lead to increased extracellular levels of Aβ, although the authors do their best to uncover them. One by one, the authors eliminate several common-sense explanations, such as the differential clearance of Aβ in different brain regions, or local increase in the precursors of Aβ, the Aβ precursor protein (APP) and the C-terminal fragment β (CTFβ); their variability across the brain cannot explain the variability in the level of soluble Aβ. The authors speculate that the increased levels of Aβ in the regions with high network activity could result from import of Aβ produced in the soma of distant neurons located elsewhere in the brain, which project into the vulnerable regions. Indeed, recent studies cited in this paper suggest that Aβ pathology may spread through interconnected neural networks. Several years ago, we proposed that the Aβ, produced in neurons in subcortical regions such as the locus coeruleus (LC), could be provided to remote brain regions through the processes of the LC neurons, which project in the AD vulnerable regions (1-3). More recent studies proposed a similar mechanism for the delivery of pyroglutamate-Aβ, produced in LC neurons, to the AD vulnerable brain regions (4,5). These studies fully support the idea that the Aβ present in the vulnerable regions could be produced elsewhere in the brain.

    We would like to propose yet another possible mechanism by which neuronal activity could release soluble Aβ in the extracellular space, from intraneuronal pools. Recent data indicate that Aβ co-resides with catecholamine neurotransmitters, and that Aβ and catecholamines undergo co-secretion (6). Also, our recent results show that a significant fraction of the intraneuronal Aβ is produced in recycling endosomes (7), which are known to be a source for synaptic vesicle recycling. It is conceivable that this intraneuronal Aβ is incorporated in the recycled synaptic vesicles, and becomes secreted during synaptic activity. These mechanisms could thus provide an explanation for the accumulation of Aβ in regions of high neural network activity.

    References:

    See also Muresan Z, V Muresan V. Brainstem Neurons Are Initiators of Neuritic Plaques. SWAN Alzheimer Knowledge Base. Alzheimer Research Forum. 2008; Hook, V.Y., et al., Regulated secretory vesicles contain beta-amyloid peptide forms with neuropeptides and catecholamines that undergo co-secretion. Annual Meeting of the Society for Neuroscience, San Diego, November 13-17, 2010; and Muresan, V., B.T. Lamb, and Z. Muresan, DISC1 is required for the formation of intracellular Aβ oligomers, suggesting a link between schizophrenia and Alzheimer’s disease. Annual Meeting of the Society for Neuroscience, San Diego, November 13-17, 2010.

    View all comments by Virgil Muresan

References

News Citations

  1. Tracing Alzheimer Disease Back to Source
  2. Cortical Hubs Found Capped With Amyloid
  3. Network Diagnostics: "Default-Mode" Brain Areas Identify Early AD
  4. BOLD New Look—Aβ Linked to Default Network Dysfunction
  5. Brain Aβ Patterns Linked to Brain Energy Metabolism
  6. Paper Alert: Synaptic Activity Increases Aβ Release
  7. Link Between Synaptic Activity, Aβ Processing Revealed
  8. Soluble Aβ—Bane or Boon? Real-time Data in Humans Yield New Insight
  9. ApoE4 Linked to Default Network Differences in Young Adults
  10. Sleep Deprivation Taxes Neurons, Racks Up Brain Aβ?
  11. Can Neurons Be Dozing in a Wakeful Mind?

Paper Citations

  1. . Molecular, structural, and functional characterization of Alzheimer's disease: evidence for a relationship between default activity, amyloid, and memory. J Neurosci. 2005 Aug 24;25(34):7709-17. PubMed.
  2. . Cortical hubs revealed by intrinsic functional connectivity: mapping, assessment of stability, and relation to Alzheimer's disease. J Neurosci. 2009 Feb 11;29(6):1860-73. PubMed.
  3. . Default-mode network activity distinguishes Alzheimer's disease from healthy aging: evidence from functional MRI. Proc Natl Acad Sci U S A. 2004 Mar 30;101(13):4637-42. PubMed.
  4. . Disruption of functional connectivity in clinically normal older adults harboring amyloid burden. J Neurosci. 2009 Oct 7;29(40):12686-94. PubMed.
  5. . Amyloid deposition is associated with impaired default network function in older persons without dementia. Neuron. 2009 Jul 30;63(2):178-88. PubMed.
  6. . Regional aerobic glycolysis in the human brain. Proc Natl Acad Sci U S A. 2010 Oct 12;107(41):17757-62. PubMed.
  7. . Spatial correlation between brain aerobic glycolysis and amyloid-β (Aβ ) deposition. Proc Natl Acad Sci U S A. 2010 Oct 12;107(41):17763-7. PubMed.
  8. . Synaptic activity regulates interstitial fluid amyloid-beta levels in vivo. Neuron. 2005 Dec 22;48(6):913-22. PubMed.
  9. . Endocytosis is required for synaptic activity-dependent release of amyloid-beta in vivo. Neuron. 2008 Apr 10;58(1):42-51. PubMed.
  10. . Amyloid-beta dynamics correlate with neurological status in the injured human brain. Science. 2008 Aug 29;321(5893):1221-4. PubMed.
  11. . Lactate production and neurotransmitters; evidence from microdialysis studies. Pharmacol Biochem Behav. 2008 Aug;90(2):273-81. PubMed.
  12. . Distinct patterns of brain activity in young carriers of the APOE-epsilon4 allele. Proc Natl Acad Sci U S A. 2009 Apr 28;106(17):7209-14. PubMed.
  13. . Amyloid-beta dynamics are regulated by orexin and the sleep-wake cycle. Science. 2009 Nov 13;326(5955):1005-7. PubMed.

Other Citations

  1. Tg2576 AD mice

External Citations

  1. APP/PS1 mice

Further Reading

Papers

  1. . APOE predicts amyloid-beta but not tau Alzheimer pathology in cognitively normal aging. Ann Neurol. 2010 Jan;67(1):122-31. PubMed.
  2. . APOE4 allele disrupts resting state fMRI connectivity in the absence of amyloid plaques or decreased CSF Aβ42. J Neurosci. 2010 Dec 15;30(50):17035-40. PubMed.

Primary Papers

  1. . Neuronal activity regulates the regional vulnerability to amyloid-β deposition. Nat Neurosci. 2011 Jun;14(6):750-6. PubMed.