Brain Aβ Patterns Linked to Brain Energy Metabolism

What makes some areas of the brain more vulnerable to Aβ? A pair of related papers in the September 13 PNAS online, from research groups led by Marcus Raichle and Mark Mintun at Washington University in St. Louis, Missouri, offer a clue. These scientists found that the same regions that accumulate Aβ in AD brains are distinguished in healthy young adults by their distinctive energy metabolism. Instead of burning glucose down to its end products—carbon dioxide and water—these regions make extensive use of glycolysis. This is a glucose metabolic pathway that produces less energy, but does it faster than going all the way through with oxidative phosphorylation, which was formerly thought to be the prevailing pathway from glucose in the brain. The results suggest a link between patterns of energy use in the brain and later vulnerability to Alzheimer disease.

“I think the finding is really fascinating, and potentially critical in trying to understand the mechanisms of why amyloid deposits where it does in the brain,” said Reisa Sperling at Brigham and Women’s Hospital, Boston, Massachusetts. She was not involved in the study. “I think it’s a major step forward.”

The researchers found that aerobic glycolysis (i.e., glucose metabolism over and above that explained by oxygen consumption) occurred predominantly in regions of the brain collectively known as the default mode network (DMN). The default mode network is active in resting or daydreaming brains, but dials down its activity during goal-directed tasks. The DMN is critical in multiple cognitive processes, Sperling said, particularly in memory encoding and retrieval. Intriguingly, DMN regions are some of the first to accumulate plaque in AD brains (see ARF related news story and ARF news story on Buckner et al., 2005; and ARF related news story on Buckner et al., 2009). People who carry the ApoE4 risk allele for Alzheimer’s show distinct patterns of brain activity in this network, again suggesting a relationship between metabolism in this area and susceptibility to AD (see ARF related news story on Buckner et al., 2008).

In the first paper, the authors addressed the question of what makes the default mode network special by looking at patterns of glucose utilization in the brain. About 10 to 12 percent of the brain’s glucose supply is used for glycolysis, but little was known about where in the brain glycolysis occurs. In part, this is because glycolysis is difficult to measure. The standard technique for visualizing glucose uptake is FDG-PET, positron emission tomography of a glucose analog in which an F18 isotope of fluoride replaces an oxygen atom to give the analog fluorodeoxyglucose (FDG). However, FDG-PET can’t distinguish between glycolysis and oxidative glucose metabolism, since FDG is not metabolized beyond the first step, which glycolysis and aerobic glucose metabolism share in common. To get around this limitation, first author Neil Vaishnavi compared glucose use to oxygen use and blood flow in the brains of 33 healthy young volunteers. Glycolysis was deemed to be taking place in regions where glucose consumption exceeded oxygen use.

Vaishnavi and colleagues discovered multiple areas with elevated glycolysis, and noticed that these regions corresponded quite well to the default mode network, including the medial prefrontal cortex, lateral parietal cortex, and posterior cingulate cortex, as well as to some additional areas associated with task control processes. The authors confirmed the spatial overlap between glycolysis and the DMN by using blood oxygen level-dependent (BOLD) functional MRI to map the DMN in a subset of their volunteers. Not all participants had the same levels of glycolysis, Mintun said.

Because the DMN is known to be susceptible to Aβ deposition, the authors followed up on their results by looking at the patterns of Aβ deposits in older brains. As reported in the second paper, first author Andrei Vlassenko and colleagues recruited 11 people with AD, and 14 elderly volunteers who were cognitively normal but had elevated levels of amyloid plaques in their brains. Vlassenko and colleagues measured Aβ using PET with Pittsburgh Compound B, and found that Aβ deposition in both groups of participants significantly correlated with the regions of glycolysis in young brains. Intriguingly, this relationship was much stronger in people with AD than in the cognitively normal volunteers who had amyloid deposits, suggesting a link between glycolysis and the development of AD, or even a partial explanation for why some people appear to be able to tolerate brain amyloid deposition for longer than others (see figure below).

image

False color intensity images (blue being low, red being high) of the brain suggest overlap between regions of glycolysis (upper) and plaque deposition (lower). Image credit: Mark Mintun

“This gives us a different way to think about how amyloid plaques start and grow,” Mintun said. “While this [result] in itself does not suggest a pathway for preventing amyloid plaques, it does give us some interesting ideas to pursue.” One key question, he said, is what distinguishes people who develop AD from those who don’t. “Are the people who developed plaques the ones who had the most extreme glycolysis?” Mintun said the scientists will explore this question by looking at glycolysis patterns in middle-aged adults to see if individual variations in glycolysis levels correlate with the formation of plaques. Another fruitful approach will be to investigate if this relationship between glycolysis and Aβ occurs in mouse models of AD, Mintun said, and if so, to use these models to explore the mechanisms behind the phenomenon.

Sperling finds the mechanistic questions particularly compelling. “What is it about the default mode network that requires aerobic glycolysis?” she asked. It’s not as simple as high energy demand. As Mintun and colleagues point out in the papers, the visual cortex sucks up high amounts of glucose, but it does not show much glycolysis, nor is it susceptible to Aβ deposits. Instead, Sperling said, the DMN may have a need for the very fast energy supply provided by glycolysis. “There’s evidence that this network has to modulate very rapidly when you are learning something new,” she said. The DMN also consists of regions that are highly interconnected, Sperling said, and perhaps there is something about maintaining this connectivity that is metabolically demanding.

Mintun and colleagues suggest several other hypotheses. Glycolysis might have a role in synaptic plasticity, for example, by rapidly providing a local energy source for membrane pumps that are critical in AMPA receptor turnover (see Zhang et al., 2009). Another possibility is that incomplete glucose metabolism provides small carbohydrates and derivatives needed for cellular proliferation and extension of axons and dendrites. In support of this, the authors note that glycolysis is much higher in the newborn brain than in the adult brain, accounting for 35 percent of a baby’s brain glucose use. In the adult brain, glucose metabolites could be important for synapse maintenance. Glycolysis increases during wakefulness and after the brain performs a demanding task, the authors say, implying glycolysis could play a role in learning-induced biosynthesis.

Other scientists see a synaptic connection as well. “In my opinion, this study adds to the evidence that synaptic or peri-synaptic changes are involved in the earliest predisposition to AD,” Eric Reiman of Banner Health in Phoenix, Arizona, wrote in an e-mail to ARF.

Others think that more data are needed before the implications of the new findings are clear. William Powers, at University of North Carolina in Chapel Hill, wonders if the correlation between glycolysis and Aβ deposition is a clue to the pathogenesis of AD, or just an interesting coincidence. To answer this question, Powers suggested using animal models to see if manipulating glucose metabolism affects Aβ accumulation. If so, that would imply a causal link.

Glycolysis may play a role in other neurodegenerative diseases as well. For example, Powers has found that brains with Huntington disease show reduced levels of glycolysis, although not in a regional pattern (see Powers et al., 2007). Because astrocytes are known to use glycolysis to power membrane sodium pumps, Powers said that reduced glycolysis might correlate with astrocyte dysfunction.

Importantly for biomarker and diagnostic work in the AD field, the new findings highlight a limitation of FDG-PET, in that this technique can’t distinguish between aerobic metabolism and glycolysis. Most researchers interviewed for this story felt that this was not a significant problem, and that FDG-PET will continue to be the method of choice for looking at glucose use in the brain. Powers pointed out that the oxygen metabolism studies used to detect glycolysis are quite difficult to perform, and would not be practical for regular diagnostic use.—Madolyn Bowman Rogers

Comments

  1. The studies on basic cerebral energy metabolism and its relationship to AD published in companion PNAS papers by Marc Raichle and colleagues at Washington University give us cause to stop and rethink—and dust off our copy of Lehninger. The details of the biochemistry are tedious and difficult to keep in working memory long enough to fully appreciate all of the relationships discussed in these papers. However, I think there are some manageable concepts here that could help move us forward in trying to understand AD and its treatment. First, the brain makes substantial use of aerobic glycolysis and does this in a regionally specific manner. Second, the metabolic fallout and vulnerabilities of cells—astrocytic or neuronal—that use aerobic glycolysis may be very different from those of cells that preferentially use oxidative phosphorylation. This fallout or vulnerability may be responsible for the regional overlap between high aerobic glycolysis and high amyloid-β deposition, not only in AD, but at the earliest stages of amyloid deposition in cognitively normal elderly controls. This has great importance. Perhaps these papers should be titled, “Amyloid Cascade Hypothesis—The Prequel.” That is, in our focus on the effects of amyloid-β (Aβ) accumulation, we seldom stop to discern why it is accumulating to begin with. The obvious exception to this is cases of autosomal dominant mutations in APP and presenilin. It is interesting that these forms of AD show a very different regional initiation of Aβ pathology in the striatum, another area of relatively high aerobic glycolysis.

    As a field, we have been faced with serious challenges in translating the amyloid cascade hypothesis into effective therapies. By no means is this statement meant as evidence that that hypothesis is incorrect, just that the translation of that hypothesis into therapies may require some rethinking. Many have recently suggested that interventions in the amyloid cascade may need to occur early in the pathogenesis of AD, perhaps even at a presymptomatic stage. The ideas put forth by Raichle and colleagues push us even further, posing new challenges, but perhaps opening new possibilities as well. Perhaps relatively modest interventions that would regulate the rate of aerobic glycolysis throughout mid-life—even prior to the initiation of Aβ deposition—would have large payoffs in later life. These papers are one more reason to ask ourselves, “Are we not finding effective therapies for AD because we’re looking at the wrong therapies, or because we’re looking at too late a stage of the disease to expect to be successful?”

    View all comments by William Klunk

  2. In these elegant and important studies, researchers from Washington University in St. Louis developed a novel strategy to characterize the resting pattern of “aerobic glycolysis” (reflecting the extent to which glucose metabolism exceeds that associated with oxygen metabolism) in the living human brain. To do so, they used PET measurements of cerebral blood flow, blood volume, oxygen metabolism and glucose metabolism and an innovative image-analysis strategy to compute a “glycolytic index” image in each person.

    They have shown that a group of brain regions associated with elevated aerobic glycolysis in cognitively normal young adults corresponds remarkably well to both the default mode network. This is the group of brain regions that the same research group originally found to be more active when normal individuals are not engaged in attention-demanding, goal-directed task performance, and also the group of regions associated with the most fibrillar amyloid in symptomatic and asymptomatic older adults with PET evidence of amyloid pathology.

    This study raises new questions about the different roles of glucose in neuronal synapses and peri-synaptic glial cells and the extent to which the cellular processes associated with aerobic glycolysis are related to the predisposition to amyloid pathology. Among other things, it will be interesting to characterize and compare the pattern of anaerobic metabolism in cognitively normal adults at differential risk for early- or late-onset AD to determine the extent to which alterations in these measures precede fibrillar amyloid deposition, and to identify molecular processes that could be targeted by novel treatments.

    In my opinion, this study adds to the evidence that synaptic or peri-synaptic changes are involved in the earliest predisposition to AD, at least in those individuals with late-onset AD, even preceding the earliest measurable amyloid alterations that may be involved in the pathogenesis of AD.

    I congratulate the researchers for their contributions. While I recognize the relative paucity of PET sites that are currently capable of conducting 150-based PET measurements of cerebral blood flow, blood volume and oxygen metabolism, I predict that the scientific importance of this work will continue to grow.

    View all comments by Eric M. Reiman

  3. Drs. Raichle and Mintun have reported a very interesting phenomenon, that is, amyloid deposition is related to aerobic glycolysis. But they have not clarified the reason why amyloid preferably deposits in the glycolysis portions of brain.

    I'd like to present some ideas about it. We have observed homocysteic acid as a pathogen for Alzheimer disease (1). HA induced Aβ42 (2). Also it is reported that HA induced strong glycolysis (3). Based on this evidence, we suggest that amyloid deposition in aerobic glycolysis is induced by HA.

    This suggestion should be tested. The field should consider homocysteic acid toxicity in Alzheimer disease.

    References:

    Hasegawa T, Mikoda N, Kitazawa M, LaFerla FM. Treatment of Alzheimer's disease with anti-homocysteic acid antibody in 3xTg-AD male mice. PLoS One. 2010 Jan 20;5(1):e8593. PubMed.

    Hasegawa T, Ukai W, Jo DG, Xu X, Mattson MP, Nakagawa M, Araki W, Saito T, Yamada T. Homocysteic acid induces intraneuronal accumulation of neurotoxic Abeta42: implications for the pathogenesis of Alzheimer's disease. J Neurosci Res. 2005 Jun 15;80(6):869-76. PubMed.

    Folbergrová J, Haugvicová R, Mares P. Attenuation of seizures induced by homocysteic acid in immature rats by metabotropic glutamate group II and group III receptor agonists. Brain Res. 2001 Jul 27;908(2):120-9. PubMed.

    View all comments by Tohru Hasegawa

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. Tracing Alzheimer Disease Back to Source
  2. Network News: Images of AD Brains Reveal Widespread Snafus
  3. Cortical Hubs Found Capped With Amyloid
  4. ApoE4 Linked to Default Network Differences in Young Adults

Paper Citations

  1. Buckner RL, Snyder AZ, Shannon BJ, LaRossa G, Sachs R, Fotenos AF, Sheline YI, Klunk WE, Mathis CA, Morris JC, Mintun MA. 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. Buckner RL, Sepulcre J, Talukdar T, Krienen FM, Liu H, Hedden T, Andrews-Hanna JR, Sperling RA, Johnson KA. 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. Buckner RL, Andrews-Hanna JR, Schacter DL. The brain's default network: anatomy, function, and relevance to disease. Ann N Y Acad Sci. 2008 Mar;1124:1-38. PubMed.
  4. Zhang D, Hou Q, Wang M, Lin A, Jarzylo L, Navis A, Raissi A, Liu F, Man HY. Na,K-ATPase activity regulates AMPA receptor turnover through proteasome-mediated proteolysis. J Neurosci. 2009 Apr 8;29(14):4498-511. PubMed.
  5. Powers WJ, Videen TO, Markham J, McGee-Minnich L, Antenor-Dorsey JV, Hershey T, Perlmutter JS. Selective defect of in vivo glycolysis in early Huntington's disease striatum. Proc Natl Acad Sci U S A. 2007 Feb 20;104(8):2945-9. PubMed.

Further Reading

Primary Papers

  1. Vaishnavi SN, Vlassenko AG, Rundle MM, Snyder AZ, Mintun MA, Raichle ME. Regional aerobic glycolysis in the human brain. Proc Natl Acad Sci U S A. 2010 Oct 12;107(41):17757-62. Epub 2010 Sep 13 PubMed.
  2. Vlassenko AG, Vaishnavi SN, Couture L, Sacco D, Shannon BJ, Mach RH, Morris JC, Raichle ME, Mintun MA. 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.

Annotate

To make an annotation you must Login or Register.