In a provocative paper, researchers led by Tobias Hartmann at the Center for Molecular Biology in Heidelberg, Germany, proposed in the October 9 Nature Cell Biology online that Aβ42 promotes breakdown of the lipid sphingomyelin as part of its physiological function, whereas its sister peptide Aβ40 slows down production of cholesterol. Control of this new form of lipid metabolism falls to presenilin, the scientists report, and FAD mutations in the gene for this protease complex drive cholesterol levels up and sphingomyelin levels down.

The finding harks back to a basic discovery that earned Michael Brown and Joseph Goldstein a Nobel Prize in 1985. While studying how cells regulate cholesterol metabolism, these scientists uncovered one of the first examples of regulated intramembrane proteolysis (RIP), i.e., of sterol-regulatory element binding proteins. These are cleaved to activate lipid synthesis when cells run low on sterols. Hartmann’s group wondered whether RIP processing of amyloid-β precursor protein (AβPP) might also serve to modulate lipid metabolism. Given nature’s penchant for ripping itself off, it’s easy to imagine how intramembrane proteolysis may have so evolved. Indeed, Aβ, RIPped from AβPP by the intramembrane γ-secretase, appears to be able to modulate both cholesterol and sphingomyelin metabolism.

The finding could spark a reappraisal of the connections between lipids and Alzheimer disease (AD). At one level, evidence has steadily grown that elevated cholesterol is a risk factor for sporadic or late-onset AD (see below). It’s also becoming clear that lipids regulate AβPP processing (see, for example, ARF related news story). But it’s been much less clear whether the reverse is also true, i.e., whether AβPP and/or its derivatives regulate lipid metabolism.

First author Marcus Grimm and colleagues in Hartmann’s lab (with collaborators Jie Shen at Brigham and Women’s Hospital, Boston; Ulrike Mueller at Heidelberg; and Bart de Strooper at the University of Leuven, Belgium) examined how loss of γ-secretase influences lipid metabolism. When the authors examined mouse embryonic fibroblasts devoid of presenilins 1 or 2 (either of which can provide the core protease for the intramembrane γ-secretase complex), they found that cholesterol levels were twofold higher than in normal fibroblasts. Sphingomyelin levels, too, were increased by about the same amount, but levels of phosphatidylcholine, another lipid, remained unchanged. With respect to the lipids, changes in sphingomyelin levels alone are attributable to the fact that neutral sphingomyelinase activity, which breaks down sphingomyelin specifically, is more than doubled in PS1/PS2-negative cells. In fact, Grimm and colleagues found that levels of ceramide, a product of sphingomyelinase that has been implicated in AD (see ARF related news story and recent ARF Live Discussion), are increased, although not as strongly as sphingomyelin in PS-negative cells.

To test these data in vivo, the scientists crossed PS1 conditional knockout mice developed in Shen’s lab with PS2 knockout mice. Reproducing their initial result, the scientists found that brain levels of cholesterol and sphingomyelin were about 20 percent higher than normal in the γ-secretase-depleted mice.

How might γ-secretase contribute to lipid metabolism? The protease has many other substrates besides AβPP, for example, Notch. Conceivably, many γ-secretase products could regulate fat metabolism. But Grimm and colleagues narrowed the field down to AβPP products. In embryonic fibroblasts devoid of AβPP and its homolog APLP2, inhibiting γ-secretase failed to elicit any changes in cholesterol or sphingomyelin levels. In contrast, treating wild-type cells with γ-secretase inhibitors mimicked the effect seen in PS1/PS2 knockout cells, suggesting that AβPP is necessary and sufficient for γ-secretase modulation of lipid metabolism.

The authors provide evidence for the role of Aβ peptides themselves. They report that at physiological concentrations (0.04-40 ng/ml), Aβ42 can directly stimulate neutral sphingomyelinase, while Aβ40 preferentially inhibits hydroxymethylglutaryl-CoA reductase (HMGR). That enzyme catalyzes the rate-limiting step in cholesterol production and is the target of cholesterol-lowering statins (see ARF related news story on the potential benefits of statins for AD patients).

Together with previous work, these findings paint a more complex picture of the relationship between lipids and γ-secretase/AβPP (see Figure). For example, it suggests that shifts in the ratio of Aβ40:Aβ42 would lead to changes in the cholesterol-sphingomyelin balance. Researchers know, for example, that some familial PS mutations causing early onset AD lead to an increase in the relative amount of Aβ42. This has long been interpreted as crucial to pathogenesis because Aβ42 tends to aggregate more readily than does Aβ40. The present paper raises the question of whether the role of these mutations on lipid metabolism has as much to do with pathogenesis as with amyloid aggregation. “Active involvement of APP cleavage, Aβ, γ-secretase activity, and the effect of pathological PS-FAD mutations in lipid homeostasis provides a functional context for APP processing that has direct relevance for AD and may provide a rational basis for therapy,” the authors conclude. image

γ-secretase and lipid homeostasis
Finding that Aβ42 activates sphingomyelin catabolism while Aβ40 inhibits de novo cholesterol biosynthesis indicates an important role for γ-secretase in lipid metabolism. Factors that alter γ-secretase activity, such as lipids themselves or familial presenilin mutations, might, therefore, have a profound effect on cellular lipids. [Image courtesy of Tobias Hartmann]

Several other papers out this week deepened the complexity of the cholesterol story. Miia Kivipelto and colleagues at the Karolinska Institute and collaborating laboratories in Sweden report in the October 12 Archives of Neurology that obesity and high total serum cholesterol during mid-life (44-57 years) double the risk for developing AD in later years (age 65-79). The findings were based on clinical characteristics of almost 2,300 volunteers who were examined between 1972 and 1987, then followed up in 1998. Another study, led by Gail Li at the University of Washington, Seattle, at first blush appears to contradict the Karolinska study. In the October issue of Neurology, Li and colleagues report results of a cohort study that failed to see high serum cholesterol in old age as a risk factor for AD. But Li et al. point out that they had no access to cholesterol data for their patients from prior to 1988, and were therefore unable to test for a relationship between mid-life serum cholesterol and dementia late in life. A number of recent studies have cited mid-life metabolic factors as more predictive of AD than the same factors measured in old age. (For more on this issue, see Tobias Hartmann Q&A, posted at the end of this news story.)

Finally, despite efforts by the medical community and government agencies to educate the public about the dangers of obesity and high cholesterol, warnings go largely unheeded, at least in the US. That’s the disheartening conclusion to be drawn from Margaret Carroll and colleagues’ analysis of data from the National Health and Nutrition Examination Surveys (NHANES). Reporting in the October 12 JAMA, the authors reveal that total serum cholesterol in American adults was a measly three points lower (at 203 mg/dL) during 1999-2002 than between 1988 and 1994. The difference was not even statistically significant. The percentage of adults with very high cholesterol (greater than 240 mg/dL) fell from 20 to 17 percent.

The only saving grace appears to be that men older than 60 and women older than 50 showed steeper drops in total serum cholesterol. For example, women age 50-59 lowered their cholesterol from an average of 228 mg/dL in 1988-1994, to 216 mg/dL in the years 1999-2002. These reductions reflect the use of lipid-lowering medication, suggest the authors.

A hard truth to swallow in light of Hartmann’s study is that mean triglyceride levels actually increased slightly over the same period, thanks to the growing girth of the U.S. population.—Tom Fagan

Q&A with Tobias Hartmann..

Q: There have been many reports focusing on how lipids modulate γ-secretase activity. What prompted you to examine the effect of γ-secretase activity on lipid metabolism?
A: This dates back to my Ph.D. time in 1994. I observed that fetal calf serum contains something that increases γ-secretase activity. After some tests, it turned out that this is not due to any serum proteins, so lipids became candidates. Other lipids modulate APP cleavage, including GM1 (Zha) and cholesterol (Bodovitz, Racchi, Simons, Frears, Mizuno, Refolo, Fassbender, etc.). Whenever we studied a protein with regard to APP secretion or cell viability, lipids seemed to have an effect especially on γ-secretase. In 1997, we postulated that membrane composition is the major factor regulating the Aβ40/42 ratio (Hartmann), and, indeed, the embedding within the membrane of the APP transmembrane domain supports this (Grziwa). Together, this led to the idea that the sensitivity of γ-secretase to lipids must have a functional basis. This can mean either that a substrate needs to be cleaved in a lipid-dependent way, or that γ-secretase is an active player in cellular lipid biology. From here on we focused on two areas. One was the in vivo relevance of the γ-secretase/lipid connection, the role of Aβ42 in it, and the therapeutic potential (Fassbender, Simons); the other was to clarify the functional aspect. Then we found that lipid trafficking results in a relocalization of γ-secretase to a compartment responsible for cholesterol and sphingolipid shuffling (Runz). Finally, we knew we were on the right track.

All that time, there were already indications that Aβ might be involved. However, these studies were using non-physiological Aβ levels and I was interested in the physiological role of these proteins. I doubted whether Aβ would be a physiological player in this. Especially important was Dora Kovacs’s work with cholesterol esters (Puglielli). First, it appeared contradictory to our ideas, but with later updates, extended them very well.

Q: How do your findings affect the way we view the relationship between lipids and AD?
A: There are different angles to this.

1. First, the physiological roles of the γ-secretase/APP/Aβ lipid regulatory system. The paper explains the functional context of lipids in AD and why this affects the AD risk. The conclusion (for those who think Aβ42 is the culprit) is that lipids and everything else interfering with the γ-secretase/APP/Aβ lipid regulatory system will strongly affect Aβ production. It is important to understand that it is not necessarily big changes in single lipid levels that are central here. Rather, it is how strongly and for how long the γ-secretase/APP/Aβ system attempts to counteract an imbalance in cellular lipid composition. This system works in concert with other regulatory cycles involved in lipid homeostasis and should not be studied in isolation.

In respect to Aβ, everything depends on how much of the total regulatory burden the γ-secretase/APP/Aβ system has to carry. For example, a dysfunction in the Brown and Goldstein pathway (Brown), or in cholesterol input, would place a heavy toll on the γ-secretase/APP/Aβ lipid regulatory pathway and Aβ levels would be high. But as long as the pathway is able to cope with that, one would find only slight changes in lipid levels. For neurons, indications exist that ApoE is involved (Levi). It is necessary to pay attention to the lipid profile, but watching cholesterol alone is insufficient. For example, an increase in cholesterol has little consequence for a cell when it is paralleled by increased sphingomyelin (SM), and the opposite is true when increased cholesterol is paralleled by decreased SM, as we found is the case in PS-FAD.

2. Now consider the pathological situation of huge Aβ accumulation and massive neurodegeneration. The γ-secretase/APP/Aβ lipid regulatory system appears to contribute to this, but this is not what we studied here. There is a bookshelf full of data published by other groups dealing with this aspect (e.g., Liu, Gong, Jana, Qi, and many more). We want to be very careful in extrapolating our data to pathological Aβ levels. Indeed, some of our data show that this is not easily possible, but that said, it also gives a fascinating outlook of what might be ahead. The challenge here is to obtain reliable and meaningful animal and patient data. We don’t know how Aβ42 is toxic. Maybe high Aβ levels cause toxic lipid changes, but we do not know. There have been claims in this and other directions, but I have not seen data strong enough to prove them.

3. Now to Aβ-related therapy. Since we know the function of Aβ, we can look for novel targets and evaluate existing therapeutic approaches in a new light. Statins come to mind as one example. Instead of targeting γ-secretase directly, this approach uses a naturally existing system that appears to be tailor-made to do what we want, that is, reduce Aβ production, simply by replacing Aβ with a statin in the cholesterol regulatory cycle. This should incur few side effects. NSAIDs and other molecules that shift the 40/42 ratio are another example (kudos to Claus Piertrzik, who brought up this topic). These substances may reduce cholesterol and decrease Aβ42, for the prize of increasing SM, which may further reduce Aβ production. Straight γ-secretase inhibitors in principle would increase cholesterol levels. It would be interesting to see whether the concentrations used in clinical dosing are sufficient to do this.

Even approaches that at first sight appear to be entirely unrelated, like increasing or decreasing copper, could influence this system. Copper itself is a potent regulator of lipid metabolic enzymes, and a small phase 2 trial of dietary copper salt supplementation led by Thomas Bayer at Saarland University may give us insight on that.

4. Lipid homeostasis varies strongly within the human population, and certain approaches might be more tolerable or effective for some patients than others. Some of our preliminary data suggest that there might be significant variation between different tissues or cells. Current knowledge is woefully insufficient to judge therapeutic or preventive approaches, but we now know a bit better what to look for.

5. Lastly, there’s the question of mid-life risk factors, such as obesity, high cholesterol, physical activity, and the daily breakfast eggs. All of these may have their effects on lipid homeostasis. While plasma cholesterol hardly enters the human brain, this may not be true for other lipids and factors that affect brain lipid homeostasis. When we talk about elevated plasma (LDL) cholesterol or triglyceride levels, then this is seldom due to wrong eating habits alone. Genetic factors, positive effects of physical activity, a weekly meal of sea fish, however, may very well have a direct role for lipid homeostasis in the human brain. This would all fit wonderfully into the concept of the γ-secretase/APP/Aβ lipid regulatory system, but I hesitate to draw a direct link between mechanistic and epidemiological studies. These studies can be fruitful to each other as far as speculation goes, but they never prove or disprove each other. Tempting as this might be, drawing such links is not solid science. Late-life and AD epidemiology is even more difficult to factor in, since there is little coherent mechanistic information on AD-like lipid homeostasis and no mechanistic in vivo data. Until then, living a healthy lifestyle can’t hurt even if we don’t understand exactly why.

Q: 0.04-40ng/mL is about 0.04-40 nM. That’s pretty potent activation/inhibition. Do you think lipid modulation is the major role for Aβ/APP or might this be ancillary?
A: This is not very potent. It is just the physiological Aβ concentration range. Any biological Aβ effects above this range would be non-physiological, unless confined to a specific subcellular compartment or vesicular lumen.

Is this the major role? We can answer this question only after other physiological function(s) are identified. Meanwhile, here is my catalogue of features any major physiological function must fulfill:

1. Does one find this function in every cell that expresses the protein?

2. Is the proposed role indeed a function or just an observation, which could be triggered by an independent mechanism?

3. If you remove APP (or Aβ) from the system, do you lose this function? Otherwise, you might observe a pathological but not a physiological function.

4. Define what “the major role” truly means.

I see the role of the γ-secretase/APP/Aβ system and especially of Aβ as regulator(s) of lipid homeostasis. This is not necessarily limited to HMGR and nSMase. We already showed that Aβ stimulates acidic SMases and there might be more enzymes subject to regulation by Aβ. Examples include phosphodiesterases, the enzymatic class to which SMases belong. PDE4 could be such a candidate (see Lehnart et al., 2005) and there are some indications pointing towards vascular function. Dennis Selkoe just published an amazing story on ACE (Hemming et al., 2005), and there are more links pointing in this direction. One would need to examine closely the interaction and enzyme activities. Based on what is common in lipid homeostasis, this would not be a surprise, but rather economical use of resources by nature. At the end of the day, this all falls back to complex lipid regulatory cycles.

There is room for other functions, but they are cell type-specific. Among what has been proposed, I favor Dennis Selkoe’s LTP finding (Walsh). However, we don’t know the mechanism, which could easily be lipid-mediated, and it would be a neuron-specific role of Aβ, not a general one.

The situation might be entirely different for APP. Our findings provide APP processing with a physiological function. Obviously, this clearly identifies an APP function, but APP is a complex molecule that may serve many masters. The KPI, OX-2 and L-APP alternatively spliced exons indicate that this is the case. With regard to additional brain-specific functions, I think we will learn first from the often-neglected APLP-1. With a bit of luck, this will lead us to further APP functions in brain. The Soba study on APP family member interaction might point the way on how this could be done (Soba).

Q: How do you think Aβ40 is affecting HMGR? Any thoughts on the mechanism of action?
A: Very important question; we are working on this. Stay tuned…

Q: Likewise, how might Aβ42 affect nSMase? Your data indicate that there is insufficient ROS generated to affect the enzyme. Do you have any theories?
A: Unless you want to invoke some esoteric energy transfer, Aβ needs to contact the SMase directly. ROS, based on Jana (Jana) and the work pioneered by Allan Butterfield (Hensley) and Ashley Bush (Huang), would have been the only alternative interaction. We initially thought that ROS can’t be a player, because Aβ ROS occurs only at higher Aβ concentrations, but one reviewer asked for it and now I’m glad we did it, although the results were negative.

We further know that membranes are not an obligatory factor in this. Moreover, SMases are secreted and remain active. Structurally, Aβ has everything needed to integrate itself into lipophilic grooves and it binds to lipoprotein particles, albumin, gangliosides, and maybe other lipids like the related SM. Thus, there are many possibilities, all of which are hypothetical at this stage.

Q: How can the addition of two extra amino acids turn the peptide from a potent SMase activator into an equally potent HMGR inhibitor? Does this provide a clue to the mechanism of action? Is there precedence for such a switch in nature?
A: I know of no other peptide that’s been studied as extensively as Aβ. It is established that both peptides differ quite significantly from each other. There are many examples where single amino acid substitutions make a big difference, for example, RNA editing for some channels. Regarding the details of this interaction, speculation is premature until we know more about the interaction with HMGR. The surprising differential activity was the reason why we used peptides from two independent sources and one of the reasons behind using the FAD mutants. However, biologically it makes sense. In steady state, the γ-secretase/APP/Aβ lipid regulatory system runs to keep the molar ratio between cholesterol and sphingomyelin constant. Maybe a single peptide would suffice for that. But then there are situations where it might be necessary to change the lipid ratio or to release ceramide. For this you need two independent cycles. That these cycles can indeed be independent from each other is another conclusion to be drawn from the PS-FAD results (SM decreases, cholesterol increases; when total Aβ level increase without changing the 40/42 ratio, both lipids increase).

This highlights another fascinating aspect in regard to the SMases. SMases are very important on the subcellular level, for example, for axonal and synaptic function. One function of sphingomyelin is that it binds cholesterol and thereby regulates lipid microdomain (raft) function. SM degradation is relatively fast, and locally increased Aβ42 would easily be able to fine-tune raft function, with obvious consequences for neuron function.

Q: Your findings indicate a novel mechanism whereby FAD PS mutations might affect neurons. The altered Aβ40/42 ratio would lead to higher cholesterol, lower sphingomyelin, more ceramide. How important do you think this new balance might be in the disease pathology? How about compensatory mechanisms?
A: You are hinting towards whether Aβ42-induced lipid changes are actually the disease-causing factor. Lipids, including sphingolipids, are established factors in some severe neurodegenerative diseases. Is this the case for (F)AD? Correlations won’t help here, because they trap us in a hen-and-egg cycle. Although difficult, this question can be experimentally addressed.

Regarding ceramide and apoptosis/neuronal loss, in a study with Henning Walczak, we extensively tested the knockout cells for apoptosis, sensitivity to ceramide, etc. All of this turned out to be negative. The main impact on the total cell is on SM and cholesterol; the impact on ceramide is small. Thus, compensatory mechanisms are actively keeping the ceramide amounts at a tolerable level. This may be different in already degenerating neurons.

Q: What is the role of rafts in this?
A: Rafts are now very important. Lipids modify raft function. APP is outside of the rafts, BACE goes in and out, and γ-secretase sits right in there regulating the lipid composition (cholesterol and SM are main components of rafts). The most likely mechanism is that whenever membrane composition is suboptimal, APP and BACE come together, producing C99. The C99 then enters the raft (or alternatively γ-secretase leaves the raft) and γ-secretase is able to produce Aβ. The Aβ then sets the membrane composition back to optimum levels. This causes APP and secretases to separate and γ-secretase ceases to produce Aβ until the next cycle starts.

Q: Why is the lipid regulation described by Brown and Goldstein not sufficient?
A: If you drive an expensive system, you prefer to have two regulatory cycles. This safeguards what is going on and allows for better adjustments. For example, a car uses two regulatory cycles, gas for acceleration and brakes for deceleration. Lose one and you are in trouble. I’m intrigued by how nature co-evolved two complex regulation systems by using the same technical principles for a closely related task while not reusing a single protein. That the same proteins could not be used is due to the topology of the proteins, but that the principle is identical tells us a lot about what the essential parameters are and how powerful evolution is.

Q: What do you think are the most important questions raised by your findings?
A: You covered most of them. Here are other pressing ones: What is the role of the other secretases in this? What about the other peptides? What about the other APP family members? When changing HMGR activity, not only cholesterol but isoprenoids and cholesterol precursors change. What does this do to neuronal function? Which factors could throw spanners into the regulatory cycle; constant cholesterol influx via ApoE4 could do such a thing. Are there other factors? Now that we understand the mechanism, we see how statins work to reduce Aβ production. Are there even better targets in this cycle? Then there is the issue of diet—the most simple and risk-free therapy one could think of.

We can now apply existing knowledge taken from the cardiovascular research field, apply it to certain aspects of AD (and maybe vascular dementia), and predict what should happen. This is a big move forward and should increase our ability to fight AD.

References:
Zha Q, Ruan Y, Hartmann T, Beyreuther K, Zhang D. GM1 ganglioside regulates the proteolysis of amyloid precursor protein. Mol Psychiatry. 2004 Oct;9(10):946-52. Abstract

Bodovitz S, Klein WL. Cholesterol modulates alpha-secretase cleavage of amyloid precursor protein. J Biol Chem. 1996 Feb 23;271(8):4436-40. Abstract

Racchi M, Baetta R, Salvietti N, Ianna P, Franceschini G, Paoletti R, Fumagalli R, Govoni S, Trabucchi M, Soma M. Secretory processing of amyloid precursor protein is inhibited by increase in cellular cholesterol content. Biochem J. 1997 Mar 15;322 ( Pt 3):893-8. Abstract

Simons M, Keller P, De Strooper B, Beyreuther K, Dotti CG, Simons K. Cholesterol depletion inhibits the generation of beta-amyloid in hippocampal neurons. Proc Natl Acad Sci U S A. 1998 May 26;95(11):6460-4. Abstract

Frears ER, Stephens DJ, Walters CE, Davies H, Austen BM. The role of cholesterol in the biosynthesis of beta-amyloid. Neuroreport. 1999 Jun 3;10(8):1699-705. Abstract

Mizuno T, Nakata M, Naiki H, Michikawa M, Wang R, Haass C, Yanagisawa K. Cholesterol-dependent generation of a seeding amyloid beta-protein in cell culture. J Biol Chem. 1999 May 21;274(21):15110-4. Abstract

Refolo LM, Malester B, LaFrancois J, Bryant-Thomas T, Wang R, Tint GS, Sambamurti K, Duff K, Pappolla MA. Hypercholesterolemia accelerates the Alzheimer's amyloid pathology in a transgenic mouse model. Neurobiol Dis. 2000 Aug;7(4):321-31. Erratum in: Neurobiol Dis 2000 Dec;7(6 Pt B):690. Abstract

Fassbender K, Simons M, Bergmann C, Stroick M, Lutjohann D, Keller P, Runz H, Kuhl S, Bertsch T, von Bergmann K, Hennerici M, Beyreuther K, Hartmann T. Simvastatin strongly reduces levels of Alzheimer's disease beta -amyloid peptides Abeta 42 and Abeta 40 in vitro and in vivo. Proc Natl Acad Sci U S A. 2001 May 8;98(10):5856-61. Epub 2001 Apr 10. Abstract

Hartmann T, Bieger SC, Bruhl B, Tienari PJ, Ida N, Allsop D, Roberts GW, Masters CL, Dotti CG, Unsicker K, Beyreuther K. Distinct sites of intracellular production for Alzheimer's disease A beta40/42 amyloid peptides. Nat Med. 1997 Sep;3(9):1016-20. Abstract

Grziwa B, Grimm MO, Masters CL, Beyreuther K, Hartmann T, Lichtenthaler SF. The transmembrane domain of the amyloid precursor protein in microsomal membranes is on both sides shorter than predicted. J Biol Chem. 2003 Feb 28;278(9):6803-8. Epub 2002 Nov 25. Abstract

Simons M, Schwarzler F, Lutjohann D, von Bergmann K, Beyreuther K, Dichgans J, Wormstall H, Hartmann T, Schulz JB. Treatment with simvastatin in normocholesterolemic patients with Alzheimer's disease: A 26-week randomized, placebo-controlled, double-blind trial. Ann Neurol. 2002 Sep;52(3):346-50. Abstract

Runz H, Rietdorf J, Tomic I, de Bernard M, Beyreuther K, Pepperkok R, Hartmann T. Inhibition of intracellular cholesterol transport alters presenilin localization and amyloid precursor protein processing in neuronal cells. J Neurosci. 2002 Mar 1;22(5):1679-89. Abstract

Puglielli L, Konopka G, Pack-Chung E, Ingano LA, Berezovska O, Hyman BT, Chang TY, Tanzi RE, Kovacs DM. Acyl-coenzyme A: cholesterol acyltransferase modulates the generation of the amyloid beta-peptide. Nat Cell Biol. 2001 Oct;3(10):905-12. Abstract

Brown MS, Goldstein JL. A proteolytic pathway that controls the cholesterol content of membranes, cells, and blood. Proc Natl Acad Sci U S A. 1999 Sep 28;96(20):11041-8. Review. Abstract 49513

Levi O, Lutjohann D, Devir A, von Bergmann K, Hartmann T, Michaelson DM. Regulation of hippocampal cholesterol metabolism by apoE and environmental stimulation. J Neurochem. 2005 Sep 27; [Epub ahead of print] Abstract

Liu Y, Peterson DA, Schubert D. Amyloid beta peptide alters intracellular vesicle trafficking and cholesterol homeostasis. Proc Natl Acad Sci U S A. 1998 Oct 27;95(22):13266-71. Abstract

Gong JS, Sawamura N, Zou K, Sakai J, Yanagisawa K, Michikawa M. Amyloid beta-protein affects cholesterol metabolism in cultured neurons: implications for pivotal role of cholesterol in the amyloid cascade. J Neurosci Res. 2002 Nov 1;70(3):438-46. Abstract

Jana A, Pahan K. Fibrillar amyloid-beta peptides kill human primary neurons via NADPH oxidase-mediated activation of neutral sphingomyelinase. Implications for Alzheimer's disease. J Biol Chem. 2004 Dec 3;279(49):51451-9. Epub 2004 Sep 27. Abstract

Qi XL, Xiu J, Shan KR, Xiao Y, Gu R, Liu RY, Guan ZZ. Oxidative stress induced by beta-amyloid peptide(1-42) is involved in the altered composition of cellular membrane lipids and the decreased expression of nicotinic receptors in human SH-SY5Y neuroblastoma cells. Neurochem Int. 2005 Jun;46(8):613-21. Epub 2005 Apr 7. Abstract

Ida N, Hartmann T, Pantel J, Schroder J, Zerfass R, Forstl H, Sandbrink R, Masters CL, Beyreuther K. Analysis of heterogeneous A4 peptides in human cerebrospinal fluid and blood by a newly developed sensitive Western blot assay. J Biol Chem. 1996 Sep 13;271(37):22908-14. Abstract

Walsh DM, Klyubin I, Fadeeva JV, Cullen WK, Anwyl R, Wolfe MS, Rowan MJ, Selkoe DJ. Naturally secreted oligomers of amyloid beta protein potently inhibit hippocampal long-term potentiation in vivo. Nature. 2002 Apr 4;416(6880):535-9. Abstract

Lehnart SE, Wehrens XH, Reiken S, Warrier S, Belevych AE, Harvey RD, Richter W, Jin SL, Conti M, Marks AR. Phosphodiesterase 4D Deficiency in the Ryanodine-Receptor Complex Promotes Heart Failure and Arrhythmias. Cell. 2005 Oct 7;123(1):25-35. Abstract

Hemming ML, Selkoe DJ. Amyloid beta -protein is degraded by cellular angiotensin-converting enzyme (ACE) and elevated by an ACE inhibitor. J Biol Chem. 2005 Sep 9; [Epub ahead of print] Abstract

Soba P, Eggert S, Wagner K, Zentgraf H, Siehl K, Kreger S, Lower A, Langer A, Merdes G, Paro R, Masters CL, Muller U, Kins S, Beyreuther K. Homo- and heterodimerization of APP family members promotes intercellular adhesion. EMBO J. 2005 Sep 29; [Epub ahead of print] Abstract

Hensley K, Carney JM, Mattson MP, Aksenova M, Harris M, Wu JF, Floyd RA, Butterfield DA. A model for beta-amyloid aggregation and neurotoxicity based on free radical generation by the peptide: relevance to Alzheimer disease. Proc Natl Acad Sci U S A. 1994 Apr 12;91(8):3270-4. Abstract

Huang X, Cuajungco MP, Atwood CS, Hartshorn MA, Tyndall JD, Hanson GR, Stokes KC, Leopold M, Multhaup G, Goldstein LE, Scarpa RC, Saunders AJ, Lim J, Moir RD, Glabe C, Bowden EF, Masters CL, Fairlie DP, Tanzi RE, Bush AI. Cu(II) potentiation of Alzheimer abeta neurotoxicity. Correlation with cell-free hydrogen peroxide production and metal reduction. J Biol Chem. 1999 Dec 24;274(52):37111-6. Abstract

Comments

  1. Several studies have indicated that both cholesterol and sphingomyelin metabolism can affect the generation of Aβ. In this very elegant paper, Tobias Hartmann’s group has decided to go the opposite way and analyze whether Aβ could affect cholesterol and SM metabolism. They have used several genetic and biochemical approaches to reach the unexpected conclusion that the Aβ peptide can stimulate SM hydrolysis and reduce the biosynthesis of both SM and cholesterol. These effects could potentially be explained by perturbation of the lipid bilayer produced by Aβ. However, the fact that Aβ (in physiological concentrations) can stimulate both a purified neutral SMase (nSMase) activity in vitro and the nSMase activity recovered from cell homogenates suggests a direct effect of the peptide on the enzyme rather than on the lipid environment.

    It has long been known that sphingomyelin and cholesterol like to go together (1). Increased biosynthesis of cholesterol is accompanied by increased generation of sphingomyelin. Indeed, the same transcriptional machinery (SREBP) regulates both biosynthetic pathways. The ultimate goal is to keep or cluster cholesterol at the plasma membrane (PM). Sphingomyelin is probably the best “cholesterol-binding lipid” and is highly enriched in the PM. Indeed, its stoichiometry of cholesterol binding is 3:1 (cholesterol:SM), which is extremely high considering that phosphatydylcholine (another common “cholesterol-binding lipid”) binds cholesterol with a 1:1 stiochiometric ratio (cholesterol:PC). This close relationship works the other way around, too. Cell surface hydrolysis of SM is accompanied by a fast translocation of cholesterol to the endoplasmic reticulum (ER). The retro-translocation has the ultimate effect of down-regulating cholesterol biosynthesis (through the HMG-CoA reductase) and increasing the storage of cholesterol ester (which, however, is only temporary and limited to certain cell types). In addition to the effects produced by the retro-translocation of PM-cholesterol, ceramide (one of the products of SM hydrolysis) can down-regulate the proteolysis/activation of SREBP and, therefore, reduce both biosynthesis and uptake of cholesterol (2, 3).

    Our group has recently shown that normal aging of the brain is accompanied by activation of nSMase and consequent liberation of the second messenger ceramide, which is able to induce Aβ generation (4). This age-associated effect could be blocked by nSMase inhibitors and by genetic disruption of the ligand-binding domain of the neurotrophin receptor p75NTR, which is responsible for the activation of nSMase (in the brain and during the normal process of aging). If we join the results produced by Grimm et al. and our group (4), we can envision a model in which aging activates ceramide production and Aβ generation by acting through nSMase; Aβ can further stimulate nSMase by an apparent direct interaction, fostering an additional production of Aβ. Sphingomyelin hydrolysis would have the additional effect of reducing cholesterol biosynthesis in astrocytes, affecting the secretion of lipoprotein particles required for neurons to generate/sustain their own synapses (5). In conclusion, a vicious circle might operate that leads to abnormal production of Aβ and affects synaptogenesis. Tobias’s group has shown that the nSMase inhibitor GW4869 can block Aβ production in neurons; our group has shown that a different nSMase, manumycin A, can block Aβ production in both primary neurons and mice (4). This strategy seems to work for both the age-dependent and the Aβ-mediated effect, and is predicted to act upstream of the “vicious circle.”

    I have so far considered the effects on cholesterol metabolism described by Grimm et al. as a consequence of SM hydrolysis because there is no evidence of a possible direct effect of the Aβ peptide on the HMG-CoA reductase (at the enzymatic/protein level). Indeed, even though Aβ was given to intact cells, the authors observed a decrease in the incorporation of acetate into the mevalonic pathway; a fact that implicates the HMG-CoA reductase, an ER membrane-based protein. However, we could have yet another surprise and discover that Aβ can act directly on the enzyme itself. It would be interesting to look at SREBP processing and HMG-CoA degradation under the above conditions, and at the effects of Aβ on a purified/enriched preparation of HMG-CoA in vitro.

    Finally, one can wonder how the lack of presenilins can stimulate the SM-synthase activity. In fact, in contrast to nSMase, SM-synthase is an allosteric enzyme that seems to respond to the levels of one of its own substrates, palmitoyl-CoA. Interestingly enough, both the mevalonic and the fatty acid/palmitoyl-CoA biosynthetic pathways are under the control of the SREBP family of transcription factors (6). Even though we know that the intramembrane proteolysis of SREBP does not depend on presenilins, I still wonder whether SREBPs play any role behind the curtains. Who knows? Maybe Tobias has another ace ready for us.

    See also:

    Slotte, J.P. et al. (1994). Flow and distribution of cholesterol-Effects of phospholipids. In Current Topics in Membranes. Cell Lipids (Hoekstra, D, ed.), pp. 483-502, Academic Press, San Diego, CA.

    References:

    . Unsaturated fatty acid-mediated decreases in sterol regulatory element-mediated gene transcription are linked to cellular sphingolipid metabolism. J Biol Chem. 2002 Feb 8;277(6):3878-85. PubMed.

    . Ceramide inhibits axonal growth and nerve growth factor uptake without compromising the viability of sympathetic neurons. J Biol Chem. 2001 Sep 28;276(39):36207-14. PubMed.

    . A TrkA-to-p75NTR molecular switch activates amyloid beta-peptide generation during aging. Biochem J. 2005 Oct 1;391(Pt 1):59-67. PubMed.

    . CNS synaptogenesis promoted by glia-derived cholesterol. Science. 2001 Nov 9;294(5545):1354-7. PubMed.

    . Regulation of SREBP processing and membrane lipid production by phospholipids in Drosophila. Science. 2002 May 3;296(5569):879-83. PubMed.

  2. We appreciate the interesting study by Hartmann and colleagues. A decade ago we reported that Aβ peptides modulate the cholesterol esterification rate (1). We later showed that Aβ modulates the metabolism of cholesterol and phospholipids (2-4). We studied Aβ's effects on lipid metabolism in a number of test systems, including hepatic cells (2), cultured nerve cells (3), fetal rat brain model (3), and ex vivo in rat hippocampal slices (4) and found that it is tissue and oxidation level-dependent. This is discussed in detail in our recent publication (5) that explored the effects of Aβ on synaptic plasticity and its interrelation with the neural cholesterol homeostasis modulation by Aβ.

    Our early study of Aβ's effect on cholesterol esterification was subsequently confirmed by others (6). In this regard, it is important to note that Aβ is a structure-functional component of lipoproteins (7,8,9). Aβ therefore, can affect the reverse cholesterol transport from neuronal tissue to the periphery in addition to its role in cholesterol synthesis and intracellular dynamics. This is supported by earlier studies by Michikawa et al. (10), Igbavboa et al. (11), and us (4), who reported the effects of Aβ on cellular cholesterol uptake and efflux.

    "My belief is that Aβ is involved in this interaction by modulating cellular/membrane cholesterol, so, both cholesterol and Aβ (and APP processing) affect each other," I noted three years ago during the ARF live discussion, "Cholesterol and Alzheimer's—Charging Fast but Still at a Distance from Solid Answers." I am glad Dr. Hartmann's skepticism and willingness to see more experiments "to prove this point" has now materialized in the excellent publication by Dr. Hartmann's group.

    See also:

    Igbavboa U, Avdulov NA, Chochina SV, Sun GY, Wood WG. Amyloid beta peptides and cholesterol dynamics. Neurosci Lett. S55, S25 (2000)

    References:

    . Alzheimer's peptides A beta 1-40 and A beta 1-28 inhibit the plasma cholesterol esterification rate. Biochem Mol Biol Int. 1996 Apr;38(4):747-52. PubMed.

    . Multiple inhibitory effects of Alzheimer's peptide Abeta1-40 on lipid biosynthesis in cultured human HepG2 cells. FEBS Lett. 1996 Oct 21;395(2-3):204-6. PubMed.

    . Alzheimer's Abeta1-40 peptide modulates lipid synthesis in neuronal cultures and intact rat fetal brain under normoxic and oxidative stress conditions. Neurochem Res. 2000 May;25(5):653-60. PubMed.

    . Essential role for cholesterol in synaptic plasticity and neuronal degeneration. FASEB J. 2001 Aug;15(10):1858-60. PubMed.

    . Amyloid beta protein restores hippocampal long term potentiation: a central role for cholesterol?. Neurobiology of Lipids. 2003 Sep;1(8):46-56.

    . Amyloid beta peptide alters intracellular vesicle trafficking and cholesterol homeostasis. Proc Natl Acad Sci U S A. 1998 Oct 27;95(22):13266-71. PubMed.

    . Biochemical characterization of Alzheimer's soluble amyloid beta protein in human cerebrospinal fluid: association with high density lipoproteins. Biochem Biophys Res Commun. 1996 Jun 25;223(3):592-7. PubMed.

    . The soluble form of Alzheimer's amyloid beta protein is complexed to high density lipoprotein 3 and very high density lipoprotein in normal human plasma. Biochem Biophys Res Commun. 1994 Dec 15;205(2):1164-71. PubMed.

    . The levels of soluble amyloid beta in different high density lipoprotein subfractions distinguish Alzheimer's and normal aging cerebrospinal fluid: implication for brain cholesterol pathology?. Neurosci Lett. 2001 Nov 16;314(3):115-8. PubMed.

    . A novel action of alzheimer's amyloid beta-protein (Abeta): oligomeric Abeta promotes lipid release. J Neurosci. 2001 Sep 15;21(18):7226-35. PubMed.

  3. A wealth of cellular and animal studies indicates that cholesterol regulates Aβ generation. Use of statins is currently being explored as a safe and available strategy that may help protect against Alzheimer disease. While awaiting the outcome of large clinical trials, mechanistic studies are revealing an unexpectedly complex picture of the lipid-Aβ connection. Cholesterol is no longer the only player; cholesteryl-esters, ceramide, sphingomyelin (SM), as well as isoprenoids are among the newest additions to the lipid list. Now, Tobias Hartmann and colleagues add a remarkable twist to the story. Not only do a variety of lipids regulate Aβ generation, but Aβ can also reach back and control cellular cholesterol and SM levels. This provocative conclusion is supported by solid in vitro and in vivo studies, which assign separate functions to Aβ40 (inhibition of HMG-CoA reductase, resulting in decreased cholesterol synthesis) and Aβ42 (activation of SMase, resulting in decreased SM levels). Separate functions of the two peptides are shown in a variety of systems, including in vitro activation of nSMase by Aβ42, but much less by Aβ40; down-regulation of high cellular de novo cholesterol synthesis in APP/APLP2-/- MEF cells by exposure to Aβ40, but not Aβ42; and increased cholesterol together with decreased SM in cells expressing PS1 containing FAD mutations, leading to elevated Aβ42/Aβ40 ratios. Given that cholesterol and SM are integral components of lipid rafts, it would be interesting to examine how lipid raft levels and function are separately regulated by the two peptides in cells expressing FAD mutant presenilins.

    This study is important not only for Alzheimer disease, but also for basic cholesterol biology, as Aβ may regulate either HMG-CoA reductase or the SREB pathway. Although Aβ42 appears to directly activate SMase in in vitro assays, the exact mechanism for Aβ40 remains to be elucidated. Exposure of intact cells to Aβ40 reduces the activity of HMG-CoA reductase, an enzyme with established ER localization. The intracellular localization of HMG-CoA reductase would suggest an indirect mechanism of action for exogenous Aβ40. However, extracellular Aβ40 could not normalize cholesterol synthesis in APP/APLP2-/- MEF cells, indicating that perhaps small amounts of intracellular Aβ40 or AICD may also regulate HMG-CoA reductase activity in wild-type cells. Interestingly, lack of Aγ-secretase function in PS1/2-/- MEF cells elevates cholesterol and SM levels quite strongly, while in APP/APLP2-/- MEF cells (which are derived from different mice), levels of both lipids increase more moderately. The implication is that the impact of the γ-secretase/APP/Aβ lipid regulatory system might be quite different in strength depending on which specific cells or tissues are analyzed. One can also ask the question whether, if one looks at other tissues, perhaps there are Aγ-secretase substrates in addition to APP and APLP2 which may regulate cellular cholesterol and SM levels. These and other questions raised by Tobias will further define the delicate network of the newly established reciprocal lipid-Aβ connection.

  4. Researchers have long speculated that the Aβ peptide might have a physiological function. Unfortunately, evidence of a normal role for Aβ in cellular processes has been notoriously difficult to obtain and has led to the prevailing notion that Aβ is merely a toxic byproduct of APP metabolism—nasty “junk,” if you will. Strong evidence for a physiological function of Aβ did not emerge until 2003, when work by Malinow and colleagues suggested that Aβ may act as a negative regulator of excitatory synaptic transmission (Kamenetz et al., 2003). Surprisingly little else has been published about this putative function of Aβ, for reasons that are unclear. Now, the paper by Hartmann and colleagues reports an exciting new role for Aβ in regulating both cholesterol and sphingomyelin biosynthesis, apparently via two complex feedback loops that center on γ-secretase. The evidence they present in favor of this complex feedback regulation is extensive and quite compelling. Adding a Baroque yet intriguing twist, they discovered that the C-terminus of Aβ determines which of the two lipid pathways is to be regulated. Aβ40 inhibits HMG CoA reductase and thus lowers cholesterol levels, while Aβ42 directly activates SMase and therefore lowers sphingomyelin levels. Moreover, Aβ42-raising FAD mutations in presenilin cause cholesterol levels to increase (because reduced Aβ40 levels relieve HMG CoA reductase inhibition) and sphingomyelin levels to fall (due to Aβ42-induced stimulation of SMase). In pathology, this feedback loop could lead to a vicious circle of ever-increasing Aβ42 and cholesterol levels, and could provide a plausible explanation for the observed relationships between cholesterol levels, Aβ generation, and AD. Thus, the results of Hartmann and colleagues suggest that the variable C-termini of Aβ are not just mistakes of an indiscriminate γ-secretase, but that the Aβ40/Aβ42 ratio may in fact be physiologically determined for the regulation of lipid homeostasis. This is a fascinating paper that has far-reaching implications for the entire field.

    References:

    . APP processing and synaptic function. Neuron. 2003 Mar 27;37(6):925-37. PubMed.

    View all comments by Robert Vassar

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. ACAT and Mouse—Inhibiting Former Prevents AD-like Pathology in Latter
  2. Ceramide Leads to Higher BACE Levels
  3. Statin Use and Alzheimer Disease: A Tale of Two Methodologies?

Paper Citations

  1. . GM1 ganglioside regulates the proteolysis of amyloid precursor protein. Mol Psychiatry. 2004 Oct;9(10):946-52. PubMed.
  2. . Cholesterol modulates alpha-secretase cleavage of amyloid precursor protein. J Biol Chem. 1996 Feb 23;271(8):4436-40. PubMed.
  3. . Secretory processing of amyloid precursor protein is inhibited by increase in cellular cholesterol content. Biochem J. 1997 Mar 15;322 ( Pt 3):893-8. PubMed.
  4. . Cholesterol depletion inhibits the generation of beta-amyloid in hippocampal neurons. Proc Natl Acad Sci U S A. 1998 May 26;95(11):6460-4. PubMed.
  5. . The role of cholesterol in the biosynthesis of beta-amyloid. Neuroreport. 1999 Jun 3;10(8):1699-705. PubMed.
  6. . Cholesterol-dependent generation of a seeding amyloid beta-protein in cell culture. J Biol Chem. 1999 May 21;274(21):15110-4. PubMed.
  7. . Hypercholesterolemia accelerates the Alzheimer's amyloid pathology in a transgenic mouse model. Neurobiol Dis. 2000 Aug;7(4):321-31. PubMed.
  8. . Simvastatin strongly reduces levels of Alzheimer's disease beta -amyloid peptides Abeta 42 and Abeta 40 in vitro and in vivo. Proc Natl Acad Sci U S A. 2001 May 8;98(10):5856-61. PubMed.
  9. . Distinct sites of intracellular production for Alzheimer's disease A beta40/42 amyloid peptides. Nat Med. 1997 Sep;3(9):1016-20. PubMed.
  10. . The transmembrane domain of the amyloid precursor protein in microsomal membranes is on both sides shorter than predicted. J Biol Chem. 2003 Feb 28;278(9):6803-8. PubMed.
  11. . Treatment with simvastatin in normocholesterolemic patients with Alzheimer's disease: A 26-week randomized, placebo-controlled, double-blind trial. Ann Neurol. 2002 Sep;52(3):346-50. PubMed.
  12. . Inhibition of intracellular cholesterol transport alters presenilin localization and amyloid precursor protein processing in neuronal cells. J Neurosci. 2002 Mar 1;22(5):1679-89. PubMed.
  13. . Acyl-coenzyme A: cholesterol acyltransferase modulates the generation of the amyloid beta-peptide. Nat Cell Biol. 2001 Oct;3(10):905-12. PubMed.
  14. . 3-Hydroxy-3-methylglutaryl-coenzyme A reductase inhibitors attenuate beta-amyloid-induced microglial inflammatory responses. J Neurosci. 2005 Jan 12;25(2):299-307. PubMed.
  15. . Regulation of hippocampal cholesterol metabolism by apoE and environmental stimulation. J Neurochem. 2005 Nov;95(4):987-97. PubMed.
  16. . Amyloid beta peptide alters intracellular vesicle trafficking and cholesterol homeostasis. Proc Natl Acad Sci U S A. 1998 Oct 27;95(22):13266-71. PubMed.
  17. . Amyloid beta-protein affects cholesterol metabolism in cultured neurons: implications for pivotal role of cholesterol in the amyloid cascade. J Neurosci Res. 2002 Nov 1;70(3):438-46. PubMed.
  18. . Fibrillar amyloid-beta peptides kill human primary neurons via NADPH oxidase-mediated activation of neutral sphingomyelinase. Implications for Alzheimer's disease. J Biol Chem. 2004 Dec 3;279(49):51451-9. PubMed.
  19. . Oxidative stress induced by beta-amyloid peptide(1-42) is involved in the altered composition of cellular membrane lipids and the decreased expression of nicotinic receptors in human SH-SY5Y neuroblastoma cells. Neurochem Int. 2005 Jun;46(8):613-21. PubMed.
  20. . Analysis of heterogeneous A4 peptides in human cerebrospinal fluid and blood by a newly developed sensitive Western blot assay. J Biol Chem. 1996 Sep 13;271(37):22908-14. PubMed.
  21. . Naturally secreted oligomers of amyloid beta protein potently inhibit hippocampal long-term potentiation in vivo. Nature. 2002 Apr 4;416(6880):535-9. PubMed.
  22. . Phosphodiesterase 4D deficiency in the ryanodine-receptor complex promotes heart failure and arrhythmias. Cell. 2005 Oct 7;123(1):25-35. PubMed.
  23. . Amyloid beta-protein is degraded by cellular angiotensin-converting enzyme (ACE) and elevated by an ACE inhibitor. J Biol Chem. 2005 Nov 11;280(45):37644-50. PubMed.
  24. . Homo- and heterodimerization of APP family members promotes intercellular adhesion. EMBO J. 2005 Oct 19;24(20):3624-34. PubMed.
  25. . A model for beta-amyloid aggregation and neurotoxicity based on free radical generation by the peptide: relevance to Alzheimer disease. Proc Natl Acad Sci U S A. 1994 Apr 12;91(8):3270-4. PubMed.
  26. . Cu(II) potentiation of alzheimer abeta neurotoxicity. Correlation with cell-free hydrogen peroxide production and metal reduction. J Biol Chem. 1999 Dec 24;274(52):37111-6. PubMed.

Other Citations

  1. ARF Live Discussion

Further Reading

Primary Papers

  1. . Obesity and vascular risk factors at midlife and the risk of dementia and Alzheimer disease. Arch Neurol. 2005 Oct;62(10):1556-60. PubMed.
  2. . Serum cholesterol and risk of Alzheimer disease: a community-based cohort study. Neurology. 2005 Oct 11;65(7):1045-50. PubMed.
  3. . Trends in serum lipids and lipoproteins of adults, 1960-2002. JAMA. 2005 Oct 12;294(14):1773-81. PubMed.
  4. . Regulation of cholesterol and sphingomyelin metabolism by amyloid-beta and presenilin. Nat Cell Biol. 2005 Nov;7(11):1118-23. PubMed.