Kim et al. have given us an important insight into the relationship between plasma ApoE levels and Alzheimer’s disease pathogenesis that, as the authors point out, will drive therapeutic strategies in a 180-degree different direction than currently envisaged. Rather than raising the level of a presumptively beneficial ApoE isoform, any ApoE, whether E3 or E4, should be decreased. Missing still from this discussion is the E2 isoform, as well as behavioral studies of the mice. Nonetheless, the dramatic effects on amyloid seem to raise some exciting therapeutic possibilities.

Therapeutic strategies that have focused on disrupting the amyloid cascade in symptomatic patients to date have had less than promising outcomes. Regarding neuropathological stages, one could argue that the “amyloid stage” has nearly peaked by the time a patient presents with MCI, so that, as Bray Hyman has pointed out, intervention following this stage is attacking a target that may have been relevant for disease initiation but may no longer be as relevant for disease progression (see   Read more

Kim et al. have given us an important insight into the relationship between plasma ApoE levels and Alzheimer’s disease pathogenesis that, as the authors point out, will drive therapeutic strategies in a 180-degree different direction than currently envisaged. Rather than raising the level of a presumptively beneficial ApoE isoform, any ApoE, whether E3 or E4, should be decreased. Missing still from this discussion is the E2 isoform, as well as behavioral studies of the mice. Nonetheless, the dramatic effects on amyloid seem to raise some exciting therapeutic possibilities.

Therapeutic strategies that have focused on disrupting the amyloid cascade in symptomatic patients to date have had less than promising outcomes. Regarding neuropathological stages, one could argue that the “amyloid stage” has nearly peaked by the time a patient presents with MCI, so that, as Bray Hyman has pointed out, intervention following this stage is attacking a target that may have been relevant for disease initiation but may no longer be as relevant for disease progression (see Hyman, 2011). Targeting patients early in the amyloid cascade process implies identifying patients earlier than the expression of MCI, i.e., preclinically. During this clinically silent period, there is neuropathological, neuroimaging, and even neuropsychological evidence of disease progression. Therefore, in genetically susceptible individuals with biomarker evidence of preclinical (amyloid stage) AD, interventions to lower plasma ApoE levels might provide the right treatment, for the right target, at the right time.

View all comments by Richard Caselli

The new reports by Kim et al. are very intriguing, as they provide evidence that contradicts over a decade of research indicating that ApoE is required for Aβ deposition. Given the novelty of these results, a primary concern is that the authors establish their controversial findings using a transgenic mouse model (APPSwe/PS1-L166P) with relatively little provenance in the field (it was created in 2006), without first determining or discussing the effects of this PS1 mutation on endogenous murine ApoE or the effects of the complete loss of ApoE. Regardless, the authors show diminished plaque deposition as ApoE levels decrease. The authors interpret this data as suggesting that lowering the levels of ApoE may attenuate AD progression via decreasing the level of plaques. This interpretation is interesting, given that total plaque load has recently been shown to correlate poorly with AD-associated cognitive and behavioral deficits. In addition, recent work from our lab indicates that ApoE2 may also increase plaque deposition, consistent with clinical findings by Kawas' group that...  Read more

The new reports by Kim et al. are very intriguing, as they provide evidence that contradicts over a decade of research indicating that ApoE is required for Aβ deposition. Given the novelty of these results, a primary concern is that the authors establish their controversial findings using a transgenic mouse model (APPSwe/PS1-L166P) with relatively little provenance in the field (it was created in 2006), without first determining or discussing the effects of this PS1 mutation on endogenous murine ApoE or the effects of the complete loss of ApoE. Regardless, the authors show diminished plaque deposition as ApoE levels decrease. The authors interpret this data as suggesting that lowering the levels of ApoE may attenuate AD progression via decreasing the level of plaques. This interpretation is interesting, given that total plaque load has recently been shown to correlate poorly with AD-associated cognitive and behavioral deficits. In addition, recent work from our lab indicates that ApoE2 may also increase plaque deposition, consistent with clinical findings by Kawas' group that ApoE2 increases diffuse plaques in the oldest old. Collectively, our work suggests a role for plaque morphology rather than overt deposition. Therefore, without direct comparisons to ApoE2, it may be difficult for the authors of this study to speculate fully about the true effects of ApoE modulation in AD progression.

Rather, soluble dimers and/or oligomers are the more neurotoxic Aβ species, and ApoE may have distinct functional interactions with each of these pools. Therefore, it would have been interesting if the authors had addressed not only the levels, but the solubility of ApoE and Aβ. The selective lower levels of ApoE4 may be explained by an increase in GuHCl-soluble ApoE4 due to its association with amyloid deposits, which may have led to confounding interpretations in this study.

Finally, the authors cite this data as supporting evidence that removal of ApoE protein may be a viable therapeutic approach for AD. However, mRNA levels are not different between ApoE3 and ApoE4, consistent with significant post-translational modifications that may increase or decrease ApoE stability in vivo. Modification of ApoE at the gene level is one approach, but should be undertaken with extreme caution. It would likely result in ubiquitous, undesirable, unexpected effects, due to the innumerable roles ApoE plays in different biological systems. Thus, while interesting, these data require a large amount of follow-up research, and we should interpret these findings with discretion.

View all comments by Katherine Youmans

This is an interesting paper, showing that the concentration of human ApoE, irrespective of the isoforms (ApoE3 or ApoE4), plays a role in determining the extent of brain Aβ accumulation in a mouse model of AD. The authors have tested the effect of human ApoE gene dose on amyloid pathology in APPPS1-21 mice expressing two copies of ApoE (ApoE3/3 or ApoE4/4) or one copy of ApoE (ApoE3/- or ApoE4/-). Compared to their respective homozygous mice, the ApoE3/- and ApoE4/- mice had decreased Aβ pathology and less microgliosis. These interesting data not only reinforce the ApoE isoform-specific effect on Aβ pathology (E4>E3>E2) and on the clearance of Aβ from brain (E2>E3>E4) as reported by this and other groups, but also extend our knowledge on the role of ApoE gene dose in AD. While in ApoE3/3 mice, brain ApoE levels were about twofold greater than in ApoE4/4 mice, brain Aβ levels (soluble and insoluble) were about twofold greater in the ApoE4/4 mice, confirming the role of ApoE4 in promoting accumulation of brain Aβ. In contrast, in the mice expressing one copy of these ApoE isoforms...  Read more

This is an interesting paper, showing that the concentration of human ApoE, irrespective of the isoforms (ApoE3 or ApoE4), plays a role in determining the extent of brain Aβ accumulation in a mouse model of AD. The authors have tested the effect of human ApoE gene dose on amyloid pathology in APPPS1-21 mice expressing two copies of ApoE (ApoE3/3 or ApoE4/4) or one copy of ApoE (ApoE3/- or ApoE4/-). Compared to their respective homozygous mice, the ApoE3/- and ApoE4/- mice had decreased Aβ pathology and less microgliosis. These interesting data not only reinforce the ApoE isoform-specific effect on Aβ pathology (E4>E3>E2) and on the clearance of Aβ from brain (E2>E3>E4) as reported by this and other groups, but also extend our knowledge on the role of ApoE gene dose in AD. While in ApoE3/3 mice, brain ApoE levels were about twofold greater than in ApoE4/4 mice, brain Aβ levels (soluble and insoluble) were about twofold greater in the ApoE4/4 mice, confirming the role of ApoE4 in promoting accumulation of brain Aβ. In contrast, in the mice expressing one copy of these ApoE isoforms (ApoE3/- or ApoE4/-), brain Aβ levels were similar in both groups of mice, while Aβ plaque load and fibrillar plaque load were greater in ApoE4/- mice compared to ApoE3/- mice. There was a clear ApoE dose-dependent effect on CD45 load. These data suggest that reducing human brain ApoE levels will reduce Aβ pathology and neuroinflammation in adult brains. However, further work is needed to confirm whether altering brain ApoE levels in adult mice and in humans affects Aβ levels and AD pathology.

View all comments by Rashid Deane

Usually, I’m very skeptical and critical about studies in transgenic mice, which many researchers call AD models. This study by Dr. Holtzman and fellow researchers is, however, an exception. Not because the conclusions of this study essentially confirm observations in my own recent studies in patients with sporadic AD, but because this study is one of a handful that use transgenic animals appropriately. Here, they use mice to reinforce or reject observations made in humans rather than use human studies to test observations made of animals, as is most often the case.

In general, there are two main views on the role of apolipoprotein E (ApoE). In simple terms, the first argues that expression of ApoE is reduced in the brain of ApoE4 carriers, and this is the reason this allele is an AD risk factor. This is based on ApoE protein measurement in postmortem brains and several CSF and plasma studies—one reported very recently—and subsequent in-vitro and transgenic studies (Gupta et al., 2011). Frankly, the numbers of publications that seem to support this view are strikingly in the...  Read more

Usually, I’m very skeptical and critical about studies in transgenic mice, which many researchers call AD models. This study by Dr. Holtzman and fellow researchers is, however, an exception. Not because the conclusions of this study essentially confirm observations in my own recent studies in patients with sporadic AD, but because this study is one of a handful that use transgenic animals appropriately. Here, they use mice to reinforce or reject observations made in humans rather than use human studies to test observations made of animals, as is most often the case.

In general, there are two main views on the role of apolipoprotein E (ApoE). In simple terms, the first argues that expression of ApoE is reduced in the brain of ApoE4 carriers, and this is the reason this allele is an AD risk factor. This is based on ApoE protein measurement in postmortem brains and several CSF and plasma studies—one reported very recently—and subsequent in-vitro and transgenic studies (Gupta et al., 2011). Frankly, the numbers of publications that seem to support this view are strikingly in the majority. Some researchers profess using “ApoE3 mimicking peptide” as a treatment strategy to compensate for low ApoE protein in the brain of AD.

The second view suggests that high expression/accumulation of ApoE protein, particularly in ApoE4 AD carriers, confers the risk of AD associated with this allele. To my limited knowledge, a promoter study (Law et al., 2002) and two main papers (and several posters/oral presentations at different scientific meetings) by our group sum up the overall support for this second view. I would like to mention here that at least two additional studies supporting the second view are underway, both of which are based on human subjects.

By confirming the main findings of our studies (Lane et al., 2008; Darreh-Shori et al., 2009; Darreh-Shori et al., 2009; Darreh-Shori et al., 2010; Darreh-Shori et al., 2011; Darreh-Shori et al., 2011) and the promoter study by Law, Holtzman, and colleagues address, in my opinion, the most important issue in AD, and by so doing, contribute substantially to reversing the mainstream view about ApoE4’s roles in AD. This study shows that a high level of ApoE protein is the major driving force for Aβ deposition in the AD brain.

Given the impact which the two completely opposite views on ApoE might have on choosing new preventive and or therapeutic intervention strategies, it is imperative to know which is accurate. We cannot afford anymore disappointing negative (and occasionally harmful) clinical trials in AD.

ApoE4 was recognized quite long ago as the major genetic risk factor for sporadic AD, which comprises over 90 percent of all AD cases. Later, the majority (if not all) genetic studies in AD, regardless of their population size, repeatedly churned out ApoE as the main genetic player in the etiology of AD. Still, judging from the number of publications and candidate drug trials, the strategy of reducing Aβ production dominates the field, although there is really no good evidence for increased Aβ production in the brains of sporadic AD patients. This is the main reason for my negative view about transgenic animal “models of AD.” They are, as Dr. Holtzman and colleagues very appropriately point out, simply a model of Aβ amyloidosis (at best they are a model of a specific familiar form of AD, e.g., APPswe for the Swedish genetic form of AD).

The next most important research objective, in my opinion, is to deduce the native biological activity/role of Aβ peptides in the human brain. Truly, it does not make sense that nature would spend millions of years of evolution to bring about the universal expression of Aβ precursor protein (APP) in combination with such sophisticated cleavage machinery just to prevent Homo sapiens from reaching advanced age with intact cognitive ability.

In this context, studies by the Holtzman group (Cirrito et al., 2005) and ours (Darreh-Shori et al., 2009; Darreh-Shori et al., 2009; Darreh-Shori et al., 2010; Darreh-Shori et al., 2011; Darreh-Shori et al., 2011) portray a complex interplay between cholinesterases and Aβ peptides in which high ApoE protein level seems to have a major pathological influence. Our findings suggest the presence of a molecular complex, which we have termed BAβACs (Butyrylcholineesterase/Acetylcholineesterase-Aβ-ApoE Complexes).

Returning to the paper of Dr. Holtzman and colleagues, there is, however, a peculiar observation that does not make sense, and, as a matter of fact, directly counter-argues the main conclusion of the study. The ApoE3/3 mice show twofold higher ApoE protein levels in the brain than ApoE4/4 mice. In contrast, they show about two- to threefold lower insoluble Aβ40 and 42 in the brain than ApoE4 mice. Thus, within this same study, this contradictory observation puts the main finding in question. The impact of this contradictory observation is far too crucial to let it go, despite the explanatory remarks by the authors.

Since ApoE4 rather than ApoE3 is the AD risk factor, this observation may reflect poorer response of mouse ApoE receptors to human ApoE4 than ApoE3 (mouse ApoE is structurally like human ApoE3), or it may indicate a flaw in the ApoE protein analyses.

A clue actually is present in Fig.1A, which shows that both mice have essentially identical ApoE3 and 4 mRNA expressions. With the reservation that mRNA level and protein expression do not always go hand in hand, it hints at a plausible confounding source in the ApoE protein analysis.

For instance, measuring “levels of PBS soluble ApoE” (see the legend of Fig. 1) by no means will reflect the amounts of ApoE protein in the brain, since these are by definition very lipophilic proteins. Measurement of ApoE protein in plasma may, in this case, have been a better measure of the relative differences among the genotype groups, particularly since ApoE4, due to its closed-arm tertiary conformation, may be more lipophilic than ApoE3 isoprotein. This may additionally lead to a different extraction efficiency of ApoE3 compared to ApoE4 from the brain tissue homogenates, which consist mainly of fat/lipids, as well as differential loss to the surface of tubes and pipettes, etc., during homogenization, which can be very substantial (Hesse et al., 2000).

Another unnecessary caveat in this regard is the use of iso-specific ApoE ELISA (mentioned in Fig.1). This will provide no important information, since the genotype of the animals is already known. It may, however, have potentially introduced systemic errors in comparative ApoE level analyses. Distinct antibody pairs may produce dissimilar data due to their relative affinity to the standards (which could be peptide, recombinant, or purified ApoE protein?) versus ApoE in the samples. These are important because ApoE3 is more likely to produce dimers than ApoE4, which means that one capturing antibody may retain, simultaneously, two binding epitopes for the detecting of antibody in ApoE3 ELISA, thereby giving a higher signal per available binding sites. All of these issues may be a confounding source for the striking discrepancy between the clear impact of ApoE hemizygosity on Aβ deposition, and the contradictory result of twofold higher ApoE protein but still having about two- to threefold fewer insoluble Aβ peptides in ApoE3 than ApoE4 mice. Hopefully, the authors will clarify these confounding issues in further work.

References:

Cirrito JR, Yamada KA, Finn MB, Sloviter RS, Bales KR, May PC, Schoepp DD, Paul SM, Mennerick S, Holtzman DM. 2005. Synaptic activity regulates interstitial fluid amyloid-β levels in vivo. Neuron 48:913-922. Abstract

Darreh-Shori T, Forsberg A, Modiri N, Andreasen N, Blennow K, Kamil C, Ahmed H, Almkvist O, Langstrom B, Nordberg A. 2010. Differential levels of apolipoprotein E and butyrylcholinesterase show strong association with pathological signs of Alzheimer's disease in the brain in vivo.Neurobiol Aging. 2011 Dec;32(12):2320.e15-32. Abstract

Darreh-Shori T, Modiri N, Blennow K, Baza S, Kamil C, Ahmed H, Andreasen N, Nordberg A. 2011. The apolipoprotein E epsilon4 allele plays pathological roles in AD through high protein expression and interaction with butyrylcholinesterase. Neurobiol Aging 32:1236-1248. Abstract

Darreh-Shori T, Modiri N, Lahaut P, Nordberg A. 2009. Apolipoprotein E and Butyrylcholinesterase synergistically promote Ab peptides oligomerization. Alzheimer's & Dementia: The Journal of the Alzheimer's Association 5:P225.

Darreh-Shori T, Modiri N, Nordberg A. 2009. ApoE and amyloid β deflate the cholinergic neurotransmission by boosting the activity and stability of cholinesterases in the brain. Alzheimer's & Dementia: The Journal of the Alzheimer's Association 5:P305.

Darreh-Shori T, Siawesh M, Mousavi M, Andreasen N, Nordberg A. 2011. Apolipoprotein epsilon4 Modulates Phenotype of Butyrylcholinesterase in CSF of Patients with Alzheimer's Disease. J Alzheimers Dis. Abstract

Hesse C, Larsson H, Fredman P, Minthon L, Andreasen N, Davidsson P, Blennow K. 2000. Measurement of apolipoprotein E (ApoE) in cerebrospinal fluid. Neurochem Res 25:511-517. Abstract

Lane R, Feldman HH, Meyer J, He Y, Ferris SH, Nordberg A, Darreh-Shori T, Soininen H, Pirttila T, Farlow MR, Sfikas N, Ballard C, Greig NH. 2008. Synergistic effect of apolipoprotein E epsilon4 and butyrylcholinesterase K-variant on progression from mild cognitive impairment to Alzheimer's disease. Pharmacogenet Genomics 18:289-298. Abstract

View all comments by Taher Darreh-Shori

Carriers of the detrimental ApoE4 genotype have long been hypothesized to have slower Aβ catabolism, leading to earlier onset and more amyloid burden than non-ApoE4 carriers. Critical questions involve defining the mechanisms by which ApoE4 leads to sluggish Aβ turnover so that therapeutic methods to correct this can be developed. Brad Hyman and colleagues add to the evidence that ApoE is a critical modulator of Aβ metabolism.

Using AD brains of various ApoE genotypes that were carefully matched for amyloid burden, this group characterized the soluble fraction and observed significantly more higher-molecular-weight Aβ by direct visualization on SDS gels and after separation by size exclusion chromatography (SEC). Therefore, soluble Aβ is independent of amyloid burden, and the proportion of soluble oligomeric Aβ species is higher in ApoE4 carriers. Remarkably, ApoE co-eluted with Aβ during SEC and immunoprecipitated with Aβ, and it is possible that the ApoE-Aβ association makes specific Aβ epitopes under native conditions.

In a key advance, Hyman and colleagues...  Read more

Carriers of the detrimental ApoE4 genotype have long been hypothesized to have slower Aβ catabolism, leading to earlier onset and more amyloid burden than non-ApoE4 carriers. Critical questions involve defining the mechanisms by which ApoE4 leads to sluggish Aβ turnover so that therapeutic methods to correct this can be developed. Brad Hyman and colleagues add to the evidence that ApoE is a critical modulator of Aβ metabolism.

Using AD brains of various ApoE genotypes that were carefully matched for amyloid burden, this group characterized the soluble fraction and observed significantly more higher-molecular-weight Aβ by direct visualization on SDS gels and after separation by size exclusion chromatography (SEC). Therefore, soluble Aβ is independent of amyloid burden, and the proportion of soluble oligomeric Aβ species is higher in ApoE4 carriers. Remarkably, ApoE co-eluted with Aβ during SEC and immunoprecipitated with Aβ, and it is possible that the ApoE-Aβ association makes specific Aβ epitopes under native conditions.

In a key advance, Hyman and colleagues demonstrate that physiologically lipidated ApoE4 increases Aβ oligomerization in vitro. This appears to be a specific structural property of extracellular ApoE4. They first purified secreted ApoE particles from immortalized murine astrocytes expressing human ApoE2, ApoE3, or ApoE4, mixed these with synthetic Aβ, and observed that oligomer levels were significantly greater in the presence of ApoE4, which seems to stabilize the oligomers compared with ApoE2 and ApoE3. That Aβ oligomerization follows an ApoE4->ApoE3->ApoE2 relationship was confirmed using a split-luciferase assay in which Aβ oligomerization is required to observe luminescence. Furthermore, Aβ purified from the soluble fraction of AD patient brains forms high-molecular-weight Aβ species when incubated with ApoE. Importantly, ApoA-II, which is another apolipoprotein found on brain HDL particles, did not increase Aβ oligomerization using the split-luciferase assay. No effects on Aβ oligomerization were observed in cell lysates, suggesting that ApoE specifically affects Aβ in the extracellular milieu. Finally, reducing a salt bridge in ApoE4 by converting the arginine at amino acid 61 to a threonine reduced the level of Aβ oligomerization back to the level of ApoE3. This finding provides further support that the structural differences caused by arginine 61 specifically in human ApoE4 may be at the root of ApoE4’s dysfunction.

Further experiments explored the selectivity of the effect on Aβ oligomerization by other lipoproteins. Both ApoA-I and ApoJ led to significant increases in Aβ oligomers in the split-luciferase assay, albeit far more modestly than ApoE4. Hyman’s group then demonstrated that the lipid binding C-terminal domain of ApoE is necessary and sufficient for Aβ oligomerization.

Finally, in a series of elegant experiments, the Hyman group demonstrated that human ApoE purified from brain tissue affects Aβ oligomerization in an isoform-dependent manner, with ApoE4->ApoE3. Immunodepletion of ApoE from these samples significantly reduced Aβ oligomerization.

What is critically important about these experiments is that the researchers used sources of ApoE that are physiologically lipidated, unlike innumerable previous studies that used either unlipidated ApoE or ApoE reconstituted with non-physiological lipids. However, the ApoE-containing HDL particles used in these experiments can be considered “baseline” with respect to their lipid content. This is an important point, as modifying the lipidation status of ApoE has been under considerable study as a therapeutic approach for AD. For example, agonists of Liver-X-Receptor (LXR) and Retinoid-X-receptor (RXR) nuclear receptors, increase the expression of genes such as ABCA1 and ABCG1 that physiologically add lipids to apolipoprotein acceptors including ApoE. These agonists consistently improve cognitive behavior in APP-expressing mice. The studies vary in terms of the efficacy of these agonists to reduce amyloid or Aβ levels, or to shift Aβ from the insoluble to the soluble fraction. Furthermore, ABCA1, which is the rate-limiting step in apolipoprotein lipidation, is a key player in Aβ metabolism in vivo, as deletion of ABCA1 slows Aβ turnover, whereas selective overexpression of ABCA1 in the brain nearly eliminates amyloid deposits. What is not yet known is the relationship between ApoE lipidation and ApoE structure, particularly with respect to the argining 61 salt bridge. Whether LXR/RXR agonists or genetic manipulation of ABCA1 activity affect ApoE structure, thereby affecting Aβ oligomer formation, stability, or degradation, is not fully understood.

An important caveat to the studies using LXR agonists and genetically modified ABCA1 murine lines is that they have thus far been conducted on murine ApoE. Whether these pathways rescue the deficits in human ApoE4 with respect to Aβ metabolism, or whether they enhance the detrimental effects of ApoE4, is now a crucial question to address.

View all comments by Cheryl Wellington