This is Part 2 of a three-part series. See also Part 1 and Part 3.
2 July 2010. ApoE stands out for a frustrating mark of distinction. Despite its established status as the strongest genetic risk factor for late-onset Alzheimer disease, scientists digging for the why and the how have unearthed a myriad of potential interactions but no consensus mechanism. Indeed, 17 years after the gene discovery, figuring out how ApoE4 drives up a person’s susceptibility remains a tall order. At a one-day symposium called “ApoE, ApoE Receptors and Neurodegeneration,” held 7 June 2010 at Washington University, St. Louis, Missouri, researchers from across the U.S. and abroad took stock of the latest on potential mechanisms. Receptors, including ApoE receptor 2 (see Part 1 of this three-part series), topped the hit parade of ApoE accomplices that might mediate nefarious actions of the risky isoform, but participants paid due consideration to ApoE’s influence on lipid and Aβ metabolism as well (see also Part 3).
In the brain, astrocytes produce and release ApoE, which finds its way to receptors on the surface of neurons. Those receptors include the family of low-density lipoprotein receptors. It is not proven that these receptors dovetail with ApoE’s involvement in AD pathology, but one of them, the low-density lipoprotein-related protein 1 (LRP1), does bind Aβ and even APP among a variety of ligands. In his presentation, Guojun Bu of WashU, showed how his lab members used the Cre/Lox system, under the Cam kinase promoter, to conditionally delete LRP1 from the forebrain of adult mice. Bu reported that getting rid of LRP1 boosts brain ApoE levels; it also suppresses the synaptic markers synaptophysin and PSD-95, as well as the number of dendritic spines in the forebrain. On top of this, the animals exhibit glial activation, neural inflammation, and deficits in their memory and motor function. These data come on the heels of similar in vitro results, when Bu knocked LRP1 down with a lentiviral approach. In these cultures, neurites degenerated and neurons died off (see, e.g., Fuentealba et al., 2009).
How does a loss of LRP lead to the effects, which, after all, model aspects of AD? Bu said the most likely mechanism works through perturbation of brain lipid metabolism. The Bu team found that LRP1 deletion in neurons correlates with a dearth of brain cholesterol, sulfatides, and cerebrosides, among other lipids. After these early events, dendritic spines degenerate and synapses disappear as animals age, and eventually neurons degenerate as these animals grow very old. Bu also noted that LRP1 knockouts lose NMDAR1 and GluR1 glutamate receptor subunits. That could be related to the synapse loss, or it could be an independent effect, he said. Is this relevant to AD? On this, Bu noted previous work from Eddie Koo’s lab that suggested LRP levels in the AD brain plummet by about 50 percent (see Kang et al., 2000) and said he is trying to confirm that data. But in addition, LRP1 levels are lowest in the brains of mice that have targeted replacement (TR) of their endogenous ApoE gene with that for human ApoE4, indicating that LRP1 not only regulates ApoE levels, but vice versa.
The finding suggests that ApoE4 might increase susceptibility to AD by suppressing LRP1. In keeping with this, Bu noted findings from two other meeting presenters. Joachim Herz, University of Texas Southwestern Medical Center, Dallas, reported that ApoE4 can ablate ApoE receptor 2 (ApoER2) from the cell surface by sequestering it intracellularly, while Hyang-Sook Hoe and colleagues at Georgetown University, Washington, DC, found reduced spine density in ApoE4 TR mice (see Part 1 of this series). Whether LRP1 levels fall and synapses fade away in ApoE4 mice because the apolipoprotein sequesters LRP1 in the same way as it does ApoER2 is not clear.
If reduction of LRP1 might explain, at least partly, why ApoE4 increases risk of AD, then would targeting the receptor help? Bill Rebeck of Georgetown University, Washington, DC, showed that it might, though not by improving synaptic biology. Rebeck showed that an ApoE fragment comprising the region of the protein that binds apolipoprotein receptors (amino acid 141-149) protects in vitro against glial activation/inflammation, which also occurs in the AD brain.
Lipopolysaccharide (LPS) is commonly used to induce an inflammatory glial response, and it elevates nitric oxide (NO) production and suppresses ApoE. In St. Louis, Rebeck reported that adding the ApoE141-149 peptide counteracted both these effects as well as activation of Jun-terminal kinase (JNK), one of two glial kinases LPS activates (the other being ERK). The peptide had no effect in LRP1-negative microglia. Rebeck concluded that LPS works by activating JNK in an LRP1-dependent manner. If that is true, then blocking JNK activation should relieve LPS-induced ApoE suppression. In fact, Rebeck showed that the JNK inhibitor SP600125 increases ApoE levels when injected into mouse brain. He suggested that blocking JNK could be a strategy for raising ApoE levels in people with AD. In the case of ApoE2/3 carriers, this might be beneficial, since those two isoforms enhance synaptic plasticity, which is essential for learning and memory (see Part 1).
The ApoE141-149 peptide may have other uses as well. Rebeck showed that activation of another receptor, ApoE receptor 2, puts the brakes on APP processing, as measured with a luciferase reporter system that quantifies production of the APP intracellular domain (AICD). Dimers and trimers of the 141-149 peptide worked similarly, suppressing AICD, trapping APP on the cell surface, and holding down Aβ production in PS70 cells as well as in primary neuronal cultures. Rebeck reported that a single injection of the peptide dimer into the hippocampus of wild-type mice increases α-secretase cleavage of APP and reduces Aβ levels. Rebeck said the findings, some recently reported (see Minami et al., 2010), could spur the development of small molecules that mimic the effect of the peptide and protect the brain against amyloidogenic processing of APP.
Receptor activation may be but one side of the ApoE coin. In the extracellular space, the apolipoprotein can have other effects, such as on Aβ oligomerization and clearance (see Part 3 of this series) or on cholesterol metabolism. At the conference, Cheryl Wellington, University of British Columbia, Vancouver, addressed the latter. She focuses on the cholesterol transporter protein ABCA1 as a potential therapeutic target. ABCA1 transfers cholesterol from the cell surface to ApoE, forming large lipoprotein particles. Wellington and colleagues showed that ABCA1 deficiency leads to a marked loss of ApoE and an increase in amyloid in mouse brain (see ARF related news story on Hirsch-Reinshagen et al., 2005). In contrast, transgenic mice overexpressing ABCA1 build up less amyloid (see ARF related news story). The data point to decreased clearance of Aβ when ApoE is poorly lipidated, said Wellington, and she suggested this could be used to develop a therapeutic.
One class of therapeutics that induces ABCA1 expression is the liver X receptor (LXR) agonists. Wellington shared unpublished data on the effects of the LXR agonist GW3965 on APP/PS1 transgenic mice. Her group tried the agonist both prophylactically, before the animals develop plaques, and as a treatment for mice with advanced AD-like pathology. At the lower of two doses, the agonist slightly raised ABCA1 levels in the brain, and at the higher dose, ApoE levels climbed as well. But the increase in ABCA1 did not have an effect on Aβ deposition, which only trended lower in animals treated with either dose of GW3965. Cognitive test results were more encouraging. Mice given the agonist prophylactically performed like wild-type mice in a novel object recognition task, in which untreated APP/PS1 controls performed poorly. GW3965 also restored memory when given to older mice as a treatment. GW3965 was developed by GlaxoSmithKline (see Collins et al., 2002) and is sold as a research tool by Tocris Bioscience under license.
Wellington pointed to several positive signs from this study. The effects of the agonist depended on ABCA1, that is, APP/PS1/ABCA1-negative mice did not respond. This suggests that the agonist acts through ABCA1 and not one of the many other proteins that are affected by LXR activity. This has long been a point of uncertainty with this approach. Secondly, the improvement in cognition in the older mice may bode well for using a similar strategy in people who already have AD symptoms. And lastly, that the mice appeared to become sharper without a dramatic change in Aβ accumulation suggests that ABCA1 affects a particular, crucial pool of Aβ, said Wellington. Whether these might be soluble oligomers that are now believed to be the most toxic Aβ species remains to be seen.
The importance of ApoE’s role in lipid metabolism, as it relates to AD risk, was reinforced by Patrick Sullivan, Duke University, Durham, North Carolina, who first developed the ApoE targeted replacement (TR) mice. Sullivan initially stressed that, although ApoE is a risk factor for AD, in some parts of the globe where ApoE4 genotypes are quite common, Alzheimer disease is not. Sullivan argued that environmental factors deserve more attention in AD, noting that at most there is a 60 percent concordance rate for the disease in monozygotic twins. “The environment obviously does matter,” he said.
One environmental factor that has drawn scrutiny is the Westernized diet. It could affect the lipidation status of ApoE and other lipoproteins, and in this way provide one explanation for ApoE’s link to the disease. Sullivan has looked for changes in lipid metabolism in ApoE TR mice. He found no difference in hippocampal cholesterol levels between ApoE3 and ApoE4 TR animals, but did see a difference in the fatty acid moieties attached to the glycerol backbone of phosphatidylethanolamines (PEs). In ApoE4 animals, PEs are less endowed with the long-chain fatty acids eicosapentaenoic acid and docosohexanoic acid (DHA). This may be particularly important to neurons, because the highest concentration of DHA in the entire body occurs at synapses, said Sullivan. DHA was tested in a small clinical trial for AD that showed some encouraging results (see ARF related news story), but a larger trial run by the ADCS proved negative (see ARF related news story).
Sullivan is interested in creating animal models of late-onset Alzheimer disease where the only genetic manipulation is the ApoE gene and where adding non-genetic risk factors would induce AD pathology. “We could then use metabolomics, proteomics, and lipidomics to tease out what pathways might be altered,” he said.
Perhaps the most intensely debated talk at the conference was one of the “hot topic” posters chosen for oral presentation. Katie Youmans from Mary Jo LaDu’s laboratory at the University of Illinois has coupled ApoE targeted replacement with the 5xFAD model developed by Bob Vassar’s lab. The idea was to get a mouse that features rapid acceleration of AD pathology. In these mice, deposition began in the subiculum/deep cortex, unlike in many other mouse models where it first begins in the hippocampus and superficial layers of the cortex. In this regard, the ApoE TR/5xFAD mice better recapitulate the pattern seen in the human brain, said LaDu.
Unexpectedly, Aβ deposits were actually more extensive in ApoE2 TR/5xFAD mice than in their ApoE4/5xFAD counterparts. Youmans told ARF that, though the deposits have the characteristic appearance of plaques (dense cores surrounded by a halo, e.g.), she has yet to fully characterize them. Some researchers, including Sullivan, were skeptical of these data, which seem to contradict most other findings that ApoE2 is protective. But when Youmans looked further, she found that in the ApoE2 crosses, most of the Aβ was extracellular, whereas in the ApoE4 crosses it was intraneuronal. In addition, by four months, the ApoE4 crosses had lost the synaptic marker synaptophysin, but the ApoE2 crosses actually had more of it, despite their greater abundance of extracellular and insoluble Aβ. “We basically figured out why ApoE2 is protective, and we debunked the amyloid hypothesis at the same time,” said LaDu. She clarified, saying that her data suggest amyloid deposits are not the driving force behind AD. Instead, soluble intraneuronal species of Aβ are, and ApoE4 drives their accumulation. This is in keeping with recent advances in the field that point to soluble Aβ as the most toxic species. Whether LaDu’s preliminary data will hold up remains to be seen, but she stressed that they had already repeated the analysis three times and come up with the same findings.—Tom Fagan.
This is Part 2 of a three-part series. See also Part 1 and Part 3.