GABA Neurons Blamed for Memory Loss in ApoE Mice

Considering the dramatic impact of its E4 allelic variant on Alzheimer’s disease risk, apolipoprotein E arguably has drawn scant attention from AD drug developers, but recent research may help turn the tide. At a workshop organized by Yadong Huang of the Gladstone Institute of Neurological Disease in San Francisco, and held 30 January to 1 February, several researchers presented on this topic. Kelly Bales of Pfizer Global Research and Development, Groton, Connecticut, told attendees about the company’s efforts to tweak brain levels in AD mice with small molecules. Nga Bien-Ly, a postdoc in Huang’s lab, reported finding less Aβ buildup in AD mice with halved levels—whether the reduction was present since birth or induced during adulthood. And in a widely noted Science paper published online on February 9, researchers report that AD mouse models clear amyloid-β and regain cognitive function within hours when given an approved cancer drug that enhances endogenous transcription (see ARF related news story on Cramer et al., 2012). The overall picture presents a riddle. Pharmacological data suggest that raising compounds benefits AD mice, while genetic studies indicate that less is better. Even so, scientists are getting a better handle on ApoE’s physiological role in the brain, and potential therapeutic strategies for changing brain ApoE levels are emerging.

Based on the promising mouse data reported in Science, the cancer drug bexarotene is headed for a Phase 1b biomarker trial in healthy adults. Bexarotene promotes ApoE transcription by activating retinoid X receptor (RXR)—a protein that hooks up with peroxisome proliferator-activated receptor γ (PPARγ) or liver X receptors (LXRs) to form heterodimeric transcription factors.

At the Gladstone workshop, Bales described her team’s characterization of a histone deacetylase (HDAC) inhibitor that was among the top hits from a screen for small molecules that boost brain ApoE. By blocking the enzymes that compress DNA into a less accessible state, the compound (MS275) stimulates transcription of ApoE, causing three- to fourfold increased protein levels at an effective concentration of ~200 nM, Bales said. When they gave the compound orally to knock-in mice expressing human ApoE3 at the endogenous mouse locus, the researchers saw higher ApoE levels in plasma and hippocampus, and a trend for increased ApoE in CSF, Bales reported. Preclinical development with ApoE-raising compounds is ongoing, Bales said.

Meanwhile, new work out of the Huang lab adds to a emerging body of evidence suggesting that AD mice do better with less ApoE, not more. In a recent report by David Holtzman and colleagues at Washington University, St. Louis, Missouri, APP/PS1(L166P) mice with just one copy of human ApoE (either E3 or E4) had less amyloid pathology than AD mice with both copies of the human gene (ARF related news story on Kim et al., 2011). In San Francisco, Bien-Ly reported similar results in J20 mice, a different APP-overexpressing AD model. The work has been submitted for publication.

Bien-Ly and colleagues crossed J20 animals with human ApoE3 or ApoE4 knock-in mice, and measured Aβ burden in ApoE hemizygous and homozygous offspring at six or 12 months of age. At six months, having less ApoE hardly seemed to matter—hemizygotes and homozygotes had comparable levels of soluble and insoluble brain Aβ, suggesting that a single copy of ApoE does fine for Aβ clearance, Bien-Ly said. That was not the case for older AD mice. At 12 months of age, ApoE4 homozygotes accumulated more amyloid than J20 mice with a single copy of ApoE4, and the homozygotes had more brain Aβ42 and more plaques than did J20/ApoE3 animals. “In aged mice, you see not only isoform-specific effects, but also a gene-dose effect [on Aβ buildup],” Bien-Ly reported.

Bien-Ly and colleagues took things a step further by tweaking the knock-in strains to have loxP sites flanking the human ApoE transgene. This allowed the researchers to dial down ApoE levels during adulthood using Cre recombinase. When they injected Cre-expressing adenoviruses into the hippocampus of six-month-old ApoE4-expressing J20 mice, ApoE levels fell 10 to 15 percent, and a month after injection, insoluble Aβ42 levels had plunged by half. These data suggest that decreasing brain ApoE4, or even ApoE3, could be a potential therapeutic strategy for AD, Bien-Ly said. Thus far no behavioral assessments have been reported for J20 or APP/PS1(L166P) ApoE hemizygotes.

There remains an apparent discrepancy between genetic manipulation of ApoE, which suggests that more ApoE is bad, and pharmacological agents that enhance Aβ clearance and improve cognition while increasing brain ApoE. With regard to Aβ burden, the genetic data argue that “you would probably want to lower ApoE, especially E4, not increase it,” noted Joachim Herz of the University of Texas Southwestern, Dallas, in an e-mail to ARF. Huang agreed. “No matter whether you think about Aβ-related or Aβ-independent effects, increasing ApoE4 is bad,” he said. When neurons turn on ApoE4, they generate toxic E4 fragments that can lead to tau pathology, mitochondrial impairment, and eventually cognitive loss and neurodegeneration (Huang, 2010), he said. And now, these recent data suggest there may be bad consequences, e.g., more Aβ deposition, even with ApoE3.

There is another way to consider the confusion between genetic and pharmacologic data. In the brain, ApoE “only functions in the context of high-density lipoproteins (HDLs),” said Gary Landreth, Case Western Reserve University, Cleveland, Ohio. Landreth was the principal author on the recent Science paper linking an ApoE boost to improved pathology and cognition. “More ApoE is not necessarily a good thing unless it is lipidated,” he told Alzforum. Transport proteins such as ATP-binding cassette transporter (ABCA1) load membrane lipids and ApoE onto HDL particles, which promote Aβ degradation (see ARF related news story on Jiang et al., 2008). Bexarotene and other LXR/RXR agonists not only turn on ApoE expression, but also upregulate ABCA1 and other proteins that help assemble HDL particles. “Simply overproducing ApoE without inducing the machinery to make HDLs appears not to be a great thing,” Landreth said, based on recent genetic data showing ApoE dosage effects.

Ascribing harm or benefit to ApoE modulation is challenging also because not all effects are isoform-specific. For example, complete lack of ApoE impairs Aβ clearance (see Terwel et al., 2011), arguing against decreasing ApoE as a therapeutic strategy. However, the E4 isoform worsens long-term potentiation and Aβ-induced neurotoxicity (see Trommer et al., 2005; Trommer et al., 2004; Manelli et al., 2007). “The implications of ApoE4 having some additional gain-of-toxic-function effects may need to be addressed before we can get therapeutics to work,” commented Greg Cole of the University of California, Los Angeles. Mary Jo LaDu and Leon Tai of the University of Illinois, Chicago, expressed similar concerns in an e-mail to ARF (see full comment).

And yet, amid the debate about whether to modulate brain ApoE levels up and down, some think the real issue lies elsewhere. “I do not view either the raising or lowering of ApoE levels as a viable therapeutic approach,” wrote Allen Roses of Duke University in Durham, North Carolina, in an e-mail to ARF. In his view, therapeutic compounds should target interactions between ApoE fragments and intraneuronal mitochondria, since these associations set off a cascade of mitochondrial problems that contribute to neuronal death (see Chang et al., 2005; Roses et al., 2007).

“It boils down to the question of whether Aβ clearance and plaque formation are primary determining factors for ApoE4-mediated neurodegeneration, or whether a combination of other factors, including neuroprotective effects of LXR/RXR agonists and HDAC inhibitors, will dominate,” Herz wrote.

People carrying one copy of the apolipoprotein E4 gene variant are up to four times more likely to develop Alzheimer’s disease, and that risk rises up to 15-fold in E4 homozygotes—but what makes the protein so harmful? Research suggests that ApoE4 contributes to AD in various ways, some via Aβ, others independent of it (Huang, 2010). At an ApoE workshop held 30 January to 1 February at the Gladstone Institute of Neurological Disease in San Francisco, speakers reported new data along both fronts. Conference organizer Yadong Huang of Gladstone spoke about his lab’s characterization of aging ApoE4 knock-in mice. The data squarely blame the animals’ cognitive deficits on a specific set of cells—GABAergic interneurons in the hilus of the dentate gyrus. For his part, Steven Paul of Weill Cornell Medical College in New York City reported that, surprisingly, ApoE4 speeds neurofibrillary tangle formation and triggers possible neurodegeneration in old AD transgenic mice. Paul found that ApoE4 knock-in mice have abnormal electrophysiology, even seizures. Together, the studies help flesh out ApoE4’s harmful effects at the cellular level. Other researchers reported on the consequences of genetic or chemical modulation of ApoE on cells and animal models, including changes in Aβ dynamics (see Part 1 of this series).

In a prior study, Yaisa Andrews-Zwilling of the Huang lab, and colleagues, reported that female ApoE4 knock-in mice (which express human ApoE4 in place of the endogenous mouse gene) lose hilar GABAergic interneurons as they age (Andrews-Zwilling et al., 2010). By 16 months, cell counts dropped 30 percent. The reduction correlated with learning and memory deficits, which were rescued by a drug that activates GABAA receptors. At the time, the researchers did not know if the neuronal loss was the cause of the cognitive decline or an effect of a broader pathology. However, at the Gladstone meeting, Huang noted that old male ApoE4 knock-in mice have normal numbers of hilar GABAergic interneurons and do not show cognitive deficits. This otherwise unexplained sex difference suggests to Huang that retaining these neurons is key to staying cognitively sharp, and may make males resistant to ApoE4-induced problems.

To address whether loss of hilar GABAergic interneurons in fact causes the cognitive decline, Andrews-Zwilling and Anna Gillespie of the Huang lab used optogenetics to deactivate these cells in living wild-type mice (see Fenno et al., 2011 for a review of the technology). The researchers expressed a light-sensitive chloride pump in hilar GABAergic interneurons; the pump responds to yellow laser light, hyperpolarizing the cells and thereby shutting down their activity. A one-minute pulse of light—shone through glass fibers inserted in the dentate gyrus of freely moving animals—was enough to make wild-type mice perform as poorly as aging ApoE4 knock-in mice on spatial learning and memory tests.

What role do these interneurons play in memory and learning? Huang’s team tested whether optogenetic inhibition of the hilar GABAergic interneurons impairs memory retention, or retrieval, or both. They found that mice expressing the light-sensitive chloride pump learn the location of the hidden platform of the Morris water maze just fine if the laser light is kept off during five days of training trials. The mice forget the location if the light is switched on 24 hours after the last learning trial; but, the animals recall the platform location 48 hours later with the laser turned off. “That means there is a problem with memory retrieval but not retention. In Huang’s view, this sort of cognitive malfunction resembles the predicament of early AD patients, “who cannot remember something today, but tomorrow might remember it,” said Huang.

As a first step toward addressing whether the neuronal deficits in aging ApoE4 mice resemble AD, Huang’s team used induced pluripotent stem (iPS) cells to generate human neurons with ApoE3/E3 or ApoE4/E4 genotypes. The procedure is slow, requiring six to eight weeks to culture the iPS cells and differentiate them into neurons. However, immunostaining and Western blotting suggest that ApoE4 iPS cells make fewer GABAergic interneurons and have increased tau phosphorylation, said Huang.

The theme of tau pathology stood out in Paul’s presentation as well. In collaboration with Patrick Sullivan at Duke University School of Medicine, Durham, North Carolina, Paul and colleagues found evidence for hyperphosphorylated tau, neurofibrillary tangles, and even neurodegeneration in the hippocampus of 15-month-old PDAPP mice expressing human ApoE4—but not in PDAPP mice expressing mouse ApoE or human ApoE3 or E2, or in wild-type mice just expressing human ApoE isoforms (including E4). “Thus, it appears that ApoE4 has amyloid-independent effects in facilitating tau pathology, but only in the presence of human Aβ,” Paul noted in an e-mail to ARF.

In a separate set of studies, Paul reported that about 60 percent of female ApoE4 knock-in mice, and 40 percent of males, have seizure behaviors. These behaviors are mostly mild, for example, head jerks. Nearly a quarter of ApoE4 animals—but not E3 or E2 knock-in mice—have severe clonic (i.e., rhythmic muscle contraction/relaxation) seizures, as well as a lower threshold for pentylenetetrazol-induced seizures and abnormal brain electrical activity measured by electroencephalography (EEG). “This may be the best model of AD-associated seizure disorder, and it is an amyloid-independent behavior,” Paul noted. Scientists have reported epileptiform activity and network instability in several strains of AD transgenic mice (see ARF related news story on Palop et al., 2007; ARF related conference story) and in people with AD or mild cognitive impairment (see ARF related news story).