Among hundreds of genes that might nudge your risk of Alzheimer's up or down, Apolipoprotein E (APOE) has the strongest effect. Scientists discovered a quarter century ago that people with the APOE ε4 version of this gene are four to 15 times more likely to develop Alzheimer’s, a deadly brain disorder that afflicts more than five million Americans. Yet how APOE actually sets off dementia has been somewhat of a mystery—and efforts to use it as a drug target have floundered.

The field’s attention has focused on another “A” word—amyloid beta (Aβ). This protein can unwittingly accrue in the brain for years, disrupting nerve connections essential for thinking and memory. APOE has been thought of as a co-conspirator in this process, but finding ways to undermine its collusion have proved challenging.

Anti-amyloid drugs have consumed the labors of pharmaceutical companies. If a drug could break those insidious clumps of protein or keep them from forming, drug developers reasoned, it could in theory halt the progression of the disease. But billions of dollars have poured into large-scale clinical trials of amyloid-lowering therapies that so far have failed.

APOE has hovered on the periphery as far as drug development, but this could soon change. Connections have emerged between the functioning of APOE and Aβ. In 2012 Boston scientists studying autopsy tissue from Alzheimer’s patients found APOE ε4 individuals had unusually high levels of brain Aβ. And they noticed Aβ clumped more readily in test tubes if mixed with ApoE proteins, especially ApoE4. Mouse data from teams at Washington University in Saint Louis and the University of California, San Francisco—suggested a similar relationship. Each lab worked with existing Alzheimer’s mouse models and further modified their genomes to make different types and amounts of ApoE proteins. In both studies animals with less ApoE had fewer Aβ plaques in their brains.

But the story grew more complicated. Although amyloid clogs the brain early on, memory loss tracks much better with a different protein, tau, which forms so-called “tangles” within nerve cells. Still, many in the field remained amyloid-centric, figuring tau would not accumulate without amyloid setting things awry first. So Yang Shi and her PhD advisor, neurologist David Holtzman at Washington University were in for a surprise when they peeked at a set of brain slices from mice engineered to produce tau pathology. The slides showed tangle production had nothing to do with amyloid and everything to do with APOE. ApoE4-producing mice racked up so many tangles in their brains that neurons died en masse—enough to see without a microscope. “When the brain’s smaller, it’s like wow, this is really obvious,” Holtzman says about inspecting the rodents. Another conclusion of the researchers was equally startling and may ultimately provide a clue for drug developers: If mice were genetically rigged to lack ApoE, their brains looked fine.

Other researchers took notice. These findings “fundamentally change the conversation,” says Gary Landreth, a neuroscientist at Indiana University, who was not involved with the research.  Reported September 20 in Nature, the new results raise ApoE’s profile in the mix of molecular events underlying Alzheimer’s and give strong support for a drug therapy that lowers brain ApoE.

A key caveat must be noted, though. The mice don’t quite model Alzheimer’s. A key goal of the study was to see if APOE drives tau pathology apart from its already known effects on amyloid. So the researchers chose a mouse model that develops tau pathology and neuron loss—but no Aβ plaques. These symptoms arise because the mice contain a tau mutation that normally causes a related degenerative disease called frontotemporal dementia, which affects cognition and behavior. No tau mutations are known to cause Alzheimer’s. Tau accumulates in brain areas affected in both dementias, though, so it’s likely some of the new findings will apply broadly to tau-driven brain disorders.

As far as the study’s therapeutic implications, one thing seems clear: If new drugs are to target ApoE in the brain, they should bring its levels down, not up. This turns out to be a key insight because prior observations had supported the opposite rationale. The confusion arose from human studies. When researchers measured ApoE protein content in human spinal fluid or brain tissue, they noticed APOE ε4 individuals consistently have less of the protein than people who carry other versions of the gene—APOE ε3 or ε2. Mouse data also supported this thinking. When Landreth and colleagues treated an Alzheimer’s mouse model with a cancer drug that boosts ApoE production, the mice cleared brain Aβ and regained cognitive function.

So it seemed reasonable to think APOE ε4 carriers might be rescued by raising ApoE protein levels. Indeed, one small trial gave Alzheimer’s patients the same cancer drug that looked promising in mice—but it failed. APOE ε4 is found in 25 to 30 percent of the population and in about 40 percent of the late-onset form of disease that makes up the vast majority of Alzheimer’s cases. (A rare early-onset form of the disease guarantees a person develops Alzheimer’s at a young age after inheriting even one copy of certain genes.)

The current study also touches on a much-debated aspect of neurodegeneration—the inflammatory response, which ultimately seems to worsen the disease. Neuroinflammation “was virtually blocked in animals with no ApoE,” Holtzman says. And in APOE ε4 mice, proinflammatory genes were way higher than in mice with other APOE variants.

Based on the new findings, it’s conceivable an ApoE-lowering drug could provide a triple punch. If you lower ApoE early in life, it could prevent or slow amyloid deposition, Holtzman says. If given later, the intervention might not do much for amyloid but could potentially have a big impact on tau pathology and inflammation—and there “you may have a longer window,” he adds. “If ApoE is mediating the inflammatory response, that’s something you theoretically would be able to decrease at any time.”

In reality it might not be so simple. Whether inflammation slows or speeds the disease process—and thus, whether drugs should boost it or shut it down—is a long-standing debate. Some studies argue the inflammatory response is initially protective: It revs up immune cells to clear misfolded proteins, including amyloid and tau. Prolonged inflammation, however, leads to the release of harmful chemicals that can kill cells and exacerbate disease. So for therapeutics, it might come down to timing. A drug that ramps up the inflammatory response could help early in the disease, but applying a boost later might make things worse, suggests Yadong Huang, a U.C.S.F. neuroscientist who studies ApoE but was not part of the Nature study.

Two years ago Huang and a U.C.S.F. colleague co-founded a biopharmaceutical company to develop ApoE-lowering therapies for neurodegenerative disease. Several other companies are also working on strategies that target ApoE4, Huang says. All of this work is preclinical—in cells and animal models—thus far.

Potential therapies could take several forms. Some approaches could work at the protein level—by stimulating ApoE turnover or clearing ApoE with antibodies. Other therapies might slow gene transcription so cells make less ApoE protein to begin with. With newer gene-editing tools such as CRISPR–Cas9, researchers can now make these kinds of DNA modifications with greater speed and precision.

But what about safety? Even if it’s technologically feasible to make a drug that lowers ApoE, isn’t there concern about shutting down a protein that performs useful functions in the body? ApoE helps carry cholesterol and other fats through the bloodstream. People who lack the APOE gene can develop dangerously high cholesterol levels and face increased risk of heart attack and stroke. Nevertheless, these individuals appear to be cognitively normal. Ideally a therapy would lower ApoE in the brain but not in the blood, Holtzman says.

As a first step, his team is testing whether it’s possible to stop or slow tau-driven neuron loss and inflammation by lowering ApoE in the early life of laboratory rodents. This scheme mimics a human scenario better than the recent study, which analyzed mice that express or lack APOE from birth. “The implication here, with the recent tau findings, is that you’d really block the neurodegeneration that leads to cognitive decline,” Holtzman says.