In a clever bit of sleuthing, Karen Ashe and colleagues appear to have solved the case of the mysterious memory loss in young mice engineered to express a mutated form of the human amyloid precursor protein (APP). Mysterious because the mice display memory problems as early as 6 months of age, in the absence of any neuronal loss and long before amyloid deposits appear in their brains. From Ashe’s previous work, and work from other labs (see review by Walsh and Selkoe, 2004, and Barghorn et al., 2005), all signs pointed toward a soluble oligomeric form of Aβ as the culprit, but no one knew exactly what that form might be in vivo.
Now, Ashe and colleagues at University of Minnesota in Minneapolis, along with collaborators from several other U.S. institutions, report the identification of a soluble, extracellular Aβ12-mer that appears responsible for erasing memory in their mice. The oligomer, which they call Aβ*56 (Aβstar56), appears in the brain precisely when memory impairment commences, and its levels in individual mice correlate with their performance in a test of recall. To prove that the oligomer caused memory loss, the researchers purified the species from mouse brain and showed that it could elicit transient memory impairment in rats when it was directly injected into their brains. The work, which Ashe first presented in a talk at last year’s Society for Neuroscience meeting (see ARF related news story), appears in this week’s Nature.
The results add to the growing understanding of the early toxicity that results from Aβ overproduction, independent of neuronal loss and amyloid plaque formation. The results suggest that, if a similar type of Aβ is found in humans with Alzheimer disease, it could serve as an early diagnostic marker, and a target for early interventions to treat, or even prevent, memory loss. Ultimate proof that Aβ*56 indeed causes the memory loss in humans will require that a drug be found that can reverse, or prevent, its effect.
To look for the causative agent of early cognitive changes in AD mice, Ashe took advantage of her Tg2576 mice. The mice, which carry a human APP transgene with the Swedish mutation, have a biphasic decline in memory with age. The first noticeable impairment shows up at 6 months, after which performance stabilizes for 6-9 months, and then undergoes a second decline. First author Sylvain Lesné led the effort to look for a specific Aβ species whose appearance coincided with the earliest stage of memory loss.
After developing a subcellular fractionation procedure to separately look at extracellular, intracellular, membrane-bound, and insoluble materials, the researchers used immunoblotting to detect Aβ. In particular, two multimeric forms of Aβ, a 40 kDa putative nonamer and a 56 kDa putative dodecamer appeared at 6 months, and then remained at steady levels thereafter. Monomers, trimers, and hexamers were also detected, but they appeared earlier, and thus seemed unlikely to be a primary cause of the memory problems.
The researchers capitalized on a natural variation among animals in the amounts of the different Aβ species to ask in another way which, if any, of these species were associated with memory impairment. To do this, they compared the performance of individual young mice in the Morris water maze with the levels of Aβ monomers, dimers, trimers, etc. The best correlation was seen with the 56 kDa species—as its level rose, memory function declined in 6-month-old mice. A weaker correlation was seen for the less abundant 40 kDa species, and no correlation was evident for monomers, trimers, or hexamers. Curiously, the 56 kDa species never increased any further in old mice, even as progressive memory loss occurred. The researchers speculate that perhaps other mechanisms take over later, when dystrophic neurites and other abnormalities begin to appear.
From these experiments, Lesné et al. concluded that the most likely candidate for causing memory loss was Aβ*56. To strengthen their hypothesis, they did a painstaking set of behavioral experiments in collaboration with Ming Teng Koh and Michela Gallagher at Johns Hopkins University in Baltimore. After purifying the Aβ*56 complex from brains of impaired mice, they administered it to young, healthy rats via a cannula into the lateral ventricle. Giving Aβ*56 two hours before the rats were tested in a Morris water maze had no effect on their training to find a hidden platform. But if the rats were dosed again 24 hours later and put into a pool with no platform, the treated animals showed no spatial memory; they swam randomly, whereas the vehicle-infused rats spent more time in the quadrant where the platform had been. These experiments indicated that Aβ*56 impaired long-term memory, but not the initial acquisition of spatial information. The effect of Aβ*56 on memory was reversible, since both groups performed equivalently 2 weeks later, when no Aβ*56 was given.
What exactly is Aβ*56? It was identified by immunoblotting with the anti-Aβ antibodies 6E10 and 4G8. Subsequently, the researchers showed it also reacted with the A11 antibody made by coauthors Charlie Glabe and Rakez Kayed at University of California, Irvine. This antibody only recognizes soluble Aβ oligomers larger than tetramers, not fibrillar forms (see ARF related news story). Non-denaturing size exclusion chromatography showed that the multimer was a native form, not an artifact generated by gel electrophoresis. The structure was urea-resistant, but did denature in the strong hydrogen-bonding solvent HFIP. Purified complexes contained Aβ as a major component when analyzed by mass spectrometry. The data all add up to a highly stable, hydrogen-bonded Aβ structure, most likely a tetramer of Aβ trimers. A recent study showed a similar dodecameric form of Aβ detected in Tg2576 mice was reduced by passive immunization (Ma et al., 2005).
The study raises many new questions, as Richard Morris, who invented the Morris water maze, and Lennart Mucke, who has pioneered research into cognitive effects of soluble Aβ species, point out in their accompanying News and Views piece. Is the Aβ*56 present in other mouse models of AD with memory deficits? Is Aβ*56 related to the oligomers detected in the CSF of AD patients? Will these or other Aβ assemblies turn out to be reliable biomarkers for early diagnosis, or targets for early intervention (see ARF related news story)? And last but not least, they write, how do they actually derange normal function?
Ashe is already on the hunt for answers to those questions, saying last fall that her lab is searching for the receptor for Aβ*56. She told National Public Radio that she has already determined that patients with AD have detectable amounts of Aβ*56, while people without AD do not (listen to the report). Now, Ashe and her colleagues are looking for traces of the offending complex even earlier in AD.—Pat McCaffrey