The idea that not all oligomers are obligate precursors to fibrils, but instead follow their own, separate pathway has been around for at least a decade, when the first antibodies against such stable species were found, noted Charles Glabe, University of California, Irvine, (see Yang et al., 1995). More recently, it began to gain currency in a broader way. For example, research from other neurodegenerative diseases involving deposition proteins such as α-synuclein and huntingtin suggests that chaperones importantly influence which pathway the monomeric protein will enter (Wacker et al., 2004; Muchowski and Wacker, 2005).

Additional data on Aβ oligomers came from a team at Merck Research Labs in West Point, Pennsylvania. Merck last year licensed Klein’s ADDL technology from the biotechnology company Acumen in South San Francisco, California, and in Washington, D.C., last month, the Merck/Acumen Joint Research Team presented a series of posters and a talk. For example, the scientists developed an assay to compare and quantify how well various antibodies can block the binding of ADDLs to cultured hippocampal neurons. The A11 antibody did not block ADDL binding in this assay, while others, including 6E10 and 4G8 partially blocked it, and the WO-2 antibody, made by Lars Lannfelt’s group while he was still at Karolinska Institute in Huddinge, Sweden, blocked more strongly. Curiously, despite this difference, the immunocytochemical binding of these antibodies to ADDLs looked similar. Cross-binding studies such as these are highlighting that the different groups are each looking at slightly different forms of oligomer. (On this note, the Abbott researchers said that A11 binds their globulomer at 1000-fold lower affinity than do their own antibodies, but Ashe said A11 does recognize Aβ*56.)

Faced with the task of characterizing ADDLs quantitatively, the scientists from Merck/Acumen brought in some analytical prowess. They compared data from standard SDS-PAGE, which had previously indicated the presence of trimers, tetramers, and several other species, with data gathered from solution-based methods such as high-performance size-exclusion chromatography coupled with multi-angled laser light scattering, and with analytical ultracentrifugation and atomic force microscopy. This analysis indicated that ADDLs occur in a much wider range of sizes than previously thought, the team reported.

The Merck researchers also showed data suggesting that ADDL binding to cultured primary hippocampal neurons leads to an increase in phosphorylated tau in those neurons several hours later. This dovetails with data presented by Sally Frautschy and Greg Cole’s group at University of California, Los Angeles. These investigators infused an antibody against the Aβ1-15 epitope into the brains of Tg2576 mice and showed that this experimental passive vaccination reduced not only the levels of Aβ oligomers but also of phosphorylated tau and active GSK3β, one of the kinases known to phosphorylate tau. In subsequent in-vitro and cell-based experiments, Aβ42 oligomer preparations activated GSK3β. The A11 antibody, which recognizes Aβ oligomers but not monomers or fibrils, counteracted that activation. Furthermore, Frautschy showed that CNS infusion of Aβ oligomers stimulated GSK3beta activation and caused cognitive deficits in the Morris water maze, which were blocked by specific GSK3 inhibition. Taken together, these findings strengthen an emerging realization that Aβ oligomers act upstream of tau, and suggest that this might happen in part via GSK3β. They also imply that if a safe and effective vaccine against oligomeric Aβ could be found, it might have an indirect effect on the tau arm of AD pathogenesis, as well. (For summaries of similar experiments by other groups, see ARF SfN tau series and ARF SfN intraneuronal Aβ series.)

The debate around whether plaques/tangles or their smaller predecessors are the primary neurotoxic species in Alzheimer disease has been roiling for years, and it still does. Consider the following two news tidbits, one for each side of the issue. The SfN meeting saw growing evidence suggesting that, at least in transgenic mice, Aβ oligomers are responsible for distinct synaptic and cognitive deficits before mature pathology has deposited. Besides Ashe’s data (see part 1 of this story), conference-goers also noted the work of Irene Cheng in Lennart Mucke’s lab at the University of California, San Francisco. On a poster crowded with visitors, she presented new data on a mouse model of the Arctic mutation that was first discovered by Lannfelt’s group (see ARF SfN story). Residing within the sequence of Aβ itself, this mutation makes the peptide more prone to aggregate. Cheng’s mice expressing this mutation predictably formed neuritic amyloid plaques faster than did transgenics expressing wild-type Aβ (Cheng et al., 2004). At the conference, Cheng showed how she has compared their learning and memory in the Morris water maze to that of mouse strains expressing human wild-type Aβ, which perform poorly in this test. Alongside the behavioral tests, she also compared the concentration, in hippocampal dentate gyrus neurons, of synaptic proteins whose expression depends on the level of activity. These proteins are relatively depleted in the mice expressing human wild-type Aβ. By contrast, mice expressing the Arctic mutation performed normally and had normal levels of the activity-dependent synaptic proteins, even though they had a head full of plaques. There were differences between a low- and high-expressing line, but in a nutshell, the scientists suggest that the onset of cognitive deficits is delayed in a situation where a particularly fibrillogenic form of Aβ drives the peptide into plaque formation so rapidly that relatively fewer of the neurotoxic Aβ oligomers are around to mess with synaptic function.

After barraging our esteemed readers with news on Aβ oligomers, and partly overlapping news on intraneuronal Aβ (see ARF SfN intraneuronal Aβ series), it seems prudent to end on a contrarian’s stance as an antidote to irrational exuberance. Lest oligomerophiles get carried away believing that cognitive dysfunction in Alzheimer disease is all about these small villains, this story will leave them with a final note on plaques. Curious minds may want to stay on the lookout for the publication of an elegant presentation by Edward Stern in Brad Hyman’s group at Massachusetts General Hospital. Stern and colleagues used intraneuronal recordings in live Tg2576 and control mice to address the question of whether plaque deposition distorts the transmission of nerve impulses in cortex. In a prior paper, Stern had recorded from neocortical pyramidal neurons and shown that the synaptic response to stimuli coming from across the corpus callosum was severely impeded when the signal had to travel through plaque-loaded brain tissue, and they had suggested that this might be a functional consequence of dystrophic neurites. Importantly, the cortical neurons were responsive, but plaques distorted their convergent inputs so that the cortical neurons could not integrate them properly (Stern et al., 2004).

In Washington last month, Stern presented a follow-up study in which he took advantage of the whisker barrel system. In this beautiful neuroanatomical array, each whisker on a mouse’s snout has a receptive field that maps first to the thalamus and from there to a distinct column of neurons in the somatosensory cortex. Bending a particular whisker prompts cortical neurons in its column to fire; tweaking neighboring whiskers from within that given receptive field will elicit a weaker yet defined and measurable response. Using this system, Stern and Hyman found that in Tg2576 mice, only the primary whisker prompted a consistent and measurable response in somatosensory cortex. The farther away from the primary whisker Stern stimulated, the more variable and disjointed the response became, in essence narrowing the de facto size of each whisker’s receptive field. In short, the precision and reliability of the cortical response decreased as the distance that convergent inputs had to propagate through plaque-littered areas increased. This possibly occurs as plaques distort the morphology of neurites. One implication is that convergent inputs impinge on their target neurons without the necessary synchrony, and this, in turn, might help explain why association cortex, which depends heavily on integration of inputs from multiple areas, is more affected in AD than is primary motor cortex, for example. Maybe the motto should be: “It’s Both, Stupid!” Final coinage of a slogan will have to await decisive data; meanwhile, pithy suggestions are always welcome.—Gabrielle Strobel.