A Direct Comparison
20 November 2007. A major point of confusion in this field has been that different research groups are working with different kinds of Aβ oligomer. The sources, the methods, and the tools used to characterize them differ enough to leave people scratching their heads about which kind of Aβ does what, and what best to work with themselves as they enter the field. Now, labs are beginning to report a side-by-side comparison of some of those species, and one such presentation is summarized below (see reference to a second, published one, further below). Miranda Reed, a postdoctoral fellow in Karen Ashe’s group at University of Minnesota Medical School in Minneapolis, presented data from a collaborative study with Jim Cleary’s lab at the Minneapolis VA Medical Center, and with MaryJo Ladu at the University of Illinois at Chicago. Reed started by noting that oligomers made from synthetic Aβ as used in ADDLs and other synthetic forms, naturally occurring Aβ derived from the transfected 7PA2 cell line, and transgenic mouse-derived Aβ*56, as used by Lesne et al. had largely driven the public literature on oligomeric Aβ species and raised two simple questions: 1) Does source matter? and 2) Does size matter?

To address these questions, the scientists compared the three types of Aβ in the same experiment. It involves training healthy young rats on an alternating lever-pressing task, performing surgery to infuse the respective Aβ preparation into their brains, and then testing the rats to assess how the injection has affected their performance (Cleary et al., 2005). Some scientists consider this a more sensitive measure of learning and memory than tests of transgenic mice in mazes (partly because rats are smarter than mice). Reed measured two types of error: approach errors and perseveration errors.

Reed reported that a dimer-enriched size exclusion chromatography fraction from conditioned medium of the 7PA2 cells caused the highest rate of both types of errors from that source, while a fraction enriched in trimers showed a non-significant trend toward increased errors. Their potency was in the nanomolar range. Among different fractions isolated from Tg2576 mouse brain, a purified trimer fraction increased approach errors, but the effect was weak and variable. Aβ*56 showed the largest increase in both types of error and was the only species to show a concentration-response relationship. The overall effect was less potent than fractions from 7PA2 medium, as both species were tested at micromolar concentrations. Synthetic Aβ was tested as an undefined mixture of oligomers that ran as monomers, trimers, and tetramers but not dimers on SDS-PAGE, which breaks down higher-order synthetic Aβ oligomers. It, too, produced both types of error when applied in the micromolar range. In toto, all preparations elicited errors except for monomers; the main difference lay in their potency and reliability.

The laboratories of Bill Klein at Northwestern University, Chicago, and Jorge Busciglio at University of California, Irvine, have shown that Aβ oligomers localize to synaptic sites, and in San Diego they reported on their continuing research on which processes might end up being disrupted there (see ARF related news story). Working with his colleagues Atul Deshpande, Erene Mina, and Charlie Glabe, all at UC Irvine, Busciglio last year published a side-by-side comparison of the effects that micromolar and nanomolar concentrations of ADDLs, and other high-n oligomers generated in Glabe’s lab, exerted on human fetal cortical neurons in culture. These oligomers, as well as ADDLs, are a heterogenous mixture and range widely in molecular weight. ADDLs appear to partially overlap in several species with the oligomers made in Glabe’s lab, including the 56 kDa dodecamers, Busciglio clarified by e-mail. The 90 kDa species also are a significant component, and those appear more similar in size to the protofibrils Lannfelt’s group is studying. At the higher concentrations, both these forms of Aβ oligomer were toxic at multiple levels, ranging from speeding up calcium influx to activating a mitochondrial death pathway. They were toxic not only at the synapses but also at other cellular membranes. The main difference between the high-n oligomers and the ADDLs was that the former tended to kill the neurons within a day, whereas the latter took up to a week to do so. Nanomolar concentrations of both preparations caused a similar but milder toxicity (Deshpande et al., 2006).

Why Do Oligomers Bind Synapses?
In San Diego, Busciglio addressed the related question of what draws soluble Aβ oligomers to synapses preferentially. He hypothesized that synaptic transmission might be at play, and described experiments in rat hippocampal slices and human cortical neurons that suggest that more Aβ oligomers cluster on the synapses when synaptic activity is stimulated pharmacologically. In contrast, blocking synaptic firing with the poison TTX reversed this effect. NMDA receptors appear to mediate this localized accumulation, as the oligomers colocalized with NR2B subunits after stimulation. Busciglio’s data also implicate nicotinic acetylcholine receptors in this synaptic receptor targeting. Furthermore, Busciglio reported that when these scientists added monomers and stimulated the preparations, they actually observed oligomeric formation at synapses, implying that overactivity at excitatory synapses may induce oligomer formation from monomers secreted at synaptic sites.

Metal ions such as zinc (Zn2+) and copper (Cu2+) may attract and further aggregate Aβ to the synaptic cleft since they are released at glutamatergic synapses during neurotransmission, Busciglio said. In these experiments, adding the chelator drug clioquinol reduced Aβ’s synaptic accumulation. (See Ritchie et al., 2003 for clinical trial results on clioquinol, and Caragounis et al., 2007 and Lau et al., 2007 for data on clioquinol’s mode of action, published this fall.)

Among the antibodies the scientists used is the polyclonal “officer” (OC) from Glabe’s lab, which labels fibrillar, donut-shaped oligomers (Kayed et al., 2007; see ARF Eibsee report). The neuron does not “eat” the “donuts”— that is, they remain on the cell surface and are not endocytosed or otherwise internalized during this experimental period of neurotransmission, Busciglio said. Rather, they appear to exert a toxic effect from the outside. As seen with immunohistochemistry on postmortem brain tissue of people who had had AD, these oligomers colocalized with the presynaptic marker synaptophysin, as well as with postsynaptic markers PSD 95 and NR2B, Busciglio showed.

Synapses Let Go, Then Shrivel
A welcome new development at the Neuroscience conference was a broadening of the research effort in that a growing number of laboratories were joining the fray. For example, Barbara Calabrese of the Scripps Research Institute in La Jolla, California, showed a series of elegant experiments visualizing what Aβ oligomers do to both sides of the synapse. Calabrese works with Shelley Halpain at Scripps, who is known for her basic research on spine dynamics and the role of the cytoskeleton in synaptic biology. In a collaboration with Eddie Koo at the University of California, San Diego, these investigators started testing the effects of Aβ oligomers secreted into the culture medium of 7PA2 cells (Calabrese et al., 2007). “This cell line allowed us to use picomolar concentrations of Aβ oligomer and to look at early effects on synapses separately from neurotoxicity. The changes we see do not kill the neurons,” Calabrese told ARF.

In her talk, Calabrese showed how she applied 40 to 80 picomolar concentrations of Aβ oligomers to primary hippocampal neurons from embryonic rats co-cultured with glia. These mixed cultures form synapses and stay alive for weeks. Calabrese showed that markers of presynaptic vesicles became fewer and smaller in glutamatergic synapses, but not in GABAergic terminals. The latter were impervious to Aβ oligomers. This finding might help explain the old observation that GABAergic, i.e., inhibitory, neurons are selectively spared in AD, and it suggests that neural circuits might shift to a more quiescent state in the presence of Aβ oligomers, Calabrese said. (Morphologically speaking, glutamatergic synapses tend to sit on the spine head, whereas GABAergic terminals generally synapse onto the dendrite’s shaft in between the spine protrusions.)

These changes happened quickly, within a couple of hours of application. This puts them in the right time frame for representing a potential cellular substrate for impaired synaptic plasticity, Calabrese added.

Calabrese noted not only a drop in number and size of glutamatergic nerve terminals and spines, but also that the remaining spines changed shape. Time-lapse imaging brought to light a new form of structural synaptic plasticity: in these movies, some previously normal-looking spines developed long, thin necks but stayed attached to their presynaptic sides, whereas other spines plain separated and collapsed. Yet other spines first elongated, then collapsed. These data address one debate in the field, namely, the question of whether Aβ oligomers affect the pre- or post-synaptic side first. Calabrese first saw the two sides uncouple, then the post-synapse collapse. This phenomenon must be understood in the biological context of “morphing,” she added, a term that describes that the synaptic connection constantly and normally undergoes some level of morphological change, even in the absence of Aβ oligomers.

Interestingly, these experiments also revealed that the spines are dynamic in more ways than that. Not only do they reappear a day after washout, they also somehow become resistant to the oligomers if those oligomers stay in the culture medium for 2 days. It’s not that the oligomers degrade in that time—transferring the same medium to a fresh batch of neurons will make the spines on those naïve neurons cave in. Rather, the spines somehow compensate for the continued presence of the oligomers. But they only do so for a time. When Calabrese washes out the oligomers, then re-applies them, the synapses shrink once again, and when she continuously applies them for 10 days, the spines do not recover spontaneously anymore. Even under this sustained assault, however, the neurons do not die. The effect is restricted to their synapses, which rapidly lose strength in electrophysiological recordings, but by itself it is not toxic to the neuron as a whole, Calabrese reported.

How this works is unclear at present. Calabrese and colleagues noticed that the spine changes disappeared when they added either a drug that blocks all types of nicotinic acetylcholine receptors, or the NMDA receptor antagonist memantine (aka Namenda). On this red-hot question, many groups are honing in on potential receptor targets of oligomeric forms of Aβ, as well as on receptor-independent mechanisms involving membrane integrity.—Gabrielle Strobel.

This concludes our report of presentations about Aβ oligomers at the Society for Neuroscience conference held 3-7 November. See also Parts 1 and 2.