A riddle wrapped in a mystery inside an enigma. The phrase could just as easily apply to Alzheimer disease, and to Aβ in particular, which is reported to punch holes in lipid membranes, generate reactive oxygen species, poison synapses, hijack signal transduction pathways, and kick start apoptosis. How could all this be true of one little peptide? Maybe the answer lies in where Aβ is. It has been spotted inside the cell, outside the cell, and even in between. Last week saw three different reports describing three different toxicities for Aβ, one for each location. In the January 24 Journal of Neuroscience, Subhas Biswas and colleagues at Columbia University, New York, report that Aβ ends up in intracellular cahoots with the morbidly named Bcl-2 interacting mediator of cell death (Bim), and induces neuronal apoptosis. In the same journal, William Klein and colleagues at Northwestern University, Evanston, Illinois, report that soluble Aβ-derived diffusible ligands (ADDLs, aka Aβ oligomers) attack neurons from without, binding and damaging dendritic spines. And in the January 24 Journal of Biological Chemistry online, Paul Axelsen and colleagues at the University of Pennsylvania, Philadelphia, report that Aβ can promote membrane lipid oxidation, which in turn promotes Aβ fibrillization. Here are some of the gory details.
The Enemy Within
First, to cell death. Bim is a proapoptotic factor that is upregulated in neurons when they are deprived of the trophic factors that normally keep them healthy. Previous studies had indicated that Bim- and Aβ-induced neuronal death share some things in common, including the activation of normally quiescent cell cycle proteins (see, for example, Park et al., 1998). To probe this relationship further, Biswas and colleagues looked for Aβ-induced changes in Bim expression. They found that aggregated Aβ42 upregulates the apoptotic protein in cultured hippocampal and cortical neurons and that cell death follows shortly thereafter. The findings suggest that Aβ toxicity may revolve around induction of Bim. The researchers examined postmortem brain tissue from AD patients and found elevated levels of Bim in the entorhinal cortex. The protein was most prevalent in areas of Aβ deposits, and Bim-positive neurons also tested positive for the cell cycle protein Cdk4 and its substrate retinoblastoma protein. Mechanistically, the authors found that Bim seems to be required for Aβ-induced neuronal cell death because shRNAs directed against Bim transcripts protected against such death. Biswas and colleagues also found that the Cdk inhibitors flavopiridol and roscovitine protected against Aβ-induced neuronal death, supporting the role of miscreant cell cycle proteins in Aβ toxicity. Re-entry of terminally differentiated neurons into the cell cycle has been a recurrent theme in AD research (see our recent live discussion and live discussion on the cell cycle and AD).
Attack from Outer Space
Klein’s group took a similar experimental approach, treating highly differentiated cultures of hippocampal neurons with ADDLs/oligomers, but in this case it appears that the major toxic effects were initiated from without. First author Pascale Lacor and colleagues found that ADDLs bound exclusively to neurons that express NMDA-type glutamate receptors. Inhibitory GABA-ergic neurons failed to bind the protein oligomers, suggesting that Aβ has a particular grudge against NMDA neurons. They also found that ADDLs bound primarily to postsynaptic sites. Synaptosomes isolated from ADDL-treated neurons using ADDL antibodies contained postsynaptic density 95 and the NR1 and NR2A/B subunits of the NMDA receptor. They did not contain the presynaptic protein syntaxin.
ADDLs had a major effect on the plasticity and morphology of dendritic spines. Lacor and colleagues found that within three hours of ADDL treatment, levels of NR1 and NR2B had fallen by 78 and 70 percent, respectively. At 6 hours post-treatment, levels of the NMDA receptor-associated protein EphB2 had fallen by 60 percent. The findings are reminiscent of earlier work from Paul Greengard and Gunnar Gouras’s labs at Rockefeller University and Cornell University, respectively (see ARF related news story) and of work from Roberto Malinow’s lab at Cold Spring Harbor Laboratories, also in New York, showing that soluble Aβ can bring down AMPA-type glutamate receptors (see ARF related news story).
Lacor and colleagues found gross morphological changes, as well. The number of spines, as judged by the spine marker drebrin, fell by about half 24 hours after treatment, while their average length roughly doubled. “The change in spine appearance caused by ADDLs is especially interesting because the elongated shape resembles that of immature spines or of disease spines found in mental retardation and prionoses,” the authors wrote. A similar spine loss is emerging in other preparations, as well (see ARF Eibsee conference report). Intriguingly, Klein’s team found that the NMDA receptor blocker memantine, which has been approved for treatment of early AD, protected against ADDL-induced drebrin loss, suggesting that the drug may have pleiotropic effects.
Dial M for Membrane
And finally, there’s Aβ’s effect on lipids. Oxidation of lipids has long been considered a potential harbinger of problems in neurons, and 4-hydroxynonenal (HNE) in particular has been studied in the context of AD. This byproduct of lipid oxidation can cross-react with protein side chains and is elevated in AD (see ARF related news story). Oxidized lipids have also been shown to increase Aβ aggregation and Aβ to promote lipid oxidation, but it was not clear how the two were related. Now, Axelsen and colleagues report that HNE constitutes a missing link.
First author Ian Murray and colleagues report that Aβ stimulates production of HNE from synthetic lipid vesicles and that this reaction requires the presence of copper (see related live discussion on the role of Aβ/Cu in oxidation reactions). Murray and colleagues also found that Aβ can promote HNE formation in lipid extracts from human brain, suggesting that the chemical reactions may be physiologically relevant. They also report that HNE covalently modifies Aβ40 and Aβ42 on histidine side chains. These modified Aβs not only have a greater tendency to aggregate, but they also have an increased affinity for lipid membranes, indicating that they may produce even more HNE.—Tom Fagan