Intracellular injection of Aβ1–42 induces neuronal cell death and TUNEL staining (green).

The most obvious symptom is not always the most relevant. In Alzheimer's disease (AD), for example, post-mortem brain analyses show large extracellular amyloid plaques of fibrillar Aβ protein. But the plaques arise late, whereas an intracellular version of the toxic protein fragment (Aβ1–42, a fragment of the amyloid precursor protein) is seen earlier, coincident with the first cognitive defects. Now Zhang et al. (page 519) show that minute amounts of this intracellular Aβ1–42 are toxic to primary neurons, even when injected directly into the cytoplasm.The final concentration of injected Aβ1–42 is 1–100 pM, for a total of somewhere between 1,500 and 150,000 molecules per neuron. In contrast, the same neurons are not significantly affected by up to 10 μM extracellular Aβ1–42, nor by intracellular Aβ1–40, perhaps because the Aβ1–40 cannot multimerize. Death requires transcription and translation, and the activities of p53, Bax, and caspase 6, probably in that order. The pathway from Aβ1–42 to p53 remains a mystery. Zhang et al. are now looking at kinases that are known to regulate p53 activity.

What goes wrong first in AD is also not clear. If Aβ1–42 exerts its activity in the cytoplasm it must first escape from the secretory pathway. Zhang et al. suggest this may happen by either reverse translocation from the ER or leakage from endosomes. As the details of these pathways are fleshed out, both the trafficking and apoptotic pathways may become excellent targets for AD drugs. ▪