Some guiding assumptions and hypotheses

1. Protein aggregation poses one of the most universally observed constraints on longevity, in that the burden of accumulated aggregates increases with age in each animal and every tissue we have examined: human brain, muscle and arterial plaque; mouse heart, kidney and cerebrum; human cells in culture; and C. elegans worms.  We envision a complex of positive-feedback loops, wherein protein aggregation congests all layers of the misfolded-protein clearance pathways (refolding chaperones, the ubiquitin-proteasome system, and autophagy), and the accrual of aggregates is inflammatory.  Systemic inflammatory signals (cytokines, NFkB, etc.) activate peripheral macrophages and CNS microglia, increasing levels of oxidative damage in all tissues.  Oxidized proteins are especially prone to aggregation, completing the loop.  Figure 1 summarizes this schema for humans and mice (showing just the tissues we have studied).

2. Many dynamic (nonstructural) proteins have evolved to tread a narrow path between stability and instability, providing maximal structural versatility and variety of interactive partners.  The price paid for this is that post-translational modifications (oxidation or excessive phosphorylation)can push these proteins into transient or irreversible misfolding, favoring their ionic or hydrophobic interaction with accidental partners, and thus protein aggregation.  Because that price is only exacted late in life, when the force of natural selection is weak or nonexistant, there is really no selective pressure limiting the tendancy toward proteins with narrow margins of stability.  We have observed increased levels of protein oxidation and phosphor- ylation in Alzheimer's aggregates and with aging of mouse tissues. Molecular-dynamic modeling in a sophistocated atomic-array simulation program (GROMACS) allows us to predict the effects of modifications on protein-protein interactions, with results remarkably close to those seen in vivo