Bacskai and Xie used a modified green fluorescent protein as a sensor of cellular oxidation potential. The protein, called roGFP, developed by Jim Remington at the University of Oregon, contains two cysteine residues that form a disulfide bond under oxidative conditions. When the bond forms, the fluorescent properties of the protein change, such that it and the reduced GFP glow best when excited by light of different wavelengths. Bacskai and Xie targeted roGFP constructs to neurons in normal mice and in a double-transgenic mouse model of AD (APPSw/PS1ΔE9), which begins to form amyloid plaques at about four to five months of age. They aimed a two-photon microscope through a transcranial window, in essence visualizing cellular oxidation in the living brain.

The microscopy revealed both good news and bad news. The good news is that very few cells in normal mice are sufficiently stressed to activate the reporter. Of some 50,000 neurons analyzed, only 0.13 percent glowed green from oxidized roGFP. Analysis of the APP/PS mice was promising as well, in that the total number of oxidized cells was still relatively small (about 0.42 percent). The bad news was what happens to neurons in the vicinity of amyloid plaques. Bacskai showed that a halo of oxidation surrounds plaques, where almost 60 percent of neurites glowed green. Over a two- to three-month period, the oxidative stress only got worse, even in the absence of plaque growth.

What happens to an oxidized cell? To answer this, Bacskai and Xie imaged individual neurons longitudinally. Over two months, a substantial number of oxidized cells disappeared in the vicinity of plaques. Curious to know how quickly neurons succumb, they looked more frequently and were surprised to see that some cells that became oxidized vanished within a few hours. In fact, Bacskai said that once a neuron becomes oxidized, it is doomed to die, and does so within 24 hours. After his talk, he told ARF that he is not sure if new neurons can replace those lost. This could be important in both AD and in normal mice. Even with oxidation rates as low as 0.13 percent, it would not take long before there was massive neuron loss in the normal mouse brain if oxidation precipitated cell death within a day.

Bacskai said he was confident that the oxidized cells were, in fact, dying and not just losing their reporter or modulating their redox potential. Co-staining for nuclei and caspase activity showed that oxidized cells underwent apoptosis, or programmed cell death, losing their nuclear material into the bargain. In fact, stopping the cell death program was difficult. Molecules that trap free radicals and reduce oxidative stress failed to prevent oxidation when given daily for a month. Antibodies to amyloid-β fell short, too, though these were only applied for an hour. Perhaps a longer treatment might have some effect, Bacskai said. One thing that did work well when applied directly to the brain was dithiothreitol (DTT), a reducing agent that breaks disulphide bonds.

This talk generated many interesting questions. Some noted that the APP/PS model shows little spontaneous neuronal loss, and wondered if this was the best model of oxidative stress and cell death for AD. Bacskai agreed that another model could prove interesting, but stressed that this was simply to test what happens during amyloidosis. Others questioned the use of the roGFP reporter itself. It seems to respond only to extremely high oxidation potentials (another reporter might detect milder oxidative stress), and once oxidized it is not clear if it can be reduced in vivo by naturally occurring reducing agents. Some wondered whether oxidative stress and death were related to the depth of a cell in the cortex. Since the two-photon microscopy does not penetrate deeply, Bacskai has only looked at cortical layers I and II; he has not looked for correlations between depth and apoptosis.

Jim Surmeier, Northwestern University, Chicago, Illinois, asked about the relationship between calcium and oxidative stress. Bacskai said it would be interesting to look into this, particularly which comes first. Scientists have long suspected that calcium poisons neurons in AD (see ARF related news story). Surmeier reported some years back that a particular type of voltage-gated calcium channel could explain why dopaminergic neurons in the substantia nigra are particularly susceptible in PD (see ARF related news story). At Taos, Surmeier outlined results published late last year in Nature based on targeting the roGFP reporter to the mitochondrial matrix of dopaminergic neurons. He found that calcium entry through L-type calcium channels during normal pacemaking increased mitochondrial oxidant stress in vulnerable dopaminergic substantia nigra neurons, and that loss of DJ-1 exacerbated this oxidant stress (see ARF related news story). DJ-1 loss-of-function mutations are associated with early onset Parkinson’s disease.—Tom Fagan.