Historically, reactive oxygen species (ROS) have been viewed as toxic molecules that cause oxidative stress. However, this view has been challenged by studies from a number of laboratories in recent years. It is becoming increasingly clear that many cell types, including neurons, employ ROS as cellular messengers to modulate signal transduction pathways that are necessary for normal physiological processes. Consistent with this notion, in previously published studies several laboratories, including ours, have shown that ROS are critical for hippocampal long-term potentiation (LTP) in area CA1. In addition, we, and others have shown that LTP and hippocampus-dependent memory are impaired in transgenic mice that overexpress either cytoplasmic superoxide dismutase (SOD-1) or extracellular superoxide dismutase (EC-SOD). Taken together, these findings suggest that, in contrast to the historical view of ROS as toxic molecules, ROS can act as signalling molecules that are critical for normal synaptic plasticity and memory function.

If ROS are critical for hippocampal synaptic plasticity and hippocampus-dependent memory, then what signalling cascades are impacted by ROS? One candidate is the extracellular signal-regulated kinase (ERK) cascade, which has been shown to be both activated during and necessary for NMDA receptor-dependent LTP. In addition, incubation of hippocampal slices with ROS has been shown to result in ERK activation. Finally, NMDA receptor activation has been shown to result in the production of ROS. Therefore, we hypothesized that NMDA receptor-dependent activation of ERK requires ROS. We have found that NMDA receptor-dependent activation of ERK in hippocampal area CA1 is blocked by both superoxide and hydrogen peroxide scavengers. The NMDA receptor-dependent activation of ERK also was blocked by NADPH oxidase inhibitors, which suggests that NADPH oxidase is the source of ROS required for ERK activation. These results are consistent with the possibility that the ERK signalling cascade is impacted by ROS during NMDA receptor-dependent LTP.

We also have begun to investigate how ROS are produced during LTP and hippocampus-dependent memory. We are currently conducting studies with gp91phox and p47phox (two of the necessary protein components of NADPH oxidase) knockout mice to determine whether NADPH oxidase is necessary for either LTP and/or hippocampus-dependent memory. In addition, we are conducting studies with transgenic mice that overexpress MnSOD (SOD-2) to determine whether mitochondrial ROS production is necessary for LTP and/or hippocampus-dependent memory. The results of these studies should provide insight into how ROS are produced during hippocampal LTP and hippocampus-dependent memory.

This work was supported by NIH NINDS (NS34007).