Benzodiazepines protect hippocampal neurons from degeneration after transient cerebral ischemia: an ultrastructural study.

The ability of full and partial benzodiazepine receptor agonists to prevent DNA fragmentation and neuronal death after transient cerebral ischemia was investigated in the Mongolian gerbil. Diazepam (10mg/kg, i.p.) or the partial agonist imidazenil (3mg/kg, i.p.) was administered 30 and 90min after transient forebrain ischemia produced by occlusion of the carotid arteries for 5min. Treatment with diazepam completely protected CA1b hippocampal pyramidal neurons in 94% of the animals and partially protected pyramidal neurons in 6% of the animals, as assessed with a standard Nissl stain three and four days after ischemia. DNA fragmentation was examined by the terminal dUTP nick-end labeling (TUNEL) reaction. Prior to cell death, there were no TUNEL-positive neurons in area CA1b. By three days after ischemia, when neuronal degeneration was nearly complete, 14 out of 16 gerbils exhibited a positive TUNEL reaction throughout area CA1b stratum pyramidale. In 13 out of 14 gerbils treated with diazepam, no TUNEL-positive neurons were observed in this region. Imidazenil was less effective than diazepam with respect to both neuroprotection and prevention of DNA fragmentation. Three days after ischemia, six out of eight gerbils treated with imidazenil showed partial to complete neuroprotection. Imidazenil completely prevented DNA fragmentation in only one of the animals; varying degrees of TUNEL reaction persisted in the remainder. To determine whether the neurons protected by diazepam had a normal ultrastructure, gerbils were killed two to 30 days after ischemia and the hippocampal neurons in area CA1b were examined by electron microscopy. Within the first 48h after ischemia, early cytoplasmic changes of varying degrees (e.g., vacuolation, rough endoplasmic reticulum stacking, swollen mitochondria) and electron-dense dendrites were observed in gerbils not treated with diazepam. Degeneration was nearly complete by three days after ischemia. In contrast, pyramidal neuron ultrastructure appeared normal in gerbils that exhibited complete area CA1b neuroprotection (defined at the light microscope level) by diazepam when studied two, seven or 30 days after ischemia. In gerbils with partial protection of area CA1b, most of the remaining neurons exhibited varying degrees of necrosis when studied 30 days after ischemia. No apoptotic bodies were observed. We conclude that: (i) diazepam can fully protect CA1 pyramidal cells from the toxic effects of transient cerebral ischemia; (ii) when diazepam affords only partial neuroprotection, the residual CA1 pyramidal cells exhibit ultrastructural abnormalities consistent with necrotic damage; and (iii) diazepam is a more efficacious neuroprotectant than the partial benzodiazepine receptor agonist, imidazenil.