Stroke most commonly results from occlusion of a major artery in the brain and typically leads to the death of all cells within the affected tissue. Mitochondria are centrally involved in the development of this tissue injury due to modifications of their major role in supplying ATP and to changes in their properties that can contribute to the development of apoptotic and necrotic cell death. In animal models of stroke, the limited availability of glucose and oxygen directly impairs oxidative metabolism in severely ischemic regions of the affected tissue and leads to rapid changes in ATP and other energy-related metabolites. In the less-severely ischemic "penumbral" tissue, more moderate alterations develop in these metabolites, associated with near normal glucose use but impaired oxidative metabolism. This tissue remains potentially salvageable for at least the first few hours following stroke onset. Early restoration of blood flow can result in substantial recovery of energy-related metabolites throughout the affected tissue. However, glucose oxidation is markedly decreased due both to lower energy requirements in the post-ischemic tissue and limitations on the mitochondrial oxidation of pyruvate. A secondary deterioration of mitochondrial function subsequently develops that may contribute to progression to cell loss. Mitochondrial release of multiple apoptogenic proteins has been identified in ischemic and post-ischemic brain, mostly in neurons. Pharmacological interventions and genetic modifications in rodent models strongly implicate caspase-dependent and caspase-independent apoptosis and the mitochondrial permeability transition as important contributors to tissue damage, particularly when induced by short periods of temporary focal ischemia.

Two procedures are described for isolating free (nonsynaptosomal) mitochondria from rat brain. Both procedures employ a discontinuous Percoll gradient and yield well coupled mitochondria which exhibit high rates of respiratory activity and contain little residual contamination by synaptosomes or myelin. The procedures are considerably more rapid than methods described previously for the isolation of brain mitochondria and do not require an ultracentrifuge or swing-out rotor. The first method separates mitochondria by gradient centrifugation from a P2 (crude mitochondrial) fraction and is likely to be widely applicable for studies in which at least 500 mg of tissue are available as starting material. In the second method, the unfractionated homogenate is subjected directly to gradient centrifugation. This method requires the preparation of more gradients (per gram of tissue) than the first method and yields a subcellular fraction with slightly more synaptosomal contamination. However, this second procedure is more rapid, requires less manipulation of the tissue, and is suitable for obtaining mitochondria with well preserved metabolic characteristics from subregions of single rat brains.

We have developed procedures that combine differential centrifugation and discontinuous Percoll density gradient centrifugation to isolate mitochondria from rat forebrains and brain subregions. The use of Percoll density gradient centrifugation is central to obtaining preparations that contain little contamination with synaptosomes and myelin. Protocols are presented for three variations of this procedure that differ in their suitability for dealing with large or small samples, in the proportion of total mitochondria isolated and in the total preparation time. One variation uses digitonin to disrupt synaptosomes before mitochondrial isolation. This method is well suited for preparing mitochondria from small tissue samples, but the isolated organelles are not appropriate for all studies. Each of the procedures produces mitochondria that are well coupled and exhibit high rates of respiratory activity. The procedures require an initial setup time of 45-75 min and between 1 and 3 h for the mitochondrial isolation.