Journal of Cerebral Blood Flow & Metabolism

Using autoradiographic image-averaging strategies, we studied the relationship between local glucose utilization (LCMRglc) and blood flow (LCBF) in a highly reproducible model of transient (2-hour) middle cerebral artery occlusion (MCAO) produced in Sprague-Dawley rats by insertion of an intraluminal suture coated with poly-L-lysine. Neurobehavioral examination at 60 minutes after occlusion substantiated a high-grade deficit in all animals. In two subgroups, LCBF was measured with ¹⁴C-iodoantipyrine at either 1.5 hours of MCAO, or at 1 hour of recirculation after suture removal. In two other matched subgroups, LCMRglc was measured with ¹⁴C-2-deoxyglucose at 1.5 to 2.25 hours of MCAO, and at 0.75 to 1.5 hours of recirculation after 2 hours of MCAO. Average image data sets were generated for LCBF, LCMRglc, and the LCMRglc/LCBF ratio for each study time. Middle cerebral artery occlusion for 2 hours induced graded LCBF decrements affecting ipsilateral cortical and basal ganglionic regions. After 1 hour of recirculation, LCBF in previously ischemic neocortical regions increased by 40% to 200% above ischemic levels, but remained depressed, on average, at about 40% of control. By contrast, frank hyperemia was noted in the previously ischemic caudoputamen. Mean cortical LCBF values during MCAO correlated highly with their respective LCBF values after 1 hour of recirculation (R = 0.93), suggesting that postischemic LCBF recovery is related to the depth of ischemia. Despite focal ischemia, LCMRglc during ~2 hours of MCAO was preserved, on average, at near-normal levels; but following ~1 h of recirculation, LCMRglc became markedly depressed (on average, 55% of control in previously densely ischemic cortical regions). Regression analysis indicated that this depressed glucose utilization was determined largely by the intensity of antecedent ischemia. By pixel analysis, the ischemic core (defined as LCBF 0% to 20% of control) comprised 33% of the ischemic hemisphere, and the penumbra (LCBF 20% to 40%) accounted for 26%. The penumbra was concentrated at the coronal poles of the ischemic lesion and formed a thin shell around the central ischemic core. During 2 hours of MCAO, the LCMRglc/LCBF ratio within the ischemic penumbra was increased four-fold above normal (average, 179 umol/100 mL). In marked contrast, after ~1 h recirculation, this uncoupling had almost completely subsided. The companion study (Zhao et al., 1997) further analyzes these findings in relation to patterns of infarctive histopathology.

We conducted a pixel-based analysis of the acute hemodynamic and metabolic determinants of infarctive histopathology in a reproducible model of temporary (2-hour) middle cerebral artery occlusion (MCAO) produced in rats by an intraluminal suture. Three-dimensional averaged image data sets of local cerebral blood flow (LCBF) and glucose utilization (LCMRglc) acquired in the companion study (Belayev et al., 1997) either at the end of a 2-hour period of MCAO or after 1 hour of recirculation were comapped (using digitized atlas-templates) with data sets depicting the frequency of histological infarction in a matched animal group (n = 8) in which 2 hours of MCAO was followed by 3-day survival, sequential neurobehavioral examinations, and perfusion-fixation and paraffin-embedding of brains for light-microscopic analysis. All rats developed marked postural-reflex and forelimb-placing deficits at 60 minutes of MCAO, signifying high-grade ischemia. Tactile placing deficits persisted during the 72-hour observation period while visual placing and postural-reflex abnormalities variably improved. Comapping of LCBF and histopathology showed that in those pixels destined to undergo infarction, LCBF measured at 2 hours of MCAO showed a sharp distributional peak centered at 0.14 mL/g/min. In 70% of pixels destined to infarct, LCBF at 2 hours of MCAO was 0.24 mL/g/min or below, and in 89% LCBF was below 0.47 mL/g/min (the upper limits of the ischemic core and penumbra, respectively, as defined in the companion study [Belayev et al., 1997]). Local cerebral glucose utilization measured at ~1 hour after 2 hours of MCAO was distributed bimodally in the previously ischemic hemisphere. The major peak, at 22 μmol/100 g/min, coincided exactly with the distribution peak of pixels destined to undergo infarction, while in pixels with a zero probability of infarction, LCMRglc was higher by 12 to 13 μmol/100 g/min. These results indicate that local blood flow at 2 hours of MCAO is a robust predictor of eventual infarction. Pixels with ischemic-core levels of LCBF (0% to 20% of control) have a 96% probability of infarction, while the fate of the penumbra is more heterogeneous: below LCBF of 0.35 mL/g/min, the probability of infarction is 92%, while approximately 20% pixels in the upper-penumbral LCBF range (30% to 40% of control) escape infarction. Our data strongly support the view that the likelihood of infarction within the ischemic penumbra is highly influenced by very subtle differences in early perfusion.

Postischemic brain reperfusion is associated with a substantial and long-lasting reduction of protein synthesis in selectively vulnerable neurons. Because the overall translation initiation rate is typically regulated by altering the phosphorylation of serine 51 on the α-subunit of eukaryotic initiation factor 2 (eIF-2α), we used an antibody specific to phosphorylated eIF-2α [eIF-2(αP)] to study the regional and cellular distribution of eIF-2(αP) in normal, ischemic, and reperfused rat brains. Western blots of brain postmitochondrial supernatants revealed that ~1% of all eIF-2α is phosphorylated in controls, eIF-2(αP) is not reduced by up to 30 minutes of ischemia, and eIF-2(αP) is increased ~20-fold after 10 and 90 minutes of reperfusion. Immunohistochemistry shows localization of eIF-2(αP) to astrocytes in normal brains, a massive increase in eIF-2(αP) in the cytoplasm of neurons within the first 10 minutes of reperfusion, accumulation of eIF-2(αP) in the nuclei of selectively vulnerable neurons after 1 hour of reperfusion, and morphology suggesting pyknosis or apoptosis in neuronal nuclei that continue to display eIF-2(αP) after 4 hours of reperfusion. These observations, together with the fact that eIF-2(αP) inhibits translation initiation, make a compelling case that eIF-2(αP) is responsible for reperfusion-induced inhibition of protein synthesis in vulnerable neurons.

Preischemic hyperglycemia or superimposed hypercapnia exaggerates brain damage caused by transient forebrain ischemia. Because high regional levels of brain-derived neurotrophic factor (BDNF) protein correlate with resistance to ischemic damage, we studied the expression of BDNF mRNA using in situ hybridization in rats subjected to 10 minutes of forebrain ischemia under normoglycemic, hyperglycemic, or hypercapnic conditions. Compared with normoglycemic animals, the increase of BDNF mRNA in dentate granule cells was attenuated and that in CA3 pyramidal neurons completely prevented in hyperglycemic rats. No ischemia-induced increases of BDNF mRNA levels in the hippocampal formation were detected in hypercapnic animals. Hyperglycemic and hypercapnic rats showed transiently decreased expression of BDNF mRNA levels in the cingulate cortex, which was not observed in normoglycemic animals. The results suggest that suppression of the BDNF gene might contribute to the increased vulnerability of the CA3 region and cingulate cortex in hyperglycemic and hypercapnic animals.

The possibility that adenosine and ATP-sensitive potassium channels (K_ATP) might be involved in the mechanisms of the increases in cerebral blood flow (CBF) that occur in insulin-induced hypoglycemia was examined. Cerebral blood flow was measured by the [¹⁴C]iodoantipyrine method in conscious rats during insulin-induced, moderate hypoglycemia (2 to 3 mmol/L glucose in arterial plasma) after intravenous injections of 10 to 20 mg/kg of caffeine, an adenosine receptor antagonist, or intracisternal infusion of 1 to 2 μmol/L glibenclamide, a K_ATP channel inhibitor. Cerebral blood flow was also measured in corresponding normoglycemic and drug-free control groups. Cerebral blood flow was 51% higher in untreated hypoglycemic than in untreated normoglycemic rats (P < 0.01). Caffeine had a small, statistically insignificant effect on CBF in normoglycemic rats, but reduced the CBF response to hypoglycemia in a dose-dependent manner, i.e., 27% increase with 10 mg/kg and complete elimination with 20 mg/kg. Chemical determinations by HPLC in extracts of freeze-blown brains showed significant increases in the levels of adenosine and its degradation products, inosine and hypoxanthine, during hypoglycemia (P < 0.05). Intracisternal glibenclamide had little effect on CBF in normoglycemia, but, like caffeine, produced dose-dependent reductions in the magnitude of the increases in CBF during hypoglycemia, i.e., +66% with glibenclamide-free artificial CSF administration, +25% with 1 μmol/L glibenclamide, and almost complete blockade (+5%) with 2 μmol/L glibenclamide. These results suggest that adenosine and K_ATP channels may play a role in the increases in CBF during hypoglycemia.

Hypoxemia and anemia are associated with increased CBF, but the mechanisms that link the changes in Pao₂ or arterial O₂ content (Cao₂) with CBF are unclear. These experiments were intended to examine the contribution of nitric oxide. Cao₂ in pentobarbital-anesthetized rabbits was reduced to approximately 6.5 mL O₂/dL by hypoxemia (Pao₂ approximately 24 to 26 mm Hg) or hemodilution with hetastarch (hematocrit approximately 14% to 15%). Animals with normal Cao₂ (approximately 17.5 to 18 mL O₂/dL) served as controls. In part I, each animal was given 3, 10, and 30 mg/kg Nω-nitro-l-arginine methyl ester (l-NAME) intravenously (total 43 mg/kg) to inhibit production of nitric oxide. Forebrain CBF was measured with radioactive microspheres approximately 15 to 20 minutes after each dose. Baseline CBF was greater in hypoxemic rabbits (111 ± 31 mL·100 g⁻¹·min⁻¹, mean ± SD) than in hemodiluted (70 ± 22 mL·100 g⁻¹·min⁻¹) or control animals (39 ± 12 mL·100 g⁻¹·min⁻¹). l-NAME (which reduced brain tissue nitric oxide synthase activity by approximately 65%) reduced CBF in hypoxemic animals to 80 ± 23 mL·100 g⁻¹·min⁻¹ (P < 0.0001), but had no significant effect on CBF in either anemic or control animals. In four additional rabbits, further hemodilution to a Cao₂ of approximately 3.5 mL O₂/dL increased baseline CBF to 126 ± 21 mL·100 g⁻¹·min⁻¹, but again there was no effect of l-NAME. In part II, animals were anesthetized as above, and a closed cranial window was prepared. The cyclic GMP (cGMP) content of the artificial CSF superfusate was measured under baseline conditions, and then after the reduction of Cao₂ to approximately 6.5 mL O₂/dL by either hypoxemia or hemodilution. Concentrations of cGMP did not change during either control conditions or after hemodilution. However, cGMP increased significantly with the induction of hypoxemia. The cGMP increase in hypoxemic animals could be blocked with l-NAME. These results suggest that nitric oxide plays some role in hypoxemic vasodilation, but not during hemodilution.

Laser-Doppler flowmetry (LDF) is a reliable method for estimation of relative changes of CBF. The measurement depth depends on wavelength of the laser light and the separation distance of transmitting and recording optical fibers. We designed an LDF probe using two wavelengths of laser light (543 nm and 780 nm), and three separation distances of optical fibers to measure CBF in four layers of the cerebral cortex at the same time. In vitro comparison with electromagnetic flow measurements showed linear relationship between LDF and blood flow velocity at four depths within the range relevant to physiologic measurements. Using artificial brain tissue slices we showed that the signal for each channel decreased in a theoretically predictable fashion as a function of slice thickness. Application of adenosine at various depths in neocortex of halothane-anesthetized rats showed a predominant CBF increase at the level of application. Electrical stimulation at the surface of the cerebellar cortex demonstrated superficial predominance of increased CBF as predicted from the distribution of neuronal activity. In the cerebellum, hypercapnia increased CBF in a heterogeneous fashion, the major increase being at apparent depths of approximately 300 and 600 μm, whereas in the cerebral cortex, hypercapnia induced a uniform increase. In contrast, the CBF response to cortical spreading depression in the cerebral cortex was markedly heterogeneous. Thus, real-time laminar analysis of CBF with spatial resolution of 200 to 300 μm may be achieved by LDF. The real-time in depth resolution may give insight into the functional organization of the cortical microcirculation and adaptive features of CBF regulation in response to physiologic and pathophysiologic stimuli.

The effect of interleukin-1β (IL-1β) on a cerebral artery was investigated in anesthetized dogs. Intracisternal administration of IL-1β (0.03 and 0.3 μg) dilated the canine basilar artery in a dose-dependent manner, without affecting systemic blood pressure or heart rate. The increase in diameter induced by 0.3 μg of IL-1β was 28.4% ± 13.4% of control at 2 hours and was inhibited by 30 μg of the IL-1β receptor antagonist, zinc protoporphyrin (4.5% ± 13.5%, P < 0.05). Interleukin-1β did not affect the concentration of nitric oxide metabolites in CSF. However, there was an increase in the concentration of eicosanoids in CSF, and the elevation of 6-keto-PGF1_α paralleled the vasodilation. Pretreatment with 30 μg of the selective inducible cyclooxygenase (COX-2) inhibitor NS-398 also inhibited the IL-1β-induced vasodilation significantly (5.9% ± 9.4% at 2 hours, P < 0.01). Western blot analysis revealed the expression of a 68-kD COX-2-like protein in basilar artery extracts. These findings suggest that the IL-1β-induced vasodilator effect is linked to the prostaglandin cascade, predominantly to prostaglandin I₂, by induction of COX-2, but not to the stimulation of nitric oxide metabolism.