Stretch-Induced Injury of Cultured Neuronal, Glial, and Endothelial Cells
Effect of Polyethylene Glycol–Conjugated Superoxide Dismutase
Background and Purpose There is abundant evidence that after in vivo traumatic brain injury, oxygen radicals contribute to changes in cerebrovascular structure and function; however, the cellular source of these oxygen radicals is not clear. The purpose of these experiments was to use a newly developed in vitro tissue culture model to elucidate the effect of strain, or stretch, on neuronal, glial, and endothelial cells and to determine the effect of the free radical scavenger polyethylene glycol–conjugated superoxide dismutase (PEG-SOD; pegorgotein, Dismutec) on the response of each cell type to trauma.
Methods Rat brain astrocytes, neuronal plus glial cells, and aortic endothelial cells were grown in cell culture wells with 2-mm-thick silastic membrane bottoms. A controllable, 50-millisecond pressure pulse was used to transiently deform the silastic membrane and thus stretch the cells. Injury was assessed by quantifying the number of cells that took up the normally cell-impermeable dye propidium iodide. Some cultures were pretreated with 100 to 300 U/mL PEG-SOD.
Results Increasing degrees of deformation produced increased cell injury in astrocytes, neuronal plus glial cultures, and aortic endothelial cells. By 24 hours after injury, all cultures showed evidence of repair as demonstrated by cells regaining their capacity to exclude propidium iodide. Compared with astrocytes or neuronal plus glial cultures, endothelial cells were much more resistant to stretch-induced injury and more quickly regained their capacity to exclude propidium iodide. PEG-SOD had no effect on the neuronal or glial response to injury but reduced immediate posttraumatic endothelial cell dye uptake by 51%.
Conclusions These studies further document the utility of the model for studying cell injury and repair and further support the vascular endothelial cell as a site of free radical generation and radical-mediated injury. On the assumption that, like aortic endothelial cells, stretch-injured cerebral endothelial cells also produce oxygen radicals, our results further suggest the endothelial cell as a site of therapeutic action of free radical scavengers after traumatic brain injury.
There is abundant evidence that after in vivo traumatic, hypertensive, or ischemic injury, oxygen radicals contribute to observed changes in cerebrovascular reactivity,1 2 3 cerebral blood flow,4 blood-brain barrier permeability,5 6 infarct size,7 brain edema,8 and lipid hydroperoxides.9 Pharmacological therapy with exogenous SOD, catalase, deferoxamine, and novel antioxidant free radical scavengers such as Lazaroids10 has been effective in reducing the consequences of injury. The role of free radicals in brain injury has been even further supported by the demonstration that transgenic animals with increased endogenous levels of SOD have reduced deficits after brain injury.11 12
Despite the extensive evidence that free radical scavenging after brain injury reduces deficits, the exact cellular sites of action of these agents remain uncertain. With respect to experimental fluid percussion brain injury, increasing evidence supports a vascular site of action.13 Overall, however, there is little information as to the role of free radicals in traumatic injury of various brain cell types. Elucidation of the site or sites of action of free radical scavengers may help to target therapy with novel pharmacological approaches. The purpose of these studies was to use a new model of stretch-induced injury of cells in culture to probe the possible active site, or sites, of free radical scavengers in ameliorating the cascade of injurious events initiated by trauma. We have examined the response of mixed neuronal plus glial cells, astrocytes, and aortic endothelial cells to stretch-induced injury and determined the effect of PEG-SOD (pegorgotein) on injury. Our results further implicate endothelial cells as the site of free radical generation and therapeutic efficacy of oxygen radical scavengers.
Materials and Methods
Cells were cultured on Flex Plates (Flexcell International) with 2-mm-thick, collagen-coated silastic membrane bottoms, as previously described in detail.14 On gaining confluence, the cells were injured using a model 94A cell injury controller (Commonwealth Biotechnology Inc, Richmond, Va). With the cell injury controller and its attachments, the flex plate membranes were deformed by a 50-millisecond pulse of gas to produce various magnitudes of membrane deformation and cell injury. The cell injury controller produces a rapid, reversible deformation of the silastic membrane and stretches the cells growing on the membrane. Injury is assessed by nuclear uptake and binding of PrI, a fluorescent dye that does not penetrate cells with normal (unstretched) membranes. In our previous studies, the cells taking up PrI were directly counted visually with a fluorescence microscope. However, in the present studies we used an imaging system (Image Explorer, Signal Analytics) to count the number of cells that contained PrI. The injury was expressed as the number of cells binding dye per milligram of cell protein, the latter being measured after cell counting.
Astrocytes were prepared as we have previously reported.15 Briefly, cortices were isolated from 1- to 2-day-old rats, cleaned of white matter and meninges, minced, and trypsin-digested for 10 minutes. Next, the tissue was diluted into feeding medium, washed, triturated, counted, and placed into 75-cm2 flasks (1 to 2×106 cells per flask) for 10 to 14 days. When confluent, the cells were plated into tissue culture wells with silastic bottoms. To accomplish this, confluent cells were treated with trypsin/EDTA (0.25% trypsin/0.02% EDTA in saline) for exactly 1 minute. The trypsin/EDTA was then aspirated, and the cell monolayer, which was still attached to the flask, was covered with 10 mL DMEM and incubated at 37°C for approximately 10 minutes. The lifted cells were then transferred to a plastic tube, washed, centrifuged, and counted, and 2×105 cells in 1 mL DMEM containing 10% FBS were plated onto collagen-coated 25-mm-diameter silastic membranes, which are 2 mm thick and form the bottom surface of each well in a six-well tissue culture flex plate. The cells were characterized for purity as previously15 and found to be >98% pure. The cells were used for experiments at a total of 4 weeks after removal from the rat.
Neuronal Plus Glial Cells
Neuronal plus glial cell cultures were prepared from 1- to 2-day-old maximum barrier–maintained viral antibody–free Sprague-Dawley rats (Zivic Miller, Allison Park, Pa). After decapitation, whole brains were removed aseptically and placed in sterile dissecting saline (33 mmol/L glucose, 44 mmol/L sucrose, 137 mmol/L NaCl, 5.3 mmol/L KCl, 0.17 mmol/L Na2HPO4 · 7H2O, 0.22 mmol/L KH2PO4, 10 mmol/L HEPES, 0.0012 g/L phenol red, adjusted to pH 7.3 and 325 mOsm). Cerebral hemispheres, olfactory bulbs, hippocampal formations, basal ganglia, and meninges were removed, thus leaving the neopallium. The neopallia were transferred into 5 mL of dissecting saline containing 0.125% porcine-derived trypsin (Sigma Chemical Co) and cut with sterile scalpels into small cubes (≈1 mm3). The tissue was kept in trypsin for 10 minutes and was then transferred to a tube containing 5 mL of culture medium. The culture medium was DMEM containing glucose (4.5 g/L) supplemented with 10% FBS, Pen/Strep (penicillin 100 U/mL, streptomycin 100 μg/mL), and 2 mmol/L l-glutamine. The tissue was washed by dispersing it up and down with a 10-mL sterile plastic pipette. The supernatant was removed, and 5 mL of culture medium was added for a second wash. The supernatant was discarded, and 5 mL of culture medium was added. The tissue fragments were dissociated by trituration with a 9-in-long, cotton-plugged, sterile glass Pasteur pipette until no large pieces were apparent. Second and third triturations were done with flame-narrowed glass pipettes with successively smaller bores. The suspension was centrifuged for 10 minutes at 200g. The supernatant was removed, and 5 mL of culture medium was added. The suspension was triturated twice as described above. The suspension was filtered with a 70-μm nylon cell strainer (Falcon), diluted with DMEM, and counted with a hemocytometer. Aliquots of the cell suspension (1 mL, containing ≈1×106 cells) were seeded into each well of a collagen-coated flex plate. The cells were incubated at 37°C in a 95%/5% mixture of air and CO2. After 6 to 7 days, half of the medium was removed from each well, and a neuronal growth medium containing 1× MEM (Gibco), 10 mmol/L glucose, Pen/Strep (100 U/mL:100 μg/mL), and 5% horse serum (Gibco) was added. The cells were used within 10 to 14 days of culturing. The cultures consisted of a mixed population of neuronal and glial cells. Glial cells typically adhered to the membrane substrate, and the neuronal cells, often growing in clusters, adhered to the glial cells. Neuronal processes and interconnections between the various neuronal cell clusters were readily apparent.
Rabbit aortic endothelial cells. Endothelial cells were prepared according to the method of Santilli et al,16 with some modifications. Young adult male New Zealand White rabbits (2.5 kg) were anesthetized with sodium pentobarbital; thoracic aortas were removed to a sterile dish. The aortas were washed twice with HBSS (Gibco), the adventitia was removed, and the aortas were washed again and then opened longitudinally. Next, the aortas were immersed in 0.1% collagenase (Sigma) in HBSS for 20 minutes, and the intima was gently scraped with a sterile scalpel. The artery wall tissue was discarded, and the enzyme solution and endothelial tissue were passed through a 100-μm nylon filter and collected into an equal amount of M199 medium containing 10% FBS. Endothelial cells were centrifuged at 200g for 10 minutes, the supernatant was removed, and the cells were resuspended in M199 and 10% FBS and centrifuged again. The supernatant was removed, and the endothelial cells were seeded with endothelial medium (M199, Gibco) containing 25 mmol/L HEPES and supplemented with gentamicin (Gibco, 50 μg/mL), fungizone (Gibco, 2.5 μg/mL), heparin (Sigma, 100 μg/mL), endothelial cell growth supplement (Sigma, from bovine neural tissue, 20 μg/mL), glutamine (Gibco, 2 mmol/L), and FBS (Hyclone, 20% vol/vol). The maintenance medium contained only 10% FBS. The cells were plated onto fibronectin-coated (Gibco, 2 μg/cm2) four-well plates (Falcon). When the cells became confluent, they were washed twice with HBSS without Ca2+ or Mg2+ and detached with 0.025% trypsin in 0.01% EDTA (a 1:1 ratio of 0.02% EDTA solution in physiological saline and HBSS without Ca2+ or Mg2+) at 37°C for approximately 10 minutes. When the cells lifted, the suspension was diluted with an equal volume of M199 medium plus 10% FBS, centrifuged at 150g for 5 minutes, resuspended in plating medium, and transferred to fibronectin-coated 35-mm or 60-mm tissue culture dishes at a split ratio of 1:5. At confluence, the cells were removed as above and plated onto collagen-coated flex plates at a density of 2×105 cells/mL. The cells were usually used at passage 2 at approximately 3 to 4 weeks after removal from the aorta.
Bovine aortic endothelial cells. Four adult bovine thoracic aortas (Dinner Bell Meat Products) were placed into a pan containing HBSS without Ca2+ or Mg2+, fungizone (2.5 μg/mL), and gentamicin (50 μg/mL) and cut into approximately 3-in-square pieces. As each piece was cut, it was transferred to sterile saline with fungizone and gentamicin. Next, each piece of aorta was placed in a sterile 100-mm dish, and the endothelial surface was scraped with a sterile scalpel. The scraped cells were placed into another 100-mm dish containing 10 mL M199 and 10% FBS. After all pieces were scraped, the solution was centrifuged at 100g for 5 minutes. The supernatant was aspirated, and the pellet was gently resuspended in 16 mL of plating medium and cultured in the same way as the rabbit aortic endothelial cells (see above).
The identity of the rabbit and bovine endothelial cells was confirmed by staining with anti-human von Willebrand factor (Sigma) or by endothelial cell uptake of acetylated low-density-lipoprotein (BODIPY Fl conjugate, Molecular Probes). Monoclonal anti-myosin (Sigma) was used to test for the presence of contaminating fibroblasts. We used only endothelial cell preparations that appeared to be at least 90% endothelial cells.
Protocols and Analysis
The first objective for each type of culture was to examine the effect of the degree of deformation and time after injury on cell uptake of PrI. Cells were injured at increasing intensities corresponding to 5.7-, 6.3-, 7.5-, 8.6-, 9.5-, 11.4-, or 13-mm deformation. This translates into 31%, 38%, 54%, 72%, 89%, 130%, or 170% stretch, respectively. Not all cell types were injured with all the above intensities. Mixed cultures and astrocytes were injured with the four lowest deformations; endothelial cells, for reasons outlined in “Results,” were injured with 7.5-, 9.5-, 11.4-, and 13-mm silastic membrane deformations. We studied the cell response at various times after injury, as in our previous studies in astrocytes. Cells were labeled with PrI for 10 minutes immediately after injury or returned to the incubator and examined with dye added at 2, 6, or 24 hours after injury. This allowed us to determine whether injury stayed the same or worsened or cells repaired with time after injury.
Once the basic response to injury was established, we examined the effect of pretreatment with PEG-SOD (pegorgotein, Dismutec), generously donated by Sanofi Winthrop Inc. PEG-SOD, unlike native SOD, is resistant to enzymatic breakdown and has a half-life of 4 days. Therefore, SOD enzyme activity was maintained over the duration of our experimental observations. The cell cultures had 100 to 300 U pegorgotein added to each milliliter of culture medium 30 minutes before stretch-induced injury. The pegorgotein and normal culture medium were kept in the wells throughout the period of experimental observation. The SOD concentrations used in these experiments were equal to or in excess of those concentrations that we have shown to be effective when applied topically to the brain in vivo to prevent cerebrovascular abnormalities produced by fluid percussion brain injury1 or to prevent oxygen radical–induced arteriolar dilation in response to topical administration of bradykinin.17 The amount of protein contributed to each milliliter of medium by addition of 300 U/mL PEG-SOD was 0.143 mg. The protein content of the normal FBS-containing medium was 3.99 mg/mL. Therefore, the protein contributed to the experimental medium by the PEG-SOD was only 3.4% of the total protein content.
With neuronal plus glial cell and astrocyte cultures, the effect of PEG-SOD was examined immediately after injury and at 2 and 24 hours after injury. With endothelial cells, the effect of PEG-SOD was examined only immediately after injury because the cells spontaneously began to regain their capacity to exclude dye within 2 hours of injury. Data were examined with one-factor ANOVA followed by Fisher’s test for least significant difference, using the Superanova data-testing module (Abacus Concepts).
Neuronal Plus Glial Cells
As previously reported for astrocytes,14 the neuronal plus glial cell cultures remained adherent to the silastic membrane after injury. Light microscopic examination of the cells showed astroglial retraction, also as previously reported. After stretch-induced injury, axonal processes appeared loose instead of having their normal tight, cabled appearance. Using fluorescence microscopy and PrI uptake, we were unable to distinguish between neuronal versus other cell types. The pattern of dye uptake, however, was not exactly identical to that previously seen in pure astrocytes. Instead of all injured nuclei appearing oval and similarly sized, there was more diversity in the size and shape of the nuclei, as might be expected with mixed cultures. With fluorescence, the general location of the neuronal cells could be approximated by the fact that the neuronal cell bodies tend to grow in clusters, and therefore their injured nuclei also were clustered.
As shown in Fig 1⇓, neuronal plus glial cells showed uptake of PrI that increased with the severity of membrane deformation. When cells were examined at 2, 6, and 24 hours after the 5.7-, 6.3-, or 7.5-mm deformation, there was a trend for decreased dye uptake with increasing time after injury. With the most severe injury, there was a significant (P<.05) decrease in dye uptake at 6 and 24 hours after injury, indicating cell repair.
In separate experiments, we examined the effect of pretreatment with PEG-SOD on dye uptake immediately and 2 and 24 hours after injury with a moderate 6.3-mm deformation. This degree of injury was chosen so that any effects of PEG-SOD, to either worsen or lessen injury, might be detected. As shown in Fig 2⇓, PEG-SOD had no consistent effect on injury of neuronal plus glial cells, as assessed by uptake of PrI.
Fig 3⇓ shows the previously reported effect of degree of stretch and time after injury on astrocyte uptake of PrI.14 Similar to neuronal plus glial cells, astrocytes displayed the same stretch-dependent uptake of dye, with significantly reduced (P<.05) dye uptake at 6 and 24 hours after injury induced by a 7.5- and 8.6-mm deformation. Also, the absolute numbers of cells per milligram of protein that displayed dye uptake were similar to those observed in the mixed neuronal plus glial cell cultures.
After injury of the various cultures using the present experimental parameters, we did not detect any readily visible major detachment of the cells from the silastic membrane. However, to more rigorously examine this possibility, we performed separate experiments on three groups of confluent astrocytes (n=8 per group) injured with a 7.5-mm membrane deformation. We counted attached astrocytes in uninjured controls and at 2 and 24 hours after injury. Cells attached to the membrane were removed, dispersed, and counted by light microscopy. The total numbers of cells per culture well were 367 604±13 429, 315 703±12 421, and 347 422±117 986 (x̅±SEM) in uninjured and in 2-hour and 24-hour postinjury cultures, respectively. Thus the cell count was decreased by only 14% (P<.05) and only at the 2-hour postinjury assessment. These experiments imply a relatively minor cell detachment and cell division after injury.
In additional experiments, we used the 6.3-mm deformation to examine the effect of PEG-SOD on astrocyte injury. Fig 4⇓ shows the same trend for decreased cell dye uptake with increasing time after injury and that PEG-SOD had no consistent or statistically significant effect on astrocyte PrI uptake.
Endothelial cell culture was much more difficult and problematic than neuronal or astrocytic culture. Several pilot experiments (data not shown) with cerebral endothelial cells from rats produced a stretch-dependent injury similar to that shown for aortic endothelial cells in Fig 5⇓. However, we were unable to reliably produce primary pure cerebral endothelial cell cultures that met our purity criteria; we therefore turned to aortic tissue as a source of endothelial cells. Bovine aortic tissue was used for our studies of the effect of the degree of stretch and time after injury on PrI uptake.
As shown in Fig 5⇑, there were some striking differences in the way endothelial cells responded to stretch compared with astrocytes or neuronal plus glial cells. Endothelial cells were much more resistant to stretch-induced injury, and greater deformations were needed to produce the same amount of injury as in astrocytes and mixed cultures. The 7.5-, 9.5-, 11.4-, and 13-mm deformations represent a 54%, 89%, 130%, and 180% stretch or increase in diameter of the silastic membrane. The second obvious difference with endothelial cells was that after injury their repair and consequent exclusion of dye occurred much sooner after injury. Cells injured with 11.4 and 13 mm of deformation regained a major amount of their capacity to exclude dye by only 2 hours after injury.
We next examined the effect of PEG-SOD on the response of endothelial cells to stretch. These experiments were performed on rabbit aortic endothelial cells. These experiments were conducted on rabbit tissue because our local supplier of fresh bovine aortic tissue became unreliable. We therefore switched to rabbits because they were commercially available and are routinely used for studies of cerebrovascular reactivity. Fortunately, the rabbit aortic endothelial cells response to injury was similar to that of bovine aortic endothelial cells (see Fig 6⇓). The 7.5-mm deformation of rabbit aortic endothelial cell cultures produced generally similar numbers of injured cells as the same insult to bovine aortic endothelial cells. We used the 7.5-mm deformation to study the effect of PEG-SOD on injury because it produced approximately the same number of injured cells as the 6.3-mm deformation of astrocytes or mixed neuronal plus glial cells. Because endothelial cells spontaneously regained their capacity to exclude dye by 2 hours after injury, our observations of the effect of PEG-SOD on endothelial cells were limited to the period immediately after injury. As can be seen in Fig 6⇓, PEG-SOD reduced dye uptake by 51% (P<.06, unpaired t test) immediately after injury. This implies that endothelial cell–produced free radicals contribute to injury immediately after injury.
All cultures showed deformation-dependent injury with subsequent repair. Astrocytes and mixed cultures behaved similarly in response to injury, which is not surprising since mixed neuronal plus glial cultures contain large numbers of astrocytes. With the present methods, we were not technically able to discern between individual astrocytes or neurons with respect to injury or repair in mixed cultures. However, since the time course for repair was similar in astrocytes and mixed cultures, this implies neuronal repair had a similar time course to that of astrocytes and other glial cells.
In other ongoing studies, we have used the patch-clamp technique to examine neuronal resting membrane potential in mixed cultures injured with a 5.7-mm deformation.18 A decrease in resting membrane potential occurs by 1 hour after injury, and the resting membrane potential returns to normal by 24 hours after injury. This physiological parameter also implies that a transient injury occurs in neurons and correlates with our PrI data, indicating that the ability to exclude PrI is regained.
PEG-SOD did not affect astrocyte or neuronal plus glial cell stretch-induced injury with use of the present experimental parameters. This does not rule out, however, that free radicals may be important in other types or models of injury to astrocytes or neurons. For example, exogenous arachidonic acid has been shown to induce free radical production in astrocytes.19 Whether this arachidonic acid–induced free radical generation in astrocytes would result in alteration of PrI uptake is uncertain. It should be emphasized that the cascade of intracellular events induced purely by the addition of exogenous chemicals alone is likely to be different from that induced by traumatic injury. In support of an effect of the endogenous content of arachidonic acid in an astrocyte response to stretch-induced injury is our recent finding that supplementation and greater incorporation of arachidonic acid into astrocyte membranes is associated with a greater postinjury uptake of [3H]choline, an indicator of membrane injury and repair.20 This same increased uptake of [3H]choline in arachidonic acid–supplemented astrocytes can be prevented by addition of free radical scavengers, including SOD.20
Endothelial cells were also injured in a manner proportional to the degree of stretch. However, much greater magnitudes of stretch were necessary to induce comparable injury. This may not be unexpected when one considers that endothelial cells normally exist in an environment where pulsatile flow and local regulation of vascular tone can cause large changes in arterial diameter. For example, endothelial cells in precapillary arterioles undergo great changes in tensile strain as vessels go from maximal dilation to maximal constriction and perhaps near stoppage of flow, as seen with rhythmic vasomotion. If one assumes that an arteriole undergoes 100% dilation from normal resting tone, which is physiologically possible, the circumference around which the endothelial cell must stretch would also increase by 100%, since circumference is equal to 2π times the radius. Although large changes in endothelial cell length may occur normally in vivo, it is unlikely that under normal conditions such changes occur in the brief fraction of a second that was used in our experiments to injure the endothelial cells. However, the pathophysiological phenomenon of “breakthrough” of cerebral blood flow autoregulation occurs in brain. In this condition, arterioles are no longer capable of constricting adequately to keep blood flow constant in the face of increased arterial blood pressure. The possibility exists that when breakthrough occurs vessels no longer able to contract against rising pressure may suddenly be severely stretched. Under such conditions, rapid and severe endothelial cell injury, such as observed in the present experiments, may occur. In fact, our previous studies show that breakthrough of autoregulation, as induced by a bolus injection of norepinephrine, is shifted to higher pressures by treatment of rats with SOD.21 This suggests that excessive oxygen radical production in response to vascular strain may hasten breakthrough.
Cultured endothelial cells are known to be producers of superoxide and hydroxyl radicals, and this free radical production can be increased by challenges such as anoxia followed by reoxygenation.22 23 24 Some investigators, such as Rosen and Freeman,23 have used electron paramagnetic resonance to measure oxygen radical production in endothelial cell suspensions. Their technical approach is beyond the scope of the present investigation but may be used in future investigations of oxygen radical production by stretch-injured endothelial cells. In our experiments, PEG-SOD reduced endothelial cell injury in the period immediately after stretch-induced injury. This implies that endothelial cell free radical–generating mechanisms are immediately activated after trauma. We must caution, however, that the presently reported experiments were performed in rabbit aortic endothelial cells, and thus the results may not completely apply to cerebral endothelial cells. However, our previous studies of fluid percussion brain injury have also suggested immediate posttraumatic free radical generation. In cats subjected to fluid percussion brain injury, arterioles immediately become dilated and abnormally responsive to CO2 and acetylcholine.25 This abnormal reactivity can be prevented by topical application of SOD plus catalase. Also, the cerebral endothelial cells in these animals display SOD-inhibitable lesions. These arteriolar abnormalities can be similarly induced by sudden drug-induced elevations in mean arterial blood pressure,26 thus suggesting that vessels are injured not only by a pulse of increased intracranial pressure produced directly by the fluid percussion device but also by the rapid onset of catecholamine-induced hypertension that is induced by fluid percussion brain injury.
Recent experiments with fluid percussion brain injury in rats also suggest immediate free radical generation after trauma.4 In these experiments, we measured cortical cerebral blood flow changes with laser-Doppler flowmetry and found that flow deficits observed immediately after traumatic brain injury in rats were reduced by pretreating rats with an intravenous infusion of SOD.
In the present experiments, endothelial cell repair was evident by 2 hours after injury. Since 2 hours was the earliest measured PrI uptake after the initial observation immediately after injury, we are unable to say whether repair is apparent by time points such as 1 hour after injury. Interestingly, our above-mentioned studies of cerebral perfusion in traumatically injured rats also documented a highly transient SOD-inhibitable cerebral blood flow deficit.4 In these studies, we found that cerebral blood flow in rats not treated with SOD returned to near control levels by 1 hour after trauma. This transient nature of cerebral blood flow reduction after traumatic brain injury in rats has also been observed by others.27 28 We speculate that the transient SOD-inhibitable blood flow deficits may be due to transient dysfunction of cerebral endothelial cells.
Interestingly, Rosenblum et al29 recently reported that when mouse pial arterioles are injured with a laser-dye technique they lose their normal responsivity to acetylcholine and bradykinin. However, normal reactivity returns by 1 hour after injury, and this return of reactivity can be prevented by inhibitors of protein synthesis. These studies reinforce the rapidity with which endothelial cell repair occurs and suggest that recovery of stretch-injured cells may also be inhibited by protein synthesis inhibitors.
Several studies have shown that the increased cerebrovascular permeability or brain edema that occurs after ischemic, hypertensive, or cold-induced injury5 6 30 is reduced by SOD or PEG-SOD. We and others have speculated as to the site of radical scavenging that reduces the increase in blood-brain barrier permeability associated with these types of injury. Perhaps the most frequently speculated site of free radical attack has been the vasculature. Obviously, free radical injury of the vascular endothelial cells could increase vascular permeability and promote brain edema. Our present study, wherein endothelial cell permeability to PrI is reduced by PEG-SOD, further supports the likely role of endothelial cells as targets for posttraumatic SOD inhibition of free radical–induced increases in vascular permeability. Our results show that endothelial cells are not only a source of free radicals after stretch-induced injury but also the victims of free radical attack.
Although we used stretch to induce endothelial cell free radical production, other pathological challenges to endothelial cells may also be able to induce oxygen radical production. For example, anoxia followed by reoxygenation is a potent stimulator of superoxide and hydroxyl free radicals in cultured endothelial cells.22 24 Finally, it should be realized that endothelial cell free radical production under normal conditions may contribute to normal regulation of vascular function. In this respect, Laurindo et al31 have provided evidence that endothelial cells demonstrate flow-induced free radical production. Also, our previous studies of vascular permeability in normal rats suggested that SOD may alter vascular permeability.6 Thus, vascular production of oxygen radicals may be important for normal physiology, whereas vascular overproduction of radicals during pathological stresses may have negative consequences.
In summary, we have shown that cell strain, an important component of traumatic brain injury, causes damage and increased cell permeability that is proportional to the degree of strain. Astrocytes and neuronal plus glial cultures are similarly responsive to injury, whereas endothelial cells are more resistant. Astrocytes, neuronal plus glial cultures, and endothelial cultures are all capable of repair; however, endothelial cells do so much more rapidly. Additionally, endothelial cells have an initial component of injury that appears to be oxygen radical–mediated. Our studies further elucidate the role of oxygen radicals in the sequelae of traumatic injury and sudden hypertension-induced cerebrovascular injury and further demonstrate the utility of the in vitro model of stretch-induced injury.
Selected Abbreviations and Acronyms
|DMEM||=||Dulbecco’s modified essential medium|
|FBS||=||fetal bovine serum|
|HBSS||=||Hanks’ balanced salt solution|
|PEG-SOD||=||polyethylene glycol–conjugated superoxide dismutase|
This study was supported by National Institutes of Health grants NS-27214 and a Center of Excellence grant-in-aid from the Commonwealth of Virginia (J. Povlishock). E. Ellis is a Jacob Javits Neuroscience Investigator. We are grateful to Sallie Holt for her excellent assistance.
Review of this manuscript was directed by Donald D. Heistad, MD.
- Received October 24, 1995.
- Revision received January 19, 1996.
- Accepted January 19, 1996.
- Copyright © 1996 by American Heart Association
Wei EP, Kontos HA, Dietrich WD, Povlishock JT, Ellis EF. Inhibition by free radical scavengers and by cyclooxygenase inhibitors of pial arteriolar abnormalities from concussive brain injury in cats. Circ Res. 1981;48:95-103.
Schleien CL, Eberle B, Shaffner DH, Koehler RC, Traystman RJ. Reduced blood-brain barrier permeability after cardiac arrest by conjugated superoxide dismutase and catalase in piglets. Stroke. 1994;25:1830-1835.
Zhang X-M, Ellis EF. Superoxide dismutase reduces permeability and edema induced by hypertension in rats. Am J Physiol. 1990;259:H497-H503.
He YY, Hsu CY, Ezrin AM, Miller MS. Polyethylene glycol-conjugated superoxide dismutase in focal cerebral ischemia-reperfusion. Am J Physiol. 1993;265:H252-H256.
Chan PH, Schmidley JW, Fishman RA, Longar SM. Brain injury, edema, and vascular permeability changes induced by oxygen-derived free radicals. Neurology. 1984;34:315-320.
Hall ED, McCall JM. Lazaroids: potent inhibitors of iron-dependent lipid peroxidation for neurodegenerative disorders. In: Riederer P, Youdim MBH, eds. Iron in Central Nervous System Disorders: Key Topics in Brain Research. Wien, Austria: Springer-Verlag; 1993:173-188.
Oury TD, Piantadosi CA, Crapo JD. Cold-induced brain edema in mice: involvement of extracellular superoxide dismutase and nitric oxide. J Biol Chem. 1993;268:15394-15398.
Kamitani T, Little MH, Ellis EF. Evidence for a possible role of the brain kallikrein-kinin system in the modulation of the cerebral circulation. Circ Res. 1985;57:545-552.
Tavalin SJ, Ellis EF, Satin LS. Mechanical perturbation of cultured cortical neurons reveals a stretch-induced delayed depolarization. J Neurophysiol. 1995;74:2767-2773.
Lamb RG, Rzigalinski BA, Ellis EF. Increased phosphatidylcholine biosynthesis and arachidonic acid release in stretch-injured cultured astrocytes. J Neurotrauma. 1995;12:969. Abstract.
Zhang X-M, Ellis EF. Superoxide dismutase decreases mortality, blood pressure and cerebral blood flow responses induced by acute hypertension in rats. Stroke. 1991;22:489-494.
Zweier JL, Kuppusamy P, Lutty GA. Measurement of endothelial cell free radical generation: evidence for a central mechanism of free radical injury in postischemic tissues. Proc Natl Acad Sci U S A. 1988;85:4046-4050.
Rosen GM, Freeman BA. Detection of superoxide generated by endothelial cells. Proc Natl Acad Sci U S A. 1984;81:7269-7273.
Wei EP, Kontos HA, Christman CW, DeWitt DS, Povlishock JT. Superoxide generation and reversal of acetylcholine-induced cerebral arteriolar dilation after acute hypertension. Circ Res. 1985;57:781-787.
Kontos HA, Wei EP, Dietrich WD, Navari RM, Povlishock JT, Ghatak NR, Ellis EF, Patterson JL. Mechanism of cerebral arteriolar abnormalities after acute hypertension. Am J Physiol. 1981;240:H511-H527.
Rosenblum WI, Nelson GH, Murata S. Protein synthesis and rapid recovery of endothelium-dependent dilation after endothelial injury of pial arterioles. Am J Physiol. 1995;268:H512-H515.
Laurindo FRM, Pedro M de A, Barbeiro HV, Pileggi F, Carvalho MHC, Augusto O, da Luz PL. Vascular free radical release: ex vivo and in vivo evidence for a flow-dependent endothelial mechanism. Circ Res. 1994;74:700-709.