Evidence that infiltrating neutrophils do not release reactive oxygen species in the site of spinal cord injury

Experimental Neurology 190 (2004) 414 – 424 www.elsevier.com/locate/yexnr

Evidence that infiltrating neutrophils do not release reactive oxygen species in the site of spinal cord injury R. de Castro Jr.a,*, M.G. Hughesa, G.-Y. Xua, C. Cliftona, N.Y. Calingasanb, B.B. Gelmanc, D.J. McAdooa a

Department of Neuroscience and Cell Biology and Department of Human Biological Chemistry and Genetics, University of Texas Medical Branch, Galveston, TX 77555-1043, United States b Department of Neurology and Neuroscience, Weill Medical College of Cornell University, New York, NY 10021, United States c Department of Pathology, University of Texas Medical Branch, Galveston, TX 77555-0785, United States Received 2 November 2003; revised 2 April 2004; accepted 14 May 2004 Available online 1 October 2004

Abstract The release of reactive oxygen species (ROS) by neutrophils, which infiltrate the region of damage following spinal cord injury (SCI), was investigated to determine if such release is significant following spinal cord injury. The relationship of extracellular levels of hydroxyl radicals and hydrogen peroxide obtained by microdialysis sampling and oxidized protein levels in tissue to neutrophil infiltration following spinal cord injury was examined. Neither of the reactive oxygen species were elevated in the site of spinal cord injury relative to their concentrations in normal tissue at a time (24 h) when the numbers of neutrophils were maximum in the site of injury. Surprisingly, ablation with a neutrophil antiserum actually increased the level of oxidized proteins in Western blots. Thus, our findings are (1) that neutrophils, which infiltrate the site of damage following a spinal cord injury, do not release detectable quantities of reactive oxygen species; and (2) that the presence of neutrophils reduces the concentrations of oxidized proteins in the site of spinal cord injury. Therefore, release of reactive oxygen species by neutrophils does not contribute significantly to secondary damage following spinal cord injury. Reduced levels of oxidized proteins in the presence of neutrophils may reflect removal of damaged tissue by neutrophils. D 2004 Elsevier Inc. All rights reserved. Keywords: Hydrogen peroxide; Hydroxyl radical; Microdialysis; Neutrophils; Protein oxidation; Secondary damage; Spinal cord injury

Introduction Severe and permanent crippling caused by spinal cord injury (SCI) is caused in part by processes secondary to the initial insult. Because they are potential targets for clinical intervention, it is important to understand these secondary processes. Possible agents of secondary damage include reactive oxygen species (ROS) (Anderson et al., 1985; Demopoulos et al., 1982; Hall and Braughler, 1993; Liu et al., 1998, 1999) and infiltrating immune cells (Means and

* Corresponding author. Present address: The Burnham Institute, 10901 North Torrey Pines Road, La Jolla, CA 92027. E-mail address: [email protected] (R. de Castro). 0014-4886/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.expneurol.2004.05.046

Anderson, 1983; Taoka and Okajima, 1998). This study explores whether the latter generate the former following SCI. Polymorphonuclear leukocytes, which are mostly neutrophils, infiltrate the injured human spinal cord in large numbers (Tator and Koyanagi, 1997). There is evidence that infiltrating immune cells contribute to secondary damage following experimental SCI (Hamada et al., 1996; Taoka et al., 1997a,b), but there is also evidence that they do not (Dusart and Schwab, 1994; Holtz et al., 1989). Neutrophils may help recovery by removing damaged tissue, but they may simultaneously contribute to secondary damage by attacking viable, functional neurons. Neutrophils may also influence recovery from CNS trauma by summoning macrophages into the damaged tissue. The latter likely aid

R. de Castro Jr. et al. / Experimental Neurology 190 (2004) 414–424

in the regeneration of peripheral axons by removing damaged tissue (Blight, 1985, 1992; Streit et al., 1998) and by storing cholesterol from the myelin they ingest and then restore during remyelination of regenerating axons (Boyles et al., 1989). Like neutrophils, macrophages may also worsen damage by attacking healthy neurons (Blight, 1992). In the respiratory burst, neutrophils convert large S quantities of oxygen to the superoxide anion O2 , which they release (Jones, 1993; McNeil et al., 1989; McPhail and S Harvath, 1993; Rossi et al., 1986). O2 is a precursor to the much more reactive hydroxyl radical. Products of the respiratory burst in neutrophils oxidize phospholipids in vitro (Zimmerman et al., 1997), so this process may damage viable tissue (Fridovich, 1978). ROS release by neutrophils may participate in the destruction of infectious organisms (Klebanoff, 1974; Weiss, 1989) and/or removal of damaged tissue (McPhail and Harvath, 1993; Means and Anderson, 1983). Considerable superoxide is generated following head trauma (Kontos and Wei, 1986), cerebral inflammation (Kontos et al., 1992), and SCI (Liu et al., 1998). Administration of various superoxide dismutases, enzymes that S convert O2 to H2O2, lowers mortality from compression SCI in rats (Taoka et al., 1995), and recovery from ischemic or traumatic brain injury improves in patients treated with superoxide dismutase conjugated to polyethylene glycol (Muizelaar et al., 1993). In summary, there is considerable evidence that both neutrophil aggregation in the site of trauma and the generation of ROS are harmful following CNS trauma, suggesting that following SCI, neutrophils generate damaging ROS through their respiratory burst. The only established treatment effective for treating SCI in humans is administration of the glucocorticoid methylprednisolone; this must begin within 8 h after injury (Bracken and Holford, 1993), and the benefits thereof are modest. However, treatment of SCI often cannot commence within 8 h of trauma; and even when it does, adding a second type of treatment effective at longer times might further improve ultimate recovery. In this vein, Taoka and Okajima (1998) have proposed that methylprednisolone treatment might act synergistically with blockers of the actions of neutrophils because methylprednisolone does not influence the activation of neutrophils in their SCI model. Damaging actions of neutrophils could provide a target for delayed treatment of SCI, as neutrophil concentrations are elevated for about 4–48 h after a contusion injury to the rat spinal cord (Carlson et al., 1998; Xu et al., 1990), the type of injury inflicted in the present work. To help develop therapeutic approaches to treating SCI, it is important to characterize better the generation of ROS by neutrophils and their actions following SCI. To this end, we investigated whether ROS generation and protein oxidation correlate with neutrophil infiltration following spinal cord injury. We compared the production of ROS in injured versus normal spinal cord tissue when maximum numbers of neutrophils were present in the site of injury. In addition, we compared

415

the oxidation of proteins following SCI in normal rats to such oxidation in sites of injury in which infiltration of neutrophils was prevented by their removal from the circulation. Whether neutrophils cause secondary damage by generating ROS following CNS trauma is hitherto unaddressed by in vivo experiments.

Materials and methods Animal preparation and injury Male Sprague–Dawley rats (300–350 g; Harlan, Houston, TX) were utilized. All experiments were approved by the UTMB Animal Care and Use Committee and were conducted in accord with the recommendations of the NIH Guide for the Care and Use of Laboratory Animals. Preparatory to injury, animals were anesthetized by the intraperitoneal administration of 35 mg/kg of pentobarbital. In neutrophil depletion experiments (see below), anesthesia was initiated at 16 h after the administration of antineutrophil serum (ANS, anti-rat PMN; Accurate Chemical Company, Westbury, NY) and just before injury. Anesthesia was considered to be complete when there was no response to a foot pinch. The back of the animal was shaved and a laminectomy performed to expose spinal cord segment L2. Dorsal–vertebral processes were then rigidly clamped to stabilize the spinal cord against displacement during injury. Injury was inflicted by dropping a 10-g, 3-mm diameter brass weight 2.5 cm down a guide tube onto the exposed dorsal surface of the spinal cord (Allen, 1911). The incision was then closed with metal clips and the animal allowed to recover from the anesthesia. Neutrophil ablation Infiltration of neutrophils into the site of SCI can be prevented by depleting them with a neutrophil antiserum (Holtz et al., 1989). Neutropenia was therefore induced in pertinent experiments by intraperitoneal injection of 0.5 ml of ANS. To confirm neutropenia, circulating neutrophils were counted in a hemacytometer. Blood (25 Al) for these counts was drawn from a cut in the tail of the animal at 16 h after ANS administration and mixed with 475 Al of a counting solution (3% acetic acid and 1% crystal violet in water). Polymorphonuclear leukocytes (PMN), the majority of which were neutrophils, were identified in the samples and counted based on the morphology of their nuclei. Animals whose circulating neutrophil counts were not markedly reduced following administration of the antiserum (about one in six) were excluded from further experiments. Excluding those animals, ANS reduced circulating neutrophils by 96% and the total white blood cell count by 46% at 16 h after ANS administration. ANS depleted some other white blood cells in addition to neutrophils, as neutrophils constitute only 14–20% of circulating white blood cells.

416

R. de Castro Jr. et al. / Experimental Neurology 190 (2004) 414–424

Neutrophils remained 96% depleted from blood at 40 h post-ANS administration, the time of sampling for postinjury biochemical measurements. Administering normal rabbit serum caused a 30% increase in white blood cell counts and a 112% increase in neutrophil counts. Because of this, normal animals rather than ones to which normal rabbit serum was administered were used as controls in experiments involving neutrophil depletion. Myeloperoxidase assay At a series of times after injury, animals were reanesthetized to obtain tissue for the assay of myeloperoxidase. At the time of this anesthesia, animal body temperature was taken with a rectal probe and maintained at 37–388C. Myeloperoxidase (MPO) activity was assayed in isolated tissue samples by the procedure of Biagas et al. (1992) as a measure of the number of infiltrating neutrophils. For this assay, the reanesthetized animals were perfused transcardially with 0.9% cold saline solution containing 1000 U/l of heparin. The exposure of the spinal cord was then extended by removing bone rostrally and caudally with rongeurs and the exposed portion of the spinal cord cut out and rinsed with the perfusing solution. The cord segment was cut into 0.5 cm segments rostral, at and caudal to the injury site. The segments were placed in plastic vials, immediately frozen with dry ice and stored at 708C. When they were to be analyzed, the sections were weighed, thawed, and homogenized in 0.5 ml of a cold 50-mM potassium phosphate buffer (pH 6.0) and the homogenate transferred to a centrifuge tube. The homogenizer was washed with an additional 2 ml of the buffer; this was also transferred to the centrifuge tube. The homogenate was centrifuged (15,000  g, 10–20 min). The supernatant was decanted off and saved to be assayed for myeloperoxidase activity as described below. The pellet was washed, recentrifuged, and resuspended three times in the homogenizing buffer. The final pellet was resuspended in homogenizing buffer containing 0.5% hexyldecyltrimethylammonium bromide. This suspension was freeze–thawed three times (dry ice/ethanol-warm water cycles). After the final thawing, the suspension was centrifuged and the supernate collected. MPO activity in this last supernatant was determined by adding 100 Al of the homogenate to 2.9 ml of a solution containing 0.167 mg/ml of o-dianisidine hydrochloride and 0.0005% hydrogen peroxide. The rate of oxidation of o-dianisidine was measured over time by spectrophotometrically monitoring the change in absorption at 460 nm as a function of time. This rate reflects the activity of myeloperoxidase in the supernatant, which is in turn proportional to the number of neutrophils that were in the original sample. The number of neutrophils in a sample was determined from a standard curve of MPO activity versus number of neutrophils. Neutrophil standards were prepared by mixing various numbers of neutrophils harvested from

the peritoneum after casein injection (Lemanske et al., 1983) with a fixed weight of normal rat spinal cord tissue. Myeloperoxidase activity in the first supernatant was also assayed as just described as an indicator of whether neutrophils aggregating in the site of SCI were active because neutrophils release considerable myeloperoxidase during phagocytosis (Murphy, 1976). The same assay was performed on the supernatants from homogenates of 1  106 neutrophils collected from the peritoneum (following casein injection) and of 1  106 neutrophils added to uninjured spinal cord tissue to assess the amount of MPO released by homogenization. These samples were processed as was just described for spinal cord tissue. Neutrophil counts in tissue sections To count numbers of neutrophils in spinal cord tissue, neutrophil-ablated and non-neutrophil-ablated injured animals were perfusion fixed at 24 h after injury, the time of maximum neutrophil density in the injured tissue (Carlson et al., 1998; also, see Results). Anesthetized animals were perfused transcardially with cold saline until the liver turned pale, indicating complete blood removal. The perfusing solution was then switched to cold 4% formalin in 0.1 M phosphate buffer (pH 7.4). A segment of the spinal cord containing the site of injury was removed and postfixed for 3 h in more 4% formalin solution. The segment was then embedded in paraffin and cut transversely into 8-Am-thick sections. Sections through the lesion caused by the injury were stained with hematoxylin and eosin, and neutrophils identified as small, dark bodies with multilobular nuclei were then counted in representative sections. Neutrophil ablation was also examined via immunohistochemistry for myeloperoxidase. Rabbit anti-myeloperoxidase antibody (Chemicon, Temecula, CA) was used (0.425 mg/ml) to stain transverse sections of injured spinal cord from neutrophil-ablated and nonablated animals. Microdialysis sampling and ROS assays Superoxide rapidly dismutates to H2O2 both spontaneously (Bielski and Allen, 1977) and enzymatically (McCord and Fridovich, 1969); H2O2 can in turn be reduced to the S hydroxyl radical (HO ) by transition metal ions (Halliwell, S 1992). HO attack on salicylic acid produces similar amounts of 2,3- and 2,5-dihydroxybenzoic acid (2,3- and 2,5-DHBA) (Floyd et al., 1984; Halliwell et al., 1988). S Therefore, HO can be assayed in vivo by perfusing a salicylic acid solution through a microdialysis fiber implanted in the region of interest and analyzing 2,3- and 2,5-DHBA in the fluid collected from the outlet of the fiber S (Chiueh et al., 1992). We used this method to assay HO . In this approach, the salicylic acid diffuses out of the microdialysis fiber, is converted into DHBAs by reaction with S HO in the tissue, and the DHBAs are collected by diffusion back into the fiber.

R. de Castro Jr. et al. / Experimental Neurology 190 (2004) 414–424

We prepared microdialysis fibers by gluing polyimide inlet and outlet tubes of 122.5 Am outside diameter (MicroLumen, Tampa, FL) into short lengths of 170 Am outside diameter dialysis fiber (Spectrum Industries, Houston, TX) so as to leave a 2-mm dialysis zone. These are modified from microdialysis fibers we used previously (Liu et al., 1991) because of leakage problems experienced with the latter (Xu et al., 1998). Animals were reanesthetized at 22 h after injury. Fibers were pulled through the cord by means of a pin glued into the outlet tube; this pin was cut off after fiber insertion. One microdialysis fiber was inserted transversely through the cord at the site of injury, and one was inserted 2 cm rostral to that site. A tube going directly to a collecting vial without a dialysis zone and without passing through any tissue was also set up to collect blank samples to allow correction for background DHBA formation in the samples. The setup for sampling is illustrated in Fig. 1. Artificial cerebrospinal fluid (ACSF) of the following composition (in mM), Na+ 151.1, K+ 2.6, Mg2+ 0.9, Ca2+ 1.3, Cl 122.7, HCO3 21.0, HPO42 2.5, and glucose 3.5 containing 2.5 mM salicylic acid, was pumped through all three fibers at a flow rate of 2 Al/min to detect S HO . Administering trapping agents such as DHBA and phenylalanine through microdialysis fibers has been demonstrated to be able to detect hydroxyl radicals following spinal cord injury (Liu, 1993). The salicylic acid was recrystallized from boiling water to rid it of DHBAs present in it. To assess as accurately as possible release following injury, samples were simultaneously collected from all three tubes to correct for background and basal levels based on results from samples obtained at the same time from the same animal. High-pressure liquid chromatography (HPLC) with electrochemical detection was used to analyze 2,3- and

417

2,5-DHBA in microdialysis samples. A Bioanalytical Systems (BAS, West Lafayette, IN) high-pressure liquid chromatograph equipped with a BAS LC-4C electrochemical detector, automated sample injection, and a BAS data acquisition system was used. The detector was set to 0.650 V. A BAS Phase II ODS column packed with 3 Am diameter particles was eluted isocratically with a mobile phase consisting of 0.03 M sodium acetate, 0.03 M potassium citrate, and 0.01 M sodium chloride dissolved in water with the pH adjusted to 3.8. Salicylic acid is oxidized to 2,5DHBA by P-450 isoforms in liver microsome fractions (Ingelman-Sundberg et al., 1991), making the 2,3-isomer the preferred one for analysis when enzymatic oxidation can be a problem (Halliwell et al., 1991). However, it has also S been shown that 2,5-DHBA is a valid indicator of HO levels in microdialysis experiments in which salicylic acid is administered (Chiueh et al., 1992), so we report results for both isomers. Samples to be analyzed for H2O2, the intermediate S S between O2 and HO , were also collected over 22–24 h postinjury in separate microdialysis experiments. Perfusates were collected from within the site of injury and 2 cm rostral to that site. H2O2 was analyzed according to the method of Liu et al. (1999) by adding an equal volume of an aqueous solution of 0.2 mM FeCl2 and 2.5 mM salicylic acid to the collected microdialysates. The Fe2+ S dissociates the H2O2 in the samples to HO + HO , the first of which attacks salicylic acid to generate 2,3- and 2,5-DHBA. Reacted samples were analyzed for DHBAs by HPLC-electrochemical detection as described above. Background reaction was corrected for by subtracting the concentrations of DHBAs formed in the ACSF blanks by this procedure from those in the experimental samples. The formation of DHBAs in the ACSF demonstrates the

Fig. 1. Setup for sampling the production of reactive oxygen species following spinal cord injury. The foremost dialysis fiber samples from the site of injury, the second fiber samples from within the cord 2 cm rostral to the site of injury, and the third fiber provides control ACSF that has passed through a dialysis fiber that does not pass through tissue. This enables use of each animal as its control, thereby reducing experimental uncertainties arising from interanimal variations.

418

R. de Castro Jr. et al. / Experimental Neurology 190 (2004) 414–424

presence of an oxidizing agent in the ACSF, a potential source of artifacts in this type of measurement. This background was quite reproducible, so corrections for it could be readily made. Protein oxidation The formation of carbonyl groups in proteins is a useful indicator of oxidation in a tissue. These groups are formed by oxidation of lysine, arginine, proline, and threonine residues (Stadtman, 1993). Protein carbonyl content was determined in tissue taken 24 h postinjury from injured, neutrophil-depleted animals, and injured normal animals; tissue was also taken from uninjured normal animals. The isolated portions of the cord were homogenized in phosphate-buffered saline (PBS) at pH 7.2. To prevent proteolysis during sample workup, this solution contained 2 mM AEGSF 2, 130 AM bestatin, 1.5 AM E-64, 1 AM leupeptin, 0.3 M aprotinin, 1 mM EDTA, and 1% Triton X-100. Homogenates were spun at 15,000  g for 15–20 min and the supernatants taken. The protein contents of the supernatants were determined by the bicinchoninic acid (BCA) method (Pierce Chemical Co., Rockford, IL). Protein carbonyl formation was determined with the OxyBlot Oxidized Protein Detection Kit (Intergen, Purchase, NY). Protein (10 Ag based on the BCA assay) in the supernatant was derivatized with 2,4-dinitrophenylhydrazine (DNPH) in sodium dodecyl sulfate (6%) for 15 min at room temperature. The derivatization was stopped by adding the neutralization solution provided in the kit. The samples were then loaded into the wells of a 7.5% polyacrylamide gel and electrophoresed. After electrophoresis, the proteins were transferred from the gel to a nitrocellulose sheet with an electroblot apparatus. The blot was blocked for 1 h at room temperature with 1% bovine serum albumin in PBS (pH 7.2) containing 0.5% Tween 20 (PBS-T), incubated with an antiserum to DNPH groups (Polyclonal anti-DNPH, 1:150 in BSA/PBS-T) for 1 h at room temperature, and rinsed twice with PBS-T. The blot was then incubated with a secondary antibody (horseradish peroxidase conjugated, goat anti-rabbit IgG, 1:300 in BSA/ PBS-T) for 1 h at room temperature and rinsed five times with PBS-T. Immunoreactivity to DNPH derivatives was detected with a chemiluminescence reaction initiated by incubating the blot with the detection solution for 1 min. Then film (Eastman Kodak, Rochester, NY) was exposed to the blot for 6 min and developed. Band densities were quantified with Alpha-Digital Imaging Software, Version 4.0 (Alpha Innotech Corp., San Leandro, CA). Densities of individual bands were determined and summed to give the densities of entire lanes. The summed intensities of the signals were taken to be proportional to the total protein carbonyl content of the samples. No bands were observable in any nonderivatized sample run as controls for protein carbonyl determinations.

Results Neutrophil infiltration and depletion Both counts of neutrophils in histological sections (Fig. 2 shows MPO-immunostained sections injured spinal cords, normal nonablated, and neutrophil ablated) and the results of the myeloperoxidase assay (Fig. 3) demonstrate that injury induced infiltration of substantial numbers of neutrophils into the area of damage, with the maximum number present at about 24 h postinjury in normal animals, as observed by previous workers (Carlson et al., 1998; Chatzipanteli et al., 2000). Smaller numbers of neutrophils were present in the sections rostral and caudal to the site of injury in injured, nondepleted animals. The neutrophils in the site of injury were distributed over the tissue parenchyma (33%), leptomeninges (61%), area of hemorrhage (3%), and the vascular system (2%). Based on myeloperoxidase activity, the neutrophil antiserum reduced the number of neutrophils in the site of injury by 82%. The numbers of neutrophils counted in histological sections were reduced even more dramatically: 400 F

Fig. 2. Neutrophil invasion 24 h after injury—immunohistochemistry for myeloperoxidase in the spinal cord. Upper picture, tissue from a nondepleted animal; dense, brown, dotlike cells are neutrophils; CC = central canal. Lower picture, from a similar area in a neutrophil-depleted animal. At this time, neutrophils are the only white blood cells congregating in the site of injury. This illustrates that antineutrophil serum treatment was effective in preventing the infiltration of neutrophils into the site of spinal cord injury.

R. de Castro Jr. et al. / Experimental Neurology 190 (2004) 414–424

419

Fig. 3. Main figure—time course of neutrophil infiltration into the injured spinal cord based on the myeloperoxidase assay. MPO activity is expressed as the change in absorption of the reaction solution per minute normalized to the weight of the tissue sample. The 0 h samples were taken immediately after injury. Neutrophil infiltration was maximum at 24 h after injury. Inset—legend, I = injury site, R = rostral, C = caudal.

230 neutrophils in the site of injury of normal animals, and 8 F 8 neutrophils in injured, neutrophil-depleted animals (n = 3 for each, F SD). No other white blood cells were present in significant numbers in the site of injury at 24 h postinjury in undepleted animals, so results reported reflect only actions (or lack thereof) of neutrophils. Reactive oxygen species DHBAs, assayed as measures of the generation of hydroxyl radicals, were detected in all microdialysis samples containing salicylic acid (added prior to perfusion through the fiber (Table 1). Variation in interanimal basal levels were greater than such variation between different areas in the individual cords, so our simultaneous collection of control samples from outside the injury zone of the same cord substantially increased the precision of the results. Starting 22 h after injury was inflicted, samples were collected for four 30 min periods and then one 60-min period. No changes in concentrations of the species analyzed were observed across this time. Because such damage abates within 1–2 h after fiber insertion (Sorkin et al., 1988), the absence of any difference between the first and last samples collected

Table 1 Hydroxyl radical production at the time of maximum neutrophil infiltration Product

Injured tissue

Normal tissue

Background

2,3-DHBA 2,5-DHBA

128 F 62 (0.04) 140 F 35 (0.04)

137 F 70 (0.04) 130 F 84 (0.04)

65 F 22 80 F 14

Values are DHBA concentrations in microdialysates in nM (mean F SD from measurements on five animals). Values in parentheses are two-sided P values from the Wilcoxon signed rank test comparing injured and normal to background. P values for comparing injured and normal were 0.63 from 2,3-DHBA and 0.81 for 2,5-DHBA, so there was no effect of injury on generation of OH at maximum postinjury neutrophil concentrations.

S

demonstrates that damage due to the insertion of the fiber did not affect our results. There was no difference between DHBA concentrations in samples from microdialysis fibers that passed through the site of injury and samples from fibers that passed through normal spinal cord tissue. Concentrations of both DHBAs in the samples from the site of injury (128 F 62 nM) and normal tissue (137 F 70 nM) were significantly higher than those in the tubing that bypassed the spinal cord (65 F 22 nM) (Friedman repeated measures analysis of variance: P = 0.015 for the data from 2,3-DHBA and P = 0.022 for the data for 2,5DHBA; P = 0.04 for all pairwise comparisons of injured to normal by the Wilcoxon signed rank test; Table 1). S Therefore, detectable HO was formed in both the site of injury and in normal tissue; however, there was no S additional formation of HO by the neutrophils that infiltrated the spinal cord following injury (Table 1). The S detection of HO in the normal tissue demonstrates that our techniques would be able to detect production of a S change in HO levels by neutrophils large enough to influence the outcome of SCI. H2O2 was also analyzed as a measure of ROS production by infiltrating neutrophils. Its chemical stability and ready collection should make it a reliable indicator of ROS S generation. The O2 that neutrophils generate is rapidly transformed to H2O2 (Test and Weiss, 1984), so generation S of significant amounts of O2 should increase formation of H2O2. However, at 24 h postinjury, the concentrations of DHBAs generated from H2O2 present in microdialysis samples collected from the site of injury were either lower (D = 14 F 10 arbitrary units based on 2,5-DHBA) or not different (D = 4.2 F 22 based on 2,3-DHBA) from concentrations in samples collected 2 cm rostral to that site (Table 2), but clearly never higher (paired t test, n = 5).

420

R. de Castro Jr. et al. / Experimental Neurology 190 (2004) 414–424

Table 2 Hydrogen peroxide production at the time of maximum neutrophil infiltration Product

Injured tissue

Normal tissue

2,3-DHBA 2,5-DHBA

151 F 24 242 F 80

147 F 27 257 F 83

D 4.2 F 22 14 F 10

D = average of the differences for each animal between values for injured and normal (FSD). The significance of the difference is not apparent in the uncertainties because of interanimal variations. Accompanying values in ACSF blanks were 1.13 F 0.06 times higher for samples from injured tissue and 1.05 F 0.03 times higher for samples from control tissue (combined values for 2,3- and 2,5-DHBA). Values are peak areas in arbitrary units. The values for the 2,5-isomer are significantly different ( P = 0.030), but those for the 2,3-isomer are not ( P = 0.69) based on a paired t test (mean F SD, n = 5).

Even if the difference based on 2,5-DHBA analysis is real, it S is small (6%). Thus, as with HO , measured concentrations of H2O2 provide no evidence that the neutrophils that

infiltrate the site of injury release reactive oxygen species. Furthermore, H2O2 concentrations in collected samples were all slightly lower than in ACSF blanks, so no H2O2 was detected in either injured or uninjured tissue. The lowering of the concentration of a DHBA-generating species in the ACSF may be attributable to diffusion of a small fraction of that species out of the fiber during microdialysis. Myeloperoxidase release Myeloperoxidase activity was assayed in the supernatant from the initial homogenization of spinal cord tissue, of neutrophils, and of neutrophils added to uninjured spinal cord tissue as a measure of activity of the infiltrating neutrophils. There was substantial myeloperoxidase activity in the first supernatant from injured neutrophil-containing tissue, 36% of the total myeloperoxidase activity in that supernatant plus that subsequently extracted from the pellet (n = 5). When isolated neutrophils were added to uninjured tissue, 6% of the myeloperoxidase activity assayed was in the first supernatant and 94% in the second (n = 4). This demonstrates that at most a small fraction of the myeloperoxidase activity in neutrophil infiltrated tissue was released by the homogenizing procedure, that is, substantial myeloperoxidase was secreted by the neutrophils that infiltrate injured spinal tissue. Protein oxidation Protein carbonyl content was analyzed by gel electrophoresis followed by Western blotting. A representative blot of a gel utilized for protein carbonyl analysis in samples from the spinal cord is shown in Fig. 4. Averages (n = 3) of the traces of the densities along the length of the lanes ( yaxis = density on the gel in arbitrary units; x-axis = distance along the lane) are shown in the upper portion of Fig. 4. Table 3 lists mean total densities obtained by summing signals such as those depicted in Fig. 4 from the entire lanes. Mean total carbonyl content was highest in the neutrophildepleted ( N) samples, lowest in the samples from normal, injured (+N) animals, and intermediate in the samples from uninjured cords. The average total density was significantly greater in the N (244 F 32 arbitrary units) than in the +N samples (160 F 11) (t test, P = 0.01, n = 3). It is surprising that substantial amounts of carbonyl-containing proteins were present in control tissue, but similar observations have

Fig. 4. Representative Western blots for analysis of protein carbonyl content by gel electrophoresis. Blots were reacted with DNPH and protein carbonyl visualized by staining with an appropriate antiserum as described in Materials and methods. Bottom—three representative lanes from the actual blots. Molecular weights decrease from left to right. +N—data from injured normal animals. N—data from injured, neutrophil-depleted animals. U— data from uninjured animals. The traces represent density of staining (averages, n = 3 per group) as determined by scanning with a densitometer. Note that the peaks were less distinct and the total density lower in +N than in N animals and that some peaks (2, 8) present in normal tissue were not present in injured tissue.

Table 3 Total protein carbonyl content in spinal cord tissue Normal injured (+N)

Neutrophil-ablated injured ( N)

Uninjured normal (U)

160 F 11

244 F 32

204 F 28

Values were obtained by integrating the total densities in each lane. The values for normal injured are significantly different from those for the neutrophil-ablated injured ( P = 0.01, t test, n = 3).

R. de Castro Jr. et al. / Experimental Neurology 190 (2004) 414–424

been reported in work in which spectrophotometric rather than immunological detection of the DNP derivatives was used (Leski et al., 2001), so carbonyl-containing proteins are present in normal spinal cord tissue. The intensities of individual bands varied from sample to sample, for example, bands 4 and 7 (as labeled in Fig. 4) were most prominent in the samples from uninjured tissue; bands 10 and 11 were most prominent in the N samples; and bands 1, 3, and 9 were observable only in injured tissue. However, no band in the +N samples was ever more intense than the corresponding band in the N samples. Thus, neutrophils appear to remove more oxidized proteins than they create in the site of SCI.

Discussion Generation of reactive oxygen species It has been hypothesized that, following SCI, ROS generated by infiltrating neutrophils attack viable cells as well as their intended targets, presumably damaged tissue (Means and Anderson, 1983; Taoka et al., 1998). Some data from neutrophil ablation experiments support a role of neutrophils in secondary damage following SCI (Taoka et al., 1998). However, whether there is significant in vivo S S formation of HO from O2 generated by neutrophils in general has been questioned due to insufficient concentrations of required metal ions (Britigan et al., 1988) and the S rapid conversion of H2O2, the immediate HO precursor, to HOCl (Weiss, 1989). Previous experiments by ourselves demonstrate a lack of low molecular weight iron ions in the extracellular space over the first 4 h after spinal cord injury (de Castro et al., 1999). H2O2, if anything, was slightly lower in samples collected from the site of injury than in the normal spinal cord, demonstrating its negligible generation S through O2 in neutrophils followed by its release. The lack S of detectable generation of either extracellular H2O2 or HO by neutrophils, together with the reduced levels of oxidized protein in the presence of neutrophils following injury, demonstrates that infiltrating neutrophils do not release substantial quantities of these ROS following SCI, and so do not thereby damage neurons following SCI. This is despite the implication of the release of myeloperoxidase from neutrophils, we found in the site of SCI that the neutrophils that infiltrated spinal cord tissue were actively secreting (Murphy, 1976). It might be argued that neutrophils released ROS at a time other than that at which we took our samples. However, this is unlikely, as at 24 h neutrophils are continually infiltrating and disappearing from the site of injury. Thus, at 22–25 h after injury, the time of our measurements, neutrophils at all stages of their postinfiltration activity cycle should have been present. H2O2 production increases acutely following spinal cord injury (Liu et al., 1999). However, those increases are only to about 30% above baseline and return to basal levels at

421

about 11 h after injury, consistent with the present failure to detect elevation of ROS at 22–25 h post-SCI. Direct measurement also revealed doubling of the extracellular S concentrations of O2 following SCI (Liu et al., 1998); S however, like H2O2, O2 returned to baseline levels after about 10 h, a time of rapid neutrophil infiltration and well before the time of their maximum presence. The decline of these ROS to basal levels just when the numbers of infiltrating neutrophils was maximal is added evidence against substantial release of those ROS by neutrophils. Our failure to detect elevated production of ROS or increased protein oxidation in the presence of high levels of neutrophils is surprising, given that appropriately activated S neutrophils are capable of producing large quantities of O2 (Jones, 1993; McNeil et al., 1989; Rossi et al., 1986). S Possibly, the neutrophils produced O2 , but detection of S S H2O2 and formation of HO were prevented by the rapid conversion of H2O2 to HOCl by myeloperoxidase. Thus, neutrophils may secrete myeloperoxidase following SCI to scavenge reactive ROS before they can harm tissue; that is, under some circumstances, rather than generating ROS, neutrophils may secrete myeloperoxidase to inactivate ROS arising from other sources. That might be the reason for the slight decline in H2O2 levels if such occur in the injured cord (significant according to 2,5-DHBA but not 2,3DHBA). Neutrophil expression of heme oxygenase-1 in injured spinal cord tissue may also protect against oxidative stress (Liu et al., 2002). Another possibility is that ROS are produced exclusively S inside neutrophils and never get out, as some O2 is formed S inside neutrophils. However, it is unlikely that O2 is exclusively formed and retained within neutrophils because recent ultrastructural studies have demonstrated that by 10 S min after activation, O2 is being produced by neutrophils mostly into invaginations that are in communication with the extracellular space (Jiang et al., 2000). Neutrophils sometimes form seals against particles too large for them to ingest and release their agents into the resulting enclosed space (Campbell et al., 1982). The absence of detectable ROS generation by neutrophils in present experiments might reflect secretion into and retention of ROS in such a confined space. Release of ROS into such spaces in present experiments is highly unlikely, however, because H2O2 generated in such space should be detectable by our methods because it diffuses freely out of neutrophils (Test and Weiss, 1984). Thus, in conclusion, it appears that infiltrating neutrophils simply do not generate ROS following SCI. Although ROS appear to be generated and contribute to secondary damage following SCI, our results demonstrate that those ROS are not produced by neutrophils. There are a number of mechanisms other than generation by neutrophils S whereby O2 and other ROS derived from it might be produced following CNS trauma, for example, malfunction of electron transfer chains, catalysis by xanthine oxidase (McCord, 1985; Xu et al., 1991), and as a byproduct of

422

R. de Castro Jr. et al. / Experimental Neurology 190 (2004) 414–424

prostaglandin synthesis (Kontos et al., 1985). Contribution of ROS to the impairments caused by SCI is supported by several reports of reductions in impairments by treatment with forms of superoxide dismutase (Taoka et al., 1995) that S inactivate O2 . In addition, administration of paraquat, a S generator of O2 , into the rat spinal cord kills neurons (Liu et al., 1995). The present lack of detection of ROS at around 24 h postinjury suggests that therapeutic treatments intended to intercept ROS need to be administered within the first few hours after SCI to be effective, correlating with the time of the therapeutic window for methylprednisolone (Bracken and Holford, 1993). The undetectability of ROS production by neutrophils in our experiments suggests that other actions of neutrophils, such as release of any of a substantial variety of proteolytic enzymes (Chandler et al., 1995; Taoka et al., 1998; Weiss, 1989), may contribute to secondary damage following SCI. ROS from neutrophils may indirectly cause tissue damage by disinhibiting proteases that attack viable tissue (Weiss, 1989), but that process cannot be important when ROS are not being produced. Smedly et al. (1986) also obtained evidence favoring involvement of neutrophil elastase but not ROS in in vitro damage to endothelial cells by neutrophils. This is supported by improved recovery of motor function following SCI by inhibition of neutrophil elastase (Tonai et al., 2001). Furthermore, the concentration of a neutrophil enzyme, matrix metaloproteinase 9, increases in conjunction with neutrophil infiltration following contusion SCI (de Castro et al., 2000; Noble et al., 2002). Therefore, the role of neutrophil proteases as damaging agents following SCI merits further investigation. Protein oxidation Significantly more oxidized protein was present in the site of injury in neutrophil-depleted animals than in injured cord tissue infiltrated by neutrophils. This is good evidence that neutrophils remove oxidized proteins following SCI, perhaps in conjunction with phagocytosis. Utilization of ROS by neutrophils to oxidize endocytosed material should increase rather than decrease levels of oxidized proteins; the opposite results we obtained are further evidence against generation of ROS by infiltrating neutrophils following SCI. An increase in protein carbonyl content at 1–9 h post-SCI followed by normal levels at 48 h has been reported by Leski et al. (2001), with no measurement at 24 h. Some bands in Western blots representing carbonylcontaining proteins were more prominent in the samples from injured than in samples from uninjured animals (bands 1, 3, 9, and 10), evidence that injury induced selective production or oxidation of those proteins. However, the prominence of other bands was reduced in samples from injured animals, suggesting that the proteins they represent were selectively degraded following injury. This degradation may accompany the removal of damaged tissue.

Degradation of Coomassie-stained protein bands following SCI has been reported previously (Leski et al., 2001). Whatever the exact mechanism may be, oxidized proteins are probably degraded by proteolytic enzymes present in neutrophils. Oxidized proteins can be selectively removed by proteases (Grune et al., 1997; Huang et al., 1995). Based on present results, agents other than neutrophils probably oxidize proteins following SCI, but neutrophils may participate in removing those proteins. Taoka and Okajima (1998) have consistently found that neutrophils contribute to secondary damage following compression SCI, whereas Holtz et al. (1989) found no effect of neutrophils on recovery from a compression injury, and Dusart and Schwab (1994) concluded that neutrophils do not play a major role in secondary damage in a spinal hemisection model. Neutrophil depletion as described here may improve the outcome of SCI in a contusion model (de Castro, 1999). Some of the differences reported in the effects of neutrophils may reflect variation among injury models, as impact injury leads to infiltration of neutrophils into the parenchyma reaching maximum levels around 24 h after injury, whereas following compression SCI in the rat neutrophils are confined to the interior of blood vessels. These maximize at 3 h after compression SCI and then leave the area (Taoka and Okajima, 1998). In that instance, neutrophils appear to cause damage by attacking the endothelial cells of circulatory vessels (Taoka et al., 1997b, 1998). The differences in infiltration among the models may stem from facile crossing of a breached blood–brain barrier by neutrophils in contusion injury (present results) versus a much lower ability of neutrophils to cross a presumably intact barrier in compression injury. The substantial abundance of neutrophils throughout a hemisection lesion at 24 h postinjury (Dusart and Schwab, 1994) is consistent with this. In summary, we could not detect the production of ROS by neutrophils that infiltrate the spinal cord following injury, despite demonstrations of substantial ROS generation by neutrophils in in vitro experiments. Therefore, damage by neutrophils following SCI most likely occurs through other mechanisms, possibly phagocytosis and release of proteolytic enzymes. Thus, in vivo experiments are crucial to establishing the mechanisms of SCI, and mechanisms of damage in CNS trauma should not be considered to be established on the basis of plausibility and indirect evidence alone. Following SCI, levels of oxidized proteins are reduced in the presence of neutrophils, implying that neutrophils contribute more to their removal than to their formation. This could reflect selective removal of damaged tissue by neutrophils through phagocytosis and proteolysis but unaided by oxidation.

Acknowledgments We deeply appreciate the contributions of Debbie Pavlu and Thomas (Xia) Chen to the preparation of this manu-

R. de Castro Jr. et al. / Experimental Neurology 190 (2004) 414–424

script, help with statistics by Dr. James Grady, and technical assistance by Linghui Nie and Charles Mills. Supported by Mission Connect at the TIRR Foundation, NIH (11255).

References Allen, A.R., 1911. Surgery of experimental lesion of spinal cord equivalent to crush injury of fracture dislocation of spinal column. J. Am. Med. Assoc. 57, 878 – 880. Anderson, D.K., Saunders, R.D., Demediuk, P., Dugan, L.L., Braughler, J.M., Hall, E.D., Means, E.D., Horrocks, L.A., 1985. Lipid hydrolysis and peroxidation in injured spinal cord: partial protection with methylprednisolone or vitamin E and selenium. Cent. Nerv. Syst. Trauma 2, 257 – 267. Biagas, K.V., Uhl, M.W., Schiding, J.K., Nemoto, E.M., Kochanel, P.M., 1992. Assessment of posttraumatic polymorphonuclear leukocyte accumulation in rat brain using tissue myeloperoxidase assay and vinblastine treatment. J. Neurotrauma 9, 363 – 371. Bielski, B.H., Allen, A.O., 1977. Mechanism of the disproportionation of superoxide radicals. J. Phys. Chem. 81, 1048 – 1050. Blight, A.R., 1985. Delayed demyelination and macrophage invasion: a candidate for secondary cell damage in spinal cord injury. Cent. Nerv. Syst. Trauma 2, 299 – 315. Blight, A.R., 1992. Macrophages and inflammatory damage in spinal cord injury. J. Neurotrauma 9, S83 – S91. Boyles, J.K., Zoeliner, C.D., Anderson, L.J., Kosik, L.M., Pitas, R.E., Weisgraber, K.H., Hui, D.Y., Mahley, R.W., Gebicke-Haerter, P.J., Ignatius, M.J., Shooter, E.M., 1989. A role for apolopoprotein E, apolipoprotein A-I, and low density lipoprotein receptors in cholesterol transport during regeneration and remyelination of the rat sciatic nerve. J. Clin. Invest. 83, 1015 – 1031. Bracken, M.B., Holford, T.R., 1993. Effects of timing of methylprednisolone or naloxone administration on recovery of segmental and long-tract neurological function in NASCIS 2. J. Neurosurg. 79, 500 – 507. Britigan, B.E., Cohen, M.S., Rosen, G.M., 1988. Hydroxyl radical formation in neutrophils. New Engl. J. Med. 318, 858 – 859. Campbell, E.J., Senior, R.M., McDonald, J.A., Cox, D.L., 1982. Proteolysis by neutrophils. Relative importance of cell-substrate contact and oxidative inactivation of proteinase inhibitors in vitro. J. Clin. Invest. 70, 845 – 852. Carlson, S.L., Parrish, M.E., Springer, J.E., Doty, K., Dossett, L., 1998. Acute inflammatory response in spinal cord following impact injury. Exp. Neurol. 151, 77 – 88. Chandler, S., Coates, R., Gearing, A., Lury, J., Wells, G., Bone, E., 1995. Matrix metalloproteinases degrade myelin basic protein. Neurosci. Lett. 201, 223 – 226. Chatzipanteli, K., Yanagawa, Y., Marcillo, A.E., Kraydieh, S., Yezierski, R.P., Dietrich, W.D., 2000. Posttraumatic hypothermia reduces polymorphonuclear leukocyte accumulation following spinal cord injury in rats. J. Neurotrauma 17, 321 – 332. Chiueh, C.C., Krishna, G., Tulsi, P., Obata, T., Lang, K., Huang, S.-J., Murphy, D.L., 1992. Intracranial microdialysis of salicylic acid to detect hydroxyl radical generation through dopamine autooxidation in the caudate nucleus: effects of MPP. Free Radical Biol. Med. 13, 581 – 583. de Castro R. Jr., 1999, Some contributions by neutrophils to the pathophysiology of spinal cord injury: neutrophil ablation studies. PhD dissertation, University of Texas Medical Branch. de Castro Jr., R., Alcock, N.W., McAdoo, D.J., 1999. Sampling of low molecular weight iron by microdialysis following spinal cord injury. J. Neurosci. Res. 57, 735 – 739. de Castro Jr., R., Burns, C.L., McAdoo, D.J., Romanic, A.M., 2000. Metalloproteinase increases in the injured rat spinal cord. NeuroReport 11, 3551 – 3554. Demopoulos, H.B., Flamm, E.S., Seligman, M.L., Pietronigro, D.D.,

423

Tomasula, J., DeCrescito, V., 1982. Further studies on free-radical pathology in the major central nervous system disorders: effect of very high doses of methylprednisolone on the functional outcome, morphology, and chemistry of experimental spinal cord impact injury. Can. J. Physiol. Pharmacol. 60, 1415 – 1424. Dusart, I., Schwab, M.E., 1994. Secondary cell death and the inflammatory reaction after dorsal hemisection of the rat spinal cord. Eur. J. Neurosci. 6, 712 – 724. Floyd, R.A., Watson, J.J., Wong, P.K., 1984. Sensitive assay of hydroxyl free radical formation utilizing high pressure liquid chromatography with electrochemical detection of phenol and salicylate hydroxylation products. J. Biochem. Biophys. Methods 10, 221 – 235. Fridovich, I., 1978. The biology of oxygen radicals. Science 201, 875 – 880. Grune, T., Reinheckel, T., Davies, K.J.A., 1997. Degradation of oxidized proteins in mammalian cells. FASEB J. 11, 526 – 534. Hall, E.D., Braughler, J.M., 1993. Free radicals in CNS injury. In: Waxman, S.G. (Ed.), Molecular and Cellular Approaches. Raven Press, New York, pp. 81 – 105. Halliwell, B., 1992. Reactive oxygen species and the central nervous system. J. Neurochem. 59, 1609 – 1623. Halliwell, B., Grootveld, M., Gutteridge, J.M.C., 1988. Methods for the measurement of hydroxyl radicals in biochemical systems: deoxyribose degradation and aromatic hydroxylation. Methods Biochem. Anal. 33, 59 – 90. Halliwell, B., Kaur, H., Ingelman-Sundbgerg, M., 1991. Hydroxylation of salicylate as an assay for hydroxyl radicals: a cautionary note. Free Radical Biol. Med. 10, 439 – 441. Hamada, Y., Ikata, T., Katoh, S., Nakauchi, K., Niwa, M., Kawai, Y., Fukuzawa, K., 1996. Involvement of an intercellular adhesion molecule 1-dependent pathway in the pathogenesis of secondary changes after spinal cord injury in rats. J. Neurochem. 66, 1525 – 1531. Huang, L.L., Shang, F., Nowell Jr., T.R., Taylor, A., 1995. Degradation of differentially oxidized alpha-crystallins in bovine lens epithelial cells. Eye Res. 61, 45 – 54. Holtz, A., Nystrom, B., Gerdin, B., 1989. Spinal cord injury in rats: inability of nimodipine or anti-neutrophil serum to improve spinal cord blood flow or neurologic status. Acta Neurol. Scand. 79, 460 – 467. Ingelman-Sundberg, M., Kaur, H., Terelius, Y., Persson, J.-O., Halliwell, B., 1991. Hydroxylation of salicylate by microsomal fractions and cytochrome P-450. Biochem. J. 276, 753 – 757. Jiang, X., Kobayashi, T., Nahirney, P.C., Saz, E.G.C., Seguchi, H., 2000. Ultracytochemical study on the localization of superoxide producing sites in stimulating rat neutrophils. Anat. Rec. 258, 156 – 165. Jones, O.T., 1993. The mechanism of the production of superoxide by phagocytes. Mol. Chem. Neuropathol. 19, 177 – 184. Klebanoff, S.J., 1974. Role of the superoxide anion in the myeloperoxidasemediated antimicrobial system. J. Biol. Chem. 249, 3724 – 3728. Kontos, H.A., Wei, E.P., 1986. Superoxide production in experimental brain injury. J. Neurosurg. 64, 803 – 807. Kontos, H.A., Wei, E.P., Ellis, E.F., Jenkins, L.W., Povlishock, J.T., Rowe, G.T., Hess, M.L., 1985. Appearance of superoxide anion radical in cerebral extracellular space during increased prostaglandin synthesis in cats. Circ. Res. 57, 142 – 151. Kontos, C.D., Wei, E.P., Williams, J.I., Kontos, H.A., Povlishock, J.T., 1992. Cytochemical detection of superoxide in cerebral inflammation and ischemia in vivo. Am. J. Physiol. 263, H1234 – H1242. Lemanske Jr., R.F., Guthman, D.A., Kaliner, M., 1983. The biologic activity of mast cell granules. VII. The effect of anti-neutrophil antibody-induced neutropenia on rat cutaneous late phase reactions. J. Immunol. 131, 929 – 933. Leski, M.L., Bao, F., Wu, L., Qian, H., Sun, D., Liu, D., 2001. Protein and DNA oxidation in spinal injury: Neurofilaments - an oxidation target. Free Rad. Biol. Med. 30, 613 – 624. Liu, D., 1993. Generation and detection of hydroxyl radical in rat spinal cord by microdialysis administration of Fenton’s reagents and microdialysis sampling. J. Biochem. Biophys. Methods 27, 281 – 291. Liu, D., Thangnipon, W., McAdoo, D.J., 1991. Excitatory amino acids rise

424

R. de Castro Jr. et al. / Experimental Neurology 190 (2004) 414–424

to toxic levels upon impact injury to the rat spinal cord. Brain Res. 547, 344 – 348. Liu, D., Yang, J., Li, L., McAdoo, D.J., 1995. Paraquat—A superoxide generator—Kills neurons in the rat spinal cord. Free Radical Biol. Med. 18, 861 – 867. Liu, D., Sybert, T.E., Qian, H., Liu, J., 1998. Superoxide production after spinal injury detected by microperfusion of cytochrome. Free Radical Biol. Med. 25, 298 – 304. Liu, D., Liu, J., Wen, J., 1999. Elevation of hydrogen peroxide after spinal cord injury detected by using the Fenton reaction. Free Radical Biol. Med. 27, 478 – 482. Liu, Y., Tachibana, T., Dai, Y., Kondo, E., Fukuoka, T., Yamanaka, H., Noguchi, K., 2002. Heme oxygenase-1 expression after spinal cord injury: the induction in activated neutrophils. J. Neurotrauma 19, 479 – 490. McCord, J.M., 1985. Oxygen-derived free radicals in postischemic tissue injury. N. Engl. J. Med. 312, 159 – 163. McCord, J.M., Fridovich, I., 1969. Superoxide dismutase. An enzymic function for erythrocuprein (hemocuprein). J. Biol. Chem. 244, 6049 – 6055. McNeil, C.J., Smith, K.A., Bellavite, P., Bannister, J.V., 1989. Application of the electrochemistry of cytochrome c to the measurement of superoxide radical production. Free Radical Res. Commun. 7, 89 – 96. McPhail, L.C., Harvath, L., 1993. Signal transduction in neutrophil oxidative metabolism and chemotaxis. In: Abramson, J.S., Wheeler, J.G. (Eds.), The Neutrophil: The Natural Immune System. Oxford Univ. Press, New York, pp. 66. Means, E.D., Anderson, D.K., 1983. Neuronophagia by leukocytes in experimental spinal cord injury. J. Neuropathol. Exp. Neurol. 42, 707 – 719. Muizelaar, J.P., Marmarou, A., Young, H.F., Choi, S.C., Wolf, A., Schneider, R.L., Kontos, H.A., 1993. Improving the outcome of severe head injury with the oxygen radical scavenger polyethylene glycolconjugated superoxide dismutase: a phase II trial. J. Neurosurg. 78, 375 – 383. Murphy, P. (Ed.), 1976. The Neutrophil. Plenum Publ. Corp., New York. Noble, L.J., Donovan, F., Igarashi, T., Goussev, S., Werb, Z., 2002. Matrix metalloproteinases limit functional recovery after spinal cord injury by modulation of early vascular events. J. Neurosci. 22, 7526 – 7535. Rossi, F., Bianca, V.D., Grzeskowiak, M., Zeni, L., 1986. Mechanisms of oxygen free radicals production in granulocytes. In: Novelli, Ursini (Eds.), Oxygen Free Radicals in Shock. Karger, Basel, pp. 15 – 28. Smedly, L.A., Tonnesen, M.G., Sandhaus, R.A., et al., 1986. Neutrophilmediated injury to endothelial cells. J. Clin. Invest. 77, 1233 – 1243. Sorkin, L.S., Steinman, J.L., Hughes, M.G., Willis, W.D., McAdoo, D.J., 1988. Microdialysis recovery of serotonin released in spinal cord dorsal horn. J. Neurosci. Methods 23, 131 – 138.

Stadtman, E.R., 1993. Oxidative of free amino acids and amino acid residues in proteins by radiolysis and by metal-catalyzed reactions. Annu. Rev. Biochem. 62, 721 – 797. Streit, W.J., Semple-Rowland, S.L., Hurley, S.D., Miller, R.C., Popovich, P.G., Stokes, B.T., 1998. Cytokine mRNA profiles in contused spinal cord and axotomized facial nucleus suggest a beneficial role for inflammation and gliosis. Exp. Neurol. 152, 74 – 87. Taoka, Y., Okajima, K., 1998. Spinal cord injury in the rat. Prog. Neurobiol. 56, 341 – 358. Taoka, Y., Naruo, M., Koyanagi, E., Urakado, M., Inoue, M., 1995. Superoxide radicals play important roles in the pathogenesis of spinal cord injury. Paraplegia 33, 450 – 453. Taoka, Y., Okajima, K., Uchiba, M., Murakami, K., Kushimoto, S., Johno, M., Naruo, M., Okabe, H., Takatsuki, K., 1997a. Gabexate mesilate, a synthetic protease inhibitor, prevents compression-induced spinal cord injury by inhibiting activation of leukocytes in rats. Crit. Care Med. 125, 874 – 879. Taoka, Y., Okajima, K., Uchiba, M., Murakami, K., Kushimoto, S., Johno, M., Naruo, M., Okabe, H., Takatsuki, K., 1997b. Role of neutrophils in spinal cord injury in the rat. Neuroscience 79, 1177 – 1182. Taoka, Y., Okajima, K., Murakami, K., Johno, M., Naruo, M., 1998. Role of neutrophil elastase in compression-induced spinal cord injury in rats. Brain Res. 799, 264 – 269. Tator, C.H., Koyanagi, I., 1997. Vascular mechanisms in the pathophysiology of human spinal cord injury. J. Neurosurg. 86, 483 – 492. Test, S.T., Weiss, S.J., 1984. Quantitative and temporal characterization of the extracellular H2O2 pool generated by neutrophils. J. Biol. Chem. 259, 399 – 405. Tonai, T., Shiba, K.-I., Taketani, Y., Ohmoto, Y., Murata, K., Muraguchi, M., Ohsaki, H., Takeda, E., Nishisho, T., 2001. A neutrophil elastase inhibitor (ONO-5046) reduces neurologic damage after spinal cord injury in rats. J. Neurochem. 78, 1064 – 1072. Weiss, S.J., 1989. Tissue destruction by neutrophils. N. Engl. J. Med. 320, 365 – 375. Xu, J., Hsu, C.Y., Liu, T.H., Hogan, E.L., Perot, P.L., Tai, H.H., 1990. Leukotriene B4 release and polymorphonuclear cell infiltration in spinal cord injury. J. Neurochem. 55, 907 – 912. Xu, J., Beckman, J.S., Hogan, E.L., Hsu, C.Y., 1991. Xanthine oxidase in experimental spinal cord injury. J. Neurotrauma 8, 11 – 18. Xu, G.-Y., McAdoo, D.J., Hughes, M.G., de Castro Jr., R., 1998. Considerations in the determination by microdialysis of resting extracellular amino acid concentrations and release upon spinal cord injury. Neuroscience 86, 1011 – 1021. Zimmerman, J.J., Ciesielski, W., Lewandoski, J., 1997. Neutrophilmediated phospholipid peroxidation assessed by gas chromatographymass spectroscopy. Am. J. Physiol. 42, C653 – C661.