Traumatic Brain Injury Induces Prolonged Accumulation of ...

JOURNAL OF NEUROTRAUM A Volume 18, Number 9, 2001 Mary Ann Liebert, Inc.

Traumatic Brain Injury Induces Prolonged Accumulation of Cyclooxygenase-1 Expressing Microglia/Brain Macrophages in Rats JAN MARKUS SCHWAB, KARIN SEID, and HERMANN J. SCHLUESENER

ABSTRACT Inflammatory cellular responses to brain injury are promoted by proinflammatory messengers. Cyclooxygenases (prostaglandin endoperoxide H synthases [PGH]) are key enzymes in the conversion of arachidonic acid into prostanoids, which mediate immunomodulation, mitogenesis, apoptosis, blood flow, secondary injury (lipid peroxygenation), and inflammation. Here, we report COX-1 expression following brain injury. In control brains, COX-1 expression was localized rarely to brain microglia/macrophages. One to 5 days after injury, we observed a highly significant (p , 0.0001) increase in COX-11 microglia/macrophages at perilesional areas and in the developing core with a delayed culmination of cell accumulation at day 7, correlating with phagocytic activity. There, cell numbers remained persistently elevated up to 21 days following injury. Further, COX-11 cells were located in perivascular Virchow-Robin spaces also reaching maximal numbers at day 7. Lesion-confined COX-11 vessels increased in numbers from day 1, reaching the maximum at days 5–7. Double-labeling experiments confirmed coexpression of COX-1 by ED-11 and OX-421 microglia/ macrophages. Transiently after injury, most COX-11 microglia/macrophages coexpress the activation antigen OX-6 (MHC class II). However, the prolonged accumulation of COX-11 , ED-11 microglia/macrophages in lesional areas enduring the acute postinjury inflammatory response points to a role of COX-1 in the pathophysiology of secondary injury. We have identified localized, accumulated COX-1 expression as a potential pharmacological target in the treatment of brain injury. Our results suggest that therapeutic approaches based on long-term blocking including COX-1, might be superior to selective COX-2 blocking to suppress the local synthesis of prostanoids. Key words: bystander damage; inflammation; prostaglandin; tissue remodeling

INTRODUCTION

S

results from a complex sequence of pathophysiological events that evolve over time and space (Dusart and Schwab, 1994; Ideka and Long, 1990). Subsequent to traumatic injury, the inECONDARY TRAUMATIC INJURY

tracellular influx of the second messenger Ca21 activates phospholipase A2 and cyclooxygenases. The ensuing lipid peroxydation induces cell membrane damage and release of toxic prostanoids, and is supported by generation of reactive oxygen species (ROS). ROS overwhelm endogenous scavenging mechanisms and downstream

Institute of Brain Research, University of Tuebingen, Medical School, Tuebingen, Germany.

881

SCHWAB ET AL. conduct secondary injury (bystander damage) executed by apoptosis and inflammation (Hall, 1996; Mattson, 1998; Wilberger, 1996). Major contributors to the intrinsic inflammatory CNS response following acute injury are prostanoids (prostaglandins PGE2, PGD2, PGF2a ; thromboxanes TXA2; and prostacyclins PGI2), which are synthesized by prostaglandin H synthetase, and the rate-limiting enzyme COX (Lee et al., 1999; Minghetti and Levi, 1998; Portanova et al., 1996). Several isoforms of COX are known: the constitutively expressed COX-1, the rapidly inducible COX-2, and a recently proposed isoform termed COX-3 (Willoughby et al., 2000). Although striking differences have been observed in the structure and regulation of the COX-1 and COX-2 genes, protein structure and the enzymatic function are remarkably similar (Crofford, 1997; DuBois et al., 1998; Smith et al., 1996; Williams and DuBois, 1996). The current understanding of intrinsic reactive mechanisms after acute CNS injury, like stroke (Lee et al., 1999; Nogawa et al., 1997; Sairanen et al., 1998) and spinal cord injury (Resnick et al., 1998), identified COX2 metabolites as major neurotoxic mediators with tightly regulated expression induced by pathogenic stimuli. The role of activated microglia and proinflammatory COX-1, which shares identical enzymatic function after traumatic brain injury, remained enigmatic (Mallat and Chamak, 1994). In order to provide a pathophysiological basis for the involvement of COX-1 in traumatic brain injury, we have analyzed COX-1 expression after stab wound brain injury and in neuropathologically unaltered rats.

MATERIALS AND METHODS Animals and Cortical Stab Wound Injury Briefly, rats were anesthetized with 7.5% chloralhydrate (0.6 mL/100 g; Merck Darmstadt, Germany). After

shaving, the skin of the head was cleaned and disinfected with 90% alcohol, and a midline incision was made to expose the skull. After drilling a 2-mm hole sited 3 mm to the right and 3 mm posterior to bregma (bregma/23), the rat was fixed in a stereotactic apparatus (ASI Instruments, Germany) to perform cortical stab wound injury. The needle of a Hamilton syringe (900 series; Hamilton Company, Reno, NV) was inserted 4 mm from the pial surface vertically into the brain. The subsequent injection of 5 mL of phosphate-buffered saline (PBS), pH 7.4, was performed over a 2-min period to increase the damage. After this, the needle was slowly withdrawn and the skin was reapposed with a suture (4/0, FS-2, Ethicon; Johnson & Johnson, Brussels, Belgium). Control rats (n 5 3) received anesthesia without injuring the brain. Brain tissue was removed, postfixed overnight at 4°C, and embedded in paraffin. Coronal sections of 5 mm (61 mm) targeting the cortical injury area were used for analysis. Findings were compared with three age- and sexmatched normal control rats.

Immunohistochemistry Eighteen male Sprague-Dawley rats (Charles River, Sulzfeld, Germany) were randomly assigned to control or brain stab wound injury groups based on the method of Maxwell et al. (1990) and were perfused with 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) at 1, 3, 5, 7, 14, or 21 days later (three rats per group). After formaldehyde fixation and paraffin-embedding, rehydrated sections were boiled (in 600-W microwave oven) seven times for 5 min in citrate buffer (2.1 g sodium citrate/L, pH 6). Endogenous peroxidase was inhibited with 1% H2O 2 in methanol (1:10; 15 min). Sections were incubated with 10% normal porcine serum (Biochrom, Berlin, Germany) to block nonspecific binding of immunoglobulins. Monospecific polyclonal antibodies directed against COX-1 (Santa Cruz Biotechnology, CA)

FIG. 1. Specificity of COX-1 antibody was confirmed by selective inhibition of staining after preincubation for 3 h on ice with 20-fold excess of the COX-1 peptide (B), whereas adding of COX-2 peptide did not block staining (A). Incubating the sections without the primary COX-1 antibody abolished the immunoreactivity signal (C,D) excluding unspecific signals mediated by the secondary antibody. Compared to control rat brains (E), in lesioned rat brains, we observed a strictly lesion associated accumulation of COX-11 cells like microglia/ macrophages at the necrotic lesion site (F) along the stab channel (stars). From day 3 on, COX-11 cells were found “moving” towards a perineuronal position (G). Activated, rod-shaped COX-11 microglia phenotypes were confined to the lesion core and perilesional areas (F), whereas remote from the core we identified few COX-11 microglial phenotypes with fine elongations (H). Another significant population of COX-11 microglia/macrophages demonstrated cytoplasmatic vacuoles and large round nuclei indicating morphologically hallmarks of phagocytic, lipid loaden, “foamy” microglia/macrophages (I, arrows). In addition, augmented lesional COX-11 cells numbers were also due to accumulation of perivascular COX-11 cells forming clusters in Virchow-Robin spaces (J), predominantly draining from venous vessels. Also in perilesional areas, the numbers of COX-11 endothelial cells increased after stab wound injury (K). Double-labeling experiments identified the majority of COX-11 cells (brown, DAB) located at the lesion site coexpressing the ED1 antigen (fast blue; L,M), which were prevailingly observed in territories of phagocytic activity. No COX-11 cells (brown, DAB) were identified to coexpress W3/13 (T-lymphocytes; N) or GFAP (O) antigens (fast blue). Bar-100 mm (E,F,I,L); 50 mm (A–D,G,H,J,K,M,N,O).

882

TBI AND PROLONGED ACCUMULATION OF COX-1 were diluted (1:400) in 1% bovine serum albumin (BSA) Tris-balanced salt solution (TBS; containing 0.025 M Tris, 0.15 M NaCl) and incubated for 1 h at room temperature. Specific binding of the antibodies were detected with a secondary biotinylated swine anti-rabbit IgG F(ab)2 antibody fragment 1:400 for 30 min (Dako, Hamburg, Germany), followed by incubation with aperoxidase-conjugated streptavidin–biotin complex (Dako).

The enzyme was visualized with diaminobenzidine as a chromogen (Fluka, Neu-Ulm, Germany). Sections were counterstained with Mayer’s Hemalaun. Specificity of COX-1 antibody was confirmed by selective inhibition of staining after preincubation for 3 h on ice with 20-fold excess of the COX-1 peptide (Fig. 1B), whereas adding of COX-2 peptide did not block staining (COX-1 and -2 control peptides, C-20, Santa Cruz Biotechnology, CA;

FIG. 1

883

SCHWAB ET AL. Fig. 1A). Negative controls consisted of sections incubated in the absence of the primary antibody (Fig. 1C,D).

Double Labeling Experiments Briefly, slices were deparaffinized, irradiated in a microwave oven for antigen retrieval and incubated with nonspecific porcine serum as described above. In double-labeling experiments, we first labeled a cell-type specific antigen like glial fibrillary acidic protein (GFAP; Boehringer Mannheim, Germany), 1:100, to detect astrocytes, myelin basic protein (MBP; Dako, Glostrup, Denmark), 1:200, to detect oligodendrocytes or neurofil-

ament (Dako, Glostrup, Denmark), 1:100, for neuronal identification. Immune cells were labeled with monoclonal antibodies against ED1 (Serotec, Oxford, U.K.) 1:100, for microglia/macrophage detection using the ABC procedure in combination with alkaline phosphatase conjugates. In addition, monoclonal antibodies directed against OX-22 (Serotec), 1:100, were used for B lymphocyte and W3/13 (Serotec), 1:100, for T lymphocyte identification. To characterize the functional immunecompetence OX-6 (Serotec), 1:100, was labeled to identify MHC class II molecules. Furthermore, in order to characterize the cellular proliferative response, we used a monoclonal cell nuclear antigen (PCNA; Dako,

FIG. 2. Results from all injured rat brains (days 1–21) demonstrated a strictly lesion-associated accumulation of COX-11 cells, like parenchymal and perivascular microglial/ macrophages, endothelial, and smooth muscle cells (p , 0.0001) compared remote areas or control rats (A,C,E). The temporal COX-1 expression pattern demonstrated a significant accumulation of COX-11 cells already at day 1 (p , 0.0001), which increased further at day 5 until day 7 (p , 0.0001) (B). COX-11 vessels (MLV 6 SEM, endothelial and/or smooth muscle cells) increased from day 1 to the maximum at day 5 and were given in percent (D). Accordingly, COX-11 perivascular spaces (Virchow-Robin spaces) containing two or more COX-11 cells (in percent) were frequently observed, augmenting at day 1, with 3 reaching a maximal number at day 7 (C). In general, the accumulation of lesional COX11 cells, such as microglia/macrophages (in parenchyma and perivascular spaces) and vessels (endothelial cells and smooth muscle cells), was evident up to 3 weeks following stab wound injury as compared to control brain (p , 0.0001; B), whereas numbers of COX-11 perivascular clusters and vessels declined after 3 weeks to almost base levels (D,F).

884

TBI AND PROLONGED ACCUMULATION OF COX-1 Glostrup, Denmark), 1:100, antibody. Qualitative entity of perivascular cells in Virchow-Robin spaces was determined using monoclonal antibodies against ED2 (Serotec), 1:200, to identify resident periarterial microglia/macrophages and CD14 (Santa Cruz Biotechnology, CA), 1:100, additionally characterizing nonresident, peripheral blood-derived monocytes (PBMC; Becher and Antel, 1996) infiltrating primarily from perivenous spaces. Antibodies were added to the slices at the given solution in a 1% BSA TBS solution (containing 0.025 M Tris, 0.15 M NaCl, pH 7.5). Visualization was achieved by adding biotinylated secondary antibodies (1:400) for 30 min and alkaline phosphatase-conjugated ABC complex diluted 1:400 in TBS-BSA for 30 min. Consecutively, we developed with Fast-Blue BB salt chromogensubstrate solution, yielding a blue reaction product. Between double-labeling experiments, slices were irradiated in a microwave for 20 min in citrate buffer. Then, COX-1 was immunolabeled as described above.

Histologic Staining for Myelin and Nuclei Serial tissue sections to those used for immunohistochemistry were stained for myelin using luxol fast blue.

FIG. 2.

The area of tissue that was obviously damaged or lacked myelin was identified along the stab wound channel in the injured cortex, dentate gyrus, hilus, and hippocampal fissure. Nuclei were stained with cresyl violet (0.1%) to identify intact versus damaged areas of gray matter.

Evaluation, Statistical Analysis, and Stereological Methodology Mean numbers of COX-11 cells from injured brain territories were compared to naive control rats. COX-1 immunopositive cells were counted in 10 high-power fields (HPF, 3200 magnification with an eye-piece grid representing 0.25 mm2 ). Because a single COX-11 cell may appear on more than one section, double counting of the same cell was prevented by strictly counting of cell bodies with nucleus. Isolated cell processes were excluded. Data were calculated as means of labeled cells/mm2 (MLC 6 SEM) and compared using the two-tailed Student’s t test. Additionally, in order to evaluate presence of COX-11 cells in perivascular Virchow-Robin spaces, 10 vessels in the perilesional area were counted and considered positive if at least a minimum of two COX-11 cells were present. COX-11 vessels, revealing COX-11

Continued.

885

SCHWAB ET AL. endothelial or/and smooth muscle cells, were analyzed in the same way.

RESULTS Brains of 18 rats with stab wound injury and three control brains were evaluated for COX-1 expression by immunohistochemistry. In control rat brains without neuropathologic alterations, COX-1 immunoreactivity was detected in few microglia/macrophages with predominantly rod-shaped morphology (Fig. 1E). These cells accounted for 2–5% of all cells located predominantly in white matter territories (MLC 5 5, SEM 5 0.52), and only single cells were detected in perivascular spaces (3.3%). Only rare COX-11 cells coexpressed activation (OX-6, ,5%) or phagocytic markers (ED1, ,5%), suggesting a predominantly resting, nonphagocytic microglial immunophenotype. Further, few endothelial and smooth muscle cells were labeled. In addition, COX-1 immunoreactivity was detected in ependymal and meningeal cells and single neurons. We have analyzed the number and distribution of COX-11 cells after stab wound injury. Stab wound injury lesions produce consistent histological outcomes (Fujita et al., 1998; Maxwell et al., 1990, 1998). CNS tissue spared at the lesion center was approximately 1–2% of total cross-sectional area of the brain. We observed a strictly lesion associated accumulation of COX-11 cells like microglia/macrophages, endothelial, and smooth muscle cells. COX-11 microglia/macrophages accumulated at the necrotic lesion site and at the developing perilesional areas along the stab channel in the injured cortex, dentate gyrus, hilus, and hippocampal fissure (Fig. 2A,B) and in perivascular Virchow-Robin spaces (Fig. 2C,D). Using the Student’s t test, we detected a significantly (p , 0.0001) higher number of COX-1 expressing microglia/macrophages along the stab wound lesion site (MLC 5 26, SEM 5 1.2; Fig. 1F) than in control tissue (MLC 5 5, SEM 5 0.5; Fig. 2E,F). In these areas of both, primary and delayed secondary injury, the number of COX-11 cells started to accumulate already at day 1 and further increased at day 5 (p , 0.0001, MLC 5 31, SEM 5 3.3) until day 7 (p , 0.0001, MLC 5 40, SEM 5 1.6; Fig. 2B) and remained elevated for up to 3 weeks (MLC 5 37, SEM 5 2.4; Fig. 2B). Culmination of COX-11 cell numbers was delayed until day 7 in comparison to activation of microglia/macrophages or influx of blood-borne macrophages reaching a maximum at day 3–5 (Fujita et al., 1998). From day 3 on, COX-11 cells were found “moving” towards a perineuronal position (Fig. 1G). The numbers of COX-11 macrophages/mi-

croglia were correlated with the severity of tissue damage as determined by LFB staining, indicating myelin breakdown in damaged axon fibers. COX-1–expressing cells were characterized by morphological hallmarks of both ramified and amoeboid microglia. Activated COX11 microglia phenotypes were confined to the lesion core (Fig. 1F) and perilesional areas, but remote from the core we identified only few COX-11 microglial phenotypes with fine elongations (Fig. 1H). Amoeboid COX-11 cells prevail already at day 1 and persist during the first 2 weeks following injury. With aging of the lesion also few parenchymal ramified COX-11 cells were observed. Another significant population of COX-11 microglia/ macrophages, observable as early as day 3, also demonstrated cytoplasmatic vacuoles and large round nuclei indicating morphologically hallmarks of phagocytic, lipid loaden, “foamy” microglia/ macrophages (Fig. 1I). These cells persisted in the necrotic lesion core and represented the majority of the COX-11 cells at 3 weeks following injury. Following stab wound injury, increased numbers of COX-11 cells were also due to accumulation of pervascular COX-11 cells in Virchow-Robin spaces (p , 0.0001, 24 6 3%; Fig. 1J), as compared to control brains (3 6 3%). Already after day 1, in contrast to total COX11 cell counts, pronounced numbers of Virchow-Robin spaces (20% of counted vessels) contained COX-11 microglia/macrophages (Fig. 2D). These numbers of COX11 perivascular spaces further increased at day 5 (p , 0.0001, 33 6 6.7%) reaching maximum levels at day 7 following injury (p , 0.0001, 40 6 10%). Also in perilesional areas, the numbers of COX-11 vessels, characterized by immunopositive endothelial and/or smooth muscle cells, increased after stab wound injury (p , 0.0001, MLV 5 22.6%, SEM 5 4.4; Fig. 1K) as compared to control brains (MLV 5 3.3%, SEM 5 3.3; Fig. 2C,D). In detail, numbers of COX-11 vessels increased already at day 1 (Fig. 2D), rose further until day 5 (MLV 5 36.7%, SEM 5 21) and persisted on maximal levels until day 7 (MLV 5 36.7%, SEM 5 4.4). In contrast to linear accumulation of COX11 microglia/macrophages, COX-11 vessels and perivascular spaces demonstrated a bimodal distribution reaching a first discrete maximum at day 1 followed by a decrease at day 3 and a second maximum at day 5. In general, the accumulation of COX-11 cells such as microglia/macrophages (in parenchyma and perivascular spaces) and vessels (endothelial cells and smooth muscle cells) was evident for up to 3 weeks following stab wound injury as compared to control brains (p , 0.0001; Fig. 2A–E). In some cases, perikarya and processes of large neurons also expressed COX-1 in lesioned and normal brains.

886

TBI AND PROLONGED ACCUMULATION OF COX-1 Double-labeling experiments were performed in order to characterize the cellular origin of COX-1 expression. In addition to neurofilament labeling, neurons were distinguished from nonneuronal cells by nuclear size, shape, and presence of a nucleolus. The majority of COX-11 cells located at the lesion site coexpressed the ED1 antigen (.80%; Fig. 1L,M), OX-42 (.80%) and transiently OX 6 (MHC class II molecules; .75%). COX-11 cells coexpressing ED1 were prevailingly observed in territories of phagocytic activity, such as the pan-necrotic lesion core suggesting a predominant phagocytic immunophenotype. In contrast, coexpression with OX42 preceded ED1 coexpression and was not restricted to areas of phagocytic activity. No COX-11 cells were identified to coexpress OX-22 (B-lymphocytes), W3/131 (Tlymphocytes; Fig. 1N), GFAP (Fig. 1O) or MBP. In core regions of the lesions, a 30–50% match between COX11 and PCNA-labeled cells was observed. In order to rule out the entity of perivascular COX-11 cells in Virchow-Robin spaces, we figured out whether COX-11 cells coexpress ED-2 antigens (prevailingly periarterial scavenger cells, constitutively present) and/or CD14 antigens additionally characterizing infiltrative cells (prevailingly perivenous, not constitutively present). We observed only rare coexpression with ED2 antigens (,20%) at different time points (days 1–7), whereas up to 60% demonstrated coexpression with CD14 antigens from day 1 to 7. There was no enhanced ED2 expression at day 1 (first peak), and during the second peak (day 5–7) we observed pronounced numbers of COX-11 cells coexpressing CD14 (50–60%). Therefore, these results suggest that perivascular COX-11 , ED22 , CD141 cells were predominantly nonresident, infiltrative cells (hematogenous macrophages) draining into the brain from postcapillary perivenular spaces.

DISCUSSION In the present study, we analyzed the localization and time course of COX-1 expression. Compared to control brains, we observed a significant, presistent accumulation of COX-11 , ED-11 , OX421 microglia/ macrophages, and endothelial and smooth muscle cells in areas of primary (lesion core) and secondary injury (perilesional areas). The majority of COX-11 microglia/macrophages transiently coexpressed the activation marker OX-6 (MHC class II glycoproteins). Further, in injured brains, an accumulation of COX-1, ED22 , CD141 microglia/macrophages were observed in perivascular Virchow-Robin spaces. In addition, lesionconfined COX-11 vessels increased in numbers following stab wound injury.

To date, the COX-1 counterpart, the highly inducible COX-2, has been associated with inflammatory, neoplastic, degenerative, ischemic, and traumatic CNS pathologies (Deininger and Schluesener, 1999; Deininger et al., 1999; Sairanen et al., 1998; Seibert et al., 1994; Smith et al., 1998; Williams and Dubois, 1996). Therefore, recent brain injury studies focused on the COX-2 isoform (Resnick et al., 1998; Sairanen et al., 1998). However, the idea of a unique role for COX-2 as the only responsible, inducible inflammatory mediator has been complicated (Giroy et al., 1999; Yamagata et al., 1993) and questioned (Langenbach et al., 1995; Morham et al., 1995), and even a protective role has been discussed (Dash et al., 2000). Associated with tissue homeostasis, the expression of the noninducible COX-1 isoform as the source of physiologically important prostaglandins (PGs) did not gain much attention, although it leads to the identical proinflammatory and deleterious products. COX-1 is involved in the physiologic response modulating blood flow and the differentiation of neuronal (Kaplan et al., 1997), stem (Shillabeer et al., 1998), and monocytic cells (Hoff et al., 1993). Little data on a pathophysiologic role of COX-1 is available. COX-1 expression induces tumorgenic transformation of immortalized endothelial cells (ECV) and reduces chemotoxicity of treated C6 glioblastoma cells by modifying properties of multidrug-resistance–related pumps (Roller et al., 1999). Pathology-associated accumulation of COX-11 microglia was evident in amyloid plaques of Alzheimer’s disease (AD) and following human focal infarction, suggesting a pathology-related, localized increase of COX-1 (Schwab et al., 2000; Yermakova et al., 1999). Likewise, little is known about the elements involved in regulating COX-1 gene expression. COX-1 is a “delayed response” gene (Kaplan et al., 1997), and its transcription is inducible by bFGF, by EGF in mouse myeloblastoma (MC3T3) cells, and by NGF on neurogenic rat pheochromocytoma cells (PC12), which were used as a model to study neurite outgrowth. Reactive microglia, a sensor for pathological events in the CNS (Kreutzberg, 1996), is essential to tissue remodeling and one of the major sources of prostaglandins in the CNS (Matsuo et al., 1995; Minghetti and Levi, 1998). The role of microglia in lipid oxygenation is supported by its capacity to produce PLA2 , the vasoactive PGE 2 , TXA, and COX-1 (Matsuo et al., 1995; Mingheti and Levi, 1998; Yermakova et al., 1999). Since COX activity is important following disturbed CNS maintenance contributing to homeostatic conditions by modulation of blood homeostasis (vasoconstriction, platelet aggregation, angiogenesis, blood pressure; Crofford, 1997; DuBois, 1996; DuBois et al., 1998; Smith et al., 1996; Tsujii et al., 1998; Williams and; Yasojima et al., 1999),

887

SCHWAB ET AL. inflammation, and secondary injury (lipid peroxygenation; Dubois et al., 1998), we defined COX-1 expression following experimental brain injury. COX-1 appeared to be expressed constitutively only by some endothelial cells, neurons, and microglia/macrophages. Following brain injury, COX-11 microglia/macrophages accumulated at the lesion site. These results suggest a paracrine role of COX-1 metabolites in the modification of the postinjury microenvironment. In detail, the delayed culmination of COX-11 microglia/macrophages at day 7 and the persistence of COX-11 “foamy cells” in the necrotic lesion core indicates postactivation phagocytic activity. Furthermore, exerting the same enzymatic activity as COX-2, our data implicates COX-1 in the monocyte/macrophage mediated secondary (bystander) damage during lipid hydrolysis by converting the membrane phospholipid product, arachidonic acid (AA). COX-1 converts AA into PGG2 (bis-oxygenase activity) and further into PGH2 (peroxygenase activity), and gives rise to subsequently metabolized free radical reactive oxygens species (ROS; Minghetti and Levi, 1998; Taoka and Ojajima, 1998). COX-1 might also be implicated in other secondary injury mechanisms such as edema initiated by progressive endothelial damage and vasogenic dysfunction (Ideka and Long, 1990; Sanders et al., 1978). This is supported by the accumulation of COX-11 vessels confined to the lesion epicenter from day 3 until day 7, a crucial period where edema formation occurs (Ideka and Long, 1990). Further, accumulation of COX-11 microglia/ macrophages in Virchow-Robin spaces reached the maximum at day 3, a crucial time point for invading lymphocytes and monocytes (Dusart and Schwab, 1994; Popovich et al., 1997). Virchow-Robin spaces are of particular neuroimmunological interest as here lymphocyte drainage and monocyte transmigration occurs (Streit et al., 1999). Since COX-11 cells were present in VirchowRobin spaces and accumulate at time points of infiltrating cells, COX-11 , in significant numbers, were bloodborne. Several lines of evidence suggest a pathophysiologic, phagocytic role of postinjury COX-1 expression by microglia/macrophages. Microglia/macrophages are known to promote angiogenesis (Leibovich et al., 1987) and neuroregeneration (Perry and Brown, 1992; Rapalino et al., 1999) but also to generate toxic agents (Guilian, 1993). As microglial cells are key players in tissue reorganization following brain injury, microglial COX-1 generation might be a crucial process in tissue homeostasis. We have identified localized, accumulated COX-1 expression as a potential substrate and proinflammatory pharmacological target for NSAID action in human brain following brain injury. These results indicate that local increase in COX

activity can be due to both transient local up-regulation of COX-2 expression and to the accumulation of infiltrating COX-1 expressing inflammatory cells. These findings substantially challenge the current paradigms of a selective role of COX-2 in posttraumatic CNS injury.

ACKNOWLEDGMENTS We thank Ingrid Nagel for critically reading the manuscript and Gudrun Albrecht for phototechnical work. This work was supported by a grant of the Federal Ministry of Education, Science, Research and Technology (BMBF; no. 01K09809/8).

REFERENCES BECHER, B., and ANTEL, J.P. (1996). Comparison of phenotypic and functional properties of immediately ex vivo and cultured human adult microglia. Glia 18, 1–10. CROFFORD, L.F. (1997). COX-1 and COX-2 tissue expression: implications and predictions. J. Rheumatol. 24, 15–19. DASH, P.K., MACH, S.A., and MOORE, A.N. (2000). Regional expression and role of cyclooxygenase-2 following experimental traumatic brain injury. J. Neurotrauma 17, 69–81. DEININGER, M.H., and SCHLEUESENER, H.J. (1999). Cyclooxygenases-1 and -2 are differentially localized to microglia and endothelium in rat EAE and glioma. J. Neuroimmunol. 95, 202–208. DEININGER, M.H., WELLER, M., STREFFER, J., et al. (1999). Patterns of cyclooxygenase-1 and -2 expression in human gliomas in vivo. Acta Neuropathol. 98, 240–244. DIRNAGL, U., IADECOLA, C., and MOSKOWITZ, M.A. (1999). Pathobiology of ischemic stroke: an integrated view. Trends Neurosci. 22, 391–396. DU BOIS, R.N., ABRAMSON, S.B., CROFFORD, L., et al. (1998). Cyclooxygenase in biology and disease. FASEB J. 12, 1063–1073. DUSART, I., and 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. FUJITA, T., YOSHIMINE, T., MARUNO, M., et al. (1998). Cellular dynamics of macrophages and microglial cells in reaction to stab wounds in rat cerebral cortex. Acta Neurochir. 140, 275–279. GIROY, D.W., COLVILLE-NASH, P.R., WILLIS, D., et al. (1999). Inducible cyclooxygenase may have antiinflammatory properties. Nat. Med. 5, 698–701. GUILIAN, D. (1993). Reactive glia as rivals in regulating neuronal survival. Glia 7, 102–118. HALL, E.D. (1996). Free radicals and lipid peroxydation, in:

888

TBI AND PROLONGED ACCUMULATION OF COX-1 Neurotrauma. R.K. Narayan, J.E. Wilberger, and J.T. Povlishock (eds), McGraw-Hill: New York, pps. 1405–1419. HOFF, T., DE WITT, D., KAEVER, V., et al. (1993). Differentiation associated expression of prostaglandin G/H synthase in monocytic cells. FEBS Lett. 320, 38–42. IADECOLA, C., FORSTER, C., NOGAWA, S., et al. (1999). Cyclooxygenase-2 immunoreactivity in the human brain following cerebral ischemia. Acta Neuropathol. 98, 9–14. KAPLAN, M.D., OLSCHOWKA, J.A., and O’BANION, M.K. (1997). Cyclooxygenase-1 behaves as a delayed response gene in PC12 cells differentiated by nerve growth factor. J. Biol. Chem. 272, 18534–18537. KREUTZBERG, G.W. (1996). Microglia: a sensor for pathological events in the CNS. Trends Neurosci. 19, 312–318. LANGENBACH, R., MORHAM, S.G., TIANO, H.F., et al. (1995). Prostaglandin synthase 1 gene disruption in mice reduces arachidonic acid-induced inflammation and indomethacin-induced gastric ulceration. Cell 83, 483–492. LEE, J.M., ZIPFEL, G.J., and CHOI, D.W. (1999). The changing landscapes of ischemic brain injury. Nature 399, A7–A14. LEIBOVICH, S.J., POLVERINI, P.J., SHEPARD, H.M., et al. (1987). Macrophage-induced angiogenesis is mediated by tumour necrosis factor–a. Nature 379, 630–632. MALLAT, M., and CHAMAK, B. (1994). Brain macrophages: neurotoxic or effector cells? J. Leukocyte Biol. 56, 416–422. MATSUO, M., HAMASAKI, Y., FUJIYAMA, F., et al. (1995). Eicosanoids are produced by microglia, not by astrocytes, in rat glial cell cultures. Brain Res. 685, 201–204. MATTSON, M.P. (1998). Modification of ion homeostasis by lipid peroxydation: roles in neural degeneration and adaptive plasticity. Trends Neurosci. 21, 53–57. MAXWELL, W.L., FOLLOWS, R., ASHHURST, D.E., et al. (1990). The response of the cerebral hemisphere to the rat injury. I. The mature rat. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 328, 479–500. MINGHETTI, L., and LEVI, G. (1998). Microglia as effector cells in brain damage and repair: focus on prostanoids and nitric oxide. Prog. Neurobiol. 54, 99–125. MORHAM, S.G., LANGENBACH, R., LOFTIN, C.D., et al. (1995). Prostaglandin synthase 2 gene disruption causes severe renal pathology in the mouse. Cell 83, 473–482.

PORTANOVA, J.P., ZHANG, Y., ANDERSON, G.D., et al. (1996). Selective neutralization of prostaglandin E2 blocks inflammation, hyperalgesia, and Interleukin 6 production in vivo. J. Exp. Med. 184, 883–891. RAPALINO, O., LAZAROV-SPIEGLER, O., AGRANOV, E., et al. (1999). Implantation of stimulated homologous macrophages results in partial recovery of paraplegic rats. Nat. Med. 4, 814–821. RESNICK, D.K., GRAHAM, S.H., DIXON, E., et al. (1998). Role of cyclooxygenase 2 in acute spinal cord injury. J. Neurotrauma 15, 1005–1013. ROLLER, A., BAHR, O.R., STREFFER, J., et al. (1999). Selective potentiation of drug cytotoxicity by NSAID in human glioma cells: the role of COX-1 and MRP. Biochem. Biophys. Res. Commun. 259, 600–605. SAIRANEN, T., RISTIMÄKI, A., KARJALAINEN-LINDSBERG, M.J., et al. (1998). Cyclooxygenase-2 is induced globally in infarcted human brain. Ann. Neurol. 43, 738–747. SANDERS, K.M. (1978). Endogenous prostaglandin E and contractile activity of isolated ileal smooth muscle. Am. J. Physiol. 234, E209–E212. SCHWAB, J.M., NGUYEN, T.D., POSTLER, E., et al. (2000). Selective accumulation of cyclooxygenase-1–expressing microglial cells/macrophages in lesion of human focal cerebral ischemia. Acta Neuropathol. 99, 609–614. SEIBERT, K., ZHANG, Y., LEAHY, K., et al. (1994). Pharmacological and biochemical demonstration of the role of cyclooxygenase 2 in inflammation and pain. Proc. Natl. Acad. Sci. U.S.A. 91, 12013–12017. SHILLABEER, G., KUMAR, V., TIBBO, E., et al. (1998). Arachidonic acid metabolites of the lipoxygenase as well as the cyclooxygenase pathway may be involved in regulating preadipocyte differentiation. Metabolism 47, 461–466. SMITH, W.L., GARAVITO, R.M., and DE WITT, D.L. (1996). Prostaglandin endoperoxide H synthetases (Cyclooxygenases)–1 and –2. J. Biol. Chem. 271, 33157–33160. SMITH, C.C., ZHANG, Y., KOBOLDT, C.M., et al. (1998). Pharmacological analysis of cyclooxygenase-1 in inflammation. Proc. Natl. Acad. Sci. U.S.A. 95, 13313–13318. STREIT, W.J., WALTER, S.A., and PENNELL, N.A. (1999). Reactive microgliosis. Prog. Neurobiol. 57, 563–581. TAOKA, Y., and OKAJIMA, K. (1998). Spinal cord injury in the rat. Prog. Neurobiol. 56, 341–358.

NOGAWA, S., ZHANG, F., ROSS, M.E., et al. (1997). Cyclooxygenase-2 gene expression in neurons contributes to ischemic brain damage. J. Neurosci. 17, 2746–2755.

TSUJII, M., KAWANO, S., TSUJI, S., et al. (1998). Cyclooxygenase regulates angiogenesis induced by colon cancer cells. Cell 93, 705–716.

PERRY, V.H., and BROWN, M.C. (1992). Macrophages and nerve regeneration. Curr. Opin. Neurobiol. 2, 679–682.

WILBERGER, J.E. (1996). Pharmacological rescusitation for spinal cord injury. in: Neurotrauma. R.K. Narayan, J.E. Wilberger, and J.T. Povlishock (eds), McGraw-Hill, New York, pps. 1219–1228.

POPOVICH, P.G., WEI, P., and STOKES, B.T. (1997). Cellular Inflammatory response after spinal cord injury in SpraqueDawley and Lewis rats. J. Comp. Neurol. 377, 443–464.

WILLIAMS, C.S., and DU BOIS, R.N. (1996). Prostaglandin

889

SCHWAB ET AL. endoperoxide synthase: why two isoforms? Am. J. Physiol. 270, G393–G400. WILLOUGHBY, D.A., MOORE, A.R., and COLVILLENASH, P.R. (2000). COX-1, COX-2, and COX-3 in the future treatment of chronic inflammatory disease. Lancet 355, 646–648.

YERMAKOVA, A.V., ROLLINS, J., CALLAHAN, L.M., et al. (1999). Cyclooxygenase-1 in human Alzheimer and control brain: quantitative analysis of expression by microglia and CA3 hippocampal neurons. J. Neuropathol. Exp. Neurol. 55, 1135–1146.

YAMAGATA, K., ANDREASSON, K.I., KAUFMANN, W.E., et al. (1993). Expression of a mitogen-inducible cyclooxygenase in brain neurons: regulation by synaptic activity and glucocorticoids. Neuron 112, 371–386. YASOJIMA, K., SCHWAB, C., MC GEER, E.G., et al. (1999). Distribution of cyclooxygenase-1 and cyclooxygenase-2 mRNAs and proteins in human brain and peripheral organs. Brain Res. 830, 226–236.

890

Address reprint requests to: Dr. Jan Markus Schwab Institute of Brain Research University of Tuebingen Medical School Calwer Str. 3 D-72076 Tuebingen, Germany E-mail: [email protected]