Neuronal Nitric Oxide Synthase Expression Is Induced in Neocortical Astrocytes After Spreading Depression

Abstract

Spreading depression (SD) confers either increased susceptibility to ischemic injury or a delayed protection. Because nitric oxide modulates ischemic injury, we investigated if altered expression of nitric oxide synthase (NOS) by SD could account for the effect of SD on ischemia. Furthermore, the identity of cells expressing NOS after SD is important, since SD results in heterogeneous, cell type-specific changes in intracellular environment, which can control NOS activity. Immunohistochemical, computer-based image analyses and Western blotting show that the number of neuronal NOS (nNOS)-positive cells in the somatosensory cortex was significantly increased at 6 hours and 3 days after SD (P < 0.05 and 0.01, respectively), whereas inducible NOS expression remained unchanged. Double-labeling of nNOS and glial fibrillary acidic protein identified these nNOS-positive cells as astrocytes. The effect of altered NO production on induced nNOS expression was examined by treating rats with sodium nitroprusside or N_A-nitro-L-arginine methyl ester (LNAM) during SD. Increased nNOS expression was prevented by sodium nitroprusside and phenylephrine or phenylephrine alone, but not LNAM. Because SD increased astrocytic nNOS expression at time points correlating with both ischemic hypersensitivity and ischemic tolerance, the ability of SD to modulate ischemic injury must be complex, perhaps involving NOS but other factors as well.

Since the discovery that nitric oxide (NO) is produced by neurons (Garthwaite et al., 1988) and may function as a neurotransmitter (Bredt and Snyder, 1989; Garthwaite et al., 1989) and a mediator of brain injury (Dawson et al., 1991), much attention has been given to the cellular sources of this diffusible molecule. Cells produce NO through the action of at least three isoforms of nitric oxide synthase (NOS): neuronal (nNOS), inducible (iNOS), and endothelial (Dawson and Snyder, 1994; Zhang and Snyder, 1995). Neuronal NOS is localized to discrete regions of the brain, where expression is exclusive to neurons (Bredt et al., 1991; Schmidt et al., 1992); however, in the cerebellum and hippocampus, nNOS also is found in astrocytes (Schmidt et al., 1992; Murphy et al., 1993; Kugler and Drenckhahn, 1996). In the neocortex, nNOS is expressed in a small percentage of the total neurons, mostly residing in deeper lamina (Bredt et al., 1991). The observations that a subset of brain astrocytes can express nNOS (Schmidt et al., 1992) and that nNOS mRNA could be isolated from astrocytes derived from many brain regions (Minc-Golomb and Schwartz, 1992) suggest that astrocytes throughout the brain may have the potential to express nNOS. Knowledge of where and under what circumstances they are able to express nNOS has important implications for understanding brain function under normal and pathologic conditions.

Expression of the NOS isoforms is induced in many cell types after a variety of stimuli. Induction of iNOS expression occurs in hepatocytes after cytokine stimulation (Geller et al., 1993), in the brain after cerebral ischemia (Iadecola et al., 1995), and throughout the body after lipopolysaccharide stimulation (Liu et al., 1993). Expression of iNOS increases in astrocytes after transient global ischemia (Endoh et al., 1994) and in both astrocytes and microglia in vitro after various types of perturbations (e.g., exposure to lipopolysaccharide and cytokines; Simmons and Murphy, 1992). Expression of nNOS also is induced in neurons by events that cause cell death, such as peripheral nerve transection (Fiallos-Estrada et al., 1993; Verge et al., 1992; Wu et al., 1994) and cerebral ischemia (Zhang, et al., 1994a).

Nitric oxide synthase activity modulates ischemic brain injury (Iadecola, 1997). Intracellular conditions (e.g., Ca²⁺, pH) influence which NOS isoforms are active. Because the intracellular conditions of glial cells can change dramatically during perturbations such as spreading depression (SD) and differ significantly from surrounding cells (Kraig and Chesler, 1988; Chesler and Kraig, 1989; Kraig and Lascola, 1994; Kraig et al., 1995), knowledge of which NOS isoforms can be expressed in glial cells is essential for understanding their potential contribution to NO production in brain. If neocortical astrocytes or microglia could express multiple NOS isoforms after perturbation, this would diversify their potential roles in modulating neuronal injury. For example, damage after ischemia is reduced in nNOS-deficient mice (Huang et al., 1994; Panahian et al., 1996; Hara et al., 1996). If glia express nNOS in vivo under conditions similar to ischemia (such as SD), they become a cellular site potentially involved in the earlier observation. Furthermore, several in vitro studies show that NO produced by iNOS in astrocytes (Hewett et al., 1994) and microglia (Boje and Arora. 1992; Chao et al., 1992) mediates neuronal cell death after excitotoxic injury. Thus, identifying NOS isoform expression in glia is essential for further understanding their role in normal and pathologic brain function.

Here we demonstrate that neocortical SD induces expression of nNOS in a subset of neocortical astrocytes. Spreading depression is a benign perturbation characterized by a propagating wave of electrical silence and negative interstitial direct current potential (Leao, 1944). It is not associated with cellular injury (Nedergaard and Hansen, 1988; Kraig et al., 1991); however, it does cause both astrocytes and microglia to become reactive (Kraig et al., 1991; Gehrmann et al., 1993; Caggiano and Kraig, 1996). In addition to transforming glia into reactive species, SD induces the expression of enzymes that produce mediators of brain inflammation and injury (Caggiano et al., 1996). Since NOS manipulation has been shown to modulate neuronal injury (Iadecola, 1997), we sought to determine if SD might induce expression of the neuronal or inducible NOS isoforms, and if so, to determine the cellular source of this expression. We found that in the neocortex experiencing SD, nNOS expression was induced within 6 hours, and expression was stronger at 3 days after SD. Spreading depression-induced expression was limited to astrocytes, but not all astrocytes within any given lamina expressed nNOS. In addition, in the course of determining if NO itself effected nNOS expression, we found that expression was blocked by treatment of animals with phenylephrine (PE), an adrenergic agonist used in conjunction with agents that modulate NO production. These results demonstrate that a select population of astrocytes residing in the neocortex can express nNOS after noninjurious perturbation, and this expression can be prevented by adrenergic agonists.

METHODS

Animal preparation and recording

All procedures followed those of Caggiano and Kraig (1996) and Caggiano and others (1996) with some modifications. Animal preparation followed modified procedures of Kraig and coworkers (1991) and Moskowitz and associates (1993). Fifty Wistar male rats (250 to 450 g; Harlan Sprague Dawley, Indianapolis, IN, and Charles River, Wilmington, MA, U.S.A.) were housed individually with food and water available ad libitum and maintained on a 12-hour light-dark schedule. Animals were fasted for 16 hours and then anesthetized by inhalation of halothane (5% induction, 3% during surgery, and 1.0% to 2.5% during electrophysiologic recordings) in a 20% oxygen-nitrogen gas mixture. Animals breathed spontaneously, and body temperature was maintained at 37 ± 0.5°C with a water body jacket.

Spreading depression was induced as previously described (Caggiano and Kraig, 1996). Briefly, 0.5- to 8.0-second pulses (175 kPa) of 0.5 mol/L KCl were injected into parietal neocortex at 6 mm caudal to bregma and 4.5 mm left of the sagittal suture, every 9 minutes for 3 hours. Injections were made at a depth of 750 μm below the pial surface through glass micropipettes (#6010 glass; A-M Corp., Everett, WA, U.S.A.) drawn to tip diameters of 12 μm. Recordings were made through two glass microelectrodes (#6030 glass; A-M Corp.) filled with 150 mmol/L NaCl and with dip diameters of 2 to 8 μm. They were placed 3 mm rostral to bregma and 2 mm left and right of the sagittal suture at a depth of 750 m. Recordings were amplified with an Axoprobe A-1 system (Axon Instruments, Burlington, CA, U.S.A.), filtered at 2 Hz, and displayed on a chart recorder. The exposed skull was bathed in Ringer solution (1 to 3 mL/min) during recordings. Ringer solution contained the following (in millimolar amounts): Na⁺ 143.5; K⁺ 3.0; Ca²⁺ 1.5; Cl⁻ 115; HCO₃⁻ 26.4; gluconate 9.6; and glucose 5.0; the Ringer solution had a pH of 7.3 to 7.4 when aerated with 95% oxygen and 5% carbon dioxide (modified from Bretag, 1969).

A tail artery was cannulated for the monitoring of arterial blood gas. Carbon dioxide tension (PaCO₂), oxygen tension (PaO₂), and pH were measured every 30 to 60 minutes with a Corning blood gas analyzer (model 238; Ciba Corning Diagnostics, Medfield, MA, U.S.A.). Halothane levels were adjusted (0.5% to 1.5%) to maintain PaCO₂ between 50 to 75 mm Hg, a range consistent with adequate anesthesia (Kraig et al., 1986). Blood glucose also was measured during the recording procedures (Glucometer, Miles Laboratory, Naperville, IL, U.S.A.). After 3 hours of SD, electrodes were removed. Animals recovered for 6 hours or 3 days under environmental conditions described earlier.

Some animals were treated 1 hour before surgery with N_A-nitro-L-arginine methyl ester (LNAM; five animals), sodium nitroprusside with phenylephrine (SNP; five animals), or PE (three animals) alone. Both SNP and PE were purchased from Sigma (St. Louis, MO, U.S.A.) and LNAM from Merck Sharp and Dohme (West Point, PA, U.S.A.). The dosages were derived from the literature (Caggiano and Kraig, 1996). The LNAM was given at 30 mg/kg intraperitoneally in 0.01 mol/L phosphate-buffered 150 mmol/L NaCl (PBS; pH 7.4). All animals treated with drugs survived for 3 days after SD under conditions described earlier.

Phenylephrine and SNP were delivered according to the methods of Zhang et al. (1994b). The SNP was administered (3 mg/kg/h) at a rate of 0.51 mL/h in 150 mmol/L NaCl through a cannula in the external carotid artery directed toward the carotid bifurcation. The PE was administered (0.04%) at a rate of 0.25 to 0.51 mL/h in 150 mmol/L NaCl through a cannula in the left femoral vein to maintain blood pressure that would otherwise have fallen with the NO donation from SNP. A subcutaneous injection (0.1 mL) of a long-acting local anesthetic (Bupivacaine; Abott, Chicago, IL, U.S.A.) was administered at the incision for each catheter site. Wounds were closed with stainless steel surgical clips.

Sham-operated animals were treated identically to animals receiving SD, except that intracerebral infusions of 0.5 mol/L NaCl were substituted for 0.5 mol/L KCl. Drug control groups received the same drug dosages as their experimental counterparts, but did not receive intracerebral infusions and placement of direct current microelectrodes. Normal age-matched animals were included for comparison. The PE animals were controls for the animals receiving SNP (with PE).

Immunohistochemistry

General staining procedures followed those of Breder and others (1995). Animals were anesthetized briefly with halothane and deeply anesthetized with intraperitoneal sodium pentobarbital (50 mg/kg). After an intracardiac injection of 1000 U heparin, animals were perfused transcardially with 200 mL normal saline and 500 mL 0.01 mol/L sodium periodate, 0.075 mol/L lysine-HCl, 2% paraformaldehyde fixative in 37 mmol/L phosphate buffer, pH 6.2 (PLP). Brains were removed and postfixed for 2 hours in PLP. Brains then were cryoprotected in 10% sucrose in 0.1 mol/L phosphate buffer (pH 7.3) for 2 hours and 20% sucrose in 0.1 mol/L phosphate buffer for 48 hours. Brains were blocked into three sections and rapidly frozen in isopentane cooled to −35°C on dry ice. Tissue was stored at −80°C until use.

Frozen coronal sections of the somatosensory cortex were cut at 40 μm on a cryostat microtome (model 855; Reichert-Jung, Cambridge Instruments, Heildelberg, Germany). Sections were collected on the surface of PBS (pH 7.3) and then refixed for 5 minutes on the surface of 0.1% glutaraldehyde in 4% paraformaldehyde to reduce section curling. Sections were rinsed with PBS and then incubated in 0.3% H₂0₂ in PBS with 0.25% Triton X-100 (Sigma) for 20 minutes. Sections were rinsed again in PBS and incubated for 1 hour in 2% goat serum (Colorado Serum Co., Denver, CO, U.S.A.) in PBS with 0.25% Triton X-100 (PGT). Antibodies raised against nNOS (sc-648) and iNOS (sc-650, Santa Cruz Biotechnology, Santa Cruz, CA, U.S.A.) were diluted to 1:500 and 1:300, respectively, in PGT. Sections from each animal were incubated overnight in the primary antibody solutions at 4°C with gentle agitation.

Antibodies were visualized with the Vectastain Elite kit (Vectastain Laboratories Inc., Burlingame, CA, U.S.A.) and diaminobenzidine. The Vectastain secondary antibody was diluted five times more than recommended in the Vectastain protocol. Sections were mounted on gelatin-coated slides. Silver intensification of the diaminobenzidine precipitate was performed according to the procedures of Breder and coworkers (1992). Sections were dehydrated in increasing ethyl-alcohol concentrations and cleared in xylenes, and cover slips were applied with DPX mounting media (Poole, U.K.).

Some sections were double-labeled with the nNOS antibody and an antibody raised against glial fibrillary acidic protein (GFAP, #814-369; Boehringer Mannheim, Indianapolis, IN, U.S.A.). Sections were processed for nNOS as described earlier. After the peroxidase-diaminobenzidine reaction, sections were washed and incubated with the GFAP antibody (1:200) overnight at 4°C with gentle agitation. Sections were washed and incubated for 2 hours in a rhodamine-labeled goat anti-mouse IgG (Biosource International, Cameriool, CA, U.S.A.). The remaining steps followed those described earlier.

The NOS immunohistochemical study also was performed after absorption of the antibodies with peptides containing the sequences to which the antibodies were raised (100, 10, and 1 ng peptide/mL; Santa Cruz Biotechnology). As a control, the nNOS antibody was preabsorbed to the iNOS peptide and the iNOS antibody preabsorbed to the nNOS peptide.

Computer-based image acquisition

Neuronal nitric oxide synthase

Images from left and right neocortex (three per animal) were acquired using a CCD camera (CH250; Photometrics, Inc., Tucson, AZ, U.S.A.) at × 60 (Fig. 2, Fig. 4, Fig. 5, and Fig. 9) or × 630 (Fig. 3 and Fig. 8) on a Leica compound microscope (Leica Mikroskopie und Systeme GmbH; Wetzlar, Germany) through a 450/40 nm band pass filter (Chroma Technology Corp., Brattleboro, VT, U.S.A.) using PMIS software (version 3.0; Photometrics, Inc.). Images (12-bit and 1024 × 1024 pixels) were stored on an optical drive (Pinnacle Micro Sierra 1.3GB; Pinnacle Micro, Inc., Irvine, CA, U.S.A.). Images were prepared and analyzed with Image Pro Plus (version 1.2; Media Cybernetics, Silver Springs, MD, U.S.A.) under Windows (3.1) on a 486 AST computer (AST Research, Inc., Irvine, CA) on an uninterruptible power source (MUPSA-1000; Philtek Power Corp., Blaine, WA). Images were reference-corrected according to the equation: I_corrected = [I_original*M/I_reference] where I indicates image and M equals the mean pixel value of the corrected reference image. Images then were filtered using a “higauss” filter (two passes with 5 × 5 pixel matrix at 50% strength; Image Pro software).

Figure 2.

Digital photomicrographs demonstrating neuronal nitric oxide synthase (nNOS) (A and B) and inducible nitric oxide synthase (iNOS) (C and D) immunohistochemical features in normal (non-SD) rat somatosensory cortex. Images were acquired as described in Methods. Images A and C were taken at × 60 and represent the superficial lamina of somatosensory cortex. Scale bars = 300 μm. B and D show approximately lamina IV–VI at the same magnification. Notice the large, process-bearing nNOS-positive neurons mostly populate the deeper layers (B), whereas few nNOS cells are found in the outer lamina of neocortex (A). Background shading in these images has been increased from use of the 480/40 band pass filter, which allowed better visualization of the fine nNOS processes. The iNOS-positive cells populate all layers of neocortex, showing expression in cells with large somas and multiple processes (C and D).

Figure 4.

Photomicrographs of nNOS (A) and iNOS (B) immunohistochemical features at 6 hours after SD. Each micrograph is taken as in previous figures from the experimental, left neocortex. Images were taken at × 60. Scale bars = 300 μm. Notice in A that nNOS is strongly expressed in small cells bearing fine processes that are not seen in normal animals (compare with Fig. 2A). Computer-based cell counting revealed a significant increase (P < 0.05) in the number of nNOS-positive objects compared with normal animals (Fig. 5A). No change was observed in deeper lamina, or in the control, right neocortex. The iNOS expression was unchanged at 6 hours after SD. Some variation in iNOS background staining is seen between Fig. 2 and Fig. 4; however, these micrographs do not represent any trend toward increased staining with SD. This was confirmed by densitometric analysis.

Figure 5.

Photomicrograph demonstrating nNOS immunohistochemical features at 3 days after SD in the experimental left neocortex. Image was taken at × 60. Scale bars = 300 μm. Notice that nNOS-positive cells in the outer lamina of neocortex are not present in normal brain. This distribution was similar to that at 6 hours (Fig. 4A); however, the nNOS-positive cells were more intensely stained at 3 days than at 6 hours. Again, no changes in the number, morphologic features, or intensity of nNOS-positive cells were seen in the deeper lamina of neocortex, or in the control, right cortices. Quantitation of these changes is found in Fig. 6A. Because iNOS expression was the same in normal, 6-hour post-SD, and 3-day post-SD animals, further iNOS photomicrographs are not shown.

Figure 9.

Photomicrographs of nNOS immunohistochemistry at 3 days after SD in animals reated with sodium nitroprusside and phenylephrine (SNP), phenylephrine alone (PE), or N_A-nitro-L-arginine methyl ester (LNAM). Images were taken at × 60. Scale bar = 300 μm. Notice that the SNP and PE images are nearly indistinguishable from normal animals (Fig. 2A), whereas the LNAM image is similar to the 3-day animals without pharmacologic manipulation (Fig. 5). Computer-based object counting confirmed these impressions (Fig. 6B).

Figure 3.

Digital photomicrograph demonstrating nNOS immunohistochemical features in normal (non-SD) rat somatosensory cortex. Photomicrograph is taken from the same area as in Fig. 2A using an × 100 objective under oil immersion (total magnification × 630). Scale bar = 25 μm. Notice the faint astrocyte-appearing cell expressing nNOS. Arrows indicate (from top to bottom) what appear to be a process, cell body, and end-foot, respectively.

Figure 8.

Composite digital photomicrograph showing correspondence of nNOS (green) and glial fibrillary acidic protein (GFAP; red) staining from an animal 3 days after SD. The nNOS immunohistochemical study was performed as described in the Methods. The GFAP staining then was performed on the same sections using a rhodamine-labeled secondary antibody. A bright-field (nNOS) and a fluorescent (GFAP 570-nm excitation, 590-nm emission) image were taken of the same field, higauss filtered and superimposed. The nNOS-positive cells in the outer layers of neocortex also stained GFAP-positive in the 3-day post-SD animals. This confirms by morphologic and immunohistochemical study that the novel nNOS-positive cells are astrocytes.

Inducible nitric oxide synthase

Analysis of the iNOS image was performed according to the procedures of Caggiano and Kraig (1996). Briefly, images were acquired as described earlier, except that sections were transilluminated with an Intralux 4000 fiber-optic flat light source (Volpi MFG, Co., Auburn, NY, U.S.A.) through a 540/40 nm band pass filter. Images were acquired with the same CCD camera described earlier.

Data analysis

For nNOS images, cell counting (Image Pro Plus) was performed on each image using a minimum cell-size threshold set at 20 pixels² (at 60 magnification). This pixel area detected most cells and large processes without detecting background noise. The intensity minimum was determined by varying the minimum threshold until most to all cells were counted, but no background staining scored positively. In this manner, positive cells or their large processes were counted. Identical threshold values were used for experimental and control cortices. Values were reported as the total number of counted cells (e.g., objects with a minimum area of at least 20 pixels²) divided by the area analyzed. This computer-driven cell counting technique allows an unbiased assessment of the number of cells within a given cross-sectional area; however, with these methods, we assert no estimates of cell number in a given brain region or volume. For these reasons, comparisons always were made between left and right neocortices of a given section.

Optical density of immunoreactivity (IR) was used to analyze the iNOS images according to the procedures described in detail in Caggiano and Kraig (1996). Briefly, areas of interest were drawn around the left and right neocortex of each section, and the optical density was measured with light and dark limits set to values that encompass those found in each section. The IR for each section was reported as the log of the left side optical density divided by the right side optical density [log (l/r)]. In this manner, equal left and right staining intensities give a value of zero.

The ubiquitous expression levels of iNOS in the areas of interest allowed densitometry analyses. Variations in background staining were eliminated by comparison of left and right neocortices. The lower expression level of nNOS in the outer three layers of the neocortex meant that positively stained cells could be accurately counted. Furthermore, this lower expression level meant that densitometric analyses would be dominated by background staining. Therefore, object counting analysis was used.

The operator was blind to the experimental group of each animal during data acquisition and data analysis. Furthermore, during staining procedures, sections from different groups always were processed together and no group was processed entirely at one time. In this manner, the possibility that group differences resulted from differences in processing of tissues was minimized.

Photomicrographs were prepared using the CCD camera and software described earlier. Images were taken at × 60 (total magnification) through 630/40 (Fig. 6) or 480/40 nm (all others) band pass filters. Images were reference-corrected (as before) and higauss filtered (7 × 7 pixel matrix, 50% strength, two passes; Image Pro Plus software). Figures were compiled and labeled using Adobe Photoshop 4.0 software (Adobe Systems, Inc., San Jose, CA, U.S.A.) and printed on a dye-sublimation printer (XLT-7720; Eastman Kodak, Rochester, NY, U.S.A.). Scale bars are as indicated in figure legends.

Figure 6.

Histograms showing group averages of nNOS-positive cells per square millimeter. (A) Histogram of nNOS-positive objects in normal, 6-hour post-SD, and 3-day post-SD animal groups. Darker columns represent the group averages of the experimental left neocortex; lighter columns represent group averages of the control right neocortex. Values are means = standard deviation. Ordinate is the number of nNOS positively stained objects per square millimeter. Computer-based object counting was performed as described in the Methods, and values were compared with corresponding neocortex from the normal group. *P < 0.05; **P < 0.01. (B) Histogram of nNOS-positive objects in normal and 3-day post-SD animal groups with and without pharmacologic manipulation. Darker columns represent the group averages of the experimental left neocortex; lighter columns represent group averages of the control right neocortex. Values are means ± standard deviation. Ordinate is the number of nNOS positively stained objects per square millimeter. Computer-based object counting was performed as described in the Methods, and values were compared with corresponding neocortex from the SD group receiving no drug treatment. *P < 0.05; **P < 0.01. Notice that counts in the SNP and PE groups have been reduced to normal levels, whereas counts from the LNAM group are not significantly different from the SD group. SNP, sodium nitroprusside and phenylephrine; PE, phenylephrine alone; LNAM, N_A-nitro-L-arginine methyl ester.

Western blots

Given the novel nNOS expression pattern induced by these experimental procedures, we believed it necessary to confirm the specificity of the nNOS antibody by Western blot analysis on samples of rat neocortex. Adult rats-matched for age, size, and sex (used above)-were deeply anesthetized with halothane and decapitated. Brains were rapidly removed. The cortices were cut away and frozen in powdered dry ice.

Western blot procedures followed those described in Sambrook and coworkers (1989). Neocortical sections weighing approximately 100 mg were washed with cold PBS and then mechanically dissociated in suspension buffer containing protease inhibitors. Gel loading buffer was added, and samples were boiled for 5 minutes and underwent sonification for 2 minutes. Samples and protein markers were run on standard 7.5% SDS-polyacrylamide gel (Laemmli, 1970) using Heofer mini-gel systems (Hoefer Scientific Instruments, San Francisco, CA, U.S.A.).

Proteins were transferred to nitrocellulose membranes and incubated overnight with nNOS or iNOS antibody. Some membranes were incubated with antibodies preabsorbed with NOS peptides. Membranes were washed and incubated with an anti-rabbit HRP-labeled antibody. Protein-antibody products were visualized using the ECL Western blotting analysis system (Amersham Life Science, Arlington Heights, IL, U.S.A.) and X-OMAT XAR-5 film (Eastman Kodak Co.).

Each pair of blots was scanned into Adobe Photoshop using a Scan Maker III scanner and Twain scanning software (Microtek Lab, Inc., Redondo Beach, CA, U.S.A.). Images were adjusted for brightness and contrast (normal and control lanes were treated identically).

RESULTS

The experimental procedures described earlier reliably induced SD in the left neocortex but never induced SD in the right control neocortex. An example of the left and right neocortical electrical activity recorded during SD is shown in Fig. 1. The top trace of each pair was recorded from the left neocortex. The bottom trace is from the right neocortex. No SD were induced in sham-operated animals. After recovery from anesthesia, animals showed normal feeding, grooming, and motor activity. Blood physiologic variables (Table 1) were within expected ranges for spontaneously breathing animals anesthetized with halothane. Values were analyzed by ANOVA and post hoc Tukey. Group values differing from other groups are shown by an asterisk.

Figure 1.

Electrophysiologic recordings of spreading depressions (SD) in the left (top trace of each pair) and right (bottom trace) frontal neocortices. Direct current (DC) potentials are shown during 3 hours of recurrent SD induced by microinjection of 0.5 mol/L KCL into parietal cortex every 9 minutes. Scale bar = 20 mV. Spreading depressions were observed only in the experimental left neocortex. Sham-operated animals received 0.5 mol/L NaCl injections, and SD were never observed. After recordings, animals survived for 6 hours or 3 days and then were processed for NOS immunohistochemical study.

Table 1.

Blood physiologic variables, temperature, anesthesia, and number of SD.

Animal group	pH	PaCO2 (torr)	PaO2 (torr)	Hematocrit (%)	Blood pressure (mm Hg)	Temperature (°C)	Halothane (%)	SD (No.)	Glucose (mmol/L)
6 hours	7.34± 0.06	53 ± 9	104 ± 3	40 ± 1	105 ± 13	37.6 ± 0.1	2.0 ±0.0*	19 ± 2	5.2 ± 0.5
3 days	7.30 ± 0.06	59 ± 8	114 ± 11	43 ± 4	109 ± 9	37.5 ± 0.3	1.6 ±0.2	20 ± 1	5.2 ± 0.5
Sham	7.27 ± 0.07	64 ± 11	99 ± 9	42 ± 2	105 ± 22	37.3 ± 0.5	1.8 ± 0.1	0 ± 0*	4.8 ± 0.3
SNP	7.31 ± 0.05	55 ± 12	105 ± 22	41 ± 2	124 ± 2	37.3 ± 0.4	1.2 ± 0.6	13 ± 3*	9.3 ± 1.6*
PE	7.31 ± 0.01	59 ± 5	103 ± 14	42 ± 1	150 ± 13*	37.1 ± 0.6	1.4 ± 0.1	19 ± 1	6.4 ± 0.8
LNAM	7.31 ± 0.04	57 ± 5	107 ± 9	42 ± 4	114 ± 11	37.7 ± 0.6	1.5 ± 0.2	19 ± 3	5.3 ± 0.6

^*Indicates significantly different from other groups according to analysis of variance and post-hoc Tukey. All groups contain three to five animals.

All values are mean ± SD.

Nitric oxide synthase distribution in normal somatosensory cortex

The nNOS-positive neurons were distributed to discrete brain regions. Overall, nNOS IR filled large, multiprocessed neurons consistent with previous observations (Bredt and Snyder, 1991). In neocortex, these large neurons were distributed throughout the neocortex, but most of the nNOS-positive cells were found in deeper lamina and made up only a small percentage of the total population (Fig. 2A and B). The iNOS IR in the neocortex was localized to large, processed cells in all layers of neocortex (Fig. 2C and D). Figure 2A and B show nNOS IR in the superficial and deeper layers or neocortex, respectively, in a normal rat. Normal iNOS IR is similarly shown in Fig. 2C and D.

A higher power (×100) photomicrograph of the superficial neocortex from a normal animal processed for nNOS IR (Fig. 3) shows that nNOS is expressed at extremely low levels in cells that have the general morphologic makeup of astrocytes. The top arrow points to an astrocyte process, the middle arrow indicates the cell body, and the bottom arrow points to a possible end-foot process on a blood vessel.

Neuronal nitric oxide synthase expression is induced at 6 hours after spreading depression

Six hours after SD, a population of cells became nNOS-positive in the superficial lamina of the neocortex experiencing SD. These cells were small, finely processed, and numerous (Fig. 4A). The nNOS expression level in this population of cells was much more intense than that found in the contralateral, control neocortex (data not shown; however, for comparison see Fig. 2A and Fig. 3). Computer-based cell counting was used to show a significant elevation at 6 hours after SD in the number of positively stained objects in the superficial lamina of the left neocortex compared with either the right neocortex of SD animals or the left neocortex of normal animals (P < 0.05). Objects were defined as cell bodies or large processes above a given size and intensity threshold (see Methods). No significant differences between the number of positive cells in control neocortices of experimental animals and neocortices of sham-operated or normal animals were seen (P > 0.05; data not shown). The nNOS-positive cells found in the deeper layers of the left, experimental neocortices were not different from those found in normal animals.

Neuronal nitric oxide synthase expression is elevated at 3 days after spreading depression

At 3 days after SD, nNOS IR displayed a similar, but more intense, expression pattern to that at 6 hours after SD. Again, a population of small, finely processed cells expressed nNOS in the outer (Fig. 5) but not the deeper lamina of the neocortex. Computer-based cell counting revealed a significant difference between the number of positively stained cells in the left neocortex of experimental animals compared with either left neocortex of normal animals or right neocortex of experimental animals (P < 0.01; Fig. 6A). No left-versus-right difference was found in normal or sham-operated animals.

The iNOS expression was not induced at 6 hours (Fig. 4B) or 3 days after SD. At 3 days, expression patterns were identical to normal and sham-operated animals (data not shown). Computer-based optical density analysis confirmed this point. The widespread and even expression pattern of iNOS (compared with nNOS) allowed for densitometric analyses (see Methods: Data analysis).

Western blot and competition assays

To confirm the specificity of the antibodies, we performed Western blot analysis using the nNOS and iNOS antibodies on extracts from normal rat neocortex. The nNOS and iNOS antibodies revealed a positive band, each occurring at the appropriate weight (~ 150 kd and ~ 130 kd, respectively; Fig. 7, right-hand column). A smaller peptide (nNOS: ~ 25 kd; iNOS: ~ 40 kd) also stained positive with each antibody. To determine if this was a degradation product or if the antibody was reacting with another peptide, we preabsorbed the antibodies with peptides (100 ng/mL) from their corresponding NOS protein. Both bands detected by each antibody were completely blocked by the appropriate peptide (Fig. 7), confirming that the antibodies were specific for their intended target.

Figure 7.

Immunohistologic and Western blot demonstrations of the specificity of the nNOS and iNOS antibodies. (Top row) Photomicrographs of nNOS and iNOS immunohistochemical features without blocking peptide (a portion of the entire nNOS enzyme). Below are photomicrographs of each antibodies’ immunohistochemical features when incubated with 100, 10, or 1 ng/mL blocking peptide. Notice that nNOS immunoreactivity (IR) was completely blocked by 100 ng/mL peptide. Nearly complete and partial blocking also was achieved with 10 and 1 ng/mL peptide, respectively. Images were taken at × 60. Scale bar = 300 μm. On the far right are Western blots of normal rat neocortex extract using the nNOS (top) and iNOS (bottom) antibody. Right-hand lanes are blots using the antibodies that were preabsorbed with 100 ng/mL peptide. The nNOS antibody detected bands at ~ 150 kd and ~ 25 kd; the iNOS antibody detected bands at ~ 130 kd and ~ 50 kd. Each of these bands were almost completely blocked by preabsorption.

Furthermore, each antibody was tested by the ability of the nNOS and iNOS peptides to block their IR in the tissue sections. The nNOS IR was completely blocked by preabsorption of the antibody with 100 or 10 ng/mL nNOS peptide, and partially blocked by 1 ng/mL peptide (Fig. 7). No blocking occurred after preabsorption with 100 ng/mL iNOS peptide (not shown). Similar blocking results were obtained when iNOS antibody was absorbed with iNOS peptide (Fig. 7). No blocking of iNOS IR occurred with absorption with nNOS peptide.

Neuronal nitric oxide synthase/glial fibrillary acidic protein double-labeling

The appearance of finely processed small nNOS-positive cells (Fig. 5) suggested that they may be astrocytes. Double-staining with nNOS antibody and an antibody against GFAP revealed that most cells that stained nNOS-positive in the outer layers of neocortex after SD also were GFAP-positive. A photomicrograph showing nNOS (green) and GFAP (red) staining is shown in Fig. 8, using a × 100 objective under oil immersion (total magnification × 630). All nNOS-positive cells (other than the larger neuronal cells appearing in normal and SD sections) stained GFAP-positive; however, many astrocytes (GFAP-positive cells) did not express nNOS. Observe that there are three nNOS-positive cells in Fig. 8 (center, center right, and bottom left). Each also is GFAP-positive (red staining).

Sodium nitroprusside and LNAM effects

To determine if NO itself might effect the expression of nNOS, as has been observed with iNOS expression (Guo et al., 1995), we treated animals before SD surgery with agents that inhibit (LNAM) NO production or nonenzymatically (SNP) donate NO. The nNOS cell counts from groups treated pharmacologically (LNAM, SNP, or PE) before and during SD were compared with ipsilateral neocortical cell counts from the group receiving SD without any pharmacologic manipulation. nNOS positive cell counts in the experimental cortex of the group treated with the combination of SNP and PE were significantly less than those from the SD group (P < 0.01; Fig. 6B and Fig. 9). The group treated with PE only also had significantly lower cell counts (P < 0.05; Fig. 6B and Fig. 9). Cell counts from the LNAM group were not significantly different from the SD counts (Fig. 6B and Fig. 9). These observations are evident histologically in the photomicrographs displayed in Fig. 9. Notice in the photomicrographs taken from the SNP and PE groups that the distribution of nNOS-positive cells is indistinguishable from that of the normal group (Figs. 2A). Furthermore, nNOS-positive cells in the LNAM group appear similar in number, morphologic features, and distribution to those from the SD group (Fig. 5, Fig. 6B, and Fig. 9).

DISCUSSION

The primary observation of this work is that expression of the neuronal isoform of NOS can be induced in neocortical astrocytes by SD. Currently, nNOS expression in vivo has been demonstrated only in astrocytes residing in the cerebellum and hippocampus under normal physiologic conditions (Murphy et al., 1993; Kugler and Drenckhahn, 1996). Furthermore, whereas nNOS expression is induced in neurons after perturbations that cause cellular injury and death (Fiallos-Estrada et al., 1993; Verge et al., 1992; Wu et al., 1994; Zhang et al., 1994a), here we show that SD-a perturbation that does not cause cellular injury or death-causes a rapid and sustained expression of nNOS. Identification of the cellular localization of NOS isoforms, and thus the potential sources of NO, is becoming essential as isoform-specific inhibitors and knockout animals are developed. Furthermore, given the uniquely dynamic intracellular environment of glia during brain stimulation (Kraig and Chesler, 1988; Chesler and Kraig, 1989; Kraig and Lascola, 1994; Kraig et al., 1995), there is the potential that NOS isoforms may be active in glia at times when they are not as active in other brain cells. Thus, information of the cellular sources and isoforms of NOS should help decipher how NO influences brain function under normal and pathologic conditions.

The NOS isoforms are distributed throughout the CNS and are localized to nearly all cell types (Dawson and Snyder, 1994; Zhang and Snyder, 1995). The NOS isoform responsible for NO production in astrocytes and microglia generally has been shown to be iNOS (Murphy et al., 1993). Expression of iNOS in astrocytes has been shown to be enhanced after various neuropathologic conditions (Endoh et al., 1994; Bo et al., 1994; Nakashima et al., 1995). In the cerebellum and hippocampus, nNOS expression also has been demonstrated in astrocytes (Murphy et al., 1993; Kugler and Drenckhahn, 1996); however, in the neocortex, nNOS is expressed exclusively to a small number of neurons (Schmidt et al., 1992). The observation that nNOS mRNA could be isolated from astrocytes in many regions outside of the cerebellum (Minc-Golomb and Schwartz, 1992) suggests that astrocytes could be induced to express nNOS elsewhere in the brain. In the current results, the morphologic features of the cells that were induced to express nNOS after SD (Fig. 5) had the general appearance of astrocytes. Double-labeling with antibodies raised against GFAP and nNOS confirmed that this novel expression of nNOS was limited to astrocytes (Fig. 8). Thus, whereas a few neurons residing in the deepest three layers of neocortex strongly express nNOS and perhaps some astrocytes weakly express nNOS (Fig. 3) in normal (non-SD) brain, after SD a large population of astrocytes in the superficial layers of neocortex also strongly express nNOS.

Spreading depression induces reactive gliosis in both astrocytes (Kraig et al., 1991) and microglia (Gehrmann et al., 1993; Caggiano and Kraig, 1996). To confirm that these cells were not microglia, we double-labeled tissue sections with the nNOS antibody and the GSA IB-4 isolectin (a marker of microglia in the CNS). Cells expressing nNOS after SD do not stain positively with the isolectin, demonstrating that they are not microglia (data not shown).

Given the novel expression pattern of the nNOS antibody in neocortical astrocytes after SD, we sought to confirm its specificity for the nNOS isoform by performing Western blots on extracts from rat neocortex. The nNOS antibody was extremely clean, detecting a band at the expected ~ 150 kd and a smaller band at ~ 25 kD (Fig. 7). We confirmed that the smaller band likely was a degradation product of nNOS by demonstrating that preabsorption of the nNOS antibody with 100 ng/mL of a nNOS peptide completely blocked the interaction of the antibody with both of these bands (Fig. 7). Furthermore, this peptide was able to block nNOS reactivity in brain sections at 10 ng/mL (Fig. 7), whereas an iNOS peptide was unable to block nNOS reactivity at any concentration tested. We believe these important controls, in addition to the nNOS/GFAP double-labeling, strongly support the conclusion that the novel expression detected by the nNOS antibody after SD is indeed the neuronal isoform of NOS localized to neocortical astrocytes.

It is unclear why only a fraction of the astrocytes in the superficial lamina of the neocortex express nNOS after SD. Differential expression of receptors and channels on astrocytes occur in anatomically distinct zones of the brain (Van der Zee et al., 1993) and functionally unique regions, such as at nodes of Ranvier (Black et al., 1989). Astrocytes also display morphologic diversity based on their laminar localization (Bailey and Shipley, 1993). Here, however, we observe that only a subset of astrocytes within a given lamina and functional zone of neocortex express nNOS in response to SD, whereas at least an equal number remain nNOS-negative. After SD, most or perhaps all astrocytes within this region become reactive, as demonstrated by their change in morphologic features and GFAP IR (Kraig et al., 1991; Caggiano and Kraig, 1996). Perhaps astrocytes within the neocortex have functional subtypes. On the other hand, perhaps these nNOS-positive astrocytes are responding to localized differential activity of the immediately adjacent neurons, resulting in differential enzyme expression.

Astrocytes undergo dynamic changes in intracellular calcium concentration that may regulate the activity of the calcium-dependent nNOS. Astrocytes experience depolarization-dependent calcium fluxes during SD that exceed 1 μmol/L (Kraig et al., 1995). Changes in calcium concentration in this range have been shown to stimulate nNOS to near maximal activity (Bredt and Snyder, 1990). Intracellular calcium concentration in astrocytes also is responsive to a variety of transmitters. Glutamate and certain agonists cause increases in intracellular calcium in primary cultured astrocytes (Enkvist et al., 1989a; Jensen and Chiu, 1990 and 1991; Glaum et al., 1990; van den Pol et al., 1992) and in astrocytes in hippocampal organotypic cultures (Dani et al., 1992). Phenylephrine, carbachol (Enkvist et al., 1989b), and serotonin (van den Pol et al., 1992) also induce intracellular calcium responses in astrocytes. Thus, despite the fact that application of N-methyl D-aspartate to cultured astrocytes produces no calcium flux (Jensen and Chiu, 1990), astrocytes clearly undergo intracellular calcium changes in response to their electrical and chemical environment. These changes make them a potentially ideal site for differential regulation of nNOS, which may in turn affect how astrocytes and NO contribute to brain injury and recovery.

Induction of nNOS expression in glia may have important and probably complex implications in the pathogenesis of ischemic brain injury. Increased expression of nNOS in neurons is induced by cerebral ischemia (Zhang et al., 1994a). Furthermore, NO is thought to mediate resultant brain damage, since nNOS-specific inhibitors ameliorate ischemic damage (Zhang et al., 1996) and nNOS mutant and knockout mice are resistant to ischemic brain injury (Panahian et al., 1996; Hara et al., 1996; Huang et al., 1994; Dawson et al., 1996). Here we show that SD induces expression of nNOS in astrocytes (not neurons). Spreading depression is closely associated with injury from cerebral ischemia. For example, SD immediately before ischemia worsens the resultant cellular damage (Takano et al., 1996); however, when SD occurs days before ischemia, an “ischemic tolerance” develops, making the brain more resistant to irreversible cellular damage (Kobayashi et al., 1995; Kawahara et al., 1995; Matsushima et al., 1996). Furthermore, SD or SD-like events occurring during focal cerebral ischemia within the penumbra correlate with the magnitude of resulting neural injury (Busch et al., 1996; Mies et al., 1993). The fact that SD alone is able to induce nNOS expression (Fig. 4A and Fig. 5), an enzyme known to be a key mediator of ischemic damage (Zhang et al., 1996; Panahian et al., 1996; Hara et al., 1996; Huang et al., 1994; Dawson et al., 1996), suggests that induction of nNOS expression in astrocytes is one mechanism by which SD may affect injury from ischemia.

Spreading depression may lessen or worsen the effects of subsequent ischemia through the control of the expression of other enzymes. One candidate enzyme is superoxide dismutase. Expression of both the manganese and copper-zinc forms of superoxide dismutase were found to be decreased at 6 hours after SD, but dramatically increased at 3 days after SD (Caggiano and Kraig, 1997). The interaction of NO and superoxide is one means by which NO mediates brain injury (Lipton et al., 1994; Dawson et al., 1994; Beckman et al., 1994). Peroxynitrite (⁻OONO), formed from the combination of NO and superoxide (O₂⁻) (Huie and Padmaja, 1993), is toxic to neurons (Lipton et al., 1993). Superoxide dismutase scavenges O₂⁻, preventing the formation of ⁻OONO. Thus, it might be the combined effects and time course of altered nNOS and superoxide dismutase levels after SD that result in the initial sensitivity and subsequent tolerance to ischemia after SD (Caggiano and Kraig, 1997).

The ability to modulate the expression of NOS isoforms may allow for the regulation of their contribution to brain diseases such as ischemia. Furthermore, this information will help elucidate the signaling mechanisms that control their transcription, translation, or modification into mature form. Nitric oxide itself is one cellular signal that controls the activity (nNOS; Rogers and Ignarro, 1993) and expression (iNOS; Guo et al., 1995) of NOS. To determine if modulation of NO production in the brain during SD might alter the expression of nNOS induced in neocortical astrocytes, we treated animals with LNAM, SNP, and PE, or PE alone. LNAM (an L-arginine analogue), which should prevent the formation of NO, had no effect on the induced expression of nNOS. Treatment of animals with SNP (a nonenzymatic NO donor) plus PE completely blocked the induced expression of nNOS (Fig. 6B and Fig. 9). Phenylephrine was given to animals receiving SNP to maintain blood pressure, which would otherwise have fallen with the donation of NO. Because the SNP/PE control (PE alone) also blocked nNOS expression in astrocytes, no conclusions can be drawn regarding the influence of SNP and NO; however, we can conclude that induction of nNOS expression is prevented by the adrenergic agonist, an observation also seen in the expression and activity of iNOS (Feinstein et al., 1993). The fact that expression can be completely blocked by an agent analogous to endogenous transmitters within the brain shows encouraging potential for our ability to control NOS expression during brain diseases.

The prevention of nNOS expression by PE alone was an unexpected result of this study. There is a limited body of literature regarding adrenergic agonists and NOS expression, and we are aware of no reports examining NOS activity or expression using an alpha-1 specific agonist such as PE. Alpha-1 receptors do mediate inositolphosphate and Ca²⁺ signaling in astrocytes (Biber et al., 1996; Enkvist et al., 1989b) and thus may modulate NOS activity and, in turn, NOS expression. Astrocytosis is regulated by beta-adrenergic receptors (Griffith and Sutin, 1996; Hodges-Savola et al., 1996) and glutamine uptake by astrocytes is regulated by alpha- and beta-adrenergic agonists (Huang and Hertz, 1995). However, we have previously shown that astrogliosis is not prevented by PE or SNP/PE (Caggiano and Kraig, 1996); thus, the reduced nNOS expression is not the result of preventing astrogliosis in general. Further analysis of the possible mechanisms by which PE prevents nNOS expression is needed but beyond the scope of this report.

In conclusion, we demonstrated that expression of the neuronal isoform of NOS is able to be induced in neocortical astrocytes by SD. This induction is rapid (by 6 hours) and sustained (3 days). It can be blocked by the adrenergic agonist PE. We found no increase in iNOS expression after SD; however, Western blot analysis and competition studies demonstrated the efficacy of the antibody. These results show that astrocytes are able to express nNOS in the neocortex and introduce another factor in the surely complex mechanism by which SD alters resultant damage from ischemia.

Abbreviations

GFAP

glial fibrillary acidic protein

IR

immunoreactivity

LNAM

N_A-nitro-L-arginine methyl ester

NO

nitric oxide

NOS

nitric oxide synthase

iNOS

inducible nitric oxide synthase

nNOS

neuronal nitric oxide synthase

PBS

phosphate-buffered saline

PE

phenylephrine

PLP

sodium periodate-lysine-HCl-paraformaldehyde fixative in phosphate buffer

PGT

phosphate-buffered saline with Triton X-100

SD

spreading depression

SNP

sodium nitroprusside with phenylephrine