Main

Despite the fact that the central nervous system lacks fully developed lymphoid tissue and is isolated from the immune system by the blood brain barrier, there is increasing evidence that the immune system has access to the brain and may be involved in the pathophysiology of brain damage(1–3). Furthermore, the immature brain has been shown to respond to endotoxin(4) or IL-1β(5) with a stronger and more rapid recruitment of leukocytes than the adult brain which suggest that the neonatal brain may be highly susceptible also to the potentially damaging consequences of an inflammatory response.

The course of inflammatory events with respect to neuropathological alterations in hypoxia-ischemia (HI) has been investigated only partly in the neonatal setting(6–10). Microglial cells/macrophages and neutrophils are activated after HI which is accompanied by a marked expression of proinflammatory cytokines (IL-1β, IL-6, TNF-α)(7,9). The inflammatory process appears to be important in the pathophysiology of HI as antineutrophil serum(10) and IL-1 receptor antagonists reduce injury(9,11), and neonatal mice deficient in IL-1 converting enzyme are resistant to HI insults(12).

There is a lack of information, however, concerning the possible participation of lymphocytes and natural killer (NK) cells in the immature brain. These cells are activated after ischemia in the adult brain(13) and depletion of lymphocytes reduced injury to some extent(14). Knowledge concerning the mechanism of cell activation is sparse. Recent data suggest that a new family of chemoattractant cytokines, the chemokines, is involved(15,16). Chemokines are proposed to act in concert with different adhesion molecules (selectins, integrins, and intercellular adhesion molecules) to attract, adhere, and activate leukocytes(15,17). The effects of chemokines on leukocytes are mediated by GTP-coupled heptahelical receptors involving stimulation of these cells to produce lamellipodia, production of oxygen free radicals, and the release of proteases from cytoplasmic storage granules(17). Chemokines are divided into four families: α, β, γ, and neurotactin based on the presence and position of the conserved cysteine residues(18,19). Members of the α-family are chemotactic to granulocytes and β-chemokines are mostly chemotactic to mononuclear cells such as monocytes and lymphocytes(19). The γ-chemokines are chemotactic to lymphocytes(20), and neurotactin is chemotactic to neutrophils(18,19). Recently, it was shown that application of α-chemokines overrides the brain's intrinsic resistance to leukocyte recruitment and induces a widespread destruction of the blood brain barrier(21) supporting a critical role of these molecules. Data on expression of chemokines in relation to the temporal activation of inflammatory cells after HI is urgently needed as such information may be helpful to find few antiinflammatory and neuroprotective strategies.

We examined the inflammatory cell response and the chemokine mRNA expression with reverse transcription-polymerase chain reaction (RT-PCR) in a model of neonatal HI in 7-day-old rats. The α-chemokines growth related gene (gro) and macrophage inflammatory protein-2 (MIP-2) and the β-chemokines MIP-1α, MIP-1β, Eotaxin, and RANTES were studied. The results were temporally related to the 1) proinflammatory cytokine response (IL-1β and TNF-α) and 2) the activation of microglia/macrophages, astroglia, granulocytes, lymphocytes, NK cells, and the expression of cell adhesion molecules β2-integrin and ICAM-1.

MATERIALS AND METHODS

Protocol and operative procedures. A total of 156 Wistar rat pups of both sexes were used (Møllegaard, Skensved, Denmark). At postnatal d 7 (P7) pups were exposed to HI as follows: the pups were anesthetized with halothane (3.0-3.5% for induction and 1.0-1.5% for maintenance) in a mixture of nitrous oxide and oxygen (1:1). The left common carotid artery was dissected and cut between double ligatures of prolene sutures (6-0). The duration of anesthesia was <10 min. After the surgical procedure, the wounds were infiltrated with a local anesthetic. The pups were left to recover for at least 1 h. The litters were then placed in a chamber perfused with a humidified gas mixture (7.70 ± 0.01% oxygen in nitrogen) for 70 min. The temperature in the gas chamber was kept at 36°C. After hypoxic exposure the pups were returned to their dams. They were reared at 20°C environmental temperature with a light:dark cycle of 12:12 h and food and water ad libitum. All animal experiments were approved by the ethical committee of Göteborg (no. 1-95).

RNA preparation. Six pups at each time were killed by decapitation directly after hypoxia and at 1 h, 3 h, 6 h, 12 h, 24 h, 72 h, and 14 d of recovery after the insult and the brains were rapidly removed and frozen in liquid nitrogen. Control pups not exposed to ligation or hypoxia were decapitated on P7, P8, P10, and P21. Total RNA was extracted, from each hemisphere, by the phenol-chloroform method described by Chomczynski and Sacchi(22). The RNA was quantified by spectrophotometry at 260 nm.

RT-PCR. First strand cDNA synthesis was made, for pooled (n = 6) and for single hemisphere RNA samples, with Superscript RNase H- reverse transcriptase kit (GibcoBRL, Life Technologies, Täby, Sweden) and random hexamer primers (Boehringer Mannheim, Scandinavia AB, Bromma, Sweden). The cDNA was diluted with water and stored at -20°C.

PCR was carried out using Taq DNA Polymerase (Sigma Chemical Co., St. Louis, MO). Upstream and downstream primer sequences are shown in Table 1. The GAPDH amplification was carried out from 12 to 28 cycles at 4-cycle intervals and the chemokine and cytokine amplifications from 24 to 40 cycles at 4-cycle intervals performed on a MiniCycler (MJ Research Inc, Watertown, MA). All primers were amplified for different cycle numbers and the appropriate cycle number was chosen from the linear amplification phase to avoid the plateau phase. All pooled groups were amplified at the same number of cycles for each chemokine for further comparison. Annealing time and temperature see Table 1 for each chemokine.

Table 1 Chemokines and cytokines studied in the post-HI rat brain; GAPDH was used as a house-keeping gene for normalization of the expression of the other genes
Full size table

To visualize the result the PCR products were separated by electrophoresis on an ethidium bromide-containing agarose gel-and photographed under UV light. A 100-basepair ladder (Boehringer Mannheim) was used to verify the size of the PCR product. Negative technical controls without cDNA were used for each set of reactions.

All photographed PCR products were scanned using a flat bed scanner and quantified using the software program IPLab Gel (1.5f, Signal Analytics Corporation, Fairfax, VA). GAPDH was used for normalization of variations in product abundance due to different efficiencies in individual RT and PCR reactions. For analysis of mRNA induction we used the ratio (%) between the chemokine of interest and the corresponding RT-PCR product for GAPDH mRNA.

Brain sections. Directly after hypoxia and at 3 h, 6 h, 12 h, 24 h, 72 h, 7 d, 14 d, and 35 d of recovery after the insult six pups were decapitated at each time. Control pups, not exposed to ligation or hypoxia, were decapitated at P7, P10, P14, P21, and P42. The brains were dissected and frozen in dry ice-chilled dimethylbutane. The brains were stored at -80°C until use. Coronal cryostat sections were cut at the level of striatum and hippocampus. The 8-µm thick sections were thaw-mounted on Polysine- (Mentzel-Gläser, Buchs, Switzerland) coated slides. These sections were stained with H&E or used for immunohistochemistry.

Immunohistochemical staining. The sections were air dried for at least 30 min and circled by a wax-pen to stop the antibodies from floating out. Sections were fixed for 10 min with 5% formaldehyde in PBS (pH 7.2). After rinsing in PBS nonspecific binding was blocked for 30 min with 4% horse serum in PBS. The serum was shaken off and the primary antibody was incubated for 60 min. Primary antibodies are listed in Table 2. After rinsing in PBS three times, all sections except the ones incubated with GFAP, were incubated with a biotinylated horse anti-mouse antibody (Vector Laboratories Inc., Burlingame, CA) for 60 min followed by 5 min treatment with 3% H2O2 in PBS. The anti-GFAP antibody was followed by a secondary biotinylated goat anti-rabbit antibody (Vector laboratories Inc.) using the same procedure as described above. The sections were incubated with avidin biotin peroxidase complex for 60 min (Vector Laboratories Inc.) and put in Na acetate buffer (pH 6.0, 0.1 M) for 10 min. Finally the immunoreactivity was visualized with 0.5 mg/mL 3,3-di-aminobenzidine enhanced with 15 mg/mL ammonium nickel sulfate, 2 mg/mL β-D glucose, 0.4 mg/mL ammonium chloride, and 0.01 mg/mL β glucose oxidase (Sigma Chemical Co.) dissolved in Na acetate buffer (pH 6.0, 0.1 M). Negative controls were performed using the same procedure described above in the absence of primary antibody. These sections were devoid of immunoreactivity. Blood, intestine, and spleen from rats were used as positive controls for the antibodies against lymphocytes and granulocytes.

Table 2 Primary antibodies used for immunohistochemistry
Full size table

We tested with different protocols for fixation, incubation time, concentration of antibodies and blocking reagents, but found no expression for the CD8(β) and ICAM-1 antibody. As positive controls for the ICAM expression we used lipopolysaccharide 0.2 µL of a 1 mg/mL solution injected into the brain 1 mm caudal to bregma, 2 mm lateral, 2 mm deep of 7-day-old rats decapitated 24-30 h after the injection. The antibody HIS48 stained only a subset (about 20%) of all neutrophils visualized by H&E staining. The MOM/3F12/F2 antibody stained all unspecific components in the vessels of both control and HI brains, and no correlation could be found to granulocytes or even other blood cells visualized by H&E staining.

Statistics. Values are given as mean ± SD unless stated otherwise. The Mann-Whitney U test was used to test differences between the hypoxic-ischemic group and the respective control groups. In those situations when one group was used for multiple comparisons the p was adjusted using the Bonferroni method.

RESULTS

RT-PCR. The constitutively expressed housekeeping gene GAPDH was used for normalization of the individual samples for the concentration of total mRNA extracted and differences in RT efficiency (Fig. 1). The ratio (%) of chemokine and GAPDH PCR products are shown in a histogram for each chemokine. As shown by earlier studies, GAPDH mRNA levels in control brains does not depend on the maturational level(7) and protein levels(23) were not affected by ischemia (Fig. 1).

Figure 1
figure 1

Gro mRNA expression after HI. GAPDH mRNA and Gro mRNA in the left (L) and right (R) hemisphere in controls, and 0, 1, 3, 6, and 12 h after HI. Gels containing ethidium bromide were photographed under UV light.

Full size image

The basal levels for the chemokine gro were low in all control brains (Fig. 1 and 2A). Immediately after HI there was a rapid increase of mRNA for gro in the ipsilateral left hemisphere and the expression reached a peak 12 h after HI (Figs. 1 and 2A). The MIP-2 mRNA reached maximum expression directly after HI and was still elevated up to 24 h in the ipsilateral left hemisphere after the insult (Figs. 2B and 3A, p < 0.01). The TNF-α mRNA expression was increased 1-24 h after HI (Fig. 2C) and the mRNA for IL-1β was induced between 0 and 12 h of reperfusion (Fig. 2D). The MIP-1α chemokine was expressed in control brains and there was a discrete up-regulation of MIP-1α in the ipsilateral hemisphere 6 h to 14 d after HI (Fig. 4A). The MIP-1β expression in the ipsilateral hemisphere, however, increased considerably in response to HI at 1-6 h of recovery (Figs. 3B and 4B). The mRNA for RANTES was expressed in control brains and there seemed to be a limited induction 24 h to 14 d after HI in the ipsilateral hemisphere (Fig. 4C). The Eotaxin chemokine was not expressed at any time in the control or HI brains (Fig. 4D).

Figure 2
figure 2

mRNA expression for gro, MIP-2, TNF-α, and IL-1β after HI. The expression (RT-PCR) of mRNA for gro (A), MIP-2 (B), TNF-α (C), and IL-1β (D) are given as ratios/GAPDH in control animals (gray bars) killed at postnatal d 7 (P7) or 21 (P21) or 0, 1, 3, 6, 12, 24, 72 h or 14 d after HI. The values are given separately for the left (open bars) HI and right (solid bars) hemisphere. Data are expressed as mean ± SD.

Full size image
Figure 3
figure 3

Expression of mRNA for MIP-2 and MIP-1β after HI. The mRNA was also analyzed in brain tissue of individual animals of controls at P7 (n = 6; gray bars), left HI hemisphere (n = 6; open bars) and right hypoxic hemisphere (n = 6; solid bars) after the insult. HI induced an expression of mRNA for MIP-2 and MIP-1β. Values are given as/GAPDH ratios (mean ± SD). **p < 0.01 vs control animals, ††p < 0.01 vs contralateral hemisphere.

Full size image
Figure 4
figure 4

mRNA expression for MIP-1α, MIP-1β, RANTES, and Eotaxin after HI. The expression (RT-PCR) of mRNA for MIP-1α (A), MIP-1β (B), RANTES (C), and Eotaxin (D) are given as ratios/GAPDH in control animals (gray bars) killed at postnatal d 7 (P7) or 21 (P21) or 0, 1, 3, 6, 12, 24, 72 h or 14 d after HI. The values are given separately for the left (open bars) HI and right (solid bars) hemisphere. Data are expressed as mean ± SD.

Full size image

Hematoxylin and eosin. Few neutrophils were detected in control brains and during the first 6 h after HI. However, a considerable increase of neutrophils was seen at 12 h of reperfusion (38.7 ± 32.9 cells per ipsilateral hemisphere section, mean ± SD) (Figs. 5 and 6,A-C) being significantly higher than in control brains (p < 0.0001). As shown in Figure 6, these cells were found both intravascularly and in the tissue. In the most damaged brains we could find as much as 80 cells/ipsilateral hemisphere section. In the sections with longitudinally cut vessels example of neutrophil adhesion to the vessel walls was observed (Fig. 6C). The neutrophils were most often found within and around vessels in the damaged gray and white matter.

Figure 5
figure 5

Neutrophil count in H&E stained sections. The number of neutrophils was significantly higher at 12 h after HI compared with controls (***p < 0.0001) and the contralateral hemisphere (††p < 0.01). (mean ± SD).

Full size image
Figure 6
figure 6

(A-C) Neutrophil accumulation after HI. Photomicrographs of brain sections from animals subjected to HI and 12 h of reperfusion. (A) Black arrows indicate neutrophils and the white arrow indicates an eosinophilic cell in which the cytoplasmic granules can be seen. (B) Arrows indicate neutrophils in the damaged tissue outside the vessels. (C) Neutrophils are seen attached to vessel walls (arrows). [bar length (A-C) 10 µm].

Full size image

We found a few cells with the typical morphology of plasma cells in the border zone of the infarct core at 12 h after HI. The presence of inflammatory cells and expression of adhesion antigens in the infarct region is summarized in Figure 7.

Figure 7
figure 7

Summary of the temporal expression of inflammatory cells, astrocytes, and β2 integrin in damaged brain regions. Neutrophils were counted in H&E stained sections or with immunohistochemistry using the HIS48 antibody. Microglia/macrophages, astrocytes, lymphocytes (CD4 and CD8a) and NK cells were visualized with immunohistochemistry (Table 1). The intensity of staining was graded semiquantitatively from mild (dashed line) to extensive (thick line).

Full size image

Immunohistochemistry. Most brains of animals exposed to HI exhibited a decreased immunoreactivity of the neuronal dendro-somatic marker microtubule associated protein-2 (MAP-2) in the ipsilateral cerebral cortex, striatum, hippocampus, and thalamus (Fig. 8). The regions with loss of MAP-2 staining corresponded to regions with brain injury as visualized in H&E sections(24). At P7-P10 we found a bilateral expression of CD4+ cells in the corpus callosum, the periventricular region, and in the meningeal vessels in control brains. Three days after HI the expression of CD4 cells was increased bilaterally in the corpus callosum but also in the damaged regions that proceeded and reached a maximum level by 7 d (Figs. 9,A and B). The CD4 expression persisted in damaged areas up to 14-35 d of reperfusion and a clear correlation between CD4 expression and the extent of damage was observed.

Figure 8
figure 8

MAP 2 and GFAP expression. The approximate distribution of the cerebrocortical infarct is visualized by the loss of MAP 2 staining (left). The GFAP expression (right) was markedly enhanced in the infarct border 72 h after HI. (bar length 1 mm).

Full size image
Figure 9
figure 9

CD4 lymphocytes and NK cells after HI. (A) CD4+ lymphocytes invaded the infarcted region at 7 d after the HI insult (bar length 1 mm). (B) Higher magnification of CD4+ cells (bar length 10 µm) and (C) an NK stained cell with large cytoplasmic granules (bar length 10 µm).

Full size image

In all control brains, a few cells (1-5 per hemisphere) with meningeal vessel localization expressed the CD8 epitope. At 24 and 72 h after HI we found a discrete increase of CD8 positive cells in the corpus callosum and in the infarct region (10-30 cells per hemisphere section).

In some NK+ cells immunoreactivity was localized to large granules in the cytoplasm (Fig. 9C). We found a few (1-5 cells per hemisphere section) NK+ cells in the meninges, gray matter, and plexus choroideus in the control brains. At 12 h there was a variable increase of NK+ cells, with a moderate expression (10-15 cells ipsilaterally, 2-6 cells contralaterally) in some brains and a more pronounced accumulation (20-60 cells ipsilaterally) in others being concentrated to the infarcted region.

In the control brains at P7, glial fibrillary acidic protein (GFAP) expressing cells were found bilaterally in the white matter of corpus callosum and at P10-P14 staining also appeared in scattered cells of the cerebral cortex. After P21 the number of GFAP+ cells decreased until P42. In the HI brains the induction of GFAP expression at the infarct border started at 12 h after the insult and increased up to 72 h (Fig. 8). Thereafter, a shift in GFAP staining occurred from a localization in the border-zone to the infarct core apparent 7-35 d after HI (not shown).

The microglia/macrophage antibody OX-18 against the MHC class I was detectable bilaterally in the corpus callosum in the control P7 brain but was not seen at P10, P14, P21, or P42. Three hours after HI the number of OX-18+ cells increased bilaterally in the corpus callosum and in the plexus choroideus in the lateral ventricles. A few OX-18+ cells were found in the damaged regions in the cerebral cortex and striatum at 12 h after HI in one-third of the brains. All brains exhibited increased OX-18 staining at 24 h and the number of immunoreactive cells increased in the damaged region at 72 h. The most pronounced induction was found in severely injured brains.

In the control brains we found less than five β2-integrin positive cells in each hemisphere section. The β2-integrin immunoreactivity was somewhat enhanced 3 h after HI, with a further increase at 12 h in some brains (30-80 cells per hemisphere section). The early expression was not concentrated to the infarct but 7 days after HI there was a marked increase of β2-integrin in the damaged region (Fig. 10) that persisted up to 35 d. The number of β2-integrin positive cells was sparse (less than five cells/section) in the contralateral hemisphere.

Figure 10
figure 10

β2 integrin staining after HI. Seven days after the insult there was an increase of β2 staining cells in the infarct region of all the damaged brains. The corresponding MAP 2 stained section is shown to indicate the region of infarction (bar length 1 mm).

Full size image

DISCUSSION

This is the first demonstration of an induction of mRNA for the α and β chemokines MIP-2, MIP-1α, and gro as well as detection of lymphocyte infiltration after neonatal HI. The α-chemokine (gro and MIP-2) response was accompanied by a cytokine increase (TNF-α and IL-1β) followed by the expression of β2-integrins and accumulation of neutrophils within and outside vessels 12 h after HI in the infarcted area. The β-chemokine expression (MIP-1β and to some extent MIP-1α) preceded the accumulation of microglia/macrophages, CD4 and CD8 positive lymphocytes, and NK cells in the infarct. The activation of microglia/macrophages, CD4 lymphocytes and astroglia persisted for at least 35 d after the insult implicating a chronic state of inflammation (Fig. 7).

Neutrophils, α-chemokines, and adhesion molecules. Many neutrophils were identified as isolated cells or aggregates within blood vessels (Fig. 6A), attached to vessel walls (Fig. 6C), and at various stages of migration through diapedesis from the vessels into the infarcted region as evidenced by tissue infiltration of neutrophils (Fig. 6B). These findings do not quite agree with previous reports(8,10) that could partly be explained by the fact that the 12 h time point has not been addressed before. Contrary to our results, Hudome and co-workers(10) found no consistent difference in cell count between the injured and uninjured hemisphere which also may be attributed to the different methods used for neutrophil visualization. However, induced neutrophil depletion was found to provide protection from edema formation(10) which implies that neutrophils may participate in the cascade leading to the development of brain damage that is also supported by studies on adult brain ischemia(25–28).

Intracerebral injection of the chemokine MIP-2(21) and transgenic mice overexpressing the chemokine gro(29) have been found to provoke polymorphonuclear leukocyte recruitment into the brain. We found an immediate α-chemokine mRNA induction already at the end of HI in this study. MIP-2 peaked at 1 h after the insult although gro peaked somewhat later. This time course is in consistence with other studies demonstrating induction of mRNA and protein for the structurally related chemokine cytokine-induced neutrophil chemoattractant in adult ischemia(30,31). The α-chemokine induction preceded and the peak levels matched the invasion of neutrophils in the vessels and tissue (Fig. 5). Other studies have shown that mRNA and protein levels increase in concert indicating synthesis within the brain(32). The α-chemokines can be produced by a variety of cells including astrocytes(32), endothelium, macrophages, fibroblasts, and smooth muscle cells(30) but it is not clear at present which of these cells that produce α-chemokines under these specific conditions. One has to be cautious when trying to make causative links on the basis of temporal relationships and future studies using immunocytochemistry and in situ hybridization for chemokines are needed to settle the cellular origin and the precise regional relationship between chemokine activation and immunoinflammatory cells. This is, however, a methodological challenge as cytokines/chemokines are produced in minute amounts and exert biologically important effects at extremely low concentrations(19).

The chemokines are thought to regulate the expression of different adhesion molecules on the surface of endothelial and inflammatory cells(19). Firm adhesion and extravascular migration of neutrophils appear to require β2-integrins(33). The human gro-related chemokine IL-8 has been shown to stimulate binding activity(34) and up-regulate β2-integrin receptors(35). In our study, we found β2-integrin immunoreactive cells at 12 h after the insult, the same time as the neutrophil cell invasion was visible. The counter receptor to the β2-integrin is the ICAM ligand(1). ICAM-1 mRNA and protein has been found to be up-regulated shortly after ischemia in the adult brain(36–38). We could not find expression of ICAM-1 in response to HI at any time point using different protocols with two different ICAM-1 antibodies. According to other studies, the ICAM-1 molecule seems to be involved primarily in transendothelial migration(39) and a lack of ICAM-1 response might therefore explain why most neutrophils seen in this study were located inside the vessels (Figs. 6,A and C). There are a number of other intercellular and vascular adhesion molecules, selectins, and integrins that might be involved in this process, and we do not know which is the most important in the immature brain(40). It is important to point out that β2-integrin positive cells remained in the tissue for 35 d whereas neutrophils were found 12 h after HI supporting the view that other adhesion proteins than β2-integrin are important for neutrophil adherence. Macrophages, microglia, and possibly lymphocytes can express the β2-integrin(1). All of these cells were represented in the infarct region during the prolonged β2-integrin expression (Fig. 7). Indeed, the increase in OX-18 positive cells matched the pattern of β2-integrin staining which suggests that β2-integrin may be expressed by microglia/macrophages.

Glial cells, lymphocytes, cytokines, and β-chemokines. The data for microglia/macrophage expression in this study both in the P7 control brains and the induction after HI are supported by earlier findings by our group and others(6,8,41). Our finding of a persistently high expression in the damaged regions at 14 d after the insult is supported by our own former study(6) and the present data demonstrate sustained expression up to 35 d. In the previous two studies the microglia/macrophage expression peaked after 2-4 d and was gone after 1-2 wk(8,41). This difference could be explained by the fact that we used an antibody directed against the MHC I antigen on microglia/macrophages (OX-18) which may represent a somewhat different population of cells than that recognized by lectin staining or the ED-1 antibody used in the other studies(8,41).

So far there are very few studies investigating the role for lymphocytes after ischemic brain injury(13,14,42,43). Subsets of lymphocytes are functionally distinct populations of cells that express different membrane proteins, i.e. the cluster of differentiation (CD) molecule. In adult ischemia, Schroeter and coworkers(43) found predominantly a CD8+ cell response in and around the infarct at 3-7 d after the insult. On the contrary, in the neonatal brain lymphocyte activation involved mainly CD4 cells in the infarct whereas merely a discrete induction of CD8+ cytotoxic/cytolytic T cells at 24-72 h occurred after the insult.

NK cells or large granular lymphocytes are defined as a functional or morphologic group of cells(44). They respond to different cytokine induction and can kill target cells rapidly by a non-MHC-restricted and antigen nonspecific mechanism. The NK cell antibody used here has been suggested to react with both NK cells and granulocytes(45). This was not the case in our study as the NK positive cells started to occur in low numbers at 12 h after the insult but increased at later times when the neutrophils were not seen in the tissue.

The β-chemokines can be produced by a variety of cells such as glial cells, monocytes, mast cells, lymphocytes, and fibroblasts(19). The target cells for the β-chemokines RANTES, MIP-1α, and MIP-1β are monocytes and T lymphocytes and for RANTES and MIP-1α also NK cells. There was an early induction of MIP-1β 1-6 h after HI (Fig. 4B) followed by moderate expression of MIP-1α at 6-72 h of recovery (Fig. 4C). The MIP-1α response agrees with results found for the MIP-1α mRNA after ischemia in adult rats(46). The presently found expression of RANTES in response to HI was discrete and delayed which is in agreement with the increase of RANTES detected 24 h after stab wound brain injury in adult animals(47). In yet another study, intracranial injection of RANTES was found to attract CD4+ cells into the brain parenchyma 24-48 h after injection(21) which is in agreement with the late CD4+ cell response that we found in the infarct region in our study. CD4 T cell activation can only be induced by an antigen presenting cell (microglia/macrophage) displaying the MHC class II together with the antigen. This agrees with our findings that the microglia/macrophage activation occurs earlier than the lymphocyte infiltration. The prolonged expression of RANTES might also be of importance for the increased number of microglial cells/macrophages late after the insult. All the chemokine target cells appeared in the infarct after the start of the chemokine expression (Fig. 7). The β-chemokine MCP-1 has previously been detected 4 h and later after HI in the neonatal brain(16). We presently found a very rapid induction (1 h after HI) of another β-chemokine (MIP-1β) which matches the very early activation of microglia/macrophages that occurs after the insult(6,41).

In a former study we found a transient increase of the cytokine IL-1β at 3-6 h after the HI insult(9). Other studies support this finding and has shown the same pattern of activation for the proinflammatory cytokine TNF-α(7). We now found an even more rapid increase for TNF-α, but with somewhat delayed peak values for IL-1β and TNF-α at 6-12 h. These cytokines are probably produced by resident cells such as microglia/macrophages in the central nervous system which in turn may be activated by β-chemokines (above). However, TNF-α and IL-1β can induce the expression of transcription factors such as activator protein-1 and nuclear factor κB that regulate the transcription of most chemokines(19) suggesting that these proinflammatory cytokines may activate the production of chemokines.

GFAP is the major intermediate filament protein of reactive astrocytes(48). Data from other studies indicate that the increased local tissue GFAP immunoreactivity is a sensitive indicator of neuronal injury(49). Our data are well supported by another study of GFAP induction in this HI model(50). We found a slight increase in GFAP staining as early as 12 h after the insult in white and gray matter areas surrounding the region of MAP 2 loss and H&E-visualized affected cells, and at 24-72 h this was readily seen in all brains. By 7 d after HI we found an interesting switch of GFAP immunoreactivity from the "border zone" to the infarct core that has not been described before in this model. This astrogliosis seemed to be permanent as it persisted to early adolescence.

No increase in Eotaxin chemokine mRNA level was ever detected with RT-PCR, which was in accordance with our morphological findings in the H&E section of only one or no eosinophilic granulocytes in the brains.

CONCLUSIONS

Hypoxia-ischemia rapidly induced mRNA for α-chemokines (gro and MIP-2) which preceded the expression of β2-integrin and accumulation of neutrophils. Induction of β-chemokines (predominantly MIP-1β) preceded the appearance of lymphocytes, microglia/macrophages, and NK cells in the infarcted area. These data suggest a critical role of chemokines for immunoinflammatory activation after HI and further characterization of these mediators is warranted. Furthermore, the activation of microglia/macrophages, CD4 lymphocytes, and astroglia persisted for at least 35 d after the insult which suggest that HI induces a chronic state of inflammation.