03064522/90$3.00+ 0.00 Pergamon Press plc Q 1990IBRO

NeuroscienceVol. 35, No 1, pp. 121-132,1990 Printed in Great Britain

AN INVESTIGATION INTO THE EARLY STAGES OF THE INFLAMMATORY RESPONSE FOLLOWING IBOTENIC ACID-INDUCED NEURONAL DEGENERATION P. J. COFFEY*, V. H. PERRY~ and J. N. P. RAWLINS Department

of Experimental Psychology, University of Oxford, South Parks Road, Oxford OX1 3UD, U.K.

Alar-Inj~tion of the excitatory neurotoxin ibotenic acid into the septum produces rapid destruction of neuronal cell bodies and accompanying gliosis. We have previously shown that following ibotenateinduced cell death this may also result in damage to healthy axons en passage (Coffey et al., Neurosci. L&t. 84, 178-184, 1988). We suggested that the axonal damage resulted from non-specific damage by recruited inflammatory cells. In this study we have further examined the phenotype of the cells involved in the inflammatory response in the rat. Immunocytochemical identification of cells in the region of the lesion site identifies them as being of haematopoitic origin and most of them have the phenotype of macrophages. The dramatic increase in their number following an ibotenate lesion is sensitive to irradiation of the body providing evidence that the majority are blood derived. The inflammatory response is accompanied by a loss of myelin and a breakdown of the blood-brain barrier in the region of the lesion site. We have shown that these two effects are consequences of the inflammatory response since reduction in the inflammatory response by prior irradiation will abrogate these two effects.

A number of simple compounds which produce excitatory responses in central nervous system neurons are neurotoxic at high concentrations (see 01ney26 for review). In addition to amino acids, such as glutamate and aspartate, other compounds such as kainic acid, ibotenic acid, quinolinic acid and N-methyl-naspartate (NMDA) are neurotoxic at the appropriate concentrations. There is much interest in these compounds not only because of their possible use in producing “axon sparing” lesions in the CNS’V~~but also because high endogenous levels of excitatory substances may underlie neuronal degeneration in a number of pathological states and chronic neurodegenerative diseases.3,33 The neuronal loss produced by the injection of an excitotoxin into the CNS is accompanied by a vigorous gliosis’6-an increase in the number of non-neuronal cells at the locus of the lesion. It has been shown that injection of kainic or ibotenic acid into the brain produces a dramatic increase in the number of microglia4+” and these cells account for a large proportion of the increase in cell density.4 Our own results suggest that the intense inflammatory response is not only invohed in removal of debris from the site of the lesion but can produce demyelination and damage to healthy axons en passuge.4 *Present address: Department of Psychology, University of Sheffield, Western Bank, Sheffield. SlO 2TN. U.K. ~To whom correspondence should be addressed. Abbreviations: HRP, horseradish peroxidase; LCA, leucocyte common antigen; NMDA, ~-methyl-D-aspa~te; PBS, phosphate-buffered saline.

In the present study we have examined the phenotype of the cells involved in this inflammatory response. We have sought to provide evidence for their blood-borne origin and provide evidence that these cells are responsible for the loss of myelin at the site of an injection. In addition we have examined how this inflammato~ response affects the permeability of the blood-brain barrier. ~PE~M~T~

PROCEDURES

Surgical procedures Adult Sprague-Dawley rats of approximately 30&400 g body weight were used. The animals were anaesthetized with a mixture of chloral hydrate and sodium pentobarbital (3.0 ml/kg of solution containing 2.1 g chloral hydrate + OSg sodium pentobarbital in 50ml) and held in a stereotaxic apparatus. A single injection of ibotenic acid (Sigma, U.K.) 0.2~1 of lOpg/pl in phosphate-buffered saline (PBS) was made into the medial septum. The ibotenic acid was delivered via a modified Hamilton syringe at a rate of 0.05 $/min4 Control injections of PBS were made into other rats. After survival times varying from 1 h to seven days the animals were prepared for either immunocytochemical analysis or assessment of damage to the blood-brain barrier. Animals were anaesthetized and perfused transcardially at room temperature with fixative containing 2% paraformaldehyde, 0.075 M lysine and 0.01 M met$eriodate in 0.037 M phosphate buffer with an initial pH of 7.2-7.4. The fixative was prepared just before ~se.~ After perfusion, the brain was-blocked and the region of the injection post-fixed for a further 6 h followed by cryoprotection in 30% sucrose in phosphate buffer overnight. The tissue was embedded in Tissue Tee OCT compound (Lamb) and rapidly frozen. Coronal sections were cut at 10 pm on 121

a cryostat and mounted on gelatimzed slides for subsequent immunocytochemistry. Additionai sections were stained with Cresyl Violet for Nissl substance. or Luxol Fast Blue for myelin. The following mouse-derived monoclonal antibodies were used: OX42 directed against the type 3 complement receptor;” W3i2.5 directed against the CD4 antigen:” OX1 + OX30 directed against leucocyte common antigen (LCA) epitopes;r6 OX6 directed against the Class II major histocompatibility antigen” and OX19 directed against T lymphocytes.h Sections were incubated with the primary antibodies for I h and the primary antibodies detected by the use of the avidin-biotin-peroxida~ method” with diaminobenzidene as the chromogen. Care was taken to prevent non-specific Fc-receptor binding by preincubation with normal serum and non-specific peroxidase activity was abrogated by incubation of the sections in I% hydrogen peroxide in methanol at the appropriate stage. Immunoreactivity of cells for a particular monoclonal antibody is designated + Blood--brain barrier permeabilit> We assessed the effect of the ibotenic acid injections on the blood-brain barrier in the folIowing manner. Animals were given injections of either PBS or ibotenic acid into the medial septum as described above. One hour, 6 h, I2 h, one day, three days or seven days later animals were anaesthetized and horseradish peroxidase (HRP: 3~hringer) was slowly injected into a femoral vein at a dose of 1.5 mg/IOO g body weight in 0.5 ml of physiological saline. The anaesthetized animal was then perfused 15min later with physiological saline followed by a fixative containing 2.5% paraformaldehyde and t .25% glutaraldehyde in 0.1 M phosphate buffer pH 7.2. For cryoprotection the brain was left in 30% sucrose in phosphate buffer overnight. Frozen sections were cut at 5Opm and alternate sections reacted with a sensitive modified Hanker---Yates methodM to reveal the HRP or stained with Cresyl Violet. At Least two animals injected with either PBS or ibotenic acid were examined at each time point. Radiation studies

Whole body irradiation at a dose of 900 rads is known to deplete the progenitor cells in the bone-marrow and has been used as a tool to reduce cellular recruitment to an inflammatory stimulus in various studies including experimental allergic encephalitis.‘4 We might expect that cells in the adult central nervous system would not be sensitive to radiation since they are post-mitotic cells. However, in preliminary experiments we found that 900 rads whole body irradiation produced small numbers of pyknotic cells widespread throughout the cortex and hippocampus. in addition we found an increase in the permeability of centra1 nervous system capillaries at scattered focal sites following intravenously injected HRP (unpublished observations). Thus, because whole body irradiation alone may damage some components of the CNS we have used both whole body irradiation and body irradiation with the head shielded by a lead block. Animals received 700 or 900 rads (42.4 rads/min) whole body irradiation from a RX30/55M irradiator (Gravatom Ltd) with a “‘Cs source. Another group of animals was anaesthetized and the head was placed under a 3.5-cm-thick block of lead and the remainder of the body received 900 rads. Measurements of the shielding produced by the lead block were made with a dosimeter (Ionex, type 2500/3) and the lead block was found to reduce the radiation dose to the cranium to a total dose of less than 50 rads. Two days later the animals were given an injection of ibotenic acid as described above. Animals survived for five days prior to being killed and the brain processed for immunocytochemistry; or three days prior to receiving an intraven&s injection of HRP to assess the state of the blood--brain barrier.

To determine the relative contribution of cells i-ccruited from blood. we examined the density of different cell types at the locus of the lesion in animals with ibotenic acid lesions with and without prior irradiation five days after ibotenate injection (n = 2 for each group assessed). Cresyl Violet stained sections and sections processed for immuil~~cyt[)chemistry with monoclonal antibody OX42 and counterstained with Cresyl Violet were selected. In sections passing through the centre of the lesion, numbers of neurons, non-neuronal cells and OX42 + cells were counted in fields IOO~m x IOOpm at intervals of 2OO~m. Neurons were distinguished by their pale nucleus, Nissi substance within the cytoplasm and in some cases a clear nucleolus. The “non-neuronal” cells included all those cells which were not neurons or endothelial cells. The cytological characteristics of cells was not sufficiently clear in this material to allow a reliable distinction between the different types of nonneuronal cells. The 0X42+ cells were also counted. RESULTS General

Neurons of the medial septum are sensitive to ibotenic acid: within 24 h those within an area about 1.5 mm in diameter have largely degenerated. Pale and shrunken nuclei were still present as were a few pyknotic cells. Along the cannula track there were cells with multilobed nuclei typical of neutrophils and these cells contained peroxidase positive granules as revealed by incubation of some sections in diaminobenzidine with hydrogen peroxide (Fig. IA). Few of these cells were found more than a few hundred microns from the cannula track. A population of cells with heterochromatic nucIei with either rounded or irregular shaped nuclei were present in conspicuous numbers and these we believe, on the basis of evidence to be presented, were mostly recruited monocytes. Three days after the Iesion, the number of non-neuronal cells had greatly increased and there was evidence of cuffing around the smaller blood vessels. The density of these cells was further increased at five and seven days after the lesion. Immunocytochemistry

In the normal septum the microgha are readily stained by the monoclonal antibody 0X42. We use the term microglia to describe those cells with the characteristic morphology shown in Fig. IB. They form a relatively regular distribution across the septum, have a small heterochromatic, irregular nucleus and a variable number of fine, branched, stubbled processes. The staining with W3/25 and LCA was very weak or absent and no cells were Class II + Twenty-four hours after injection of ibotenic acid the cell population was quite different at the locus of the lesion with many rounded 0X42+- cells, some with short processes. Cells labelled with OX42 included both neutrophils and presumptive macrophages. A notabfe feature of cells in this region is shown in Fig. 1C; the cells bearing long processes typical of microglia were much fewer in number than in the normal septum. A proportion of the cells

Ibotenic acid-induced inflammation labelled with OX42 were also W3/25+, LCA and 0X6-+ and the majority of these ceils was confined to the central part of the lesion. Three days after injury the number of OX42 f cells had greatly increased. Along the cannula track the majority of cells had a relatively simple rounded morphology while those a few hundred microns away from the cannula track, but within the area devoid of neurons, were either rounded in form or are microglial-like cells (Fig. ID). It should be noted that they are distinct from the microglia of the normal septum as the cells have larger somas, and more short processes (compare Fig. 1B and D). The cells stained with W3/25 appeared similar in morphology although somewhat fewer in number. The 0X6+ cells were predominantly rounded cells and confined to the cannula track. A small number of lymphocytes as defined by their simple morphology and staining with OX19 were also present close to the cannula track. By five and seven days post-injection the number of 0X42+ cells at the main focus of the lesion had increased to such a density that it was difficult to resolve individual cells and their plasma membranes. Around the very dense central region of the lesion there were many cells with shorter, thicker processes each of which appeared to be covered in finer, hair-Iike processes (Fig. 2A) and there were few cells of the rounded form illustrated in Fig. 1D. The denser branching gave these cells a bushy appearance, quite characteristic and distinct from resident microglia. These we shall refer to as the reactive microglia. Although this te~inolo~ is consistent with other authors’ usage we would like to make it clear that we do not infer from this terminology that the cells are derived solely either from resident cells or recently recruited cells since we cannot distinguish such cells on the basis of mo~hology alone. This term is used as a morphological descriptor. A number of the cells with the morphology of activated microglia also expressed CD4 (W3/25+) and Class II (0X6+) (Fig. 2B and C) but the density of ceils labelied with these antibodies was less than that seen with 0X42. The density of W3/25+ and 0X6+ cells declined rapidly with distance from the centre of the lesion. There were no morphological criteria to distinguish between ceils labelled with the different antibodies. Blood-brain barrier

The intravenous injection of HRP in the normal rat revealed sites known to be permeable to plasma proteins, for example the sub-fornical organ and the median eminance. Occasionally, a few leaky spots were also observed scattered at random through the series of sections. The surgical procedures required to introduce a cannula into the brain might be expected to produce some mechanical damage to the vasculature and indeed this was the case in control animals injected

123

with PBS. However, the leakage was largely confined to a region about 200 pm either side of the cannula track over the first 24 h and small numbers of rounded cells were also seen to be labelled with HRP. The leakage produced by PBS injection procedures was minimal by three days after injury (Fig. 3A). The efficacy of the HRP injection and reaction could be assessed in each animal by the presence of reaction product in the choroid plexus and sub-fornical organ. The injection of ibotenic acid produced a different picture. Six hours after the ibotenic injection the permeability of the blood-brain barrier was similar to that seen in the PBS injected animals, as judged by the distance and intensity of the spread of the reaction product. However, in addition to the small round cells loaded with HRP lying along the cannula track, in ibotenate injected animals the reaction product was not homogeneously distributed: the HRP appeared as dense “puffs” (Fig. 4A). By examining sections from animals at later time points, the nature of these puffs was much clearer particularly 24 h after injection; the HRP revealed cells with many short highly branched processes (Fig. 4B). From the density of the branching of the processes and the size of the cell body it is clear that these cells are not microglia but protoplasmic astrocytes2 Some of these cells had a long process which extended beyond the dense bush of shorter processes and appeared to end close to a blood vessel (Fig. 4B). These ceils were not seen in any of the PBS injected animals in the regions associated with mechanical damage to the vasculature. Since it was possible that the response of protoplasmic astrocytes was induced by the ibotenic acid and not the inflammatory response per se we looked for evidence that these cells were labelled prior to the development of the inflammatory reaction. Two animals injected with ibotenic acid received an injection of HRP intravenously 45 min later and the brains were fixed after a further 15 min. Even at this short survival time the outlines of protoplasmic astrocytes were visible. Three days after an ibotenate lesion the spread of the HRP was conspicuously greater than that seen in PBS injected animals (Fig. 3B) and the extent of the spread of the HRP matched well with the region of intense gliosis. We could no longer see examples of the HRP labeiled astrocytes at three days after the lesion. The absence of labelling of the astrocytes can not be attributed simply to denser HRP reaction product obscuring them, since even at the edges of the HRP spread we could not resolve these cells. At seven days post-injection the vessels were still permeable to HRP and the spread was comparable with the extent of the gliosis. Thus, an ibotenate lesion produces a greater increase in blood-brain barrier permeability than can be explained by the passage of the cannula alone and the region with increased permeability has a similar size to the area in which we know there is an increase in the number of immunocytochemically identified

Ibotenic acid-induced inflammation cells of macrophage phenotype. The labelling of protoplasmic astrocytes by HRP is not a result of mechanical damage alone since labelled astrocytes were not found in PBS injected animals given intravenous HRP, even when the HRP diffused into the parenchyma. To discover whether the increase in 0X42+ cells was a consequence of recruitment from the blood we sought to reduce the recruitment using whole body irradiation. Effects of irradiation

Examination of Cresyi Violet stained sections from irradiated animals which had received an injection of ibotenic acid into the septum, showed the toxin produced a region of neuronal loss comparable with that found in normal animals with ibotenate injections. At the locus of the lesion, five days after injection, the numbers of non-neurons cells was reduced in irradiated animals when compared with non-irradiated, and the cuffing around the blood vessels was much reduced. In sections from ibotenate injected animals processed for immunocytochemistry it was apparent that the 0X42+ cells were greatly reduced in number. The 0X42+ cells present in the region of neuronal loss were stellate or rounded cells quite distinct from resident microglia. The cells with the form of resident microglia were largely absent from the central part of the lesion but present in the surrounding tissue. To quantify the effect of radiation on the numbers of non-neuronal cells and the 0X42+ cells, we performed cell counts as described in analysis and the results are shown in Fig. 5 (see also Fig. 6B and D). In animals with ibotenate Iesions alone, there is a dramatic increase in the number of non-neuronal cells and a loss of neurons. The number of 0X42+ cells at the centre of the lesion was impossible to determine with certainty due to the density of the staining (Fig. 5A). In irradiated animals, the lesion-induced increase in numbers of non-neuronal cells was greatly reduced when compared with lesioned non-irradiated animals and a large component of this reduction can be att~buted to the lack of an increase in number of OX42f cells (Fig. SB, C and D) (F = 268.26, d.f. 3, 36; P < 0.001). Increasing the dose of radiation has an increasing effect on the reduction of the numbers of 0X42+ cells. (Fig. 5B and C) (F = 268.26, d.f. 3,

125

36; P < 0.001). To control for the possible effects of radiation on the resident population, we also studied animais with the lead shield over the head. There was an increase in the numbers of 0X42+ cells when compared with animals with whole body irradiation (Fig. SC and D) (I; = 268.26, d.f. 3, 36; P < 0.001) but the radiation still greatly reduced the number of 0X42+ cells (compare Fig. 5A and D) (F = 268.26, d.f. 3, 36; P < 0.001). Thus, it appears that the majority of OX42 + macrophages and microglia seen following ibotenic acid injections were derived from the bone-marrow. The effect of irradiation in preventing the increase in 0X42+ cells cannot be due to damage of the resident cells, since in animals with their heads shielded OX42 f cells were again reduced in number. Since whole body irradiation can greatly reduce the recruitment of the myelomonocytic cells from the blood; we were interested to learn whether diminishing the inflammatory response might prevent the damage to fibres running through the septum, which we have described previously in the region of the lesion:, and also prevent the increase in blood-brain barrier permeability following an ibotenate injection. Luxol Fast Blue stained sections were used to assess damage to the myelin sheaths and sections were taken from the animals described above with and without radiation. We compared the intensity of myelin staining in the region of the lesion. It was clear that the myelin loss found in animals with ibotenic injections alone (Fig. 6A) was not present if prior radiation had been given (Fig. 6C) and the sparing of the myelin staining was accompanied by a reduction in the number of 0X42+ cells at the lesion site (compare Fig. 6B and D). Four animals were given a dose of 900 rads with the head shielded and two days later given an injection of ibotenic a&d. After a further three days they received intravenous injections of HRP and the brains were processed as described above. The ~~eability of vessels at the site of the lesion was greatly reduced in irradiated animals with ibotenic lesions when compared with animals with ibotenic lesions alone (Fig. 3C). As before the efficacy of the HRP injection and processing was assessed by comparison of the circumventricular organs in the two different groups of animals and found to be similar.

Fig. 1. Photornicrographs to illustrate cell types present near the centre of an ibotenic lesion in the medial septum. (A) Field of cells close to the canulfa track 24 h after an ibotenic lesion. The section was reacted in diaminobenzidine and counterstained with Cresyl Violet. Note the granules of myeloperoxidase in some of the cells which have the nuclear morphology of polymorphonuclear cells (arrowed). (B) 0X42+ microglia in the normal septum with branched processes and regularly spaced cell bodies. (C) 0X42+ cells 24 h after the injection of ibotenic acid, note that the cells have few processes. (D) OX42 + cells three days after the lesion. Note that there are both rounded cells and profusely branched “activated microglia” present. Scale bars: A = 10 pm, B, C and D = 20 pm. Fig. 2. Activated microglia in the central region of neuronal loss five days after an ibotenic lesion stained with (A) 0X42, (B) W3/25 and (C) 0X6. There are few rounded cells now present. Scale bars = 20 pm.

Ibotenic acid-induced inflammation

127

Fig. 4. Photomicrographs to show HRP labelled protoplasmic astrocytes labelled (A) 6 h and (B) 24 h after an ibotenic injection. In (A) the cells appear as “puffs” of HRP dense accumulations but at 24 h (B) the morphology is better defined and an occasional long process can be resolved (arrow). Scale bars = 20 pm.

Thus ‘3 loss of myelin staining and the increase in vascula r permeability that accompany ibotenic acid lesions are largely dependent on the recruitment of myelon lonocytic cells from the blood. DISCUSSlON

Ibotc :nic acid injections produce a vigorous gliosis in the I.egion of neuronal loss. In this study we have

shown that a major cell type ~ont~buting t o this increase in non-neuronal cells has the phenot: Ype of macrophages and microglia. The majority of* these cells are derived from the blood, since we can 1largely prevent their increase in number by prior deplei tion of the bone-marrow with whole body irradiation. These recruited inflammatory cells damage healthy axons passing through the site of the lesion and con1tribute to the increase in vascular permeability at the lesion

Fig. 3. Photomicrographs to show the diffusion of HRP into the brain parenchyma following intravenous injection of HRP, (A) three days after a PBS injection, (B) three days after an ibotenic injection, and (C) three days after an ibotenic injection into an animal which bad previously received 900 rads body irradiation. Scale bars = 100 pm.

Fig. 5. Figures to show the loss of neurons and differences in 0X42+ eel1 recrnitment following different irradiation protocols. Counts of neurons, non-neuronal cells and 0X42+ cells were made at 2OOpm intervals either side of the cannula track (CT), and counts from the lateral septum not included in the lesion (NL) on the left (NLL) and right sides (NLR) of the septum. The counts show the mean values from two animals in (A) after an ibotenic acid lesion alone. The arrows for the OX42 + c&s indicate the region where the staining density was too intense to permit reliable identification of individual cells. (B) Cell counts after 700 rads whole body irradiation and ibotenic acid injection. (C) After 900 rads whole body irradiation and ibotenic acid injection. (D) After 900 rads body irradiation with the head shielded

and ibotenic acid injection. The survival time following ibotenic acid injection was five days in all cases with irradiation given two days previously.

site. In addition we have presented evidence which suggests that ibotenic acid injections have effects on resident microglia and one population of macroglia, the protoplasmic astrocytes. After an injection of ibotenic acid the resident microglia are no longer present in the region of the neuronal loss, suggesting they are either sensitive to toxin or have altered their morphology in response to ibotenate or neuronal death so that they are indistinguishable from recently recruited cells. The protoplasmic astrocytes are induced by the toxin to endocytose HRP which diffuses into the brain parenchyma. We will consider each of these points in turn. It is well known that the injection of an excitatory neurotoxin into the brain produces neuronal loss accompanied by a gliotic reaction and previous stud-

ies have shown that a major component of this response is an increase in microglia or activated microglia. 4.24From our studies of the time course of the development of this response, we can trace the morphological changes from simple rounded cells to cells which develop numerous processes and come to resemble microgfia. A similar sequence of changes in morphology has been documented by others, for example following neuronal degeneration in the cerebral cortex after kainic acid application to the dura.’ As monocytes enter the developing brain, they also pass through a series of morphological transitions as they ~fferentiate to microglia but these do not look like activated microglia.23*28The differences in the morphology of macrophages and microglia responding to natural cell death in developing brain

Ibotenic acid-induced infla~ation

129

to illustrate the changes in myelin staining with Lax01 Fast Blue in an animal five days after ibotenic injection alone (A) or five days after an iboteaic injection following prior irradiation (C). Note that in A the abrupt change in the density of the my&in staining at the edge of the lesion (arrow) but in B the myelin staining shows no such change. From the same animals nearby sections were stained with OX42 to reveal the inflammatory cells. In B the field was taken near the edge of the lesion (* in A), in D the field was taken from close to the cannula track (* in CT).There.is a big difference in the density of OX42+ cells with few cells in D. Note that the majority of cells in D do not look like resident microglia. Scale bars: A and C= IOO~m, B and D = 20pm.

Fig. 6. Photomi~o~aphs

compared with those responding to neurotoxininduced neuronal death suggests that the signals generated in the two circumstances are not the same. Since the majority of cells at the lesion site express the CR3 receptor (0X42), Class II (0X6) and CD4 antigens (W3/25) and the LCA (OX1 + 0X30) this provides strong support for the idea that these cells are of haemopoietic origin and that they are of the macrophage lineage. We have previously shown that

following a smaI1 cortical lesion cells with the morphology of microglia within the brain parenchyma can express antigens not found in the normal brain29 and the evidence suggests that the expression of CD4 on rat microglia is in part regulated by exposure to plasma proteins. Other studies have provided further evidence for the regulation of mieroglia phenotype in response to injury. 1,9 We have provided additional evidence that the majority of cells found at the lesion

site are recruited from the blood, rather than division of resident cells, since we have shown that they are sensitive to whole body irradiation at a dose known to deplete the bone-marrow population. We previously showed that following ibotenic acid injections into the septum, not only were neurons destroyed but there was also damage to the myelin sheaths and axons en passage; and the fact that injection of ibotenic acid into a fibre tract did not produce demyelination suggested that the axonal damage was an unwanted side-effect of the inflammatory response.4 The results presented here show that reducing the inflammatory response prevents the conspicuous loss of myelin in the region of neuronal loss. It should be remembered, however, that this loss of myelin from axons and damage to the axons en passage through the septum may be conspicuous because these fibres are present in small dispersed fascicles. In the striatum, where the fibres are present in large fascicles and melomonocytic cells do not invade the bundles themselves, the myelin sheaths remain largely intact4 The damage to healthy axons en passage by recruited cells is likely to be found predominantly where degenerating neurons and fibres are closely intermingled. It has been shown in a number of studies that the injection of a neurotoxin and the subsequent neuronal loss is accompanied by an increase in the intensity of staining of astrocytes with glial fibrillary acidic protein and an increase in their number.‘“,” Macrophages secrete growth factors for astrocytes’ and injection of interleukin-1, a secretory product of activated macrophages, into the CNS promotes astrogliosis and angiogenesis.’ The recruitment of macrophages following ibotenic acid injections is consistent with these cells playing a role in astrogliosis. Cells with the morphology of normal microglia are greatly reduced in density in the region of the neuronal loss shortly after injection of ibotenic acid. We might consider two possible explanations: either the cells are themselves sensitive to the excitotoxin and have degenerated, or the microglia have retracted their processes, changed their morphology and are now difficult to distinguish from recently recruited cells. At the present time we do not have evidence which would allow us to distinguish between these two possibilities with any degree of confidence. It is clear, however, from the results of the irradiation experiments that the vast majority of the cells found in the region of neuronal loss after an ibotenate lesion are recruited from the blood. It is possible that toxic for excitatory neurotoxins are indeed macrophages and microglia at neurotoxic concentrations, since monocytes are known to have receptors for kainic acid, and other substances known to be excitatory neurotoxins, and will chemotax towards very low concentrations of these substances.‘” In addition to the recruitment of cells from the blood, the injection of ibotenic acid is associated with

an increase in the permeability of the blooddbrain barrier as revealed by the dilTusion of HRP into the brain. In inflammatory sites outside the CNS, it is known that polymorphonuclear cells and macrophages play a role in the increased permeability of blood vessels.‘0*2zTo examine whether recruited cells might be responsible for increasing the permeability we used irradiation to reduce the inilammatory response and found that diffusion of HRP into the brain was much more restricted in irradiated animals than in those with ibotenic injection alone. The increase in vascular permeability associated with the inflammatory response after ibotenic acid injection is in contrast to the rapid increase in permeability found after kainic acid injection. Nitsch and Hubaner2’ found that within a few hours after kainic acid injection there was a breakdown in the blood-brain barrier both at the site of injection and at distal sites within the hippocampus. The means by which kainic acid results in the breakdown in the blood-brain barrier is not known. We might have expected that in the region of increased vascular permeability, microglia and macrophages would endocytose the extravascular HRP. Although it was clear that recruited, rounded cells would readily endocytose HRP, microglia were not obviously labelled. The cells which were conspicuously labelled were protoplasmic astrocytes, as judged by their morphology. The endocytosis of plasma proteins by astrocytes has previously been demonstrated following cold lesions to the CNS,lS and it was suggested that astrocytes may play a part in the resolution of brain edema. However: in our experiments endocytosis of HRP by astrocytes is a transient phenomenon, only being observed in the first 24 h, and appears to depend upon the injection of ibotenic acid. They were not seen in regions of mechanical damage where the HRP had simply diffused into the parenchyma, as is found after PBS injections. The excitotoxic induction of endocytosis of HRP by astrocytes could be mediated by one of several routes. It may be that non-specific endocytosis is initiated in astrocytes by the direct action of ibotenic acid on an astrocyte receptor. It is known that astrocytes in primary culture are depolarized by excitatory amino acidsI but it should be noted that type 2 cerebellar astrocytes lack NMDA receptors.” On the other hand, metabotropic glutamate receptors, which activate G proteins and stimulate inositol phospholipid metabolism, are known to be present on astrocytes*’ and they may be sensitive to ibotenate. We are not aware that this has been directly tested. The possibility that HRP endocytosis comes about indirectly as a consequence of the neuronal response to ibotenic acid toxicity is plausible since there is accumulating evidence that activity in neurons may influence glial membrane channels.” The astrocytic response to excitatory neurotoxins requires further investigation.

Ibotenic acid-induced inflammation CONCLUSION

have

implications

for

the

disorders in which excitatory neurotoxins are thought to play a part.3s33The breakdown of the blood-brain barrier may be important for other reasons since this provides not only a ready route of access for inflammatory cells but also a possible means of ingress for infectious agents. neurodegenerative

These studies show that the injection of an excitatory neurotoxin into the CNS results in a complex spectrum of events beyond the well documented destruction of neurons. The inflammatory response provoked by this lesion may not only result in damage to axons en passage but may also produce an increase in vascular permeability. The non-specific damage by an i~ammato~ response in the CNg may

131

understanding

of

Acknowledgements--This

work was supported by the MRC and Wellcame Trust. V.H.P. is a Wellcome Senior Research Fellow. We thank Dr A. F. Wi&uns for antibodies and Dr S. Gordon for discussion and comments on the manuscript.

REFERENCES complex antigen expression on rat microglia 1. Akiyama H., Itagaki S. and McGeer P. L. (1988) Major hist~ompatibility following epidural kainic acid lesions. J. Neurosci. Res. 20, 147-157. 2. Cajal Ramon y S. (1909) Histologie du Systeme Neweux de I’Homme et des Vertebres, pp. 230-252. Paris. 3. Choi D. W. (1988) Glutamate neurotoxicity and diseases of the nervous system. Neuron 1, 623634. 4. Coffey P. J., Perry V. H., Allen Y., Sinden J. and Rawlins J. N. P. (1988) Ibotenic acid induced d~yelination in the central nervous system: a consequence of a local inflammatory response. Neurosci. L&t. 84, 178-184. 5. Coyle J. T. and Schwartz R. (1976) Lesion of striatal neurones with kainic acid provides a model for Huntingdon’s chorea. Nalure 263, 244246. 6. Dallman M. J., Thomas M. L. and Green J. R. (1984) MRC OX-19 a monoclonal antibody that labels rat T lymphocytes and augments in vitro proliferative responses. Eur. J. Zmmunol. 14, 260-267. 7. Giulian D. (1988) Interleukin-1 injected into the mammalian brain stimulates astrogliosis and neovascularization. .I. Neurosci. 8, 2485-2490. 8. Giulian D., Allen R. L., Baker T. J. and Tomozawa (1986) Brain peptides and ghal growth. I. Glia promoting factors as regulators of gloigenesis in the developing and injured central nervous system. J. Cell Biol. 102, 803-811. 9. Graeber M. B., Streit W. J. and Kreutzberg G. W. (1988) Axotomy of the rat facial nerve leads to increased CR3 complement receptor expression by activated microglial cells. J. Neurosci. Res. 21, 18-24. 10. Harlan J. M. (1985) Leukocyte-endothelial interactions. Blood 65, 513-525. 11. Hsu S. M., Raine L. and Fanger H. (1981) The use of avidin-biotin-peroxidase (ABC) complex in immunoperoxidase techniques: a comparison between ABC and unlabelled antibody (PAP) procedures. J. Histochem. Cytochem. 29, 577-580.

12. Isacson O., Fischer W., Wictorin K., Dawbam D. and Bjorklund A. (1987) Astroglial response in the excitotoxicaliy lesioned neostriatum and its projection areas in the rat. Neuroscience 20, 1043-1056. 13. Jefferies W. A., Green J, R. and Williams A. F. (1985) Authentic T helper CD4 (W3/25) on rat peritoneal macrophages. J. exp. h4ed. 162, 117-127. 14. Kettenmann H., Backus K. and Schachner M. (1984) Aspartate, glutamate, and y-aminobutyric acid depolarize cultured astrocytes. Neurosci. Lett. 52, 25-29. 15. Klatzo I., ChuiE., Fujiwara K. and Spatz M. (1980) Resolution of vasogenic brain edema. In Advances in Neurology, Vol. 28. Brain Edema (eds Cervos-Navarro J. and Ferszt R.). DD. 359-373. Raven Press. New York. 16. Kohler C. and Schwartz R. (1983) Comparison of ibotenate andkainate toxicity in the rat brain: a histological study. Neuroscience 8, 819-835.

17. Liesi P., Kaakkola S., Dahl D. and Vaheri A. (1984) Laminin is induced in astrocytes of adult brain by injury. Eur. molec. Biol. Org. J. 3, 683-686.

18. Malone J. D., Richards M. and Kahn A. J. (1986) Human peripheral monocytes express putative receptors for neuroexcitatory amino acids. Proc. natn. Acad. Sci. U.S.A. 83, 3307-3310. 19, Marrero H., Astion M. L., Coles J. A. and Orkand R. K. (1989) Facilitation of voltage-gated ion channels in frog neuroglia by nerve impulses. Nature 339, 378380. fixative, a new fixative for immunoelectron 20 Mclean I. W. and Nakane P. K. (1974) Periodate-lysine-paraformaldehyde microscopy. J. Histochem. Cytochem. 22, 1077-1083. 21 McMaster W. R. and Williams A. F. (1979) Identi~cation of Ia glycoproteins in rat thymus and purification from rat spleen. Eur. J. fmmunol. 9, 426433. 77 Movat H. Z. (lY85) The lnfiammatory Reaction. -. Elsevier, Amsterdam. .._. 23. Murabe Y. and Sano Y. (1982) Morphological studies on neuroglia VI. Postnatal development of micro&al cells. Cell Tiss. Res. 225, 469485. 24. Murabe Y., Ibata Y. and Sano Y. (1981) Morphological studies on neuroglia III. Macrophage “microgliocytosis” in kainic-acid induced lesions. Cell Tiss. Res. 218, 75-86.

response and

25. Nitsch C. and Hubaner (1986) Distant blood-brain barrier opening in subfields of the hippocampus after intrastriatal injections of kainic acid but not ibotenie acid. Neurosci. I.&t. 64, 53-58. 26. Olney J. W. (1984) Excitotoxins: an overview. In Excitotoxins (eds Fuxe K., Roberts P. and Schwartz R.), Vol. 39, pp. 82-96. Wenner-Gren International Symposium Series. Plenum Press, New York. 27. Pearce B., Albrecht J., Morrow C. and Murphy S. (1986) Astrocyte glutamate receptor activation promotes inositol phopholipid turnover and calcium flux. Neurosci. Lett. 72, 335-340. 28. Perry V. H., Hume D. A. and Gordon S. (1985) Immunohistochemical localization of macrophages and microglia in the adult and developing mouse brain. Neuroscience 15, 313-326. 29. Perry V. H. and Gordon S. (1987) Modulation of the CD4 antigen on macrophages and microglia in rat brain. J. exp. Med. 146, 1138-1143.

132

P. J.

COFFEY

ef al

30. Perry V. H. and Linden R. (1982) Evidence for dendritic competition in the developing retina. 8ature 297, 6X3-6X5. 31. Robinson A. P., White T. M. and Mason D. W. (1986) Macrophage heterogeneity in the rat as delineattrd by monoclonal antibodies MRC OX-41 and MRC 0X-42, the latter recognizing complement receptor type 3. Inrmunolog~ 57, 239-247. 32. Schwartz R., Whetsell W. 0. and Mangano R. M. (1983) Quinolinic acid: an endogenous metabolite that produces axon-sparing lesions in rat brain. Science 219, 316-3 18. 33. Schwarcz R. and Meldrum 3. (1985) Excitatory amino acid antagonists provide a therapeutic approach to neurological disorders. Lancet 140-143. 34. Sedgewick J., Brostoff S. and Mason D. (1987) Experiments ahergic en~phalomyeIitis in the absence of a classical delayed-type hypersensitivity reaction. J. exp. Med. 165, 10581075. 35. Usowicz M. M., Gallo V. and Cull-Candy S. G. (1989) Multiple conductance channels in type-2 cerebellar astrocytes activated by excitatory amino acids. Nature 339, 380-383. 36. Woollett G. R., Barclay A. N., Puklavec M. and Williams A. F. (1985) Molecular and antigenic heterogeneity of the rat leukocyte common antigen from thymocytes and T and B lymph~ytes. Eur. f. I~~unol. 15, 168-173. (Accrpfed

27 Nooember

1989)