GLIA 6 : 3 9 4 7 (1992)
Amoeboid and Ramified Microglia: Their Interrelationship and Response to Brain Injury SENG-KEE LEONG AND ENG-ANG LING Department of Anatomy, Faculty of Medicine, National University of Singapore, Smgapore 051 1
KEY WORDS
Microglia, Amoeboid, Ramified, Antigenicity, Brain injury
ABSTRACT Rio-Hortega’s hypothesis that transiently appearing amoeboid microglia might become ramified microglia in the adult and that the latter could differentiate into brain macrophages in the event of brain damage could not be proved because of inherent limitations in existing techniques. The present investigation used a novel method of labelling the rat supraventricular amoeboid microglia with a n enduring fluorescent marker, rhodamine B isothiocyanate, introduced intraperitoneally. Observation of their subsequent development showed that they became transformed into the ramified microglia. Both the amoeboid and ramified microglia were OX-42 positive, indicating their macrophage/monocyte lineage. Other microglia in the cerebral neocortex, which were also OX-42 positive, were not derived from any of the rhodamine-labelled cells. Rhodamine-labelled microglia did not migrate toward the site of a superficial cerebral injury. Following a deep lesion reaching the corpus callosum, greatly increased numbers of labelled amoeboid microglia were frequently observed at or near the lesion site. Large rhodamine-labelled cells, which were OX-42 positive, appeared at all lesioned sites, and such were most likely blood derived monocytes. The antigenicity of the ramified microglia became elevated when rhodamine B isothiocyanate was present intracellularly and even more so with the presence of a nearby intracerebral stab wound. R1 1992 Wiley-Liss, Inc.
INTRODUCTION Though microglia have been implicated in many important functions such as wound healing (Penfield, 1932), regulation of astrocytic differentiation, immune response (for review, see Streit et al., 1988)and amyloid deposition in Alzheimer’s disease (Haga et al., 1989; Mattiace et al., 1990; Perlmutter et al., 19901, their origins remain controversial (for review, see Jordan and Thomas, 1988).About 8 decades ago, Rio-Hortega (Penfield, 1932) hypothesized that transiently appearing amoeboid microglia might become ramified microglia in the adult and that the latter could differentiate into brain macrophages in the event of brain damage. Such hypothesis, however, could not be proved because of inherent limitations in existing techniques, diverse and exciting they might be. For example, the cell markers used by some investigators (Kaur et al., 19861, do 0 1992 Wiley-Liss, Inc.
not remain long enough in the labelled cells to determine their fate. Also, with the use of carbon particles to demonstrate the role of blood monocytes as precursors of the supraventricular amoeboid microglial cells ( S A M C ) situated in the corpus callosum above the lateral ventricles of neonatal rats, only a small number of the SAMC were carbon-labelled (Ling et al., 1980). The large majority of the SAMC were not labelled. As our previous work (Leong and Ling, 1990) showed that certain fluorescent dyes injected into the peritoneal cavity gained access into the bloodstream and labelled the choroid plexus, we speculated that some of these dyes might be secreted by the choroid plexus into the cere-
Received June 26,1991; accepted November 11,1991 Address reprint requests to Seng-Kee Leong, Dept. of Anatomy, Faculty of Medicine, National University of Singapore, Kent Ridge, Singapore 051 1
40
LEONG AND LING
4. Fluorescent supraventricular microglial cells in rats injected with rhodamine B isothiocyanate at birth and perfused 1day (Fig. 3 ) and 40
Figs. 1 4 . Fig. 1. Labelled supraventricular amoeboid microglial cells in a rat injected with rhodamine B isothiocyanate and perfused 5 days after injection. Note that in addition to the microglial cells, the choroid plexus (CP)and ependymal lining (arrowed)were also fluorescent. Fig. 2. Rhodamine-labelled cells streaming from the rostra1 extension of the cavum septum pellucidum (arrowed) along the subventricular region (indicated by arrowheads) to the supraventricular region of the corpus callosum. V: Body of lateral ventricle. Figs. 3 and
days (Fig. 4)after injection. The transformation of the microglial cells from their rounded form without processes to their flattened forms with many processes emanating from their cell bodies is obvious. Note in Figure 4 the presence of rhodamine-labelled ependymal cells, some of which are arrowed. V: body of lateral ventricle. Bars in Figures 1 4 : 0.1 mm.
bra1 ventricular system. From there, the dye could then cross the ependymal lining into the brain parenchyma where it might be taken up by the high1.y phagocytic SAMC. The subsequent development of the labelled SAMC may then be studied. Such a method, if proved successful, would avoid damaging the brain in the process of labelling cells, a complication that would certainly lead to misleading results. In our preliminary study we searched for fluorescent dyes that: ( 1 ) would be able to cross the ependymal interface and label the SMAC, (2) would not be digested by the lysosomal enzymes of the SAMC and therefore be able to retain its fluorescence for some time, and (3) would not be toxic to the SAMC. After a few trials with different fluorescent dyes, it was discovered that rhodamine latex microspheres introduced intraperitoneally did not label the SAMC. Fast blue, true blue, and fluorogold labelled the microglial cells, but their fluorescence soon faded after 2-3 days, indicating their degradation by the microglial cells. Rhodamine B isothiocyanate (RhIc), (Sigma), when administered intraperitoneally in 1%solution in normal saline, reached the choroid plexus, which it labelled (Fig. l), and was presumably secreted by the plexus into the ventricular system. From the ventricles, the dye gained access into
the brain parenchyma by crossing the ependymal lining, which it stained (Fig. 1).In the brain parenchyma, the dye was phagocytosed by the SAMC, causing them to emit a red fluorescence that showed hardly any deterioration with time. I n addition to the above route, it was possible that some dye might enter the brain parenchyma through the blood vessel walls, all of which emitted fluorescence. Electron microscopic examination of the labelled SAMC, using 6%paraformaldehyde and 1%glutaraldehyde in 0.1 M phosphate buffer as a fixative, revealed that the fluorescent cells possessed ultrastructural features typical of normal healthy looking SAMC as reported in a previous study in this laboratory (Leong e t al., 1983). Having successfully labelled the SAMC and other cells, described below, we investigated whether the SAMC (1)could be derived from any obvious source, and whether they or other labelled cells would in time migrate to the cerebral neocortex to become the resident microglia therein, and ( 2 ) would transform into flattened ramified microglia (RMC1 with advancing age and, if so, whether both of them would, like monocytes, express type 3 complement receptor (CR3) a s detected by OX-42. Finally, we aimed to study the effect of a n intracerebral stab wound inflicted at various times af-
AMOEBOID AND RAMIFIED MICROGLIA AND BRAIN INJURY
41
with a mixture of periodate-lysine-paraformaldehyde (McLean and Nakane, 1974),having a concentration of 2% or 4% paraformaldehyde. The brains obtained were kept in the sucrosed buffer overnight and frozen sections were cut at 40 km and mounted on gelatinized slides. The sections were then incubated in the monoclonal antibody OX-42 (Sera Lab, MAS 370b) diluted 1:100 with phosphate-buffered saline. After a n incubation period of 15-18 h at room temperature, the antibody was detected by using the Vectastain ABC kit against mouse IgG with 3,3’,-diaminobenzidine (DAB) a s a peroxidase substrate. The reaction product was MATERIALS AND METHODS intensified with nickel ammonium sulphate and the To label the amoeboid microglial cells, 40-50 pl of 1% sections were then lightly counterstained with methyl RhIc in normal saline was administered intraperito- green. In the above procedure, some sections were neally in 33 one-day-old rats. They were perfused from treated with triton. and others without. 10 min to 40 days after injection. To test whether the labelled microglial cells would migrate to the site of a n intracerebral stab wound, 30 rats first received 40-50 RESULTS p1 of 1%RhIc at 1-2 days after birth. From 1-26 days In addition to the choroid plexus and the associated after RhIc injection, a 21- or 281h-gauge needle was plunged unilaterally into the medial half of the middle epiplexus cells and the SAMC, the circumventricular part of the cerebral hemisphere through the cranium. organs, pia cells, some cells in the ependymal and the The animals were perfused 4 h to 40 days after the stab subventricular zone of the lateral ventricles and a conglomeration of small round cells in and around the cawound. All operations and perfusion were performed under vum septum pellucidum were also labelled (Figs. 2-4). anaesthesia, achieved by ether inhalation for neonatal Furthermore, there were labelled cells in other regions rats, and intraperitoneal administration of 0.4-0.6 rnl of the brain but such would be reported separately. As of 3.5% chloral hydrate and 0.8-1.0 ml of 7% chloral early a s 10 min after the administration of RhIc, the hydrate, respectively, for subadult and adult rats. For labelled cells emitted a low level of fluorescence, which perfusion, the animals were briefly rinsed with 0.1 M increased in intensity with time. In a rat perfused 1 day phosphate buffer (pH 7.2-7.4) and then fixed with 6% after injection, the caval cells appeared to stream diparaformaldehyde in phosphate buffer. The volumes of rectly or along the vicinity of the subventricular region fixative varied between 1 0 0 4 0 0 ml, depending on the toward the supraventricular callosal region (Fig. 2). At size of the animal. After perfusion, the brains were day 1 after RhIc injection, the SAMC were round (Fig. dissected out, kept overnight in the sucrose-buffered 3) and measured between 5 and 10 pm in diameter. fixative and sectioned a t 40 pm thickness in a cryostat. Some of these exhibited a filopodia-like process. With One out of every two (neonatal brains), three (subadult longer survival times, the cells became oval and then brains), or five (adult brains) sections were mounted on flattened or triangular, showing a variable number of gelatinized slides, air dried, and coverslipped, using the branching processes emanating from their cell bodies, nonfluoreseent Entellan as a mountant. The sections thus assuming the morphology of the typical ramified were examined and photographed in a Leitz Aristoplan or “resting” microglia (Fig. 4) observed in the adult microscope equipped with a mercury lamp for fluores- brain. In sections counterstained with cresyl fast violet, cence microscopy, using a wide band ultraviolet excita- the SAMC displayed a round nucleus and the RMC a tion filter (excitation range, 515-560 nM). After the rod-shaped nucleus. Pyknotic nuclei a t locations correabove procedure, the coverslips were removed and the sponding to those of the rhodamine-labelled cells were sections counterstained with cresyl fast violet and ex- frequently seen in older animals. There was a n obvious amined under conventional brightfield illumination. increase in the number of labelled SAMC during the For immunocytochemistry, 3 groups of animals were early postnatal period, but a s the animals increased in used. Group I consisted of 16 normal rats, aged 2 days age, their numbers were drastically reduced. to 3 months after birth. Group I1 consisted of 15 rats Other than the caval cells, which appeared to migrate injected with RhIc a t birth and perfused at 2 days, 11 to adjacent regions a s mentioned, all other labelled cells days, 27 days, and 40 days after injection. The third remained in situ with age. No labelled cells were found group of 12 rats was injected with RhIc a t birth and in the neocortex a t any of the survival times studied. received a n intracerebral stab wound 2-26 days post With advancing age, the caval cells were gradually deinjection and finally perfused 4 h to 33 days after in- pleted as the cavum septum pellucidum became obliterjury. Anaesthesia before operation and perfusion and ated. the operation procedures followed those described in In early postnatal normal rats, the SAMC demonthe preceding section. For perfusion, the animal was strated strong immunocytochemical reaction with first rinsed with Ringer’s solution, followed by fixation OX-42 (Fig. 5) a s reported previously (Ling et al., 1990). ter RhIc injection, looking specifically for any possible migration of the SAMC or RMC to the site of injury and possible activation of the antigenicity of the RMC, which could not normally be demonstrated in older rats (Jordan and Thomas, 1988; Ling et al., 1990; Sminia et al., 1987). In the course of investigation, some exciting preliminary results also prompted us to reinvestigate the immunoreactivity of the RMC in some neonatal, late postnatal, and adult normal rats.
42
LEONG AND LING
Figs. 5-7. Photomicrographs of immunolabelled supraventricular amoeboid microglial cells in a normal neonatal rat (Fig. 5) and of the same type of cells rhodamine-labelled (Fig. 6) in a rat that had received rhodamine B isothiocyanate injection a t birth and was perfused 2 days after injection. Figure 7 shows the same cells in Figure 6 immunolabelled with OX-42. Some double-labelled cells are indicated by arrows. Bars: 0.05 mm.
In day-old rats injected with RhIc and perfused with periodate-lysine-paraformaldehyde(PLP) as a fixative, the fluorescence of rhodamine-labelled SAMC could be demonstrated, but it was less intense than that of tissue perfused with 6% paraformaldehyde in phosphate buffer. Also with PLP fixative, the background nonspecific staining was stronger, Immunocytochemical staining revealed that the very cells labelled with rhodamine
Figs. 8-10. Photomicrographs of supraventricular ramified microglial cells (Fig. 8) labelled with rhodamine B isothiocyanate and of the same cells immunolabelled with OX-42 at the same (Fig. 9) and a higher (Fig. 10)magnification. Some of the double-labelled cells are indicated by arrows. The rat received rhodamine B isothiocyanate at birth, stabbed in the cerebral hemisphere 24 days after injection and perfused 30 days after the stab wound. The labelled cells were situated contralateral to the intracerebral lesion. Bars: 0.1 mm.
were OX-42 positive, as evidenced by comparing photographs of rhodamine-labelled SAMC (Fig. 6) taken before the staining procedure with those of OX-42 immunolabelled cells (Fig. 7). Rhodamine-labelled RMC (Fig. 8) were also immunopositive to OX-42 (Figs. 9, 10). There was no noticeable difference in the intensity of immunoreaction in neonatal rats perfused with PLP fixative containing either 2% or 4% paraformaldehyde. With older rats, however, it was clear that a lower percentage of paraformaldehyde in the PLP fixative gave a much stronger immunoreaction. Ramified microglia
AMOEBOID AND RAMIFIED MICROGLIA AND BRAIN INJURY
43
Figs. 11-14. Figs. 11 and 12. Immunolabelled microglial cells in the supraventricular region of the corpus callosum in a 12-day-old rat perfused with PLP fixative containing 4% paraformaldehyde. The section shown in Figure 11 was not treated with triton, but a section adjacent to it, shown in Figure 12, was treated with triton during the immunocytochemical staining. Figs. 13 and 14. Photomicrographs showing the supraventricular microglia in a normal adult rat (Fig. 13), and a rat injected with rhodamine B isothiocyanate at birth, and
subsequently received an intracerebral stab wound 26 days later and perfused 33 days after injury (Fig. 14). The two rats were perfused with PLP fixative containing 2 6 paraformaldehyde, and no triton was applied during the immunocytochemical procedure. Note that in the absence of triton, the microglia in the normal rat were poorly stained but quite distinctly stained in the experimental rat. Figures 11-14, same magnifications. Bar: 0.1 mm.
and microglia in the cerebral cortex and other parts of the brain that were not stained with PLP solution containing 4% paraformaldehyde gave distinct immunoreaction with one containing 2% paraformaldehyde. It was also noteworthy that, irrespective of the concentration of paraformaldehyde, sections treated with triton showed a more intense reaction than those without. Thus in a 12-day-oldnormal rat perfused with PLP fixative containing 4% paraformaldehyde, the supraventricular microglial cells demonstrated only weak immunoreactivity (Fig. 11) without triton treatment, but strong reaction (Fig. 12) when the sections were treated with triton. Interestingly, in rats injected with RhIc at birth and processed in like manners 11 days later, the supraventricular microglia gave intense reaction even in the absence of triton, and such was also the case in rats perfused as late as 30 days after injection. Though PLP fixative containing 2% paraformaldehyde
gave a stronger immunoreaction, it dampened the fluorescence of rhodamine-labelled cells, as mentioned earlier. This was especially noticeable in rats receiving RhIc injection at birth but perfused a few weeks later. However, the fluorescence of such cells in rats receiving intracerebral lesions was not much dimmed (Fig. 8). In such rats, the immunoreaction of the supraventricular cells was also stronger. The contrast was best demonstrated in older rats (Figs. 13,141when the immunoreactivity of the RMC in both normal (Fig. 13) and injected rats was weak in sections not treated with triton, but strong in those receiving RhIc injection followed by an intracerebral lesion (Fig. 14). The intracerebral stab wounds were either superficial or deep, situated within the medial half of the middle part of the cerebral hemisphere. The superficial lesions reached the vicinity of but not touching the supra-ventricular SAMC, whereas the deeper lesions in-
44
LEONG AND LING
Figs. 15 and 16. Rhodamine-labelled cells (Fig. 15) at the site of a superficial intracerebral stab wound and the same cells (Fig. 16) immunolabelled with OX-42 in a rat receiving rhoddmine B isothiocyanate injection at birth, stabbed 2 days following injection and perfused 2 days afterward. Some doubly labelled cells are indicated by arrows. Bars: 0.1 mm
1982; Murabe and Sano, 19821, immunocytochemistry (Ling e t al., 1990; Perry and Gordon, 1988) to tissue culture (Jordan and Thomas, 1988; Ling e t al., 1983). The subject, however, remained controversial because of inherent technical limitations. The conclusions were based mostly on circumstantial or inadequate evidences (for review, see Jordan and Thomas, 1988). In this study, using a novel method of labelling the SAMC with a n enduring cell marker without damaging the brain, we proved that (1)the SAMC and RMC were the same cells, the latter being the transformation of the former as the animal advanced in age, and (2) both of them expressed CR3 membrane receptors a s demonstrated by OX-42. As the animals matured, it became increasingly difficult to demonstrate the immunoreactivity without the use of triton and a lower percentage of paraformaldehyde. Triton being a detergent, probably had to break down certain membrane constituents to unmask the receptors for reaction with OX-42. In the case of paraformaldehyde, a lower concentration would probably exert less inhibitory effect on the receptors. It is significant to note that in older animals, CR3 receptors in the RMC, which could not be demonstrated with PLP containing 4% paraformaldehyde and without triDISCUSSION ton, could be revealed when RhIc was present intracelBy virtue of their morphology, location, and immuno- lularly. This indicated that the dye acted a s a stimulant reactivity, the labelled cells in the supraventricular cal- for the RMC in their expression of immunoreactivity. losal region could easily be identified a s the SAMC de- Such expression was further elevated in the presence of scribed previously (Leong et al., 1983; Ling et al., 1990). a n intracerebral lesion. Previous studies (reviewed by Their relation with RMC has been investigated by di- Streit et al., 1988) have also shown a n activation of verse techniques ranging from morphological observa- some intracellular and surface molecules during neution (Cammermeyer, 1970; Ling et al., 1990; Mori and ronal degeneration and regeneration. Though the origin of the SAMC before birth was not Leblond, 1969), study of phagocytic capability (Lent et al., 19851, histochemistry (Ling, 1977; Ling et al., studied, it appeared obvious that a t birth the SAMC
volved the corpus callosum. One to 4 days after a superficial lesion, a conglomeration of rather large, round rhodamine-labelled cells (Fig. 15) measuring about 15 pm in diameter and immunopositive to OX-42 (Fig. 16) appeared around the site of lesion. Such cells were not present when the animals were perfused within 8 hours after the lesion. There was no migration of the labelled microglia toward the lesioned site at any of the survival periods studied. In a deep lesion, rhodamine-labelled large round cells appeared in the immediate vicinity of the lesion. Some of them intermingled with smaller SAMC, which had earlier been labelled with rhodamine (Figs. 17-20). Compared to the normal (Fig. 211, many more SAMC were commonly seen at or near (Fig. 22) the site of lesion, especially when the lesion was made with the larger size needle. Even with 6% paraformaldehyde in phosphate buffer as a fixative, the labelled large round cells showed less intense fluorescence than the labelled SAVC, especially when the lesion was made 2-5 weeks after the introduction ofl2hIc.
AMOEBOID AND RAMIFIED MICROGLIA AND BRAIN INJURY
45
Figs. 17-20. Figure 17 shows small, round supraventricular microglial cells intermingled with large, round cells, both of which were rhodamine-labelled. The section was taken from a rat injected with rhodamine B isothiocyanate intraperitoneally within 1day after birth, stabbed in the cerebrum 1 day after injection, and subsequently perfused 4 days later. Both cell types were immunolabelled (Fig. 18)with OX-42. Figures 19 and 20 show the magnified view of the labelled cells situated within the boxes indicated in Figures 17 and 18. Arrows
indicate some doubly labelled cells. Note that Figure 18was taken at a slightly higher magnification than Figure 17, but both Figures 19 and 20 were taken at the same magnification. The immunolabelled cells in Figure 20 appeared larger than the rhodamine-labelled ones in F i y r e 19, indicating that some expansion of tissue had occurred as a result of the immunocytochemical procedure. Bars in Figures 17 and 18: 0.1 mm; in Figures 19 and 20: 0.05 mm.
were already present in great numbers and were faintly labelled within 10 minutes of introduction of RhIc. The faintly labelled cells could not be rhodamine-labelled circulating monocytes t h a t left the blood vessels for i t would be difficult to imagine so many of such cells exiting the bloodstream within such a short time. The initial increase in the number of SAMC during the first few postnatal days might be attributed to (1)their mitosis, which had been observed by Imamoto and Leblond (1978), (2) migration of monocytes from the bloodstream (Ling e t al., 1980), and (3) migration of cells from the cavum septum pellucidum, as indicated in the present study. As the animals matured in age, the number of rhodamine-labelled cells was drastically reduced. Such a reduction could most probably be attributed to cell death, observed by Imamoto and Leblond (1978). It was noted in this study that the labelled caval cells were gradually depleted with the obliteration of the cavum. Also, previous study (Tseng et al., 1983) failed to observe any degenerating cells in this region. From these two observations, we surmise that the amoeboid microglial cells contained in the cavum might be temporarily kept there for distribution to adjacent areas a t the appropriate time. Our study could not substantiate Hortega’s hypothesis that microglia in general were derived from the lep-
tomeninges of the developing brain. Cells in the pia were labelled, but they did not migrate to the cerebral cortex. Also, microglia in general could not be derived from the monocytes in the bloodstream postnatally; if so, such cells would have engulfed the RhIc present intravenously and would appear fluorescent in the cerebral cortex after leaving the bloodstream. The microglia, however, were immunopositive to OX-42, confirming a previous report on the mouse (Perry and Gordon, 1988). This indicates their macrophage/monocyte lineage, since OX-42 recognizes cell surface proteins synthesized by macrophages (Robinson et al., 1986). It is, therefore, possible that they could be derived from the bloodstream in utero and subsequently undergo proliferation and differentiation to become the resident microglia. Unlike the labelled microglia, these microglia were not phagocytic and therefore did not engulf RhIc that had diffused into the cerebral cortex. It would therefore not be unreasonable to conclude that there might be at least two populations of microglial cells in the brain. The possible existence of different populations of microglia in the brain had also been suggested by previous work (Miles and Chou, 1988). The labelled SAMC and RMC were OX-42 positive, indicating that they were of the same lineage as the microglia in the cerebral cortex. The SAMC probably
46
LEONG AND LING
Figs. 21 and 22. Figure 21 shows rhodamine-labelled supraventricular amoeboid microglial cells in a rat injected with rhodamine B isothiocyanate within 1 day after birth, and perfused 4 days later. Figure 22 shows rhodamine-labelled supraventricularmicroglial cells in another rat treated similarly, but in addition to rhodamine B isothiocyanate inject.ion, the rat also received an intracerebral stab wound inflicted 2 days after injection. The section was taken a t 900 pm caudal to the posterior end of the cerebral lesion. There is a n obvious increase in the number of labelled cells in this rat. Bars: 0.1 mm.
functioned a s macrophages to engulf degenerating axons present in the developing corpus callosum (Innocenti, 1981) or naturally occurring degenerating cells (Ling et a]., 1990; Maruyama and D’Agostino, 1967; Matsumoto and Ikuta, 1985). It is important to note that in the presence of a nearby brain lesion, the SAMC were not detracted from their primary task in the corpus callosum for they did not migrate to the site of that lesion but proliferated in situ when the corpus callosum was damaged. Having performed the phagocytic function, many of the glial cells degenerated as reported previously (Imamotoand Leblond, 1978) and a s demonstrated by the presence of pyknotic nuclei of the fluorescent cells in this study. The remaining cells became flattened and ramified, probably as a result of adaptation to the surrounding architecture as the callosal fibres began to myelinate and occupy the large intercellular space existent in the developing callosum. Rhodamine-labelled cells that appeared a t the site of injury were most likely monocytes that had left the bloodstream, as they exhibited immunopositive reaction to OX-42 and as they showed a weaker fluorescence when the lesion was made 2-5 weeks after the
introduction of RhIc. By this time, much of the RhIc present intravascularly would have been diluted with a n increase in the size of the animal and could not therefore label newly formed monocytes a s effectively. Our conclusion that the large rhodamine- and immunolabelled cells at the site of injury were most likely monocytes that had left the bloodstream is supported by previous studies (Kitamura et al., 1972; Konigsmark and Sidman, 1963; Torvik, 1975) that reported the presence of haematogenous cells at the site of cerebral injury. Our observation suggests that when a stab wound was made in a n area containing ramified microglia, labelled monocytes appeared a t the lesioned site. It appeared that the resident microglia were not able to handle the extra load imposed by the stab wound and help had to be called from the circulating monocytes. The latter were probably needed to ingest red blood cells (Penfield, 19321, stimulate angiogenesis (Hunt et a]., 1984), or release neuronotrophic factors (Nieto-Sampedro e t al., 1983) or interleukin-I, which can activate other macrophages and glial cells to produce a variety of factors (Riopelle, 1988). Finally, it might be said that as known microglial markers do not definitely distinguish between blood-derived monocytes and resident microglia, our present method of labelling blood-derived monocytes might offer a simple solution to the problem.
ACKNOWLEDGMENTS We thank Ms. C . Ang for her untiring secretarial help, the technicians of the Department of Anatomy, and Mr. Poon Keah How, Mr. Lee Eng Kiang, Ms. Joyce Koh, and other year I medical students, National University of Singapore, for their technical assistance.
REFERENCES Cammermeyer, J. (1970) The life history of the microglial cell: A light microscopic study. In: Neurosciences Research, Val. 3. S. Ephrenpreis and O.C. Solnitzky, eds. Academic Press, New York, pp. 43129. Haga, S., Akai, R., and Ishii, T. (1989) Demonstration of microglial cells in and around senile (neuritic)plaques in the Alzheimer brain. An immuno-histochemical study using a nova1 monoclonal antibody. Acta Neuropathol., 77569-575. Hunt, T.K., Knighton, D.R., Thakral, K.K., Goodson 111, W.H., and Andrews, W.S. (1984) Studies on inflammation and wound healing: Angiogenesis and collagen synthesis stimulated in viva by resident and activated wound macrophages. Surgery, 96:48-54. Imamoto, K. and Leblond, C.P. (1978)Radioautographic investigation of gliogenesis in the corpus callosum of young rats. 11. Origin of microglial cells. J . Comp. Neurol., 180:139-164. Innocenti, G.M. (1981) Growth and reshaping of axons in the establishment of visual callosal connections. Science, 212:824-827. Jordan, F.L. and Thomas, W.E. (1988) Brain macrophages: Questions of origin and interrelationship. Brain IZes. Rev., 13:165-178. Kaur, C., Ling, E.A., and Wong, W.C. (1986) Labelling of amoeboid microglial cells in rats of various ages following an intravenous injection of horseradish peroxidase. Acta Anat., 125:132-137. Kitamura, T., Hittori, H., and Fujita, S. (1972)Autoradiographic studies on histogenesis of brain macrophages in the mouse. J . Neuropath. Exp. Neurol., 31502-518.
AMOEBOID AND RAMIFIED MICROGLIA AND BRAIN INJURY Konigsmark, B.W. and Sidman, R.L. (1963) Origin of brain macrophages in the mouse. J . Neuropath. Exp. Neurol., 22643-676. Lent, R., Linden, R., and Cavalcante, L.A. (1985) Transient populations of presumptive macrophages in the brain of the developing hamster, as indicated by endocytosis of blood borne horseradish peroxidase. Neuroscience, 15:1203-1215. Leong, S-K. and Ling, E-A. (1990) Labelling of neurons with fluorescent dyes administered via intravenous, subcutaneous or intraperitoneal route. J . Neurosci. Meth., 32:15-23. Leong, S-K., Shieh, J-Y., Ling, E-A., and Wong, W-C. (1983) Labelling of amoeboid microglial cells in the supraventricular corpus callosum following the application of horseradish peroxidase in the cerebrum and spinal cord in rats. J . Anat., 136:367-377. Ling, E.A. (1977) Light and electron microscope demonstrations of some lysosomal enzymes in the amoeboid microglia in neonatal rat brain. 2.Anat., 123:637-648. Line, E.A., Kaur, C., and Wonp, W.C. (1982) Light and electron microscopic demonstration of nonyspecific esteraseh amoeboid microglial cells in the corpus callosum in postnatal rats: A cytochemical link to monocytes. J . Anat., 135:385-394. Ling, E.A., Penney, D., and Leblond, C.P. (1980) Use of carbon labelling to demonstrate the role of blood monocytes as precursors of the “amoeboid cells” present in the corpus callosum of postnatal rats. J . Comp. Neurol., 193:631-657. Ling, E.A., Kaur, C., Yick, T.Y., and Wong, W.C. (1990) Immunocytochemical localization of CR3 complement receptors with OX-42 in amoeboid microglia in postnatal rats. Anat. Embryol., 182:481486. Ling, E.A., Tseng, C.Y., Voon, F.C.T., and Wong, W.C. (1983)Isolation and culture of amoeboid microglial cells from the corpus callosum and cavum septum pellucidum in postnatal rats. J . Anat., 137:223233. Maruyama, S. and DAgostino, A.N. (1967) Cell necrosis in the central nervous system of normal rat fetuses. Neurology, 17:550-558. Matsumoto, Y. and Ikuta, F. (1985) Appearance and distribution of fetal brain macrophages in mice: Immunohistochemical study with a monoclonal antibody. Cell Tissue Res., 239:271-278. Mattiace, L.A., Davies, P., Yen, S-H., and Dickson, D.W. (1990) Microglia in cerebellar plaques in Alzheimer’s disease.Acta Neuropathol., 80:493498. McLean, I.W. and Nakane, P.K. (1974) Periodate-lysine-paraformal-
47
dehyde fixative, a new fixative for immunoelectron microscopy. J . Histochem. Cytochem., 22:1077-1087. Miles, J.M. and Chou, S.M. (1988)A new immunoperoxidase marker for microglia in paraffin section. J . Neuropath. Exp. Neurol., 47:579-587. Mori, S. and Leblond, C.P. (1969) Identification of microglia in light and electron microscopy. J . Comp. Neurol., 135:57-80. Murabe, Y. and Sano, Y. (1982) Morphologic studies on neuroglia. VI. Postnatal development of microglial cells. Cell Tissue Res., 225:469485. Nieto-Sampedro, M., Manthorpe, M., Barbin, G., Varon, S., and Cotman, C.W. (1983) Injury induced neuronotrophic activity in adult rat brain: Correlation with survival of delayed implants in the wound cavity. J . Neurosci., 3:2219-2229. Penfield, W. (1932) Neuroglia, normal and pathological. In: Cytology and Cellular Pathology of the Nervous System. W.G. Penfield, ed. Hoeber, New York, pp. 423-479. Perlmutter, L.S., Barron, E., and Chui, H.C. (1990)Morphologic association between microglia and senile plaque amyloid in Alzheimer’s disease. Neurosci. Lett., 119:32-36. Perry, V.H. and Gordon, S. (1988) Macrophages and microglia in the nervous system. TINS, 11:273-277. Riopelle, R.J. (1988) Adrenal medulla autografts in Parkinson’s disease: A proposed mechanism of action. Can. J . Neurol. Sci., 15:366370. Robinson, A.P., White, T.M., and Mason, D.W. (1986) Macrophage heterogeneity in the rat as delineated by two monoclonal antibodies MRC OX-41 and MRC OX-42, the latter recognizing complement receptor type 3. Immunology, 57239-247. Sminia, T., De Groot, C.J.A., Dijkstra, C.D., Koetsier, J.C., and Polman, C.H. (1987) Macrophages in the central nervous system of the rat. Immunobiol., 174:43-50. Streit, W.J., Graeber, M.B., and Kreutzberg, G.W. (19881 Functional plasticity of microglia: A review. Glia, 1:301-307. Torvik, A. (1975) The relationship between microglia and brain macrophages. Experimental investigations. Acta Neuropathol., S6:297300. Tseng, C.Y., Ling, E.A., and Wong, W.C. (1983) Scanning electron microscope of amoeboid microglial cells in the transient cavum septi pellucidi in pre- and postnatal rats. J . Anat., 136:251-263.
Amoeboid and Ramified Microglia: Their Interrelationship and Response to Brain Injury SENG-KEE LEONG AND ENG-ANG LING Department of Anatomy, Faculty of Medicine, National University of Singapore, Smgapore 051 1
KEY WORDS
Microglia, Amoeboid, Ramified, Antigenicity, Brain injury
ABSTRACT Rio-Hortega’s hypothesis that transiently appearing amoeboid microglia might become ramified microglia in the adult and that the latter could differentiate into brain macrophages in the event of brain damage could not be proved because of inherent limitations in existing techniques. The present investigation used a novel method of labelling the rat supraventricular amoeboid microglia with a n enduring fluorescent marker, rhodamine B isothiocyanate, introduced intraperitoneally. Observation of their subsequent development showed that they became transformed into the ramified microglia. Both the amoeboid and ramified microglia were OX-42 positive, indicating their macrophage/monocyte lineage. Other microglia in the cerebral neocortex, which were also OX-42 positive, were not derived from any of the rhodamine-labelled cells. Rhodamine-labelled microglia did not migrate toward the site of a superficial cerebral injury. Following a deep lesion reaching the corpus callosum, greatly increased numbers of labelled amoeboid microglia were frequently observed at or near the lesion site. Large rhodamine-labelled cells, which were OX-42 positive, appeared at all lesioned sites, and such were most likely blood derived monocytes. The antigenicity of the ramified microglia became elevated when rhodamine B isothiocyanate was present intracellularly and even more so with the presence of a nearby intracerebral stab wound. R1 1992 Wiley-Liss, Inc.
INTRODUCTION Though microglia have been implicated in many important functions such as wound healing (Penfield, 1932), regulation of astrocytic differentiation, immune response (for review, see Streit et al., 1988)and amyloid deposition in Alzheimer’s disease (Haga et al., 1989; Mattiace et al., 1990; Perlmutter et al., 19901, their origins remain controversial (for review, see Jordan and Thomas, 1988).About 8 decades ago, Rio-Hortega (Penfield, 1932) hypothesized that transiently appearing amoeboid microglia might become ramified microglia in the adult and that the latter could differentiate into brain macrophages in the event of brain damage. Such hypothesis, however, could not be proved because of inherent limitations in existing techniques, diverse and exciting they might be. For example, the cell markers used by some investigators (Kaur et al., 19861, do 0 1992 Wiley-Liss, Inc.
not remain long enough in the labelled cells to determine their fate. Also, with the use of carbon particles to demonstrate the role of blood monocytes as precursors of the supraventricular amoeboid microglial cells ( S A M C ) situated in the corpus callosum above the lateral ventricles of neonatal rats, only a small number of the SAMC were carbon-labelled (Ling et al., 1980). The large majority of the SAMC were not labelled. As our previous work (Leong and Ling, 1990) showed that certain fluorescent dyes injected into the peritoneal cavity gained access into the bloodstream and labelled the choroid plexus, we speculated that some of these dyes might be secreted by the choroid plexus into the cere-
Received June 26,1991; accepted November 11,1991 Address reprint requests to Seng-Kee Leong, Dept. of Anatomy, Faculty of Medicine, National University of Singapore, Kent Ridge, Singapore 051 1
40
LEONG AND LING
4. Fluorescent supraventricular microglial cells in rats injected with rhodamine B isothiocyanate at birth and perfused 1day (Fig. 3 ) and 40
Figs. 1 4 . Fig. 1. Labelled supraventricular amoeboid microglial cells in a rat injected with rhodamine B isothiocyanate and perfused 5 days after injection. Note that in addition to the microglial cells, the choroid plexus (CP)and ependymal lining (arrowed)were also fluorescent. Fig. 2. Rhodamine-labelled cells streaming from the rostra1 extension of the cavum septum pellucidum (arrowed) along the subventricular region (indicated by arrowheads) to the supraventricular region of the corpus callosum. V: Body of lateral ventricle. Figs. 3 and
days (Fig. 4)after injection. The transformation of the microglial cells from their rounded form without processes to their flattened forms with many processes emanating from their cell bodies is obvious. Note in Figure 4 the presence of rhodamine-labelled ependymal cells, some of which are arrowed. V: body of lateral ventricle. Bars in Figures 1 4 : 0.1 mm.
bra1 ventricular system. From there, the dye could then cross the ependymal lining into the brain parenchyma where it might be taken up by the high1.y phagocytic SAMC. The subsequent development of the labelled SAMC may then be studied. Such a method, if proved successful, would avoid damaging the brain in the process of labelling cells, a complication that would certainly lead to misleading results. In our preliminary study we searched for fluorescent dyes that: ( 1 ) would be able to cross the ependymal interface and label the SMAC, (2) would not be digested by the lysosomal enzymes of the SAMC and therefore be able to retain its fluorescence for some time, and (3) would not be toxic to the SAMC. After a few trials with different fluorescent dyes, it was discovered that rhodamine latex microspheres introduced intraperitoneally did not label the SAMC. Fast blue, true blue, and fluorogold labelled the microglial cells, but their fluorescence soon faded after 2-3 days, indicating their degradation by the microglial cells. Rhodamine B isothiocyanate (RhIc), (Sigma), when administered intraperitoneally in 1%solution in normal saline, reached the choroid plexus, which it labelled (Fig. l), and was presumably secreted by the plexus into the ventricular system. From the ventricles, the dye gained access into
the brain parenchyma by crossing the ependymal lining, which it stained (Fig. 1).In the brain parenchyma, the dye was phagocytosed by the SAMC, causing them to emit a red fluorescence that showed hardly any deterioration with time. I n addition to the above route, it was possible that some dye might enter the brain parenchyma through the blood vessel walls, all of which emitted fluorescence. Electron microscopic examination of the labelled SAMC, using 6%paraformaldehyde and 1%glutaraldehyde in 0.1 M phosphate buffer as a fixative, revealed that the fluorescent cells possessed ultrastructural features typical of normal healthy looking SAMC as reported in a previous study in this laboratory (Leong e t al., 1983). Having successfully labelled the SAMC and other cells, described below, we investigated whether the SAMC (1)could be derived from any obvious source, and whether they or other labelled cells would in time migrate to the cerebral neocortex to become the resident microglia therein, and ( 2 ) would transform into flattened ramified microglia (RMC1 with advancing age and, if so, whether both of them would, like monocytes, express type 3 complement receptor (CR3) a s detected by OX-42. Finally, we aimed to study the effect of a n intracerebral stab wound inflicted at various times af-
AMOEBOID AND RAMIFIED MICROGLIA AND BRAIN INJURY
41
with a mixture of periodate-lysine-paraformaldehyde (McLean and Nakane, 1974),having a concentration of 2% or 4% paraformaldehyde. The brains obtained were kept in the sucrosed buffer overnight and frozen sections were cut at 40 km and mounted on gelatinized slides. The sections were then incubated in the monoclonal antibody OX-42 (Sera Lab, MAS 370b) diluted 1:100 with phosphate-buffered saline. After a n incubation period of 15-18 h at room temperature, the antibody was detected by using the Vectastain ABC kit against mouse IgG with 3,3’,-diaminobenzidine (DAB) a s a peroxidase substrate. The reaction product was MATERIALS AND METHODS intensified with nickel ammonium sulphate and the To label the amoeboid microglial cells, 40-50 pl of 1% sections were then lightly counterstained with methyl RhIc in normal saline was administered intraperito- green. In the above procedure, some sections were neally in 33 one-day-old rats. They were perfused from treated with triton. and others without. 10 min to 40 days after injection. To test whether the labelled microglial cells would migrate to the site of a n intracerebral stab wound, 30 rats first received 40-50 RESULTS p1 of 1%RhIc at 1-2 days after birth. From 1-26 days In addition to the choroid plexus and the associated after RhIc injection, a 21- or 281h-gauge needle was plunged unilaterally into the medial half of the middle epiplexus cells and the SAMC, the circumventricular part of the cerebral hemisphere through the cranium. organs, pia cells, some cells in the ependymal and the The animals were perfused 4 h to 40 days after the stab subventricular zone of the lateral ventricles and a conglomeration of small round cells in and around the cawound. All operations and perfusion were performed under vum septum pellucidum were also labelled (Figs. 2-4). anaesthesia, achieved by ether inhalation for neonatal Furthermore, there were labelled cells in other regions rats, and intraperitoneal administration of 0.4-0.6 rnl of the brain but such would be reported separately. As of 3.5% chloral hydrate and 0.8-1.0 ml of 7% chloral early a s 10 min after the administration of RhIc, the hydrate, respectively, for subadult and adult rats. For labelled cells emitted a low level of fluorescence, which perfusion, the animals were briefly rinsed with 0.1 M increased in intensity with time. In a rat perfused 1 day phosphate buffer (pH 7.2-7.4) and then fixed with 6% after injection, the caval cells appeared to stream diparaformaldehyde in phosphate buffer. The volumes of rectly or along the vicinity of the subventricular region fixative varied between 1 0 0 4 0 0 ml, depending on the toward the supraventricular callosal region (Fig. 2). At size of the animal. After perfusion, the brains were day 1 after RhIc injection, the SAMC were round (Fig. dissected out, kept overnight in the sucrose-buffered 3) and measured between 5 and 10 pm in diameter. fixative and sectioned a t 40 pm thickness in a cryostat. Some of these exhibited a filopodia-like process. With One out of every two (neonatal brains), three (subadult longer survival times, the cells became oval and then brains), or five (adult brains) sections were mounted on flattened or triangular, showing a variable number of gelatinized slides, air dried, and coverslipped, using the branching processes emanating from their cell bodies, nonfluoreseent Entellan as a mountant. The sections thus assuming the morphology of the typical ramified were examined and photographed in a Leitz Aristoplan or “resting” microglia (Fig. 4) observed in the adult microscope equipped with a mercury lamp for fluores- brain. In sections counterstained with cresyl fast violet, cence microscopy, using a wide band ultraviolet excita- the SAMC displayed a round nucleus and the RMC a tion filter (excitation range, 515-560 nM). After the rod-shaped nucleus. Pyknotic nuclei a t locations correabove procedure, the coverslips were removed and the sponding to those of the rhodamine-labelled cells were sections counterstained with cresyl fast violet and ex- frequently seen in older animals. There was a n obvious amined under conventional brightfield illumination. increase in the number of labelled SAMC during the For immunocytochemistry, 3 groups of animals were early postnatal period, but a s the animals increased in used. Group I consisted of 16 normal rats, aged 2 days age, their numbers were drastically reduced. to 3 months after birth. Group I1 consisted of 15 rats Other than the caval cells, which appeared to migrate injected with RhIc a t birth and perfused at 2 days, 11 to adjacent regions a s mentioned, all other labelled cells days, 27 days, and 40 days after injection. The third remained in situ with age. No labelled cells were found group of 12 rats was injected with RhIc a t birth and in the neocortex a t any of the survival times studied. received a n intracerebral stab wound 2-26 days post With advancing age, the caval cells were gradually deinjection and finally perfused 4 h to 33 days after in- pleted as the cavum septum pellucidum became obliterjury. Anaesthesia before operation and perfusion and ated. the operation procedures followed those described in In early postnatal normal rats, the SAMC demonthe preceding section. For perfusion, the animal was strated strong immunocytochemical reaction with first rinsed with Ringer’s solution, followed by fixation OX-42 (Fig. 5) a s reported previously (Ling et al., 1990). ter RhIc injection, looking specifically for any possible migration of the SAMC or RMC to the site of injury and possible activation of the antigenicity of the RMC, which could not normally be demonstrated in older rats (Jordan and Thomas, 1988; Ling et al., 1990; Sminia et al., 1987). In the course of investigation, some exciting preliminary results also prompted us to reinvestigate the immunoreactivity of the RMC in some neonatal, late postnatal, and adult normal rats.
42
LEONG AND LING
Figs. 5-7. Photomicrographs of immunolabelled supraventricular amoeboid microglial cells in a normal neonatal rat (Fig. 5) and of the same type of cells rhodamine-labelled (Fig. 6) in a rat that had received rhodamine B isothiocyanate injection a t birth and was perfused 2 days after injection. Figure 7 shows the same cells in Figure 6 immunolabelled with OX-42. Some double-labelled cells are indicated by arrows. Bars: 0.05 mm.
In day-old rats injected with RhIc and perfused with periodate-lysine-paraformaldehyde(PLP) as a fixative, the fluorescence of rhodamine-labelled SAMC could be demonstrated, but it was less intense than that of tissue perfused with 6% paraformaldehyde in phosphate buffer. Also with PLP fixative, the background nonspecific staining was stronger, Immunocytochemical staining revealed that the very cells labelled with rhodamine
Figs. 8-10. Photomicrographs of supraventricular ramified microglial cells (Fig. 8) labelled with rhodamine B isothiocyanate and of the same cells immunolabelled with OX-42 at the same (Fig. 9) and a higher (Fig. 10)magnification. Some of the double-labelled cells are indicated by arrows. The rat received rhodamine B isothiocyanate at birth, stabbed in the cerebral hemisphere 24 days after injection and perfused 30 days after the stab wound. The labelled cells were situated contralateral to the intracerebral lesion. Bars: 0.1 mm.
were OX-42 positive, as evidenced by comparing photographs of rhodamine-labelled SAMC (Fig. 6) taken before the staining procedure with those of OX-42 immunolabelled cells (Fig. 7). Rhodamine-labelled RMC (Fig. 8) were also immunopositive to OX-42 (Figs. 9, 10). There was no noticeable difference in the intensity of immunoreaction in neonatal rats perfused with PLP fixative containing either 2% or 4% paraformaldehyde. With older rats, however, it was clear that a lower percentage of paraformaldehyde in the PLP fixative gave a much stronger immunoreaction. Ramified microglia
AMOEBOID AND RAMIFIED MICROGLIA AND BRAIN INJURY
43
Figs. 11-14. Figs. 11 and 12. Immunolabelled microglial cells in the supraventricular region of the corpus callosum in a 12-day-old rat perfused with PLP fixative containing 4% paraformaldehyde. The section shown in Figure 11 was not treated with triton, but a section adjacent to it, shown in Figure 12, was treated with triton during the immunocytochemical staining. Figs. 13 and 14. Photomicrographs showing the supraventricular microglia in a normal adult rat (Fig. 13), and a rat injected with rhodamine B isothiocyanate at birth, and
subsequently received an intracerebral stab wound 26 days later and perfused 33 days after injury (Fig. 14). The two rats were perfused with PLP fixative containing 2 6 paraformaldehyde, and no triton was applied during the immunocytochemical procedure. Note that in the absence of triton, the microglia in the normal rat were poorly stained but quite distinctly stained in the experimental rat. Figures 11-14, same magnifications. Bar: 0.1 mm.
and microglia in the cerebral cortex and other parts of the brain that were not stained with PLP solution containing 4% paraformaldehyde gave distinct immunoreaction with one containing 2% paraformaldehyde. It was also noteworthy that, irrespective of the concentration of paraformaldehyde, sections treated with triton showed a more intense reaction than those without. Thus in a 12-day-oldnormal rat perfused with PLP fixative containing 4% paraformaldehyde, the supraventricular microglial cells demonstrated only weak immunoreactivity (Fig. 11) without triton treatment, but strong reaction (Fig. 12) when the sections were treated with triton. Interestingly, in rats injected with RhIc at birth and processed in like manners 11 days later, the supraventricular microglia gave intense reaction even in the absence of triton, and such was also the case in rats perfused as late as 30 days after injection. Though PLP fixative containing 2% paraformaldehyde
gave a stronger immunoreaction, it dampened the fluorescence of rhodamine-labelled cells, as mentioned earlier. This was especially noticeable in rats receiving RhIc injection at birth but perfused a few weeks later. However, the fluorescence of such cells in rats receiving intracerebral lesions was not much dimmed (Fig. 8). In such rats, the immunoreaction of the supraventricular cells was also stronger. The contrast was best demonstrated in older rats (Figs. 13,141when the immunoreactivity of the RMC in both normal (Fig. 13) and injected rats was weak in sections not treated with triton, but strong in those receiving RhIc injection followed by an intracerebral lesion (Fig. 14). The intracerebral stab wounds were either superficial or deep, situated within the medial half of the middle part of the cerebral hemisphere. The superficial lesions reached the vicinity of but not touching the supra-ventricular SAMC, whereas the deeper lesions in-
44
LEONG AND LING
Figs. 15 and 16. Rhodamine-labelled cells (Fig. 15) at the site of a superficial intracerebral stab wound and the same cells (Fig. 16) immunolabelled with OX-42 in a rat receiving rhoddmine B isothiocyanate injection at birth, stabbed 2 days following injection and perfused 2 days afterward. Some doubly labelled cells are indicated by arrows. Bars: 0.1 mm
1982; Murabe and Sano, 19821, immunocytochemistry (Ling e t al., 1990; Perry and Gordon, 1988) to tissue culture (Jordan and Thomas, 1988; Ling e t al., 1983). The subject, however, remained controversial because of inherent technical limitations. The conclusions were based mostly on circumstantial or inadequate evidences (for review, see Jordan and Thomas, 1988). In this study, using a novel method of labelling the SAMC with a n enduring cell marker without damaging the brain, we proved that (1)the SAMC and RMC were the same cells, the latter being the transformation of the former as the animal advanced in age, and (2) both of them expressed CR3 membrane receptors a s demonstrated by OX-42. As the animals matured, it became increasingly difficult to demonstrate the immunoreactivity without the use of triton and a lower percentage of paraformaldehyde. Triton being a detergent, probably had to break down certain membrane constituents to unmask the receptors for reaction with OX-42. In the case of paraformaldehyde, a lower concentration would probably exert less inhibitory effect on the receptors. It is significant to note that in older animals, CR3 receptors in the RMC, which could not be demonstrated with PLP containing 4% paraformaldehyde and without triDISCUSSION ton, could be revealed when RhIc was present intracelBy virtue of their morphology, location, and immuno- lularly. This indicated that the dye acted a s a stimulant reactivity, the labelled cells in the supraventricular cal- for the RMC in their expression of immunoreactivity. losal region could easily be identified a s the SAMC de- Such expression was further elevated in the presence of scribed previously (Leong et al., 1983; Ling et al., 1990). a n intracerebral lesion. Previous studies (reviewed by Their relation with RMC has been investigated by di- Streit et al., 1988) have also shown a n activation of verse techniques ranging from morphological observa- some intracellular and surface molecules during neution (Cammermeyer, 1970; Ling et al., 1990; Mori and ronal degeneration and regeneration. Though the origin of the SAMC before birth was not Leblond, 1969), study of phagocytic capability (Lent et al., 19851, histochemistry (Ling, 1977; Ling et al., studied, it appeared obvious that a t birth the SAMC
volved the corpus callosum. One to 4 days after a superficial lesion, a conglomeration of rather large, round rhodamine-labelled cells (Fig. 15) measuring about 15 pm in diameter and immunopositive to OX-42 (Fig. 16) appeared around the site of lesion. Such cells were not present when the animals were perfused within 8 hours after the lesion. There was no migration of the labelled microglia toward the lesioned site at any of the survival periods studied. In a deep lesion, rhodamine-labelled large round cells appeared in the immediate vicinity of the lesion. Some of them intermingled with smaller SAMC, which had earlier been labelled with rhodamine (Figs. 17-20). Compared to the normal (Fig. 211, many more SAMC were commonly seen at or near (Fig. 22) the site of lesion, especially when the lesion was made with the larger size needle. Even with 6% paraformaldehyde in phosphate buffer as a fixative, the labelled large round cells showed less intense fluorescence than the labelled SAVC, especially when the lesion was made 2-5 weeks after the introduction ofl2hIc.
AMOEBOID AND RAMIFIED MICROGLIA AND BRAIN INJURY
45
Figs. 17-20. Figure 17 shows small, round supraventricular microglial cells intermingled with large, round cells, both of which were rhodamine-labelled. The section was taken from a rat injected with rhodamine B isothiocyanate intraperitoneally within 1day after birth, stabbed in the cerebrum 1 day after injection, and subsequently perfused 4 days later. Both cell types were immunolabelled (Fig. 18)with OX-42. Figures 19 and 20 show the magnified view of the labelled cells situated within the boxes indicated in Figures 17 and 18. Arrows
indicate some doubly labelled cells. Note that Figure 18was taken at a slightly higher magnification than Figure 17, but both Figures 19 and 20 were taken at the same magnification. The immunolabelled cells in Figure 20 appeared larger than the rhodamine-labelled ones in F i y r e 19, indicating that some expansion of tissue had occurred as a result of the immunocytochemical procedure. Bars in Figures 17 and 18: 0.1 mm; in Figures 19 and 20: 0.05 mm.
were already present in great numbers and were faintly labelled within 10 minutes of introduction of RhIc. The faintly labelled cells could not be rhodamine-labelled circulating monocytes t h a t left the blood vessels for i t would be difficult to imagine so many of such cells exiting the bloodstream within such a short time. The initial increase in the number of SAMC during the first few postnatal days might be attributed to (1)their mitosis, which had been observed by Imamoto and Leblond (1978), (2) migration of monocytes from the bloodstream (Ling e t al., 1980), and (3) migration of cells from the cavum septum pellucidum, as indicated in the present study. As the animals matured in age, the number of rhodamine-labelled cells was drastically reduced. Such a reduction could most probably be attributed to cell death, observed by Imamoto and Leblond (1978). It was noted in this study that the labelled caval cells were gradually depleted with the obliteration of the cavum. Also, previous study (Tseng et al., 1983) failed to observe any degenerating cells in this region. From these two observations, we surmise that the amoeboid microglial cells contained in the cavum might be temporarily kept there for distribution to adjacent areas a t the appropriate time. Our study could not substantiate Hortega’s hypothesis that microglia in general were derived from the lep-
tomeninges of the developing brain. Cells in the pia were labelled, but they did not migrate to the cerebral cortex. Also, microglia in general could not be derived from the monocytes in the bloodstream postnatally; if so, such cells would have engulfed the RhIc present intravenously and would appear fluorescent in the cerebral cortex after leaving the bloodstream. The microglia, however, were immunopositive to OX-42, confirming a previous report on the mouse (Perry and Gordon, 1988). This indicates their macrophage/monocyte lineage, since OX-42 recognizes cell surface proteins synthesized by macrophages (Robinson et al., 1986). It is, therefore, possible that they could be derived from the bloodstream in utero and subsequently undergo proliferation and differentiation to become the resident microglia. Unlike the labelled microglia, these microglia were not phagocytic and therefore did not engulf RhIc that had diffused into the cerebral cortex. It would therefore not be unreasonable to conclude that there might be at least two populations of microglial cells in the brain. The possible existence of different populations of microglia in the brain had also been suggested by previous work (Miles and Chou, 1988). The labelled SAMC and RMC were OX-42 positive, indicating that they were of the same lineage as the microglia in the cerebral cortex. The SAMC probably
46
LEONG AND LING
Figs. 21 and 22. Figure 21 shows rhodamine-labelled supraventricular amoeboid microglial cells in a rat injected with rhodamine B isothiocyanate within 1 day after birth, and perfused 4 days later. Figure 22 shows rhodamine-labelled supraventricularmicroglial cells in another rat treated similarly, but in addition to rhodamine B isothiocyanate inject.ion, the rat also received an intracerebral stab wound inflicted 2 days after injection. The section was taken a t 900 pm caudal to the posterior end of the cerebral lesion. There is a n obvious increase in the number of labelled cells in this rat. Bars: 0.1 mm.
functioned a s macrophages to engulf degenerating axons present in the developing corpus callosum (Innocenti, 1981) or naturally occurring degenerating cells (Ling et a]., 1990; Maruyama and D’Agostino, 1967; Matsumoto and Ikuta, 1985). It is important to note that in the presence of a nearby brain lesion, the SAMC were not detracted from their primary task in the corpus callosum for they did not migrate to the site of that lesion but proliferated in situ when the corpus callosum was damaged. Having performed the phagocytic function, many of the glial cells degenerated as reported previously (Imamotoand Leblond, 1978) and a s demonstrated by the presence of pyknotic nuclei of the fluorescent cells in this study. The remaining cells became flattened and ramified, probably as a result of adaptation to the surrounding architecture as the callosal fibres began to myelinate and occupy the large intercellular space existent in the developing callosum. Rhodamine-labelled cells that appeared a t the site of injury were most likely monocytes that had left the bloodstream, as they exhibited immunopositive reaction to OX-42 and as they showed a weaker fluorescence when the lesion was made 2-5 weeks after the
introduction of RhIc. By this time, much of the RhIc present intravascularly would have been diluted with a n increase in the size of the animal and could not therefore label newly formed monocytes a s effectively. Our conclusion that the large rhodamine- and immunolabelled cells at the site of injury were most likely monocytes that had left the bloodstream is supported by previous studies (Kitamura et al., 1972; Konigsmark and Sidman, 1963; Torvik, 1975) that reported the presence of haematogenous cells at the site of cerebral injury. Our observation suggests that when a stab wound was made in a n area containing ramified microglia, labelled monocytes appeared a t the lesioned site. It appeared that the resident microglia were not able to handle the extra load imposed by the stab wound and help had to be called from the circulating monocytes. The latter were probably needed to ingest red blood cells (Penfield, 19321, stimulate angiogenesis (Hunt et a]., 1984), or release neuronotrophic factors (Nieto-Sampedro e t al., 1983) or interleukin-I, which can activate other macrophages and glial cells to produce a variety of factors (Riopelle, 1988). Finally, it might be said that as known microglial markers do not definitely distinguish between blood-derived monocytes and resident microglia, our present method of labelling blood-derived monocytes might offer a simple solution to the problem.
ACKNOWLEDGMENTS We thank Ms. C . Ang for her untiring secretarial help, the technicians of the Department of Anatomy, and Mr. Poon Keah How, Mr. Lee Eng Kiang, Ms. Joyce Koh, and other year I medical students, National University of Singapore, for their technical assistance.
REFERENCES Cammermeyer, J. (1970) The life history of the microglial cell: A light microscopic study. In: Neurosciences Research, Val. 3. S. Ephrenpreis and O.C. Solnitzky, eds. Academic Press, New York, pp. 43129. Haga, S., Akai, R., and Ishii, T. (1989) Demonstration of microglial cells in and around senile (neuritic)plaques in the Alzheimer brain. An immuno-histochemical study using a nova1 monoclonal antibody. Acta Neuropathol., 77569-575. Hunt, T.K., Knighton, D.R., Thakral, K.K., Goodson 111, W.H., and Andrews, W.S. (1984) Studies on inflammation and wound healing: Angiogenesis and collagen synthesis stimulated in viva by resident and activated wound macrophages. Surgery, 96:48-54. Imamoto, K. and Leblond, C.P. (1978)Radioautographic investigation of gliogenesis in the corpus callosum of young rats. 11. Origin of microglial cells. J . Comp. Neurol., 180:139-164. Innocenti, G.M. (1981) Growth and reshaping of axons in the establishment of visual callosal connections. Science, 212:824-827. Jordan, F.L. and Thomas, W.E. (1988) Brain macrophages: Questions of origin and interrelationship. Brain IZes. Rev., 13:165-178. Kaur, C., Ling, E.A., and Wong, W.C. (1986) Labelling of amoeboid microglial cells in rats of various ages following an intravenous injection of horseradish peroxidase. Acta Anat., 125:132-137. Kitamura, T., Hittori, H., and Fujita, S. (1972)Autoradiographic studies on histogenesis of brain macrophages in the mouse. J . Neuropath. Exp. Neurol., 31502-518.
AMOEBOID AND RAMIFIED MICROGLIA AND BRAIN INJURY Konigsmark, B.W. and Sidman, R.L. (1963) Origin of brain macrophages in the mouse. J . Neuropath. Exp. Neurol., 22643-676. Lent, R., Linden, R., and Cavalcante, L.A. (1985) Transient populations of presumptive macrophages in the brain of the developing hamster, as indicated by endocytosis of blood borne horseradish peroxidase. Neuroscience, 15:1203-1215. Leong, S-K. and Ling, E-A. (1990) Labelling of neurons with fluorescent dyes administered via intravenous, subcutaneous or intraperitoneal route. J . Neurosci. Meth., 32:15-23. Leong, S-K., Shieh, J-Y., Ling, E-A., and Wong, W-C. (1983) Labelling of amoeboid microglial cells in the supraventricular corpus callosum following the application of horseradish peroxidase in the cerebrum and spinal cord in rats. J . Anat., 136:367-377. Ling, E.A. (1977) Light and electron microscope demonstrations of some lysosomal enzymes in the amoeboid microglia in neonatal rat brain. 2.Anat., 123:637-648. Line, E.A., Kaur, C., and Wonp, W.C. (1982) Light and electron microscopic demonstration of nonyspecific esteraseh amoeboid microglial cells in the corpus callosum in postnatal rats: A cytochemical link to monocytes. J . Anat., 135:385-394. Ling, E.A., Penney, D., and Leblond, C.P. (1980) Use of carbon labelling to demonstrate the role of blood monocytes as precursors of the “amoeboid cells” present in the corpus callosum of postnatal rats. J . Comp. Neurol., 193:631-657. Ling, E.A., Kaur, C., Yick, T.Y., and Wong, W.C. (1990) Immunocytochemical localization of CR3 complement receptors with OX-42 in amoeboid microglia in postnatal rats. Anat. Embryol., 182:481486. Ling, E.A., Tseng, C.Y., Voon, F.C.T., and Wong, W.C. (1983)Isolation and culture of amoeboid microglial cells from the corpus callosum and cavum septum pellucidum in postnatal rats. J . Anat., 137:223233. Maruyama, S. and DAgostino, A.N. (1967) Cell necrosis in the central nervous system of normal rat fetuses. Neurology, 17:550-558. Matsumoto, Y. and Ikuta, F. (1985) Appearance and distribution of fetal brain macrophages in mice: Immunohistochemical study with a monoclonal antibody. Cell Tissue Res., 239:271-278. Mattiace, L.A., Davies, P., Yen, S-H., and Dickson, D.W. (1990) Microglia in cerebellar plaques in Alzheimer’s disease.Acta Neuropathol., 80:493498. McLean, I.W. and Nakane, P.K. (1974) Periodate-lysine-paraformal-
47
dehyde fixative, a new fixative for immunoelectron microscopy. J . Histochem. Cytochem., 22:1077-1087. Miles, J.M. and Chou, S.M. (1988)A new immunoperoxidase marker for microglia in paraffin section. J . Neuropath. Exp. Neurol., 47:579-587. Mori, S. and Leblond, C.P. (1969) Identification of microglia in light and electron microscopy. J . Comp. Neurol., 135:57-80. Murabe, Y. and Sano, Y. (1982) Morphologic studies on neuroglia. VI. Postnatal development of microglial cells. Cell Tissue Res., 225:469485. Nieto-Sampedro, M., Manthorpe, M., Barbin, G., Varon, S., and Cotman, C.W. (1983) Injury induced neuronotrophic activity in adult rat brain: Correlation with survival of delayed implants in the wound cavity. J . Neurosci., 3:2219-2229. Penfield, W. (1932) Neuroglia, normal and pathological. In: Cytology and Cellular Pathology of the Nervous System. W.G. Penfield, ed. Hoeber, New York, pp. 423-479. Perlmutter, L.S., Barron, E., and Chui, H.C. (1990)Morphologic association between microglia and senile plaque amyloid in Alzheimer’s disease. Neurosci. Lett., 119:32-36. Perry, V.H. and Gordon, S. (1988) Macrophages and microglia in the nervous system. TINS, 11:273-277. Riopelle, R.J. (1988) Adrenal medulla autografts in Parkinson’s disease: A proposed mechanism of action. Can. J . Neurol. Sci., 15:366370. Robinson, A.P., White, T.M., and Mason, D.W. (1986) Macrophage heterogeneity in the rat as delineated by two monoclonal antibodies MRC OX-41 and MRC OX-42, the latter recognizing complement receptor type 3. Immunology, 57239-247. Sminia, T., De Groot, C.J.A., Dijkstra, C.D., Koetsier, J.C., and Polman, C.H. (1987) Macrophages in the central nervous system of the rat. Immunobiol., 174:43-50. Streit, W.J., Graeber, M.B., and Kreutzberg, G.W. (19881 Functional plasticity of microglia: A review. Glia, 1:301-307. Torvik, A. (1975) The relationship between microglia and brain macrophages. Experimental investigations. Acta Neuropathol., S6:297300. Tseng, C.Y., Ling, E.A., and Wong, W.C. (1983) Scanning electron microscope of amoeboid microglial cells in the transient cavum septi pellucidi in pre- and postnatal rats. J . Anat., 136:251-263.
