E~pp.Eye Res. (1992) 55. 101-117

Specific Transcellular Staining of Microglia Traumatic Degeneration of Carbocyanine-filled Cells SOLON

THANOS”, Research

(Received

CHRISTOS Laboratory,

Houston

PAVLIDIS,

JijRG

MEY

in the Adult Retinal AND

HANS-JiiRGEN

Department of Ophthalmology, University School of Medicine, Germany

3 June

1997 and accepted

Rat After Ganglion THIEL

of Tiibingen,

in revised form 21 October

1997)

The present work was undertaken to assess the fate of ganglion cell debris in the axotomized retina of adult rats and employed a new technique to label phagocytosing microglia via the internalized material. In the main experiment, transection axotomy was performed on the intraorbital segment of the optic nerve, and a fast-transported, vital fluorescent styryl dye (4Di-1OASP) was deposited at the ocular stump of the nerve in order to pre-label retrogradely the ganglion cells destined to die because of the axotomy. Optic nerve transection resulted in progressive degradation of ganglion cell axons, perikarya, and dendrites within the retina and in release of fluorescent material, which was then incorporated into cells identified as microglia. No other retinal cells stained, although astrocytes and Miiller’s cells also responded to neuron degeneration by accumulating glial fibrillary acidic protein. Incorporation of labelled material into microglia topo-chronologically paralleled the ganglion cell degeneration starting within the optic fibre layer (OFL) and proceeding towards the ganglion cell layer (GCL) and the inner plexiform layer (IPL) of the affected retina. Long-term labelling of microglia monitored up to 3 months after optic nerve transection indicated that labelled microglial cells persisted within the retina. Microglia displayed a strong territorial arrangement within the GCL and IPL, and staggered, bilaminated distribution in both layers. These studies directly prove that microglia in the retina can be transcellularly labelled during traumatic degeneration of ganglion cells. The findings suggest that microglial cells play an important role in axotomy-induced wound healing and removal of cell debris. Key words: rat retina; ganglion cells : axotomy : degeneration ; fluorescent dyes : microglia ; phagocytosis.

1. Introduction The consequences of injury to the central nervous system, the spinal cord, and the retina of higher

vertebrates are anterograde and retrograde neuronal degeneration and the failure of axons to regrow and re-establish connections (Cajal, 192 8 ; James, 193 3 ; Eayrs, 1952). In the retina of adult rats, retrograde ganglion cell degeneration commences a few days after intraorbital transection of the optic nerve and proceeds during the weeks and months following the axotomy, finally resulting in a depletion of the retinal ganglion cell layer (RGC) (Richardson, Issa and Shemie, 1982; Misantone, Gerschenbaum and Murray, 1984; Barron et al., 1986; Thanos, 1988b; Vidal-Sanz et al., 1988). Other cell populations surrounding the directly affected neurons, in particular astrocytes

and Miiller’s cells, respond to injuries

by accumulating glial fibrillary acidic protein (GFAP), indicating that degradation of neurons is generally accompanied by multiple glial reactions as well (Fulcrand and Privat, 19 77 ; Bignami and Dahl, 19 79 ; Streit and Kreutzberg, 1988). However, the molecular * For correspondence at: Research Laboratory, Department of Ophthalmology, University of Tiibingen. School of Medicine, Schleichstrasse 12. 7400 Tiibingen. Germany. 00144835/92/070101+

17 sos.oo/o

basis for interactions between dying neurons and glial cells remains obscure (see Perry and Gordon, 1988 for a review). Besides neurons and the macroglia (astrocytes, oligodendrocytes and certain specialized glia lie the radial glia in the brain and the Miiller’s glia in the retina), the so-called microglia are the third major group of cells within the central nervous system (CNS) and retina of mammals (de1 Rio-Hortega, 1932; Cammermeyer, 1970 ; Vaughn and Peters, 1968 ; Vrabec, 19 70). Brain microglia have been reliably identified with different techniques. Amongst these are the weak silver carbonate technique of de1 Rio-Hortega (1932), and enzyme histochemical methods, since microglial plasma membranes possess intense thiamine pyrophosphatase (TPPase) and nucleoside diphosphatase (NDPase) activities (Cammermeyer, 1970; Murabe and Sano, 1981, 1982). Microglia are localized in various areas of the developing, adult, and lesioned CNS (Mori and Leblond, 1969; Ling, 1982; Boya et al., 1987; Streit and Kreutzberg, 1988; Streit, Graeber and Kreutzberg, 1990; Perry and Gordon, 1988; Schnitzer, 1989; Schnitzer andscherer, 1990), as well as in the immature and mature retina (Vrabec, 19 70 ; Boycott and Hopkins, 198 1; Murabe and Sano, 1981, 1982; Schnitzer, 1989). A hematogeneous origin of brain and retinal microglia,

due to migration

0 1992 Academic Press Limited

102

S. THANOS

of monocytes from the leptomeninges into neighbouring brain tissue late in embryonic development (de1 Rio-Hortega, 1932; Perry, Hume and Gordon, 1985; Perry and Gordon, 1988) and their transformation into micoglia, has not been confirmed in other studies that failed to label the microglia with antibodies to blood monocytes (Oehmichen, Wiethiilter and Greaves, 1979). The function of brain and retinal microglia in the repair process at the sites of injuries is not well defined (see Perry and Gordon, 1988 for a review). Chronological observations of the microglia during neuronal cell death in the post-natal rabbit retina (Schnitzer, 1989) assigned to the microglia the role of removing degeneration-produced cell debris. Retinal microglia in the adult rabbit respond to optic nerve cut with enhanced NDPase activity (Schnitzer and Scherer, 1990). This indicates that so-called ’ resting ’ microglia identified within the normal adult rat retina (Murabe and Sano. 1981, 1982) are cells that had immigrated during the period of cell death and become structurally integrated in the retina. The coincidence of macrophage recruitment and invasion into lesioned and inflamed brain areas like the optic nerve (Stall, Trapp and Griffin, 1989) has been seen as further evidence of a functional role of macrophages at the sites of lesions (cf. Perry and Gordon, 1988: Giulian, 1990 for reviews). The present work was devoted to identifying and characterizing the spatio-temporal pattern of debris removal in the retrogradely degenerating adult rat retina. For this, the optic nerves of Sprague-Dawley rats were transected beyond the eyecup and the retinal ganglion cells, whose axons form the optic nerve, and retrogradely labelled with the fluorescent styryl dye 4Di-lOASP. This vital dye has a lipophilic component which inserts into cell membranes, thus accompanying the membrane particles after cell degradation. This feature facilitated observation of living cells and made it easier to follow the fate of fluorescent debris after cell death and degradation. The course of progressive ganglion cell degeneration in the axotomized retina (Richardson, Issa and Shemie, 1982; Misantone, Gerschenbaum and Murray, 1984; Barron et al., 1986: Thanos, 1988b) was studied from the second day to the end of the third month after optic nerve lesion, in dissected and whole-mounted sectioned retinas. We assumed that the cells which remove degradation products of dying fluorescent neurons also become labelled by internalizing the fluorescent dye.

2. Materials

and Methods

Surgical Procedures This study is based on 25 adult female SpragueDawley rats each weighing 200-250 g. In intra-

ET AL.

peritoneal chloral hydrate induced anaesthesia (42 mg per kg body weight), the left optic nerve of 22 rats was intraorbitally exposed and, after longitudinal incision of its meningeal sheath, the nerve was completely transected. Care was taken to avoid damage to the retinal blood supply. In 16 rats, crystals (0.2-0.4 mm in diameter) of the fluorescent styryl dye [D29 1, N-4(4-didecylaminostyryl)-N-methylpyridinium iodide, (4Di-lo-ASP), Molecular Probes, Oregon] were deposited immediately after transection at the ocular stump of the optic nerve as reported for DiI (Thanos, 1988a, b) in order to label retrogradely the retinal ganglion cells before they undergo degeneration. Comparison between different fluorescent probes from the carbocynine dye family (Honig and Hume, 1986 : Thanos and Bonhoeffer, 198 7 ; Lichtman, Magrassi and Purves, 198 7 ; Magrassi, Purves and Lichtman, 198 7) revealed that the most efficient and consistent labelling was obtained with 4Di-lOASP. We therefore selected this dye for the present study. In six control rats, the dye was deposited at the optic nerve stump during a second surgical intervention, 2 days prior to death, in order to label only this population of ganglion cells that survived the primary axotomy and to exclude labelling of other retinal cells. in particular via the retinal vasculature. In four control rats, the skull was opened above the visual cortex and, after aspiration of part of the cortex, the fluorescent dye was inserted into the left superior colliculus. These control animals were used to study whether the fluorescent dye is toxic to the retinal ganglion cells and whether it leaks from non-axotomized neurons. In two normal rats, the dye was injected into the vitreous body (5 ,ul of 2.5% ethanolic solution) in order to exclude the possibility that the fluorescent agent is directly taken up by microglial cells. In other controls aimed to prove whether the dye is specifically taken up by microglia and not by other retinal cells, dissected retinas were incubated with dye solution and examined for the types of cells which internalized the dye. Since deposition of the fluorescent dye at the optic nerve stump resulted in labelling of about 300/, of ganglion cells (see Results), two control rats were used to determine the maximum microglia density within the rat retina. To achieve this goal, all ganglion cells were retrogradely labelled from the superior colliculus, and the optic nerves were transected 1 week later. Glia density was examined 30 days after optic nerve transection. Following survival times of 2, 3, 8, 14. 30, 50 and 90 days, the rats of the major experimental group were killed with an overdose of 7% chloral hydrate. After intracardial perfusion with phosphate-buffered saline, the animals were fixed with 200 ml aqueous 4% paraformaldehyde and the retinas removed, incised in four quadrants and flat-mounted on filters with the optic fibre layer upwards.

SPECIFIC

TRANSCELLULAR

STAINING

OF

103

MICROGLIA

Morphometry The retinas were viewed either as whole-mounts or after cryostat sectioning (20-30 pm). Fluorescent ganglion cells and microglia were observed by means of a fluorescein filter, since the dye fluoresces greenyellowish. For quantification of the ganglion cells and microglia, the retinas were divided into three concentric areas with radii of about 1 mm (central), 2 mm (middle) and more than 2 mm (peripheral) from the midpoint of the optic nerve head as viewed in the whole-mounted retina. Ganglion cell and microglia densities were determined in each concentric field by measuring in each quadrant 3040 randomly distributed microscope fields with the 20 x lens. The data were averaged for each field to obtain densities of ganglion cells and microglia. In addition, microglia density was determined in the perivascular areas, i.e. along the large arteries and veins, and in less vascularized fields to detect vessel-dependent differences in the distribution of glia. Finally, by targeting the focus of the microscope, the microglia density was determined within the ganglion cell layer (GCL) and in the inner plexiform layer (IPL). The distances between neighbouring microglial cells were measured from photographic prints made of whole-mounted retinas 30 days and 3 months after optic nerve transection and labelling. For this, photographic prints from either layer with labelled microglia were placed on a digitizing tablet and the cell-to-cell distances were measured with a stylus and entered into a computer (Contron. Videoplan) to calculate the average distance. The areas of retinal regions devoid of microglia were also measured and calculated with the same computer. In addition, their relationship to surviving ganglion cells was determined, in order to calculate the incidence of ganglion cell survival with microglial cell appearance.

(Merck) for fluorescence microscopy. Immunohistochemistry with the Thy-l antibody (Barnstable and Drsiger, 1984) was performed on three whole-mounted retinas fixed 30 days after axotomy and 4Di-1 OASP labelling, in order to detect ganglion cell debris within the microglia. Double-labelling of 4Di-lOASP-positive retinas with the monoclonal antibody OX-42 (Dunn) directed to activated monocytes and microglia (Hickey and Kimura, 1988) was performed to prove the nature of retinal cells which were labelled upon degeneration of prelabelled ganglion cells (three retinas). 3. Results Labelling of Ganglion Cells

Deposition of the fluorescent styryl dye (4Di-lOASP) in the superior colliculus resulted in intense labelling of ganglion cells in the contralateral retina 3 days later [Fig. l(A)], showing that the dye is taken up by ganglion cell axon terminals and transported rapidly into the cell bodies. The density of ganglion cells in the central retinas of two rats was 1793 _+289 cells mm-” (60 fields) and decreased to 1489+287 cells mm-”

Enzyme and Immunohistochemistry

The nature of microglia was assured by enzyme histochemical localization of triphosphate pyrophosphatase (TPPase) activity according to the method of Murabe and Sano (1982). TPPase-positive cells were viewed in retinal whole-mounts. In addition, activated microglia were labelled with antibodies against vimentin according to the protocol of Graeber, Streit and Kreutzberg (1988). The retinal astrocytes and Miiller’s cells were stained in control retinas (n = 4) and in degenerated retinas (n = 4) with serum to GFAP (Dakopats). For this, cryostat sections (20-30 pm) and the whole-mounted retinas were post-fixed in cold methanol ( - 2O”C, 10 min), washed twice (30 min) in PBS, incubated overnight at 40°C with rabbit antiGFAP antiserum (diluted 1: 50 in PBS), washed twice (30 min) in PBS, incubated 30 min with TRIC-coupled secondary antibody to rabbit IgG (Dianova), washed twice, and mounted on glass slides in Permount

Fro. 1. Fluorescence photomicrographs showing retrogradely labelled ganglion cells in the flat-mounted rat retina 3 days (A) and 30 days (B) after injection of 4Di-1 OASP into the superficial laminae of the contralateral superior colliculus. In both casesthe cell bodies of numerous ganglion cells belonging to the various morphological types are brightly labelled. Scale bar: 50 pm.

104

S. THANOS

Fn :. 2. A, Ganglion cell labelling seen on a retina whole-mount 2 days after deposition of 4Di-1 OASP into the stt tram iectecII optic nerve. Various types of ganglion cells are fluorescent, whereas in only some of them are the dendritil corn& rletel y outlined. B. Retrogradely labelled ganglion cell bodies (arrows) and microglia seen in the whole-mounts days after optic nerve transection and deposition of 4Di-1OASP at the nerve stump. C, Most of the ganglion cells di the 1 4th day after optic nerve cut, while microglial cells become labelled. Scale bar: 50 ,um.

ET AL

of the inches :tina 8 )ear at

SPECIFIC

TRANSCELLULAR

STAINING

OF

MICROGLIA

105

FIG . 3. A, Degeneratingganglioncellsandfluorescentmicrogliaseen30 daysafter optic nerve transectionand simultam:ous label1ing. B, Ganglion cellsbut not microgliawere seen30 days after axotomy when the fluorescentdye was appliedto coptic nerve stump 2 days prior to killing. C, Only few ganglion cellsare identifiable50 days after axotomy and simultane lOUSdye applic:ation, D, In the control retina, seenalso50 daysafter axotomy, the dyewas deposited2 daysprior to deathof the aniimal. Scalebar: 50 urn.

106

S. THANOS

I

\T

I

100 Ii!+ B 0

l.

10

20

< : 60 : = 80 : ;5

ET AL.

201

T --A

30 40 50 70 Days offer optic nerve transection

90

100

FIG. 4. Course of ganglion cell degeneration (e), of microglia labelling within the ganglion cell layer ( n ), and of microglia within the inner plexiform layer (0). Microglia density peaks 14 days after optic nerve transection, when the ganglion cell density decreases rapidly to about 110 labelled cells mmm2.Decrease of GCL microglia between days 14 and 90 is paralleled by increase of IPL microglia. Bars indicate S.E.M. (60 fields) in the middle, and to 1056+ 338 cells mrne2 (60 fields) in the peripheral retina. The average density of ganglion cells on the retinal surface was 1446 f 3 76 cells mm-*, showing that the dye labelled most of the retinal ganglion cells (Perry, 1979). The fact that ganglion cell bodies of different diameters were labelled [Fig. l(A)] indicated that the various ganglion cell types were filled with 4Di-1OASP. Almost identical average densities (1500 + 402 cells mm-“, 90 fields) of ganglion cells were determined in the retinas of three animals at 30 days after application of 4Di-1OASP into the superior colliculus [Fig. l(B)], showing that the dye persists in the labelled cells for 1 month without affecting the cell morphology. In retinal sections (two retinas), the labelling was also restricted to the GCL and IPL (not shown). Ganglion Cell Degeneration Placement of the fluorescent dye into the stump of the optic nerve just after transection resulted in rapid retrograde labelling of 720 + 40 cells mm-” (Fig. 4), as microscopic examination revealed 2 days later. Some of the ganglion cells of various morphological types had completely stained dendrites [Fig. 2(A)]. The number of ganglion cells was lower than that determined in the case of labelling from the SC, presumably either because not all transected axons have access to the dye or because not all of them are able to retrogradely transport the dye after transection. Eight days after optic nerve injury and labelling, the density of fluorescent ganglion cells decreased to 605 f 2 5 cells mm-’ [Figs 2(B) and 41. Morphologically, dying ganglion cells were characterized by cell body swelling, dislocation of nuclei, pycnosis, chromatolysis and irregular dendritic varicosities

100

0’

Ii

Ii IIlrn MG (GCL)

,%

111 ~ I’,~’I!

I nm MG (IPL)

I urn MG (GCL 1

I nm MC (IPL)

FIG. 5. Graphics correlating the distribution of labelled ganglion cells (A) and microglia (B) with respect to retinal eccentricity and retinal depth 30 and 50 days after optic nerve cut. The retinal zones of eccentricity are represented as I to III in accordance with distances of less than 1 mm (I), less than 2 mm (II) and more than 2 mm (III) from the centre of the optic nerve head. There is a gradual increase in ganglion cells from the central to the peripheral retina. Microglia density in the ganglion cell layer (GCL) and in the inner plexiform layer (IPL) is higher in the central retina (I) and decreases gradually towards the periphery (III) both at 30 and 50 days after optic nerve cut.

(Lieberman, 1971) which are also typical for degenerating ganglion cells under light microscopy (Richardson, Issa and Shemie, 1982; Misantone, Gerschenbaum and Murray, 1984 ; Thanos, 1988b). Ganglion cell degeneration was visible throughout the retinal surface on the eighth day after lesion. At the end of the second week, labelled ganglion cells were characterized by irregular dendritic varicosities [Fig. 2(C)]. Their number decreased to 105 i 28 cells mm-’ [Figs 2(A) and 41. Further ganglion cell depletion was observed in retinas evaluated 30 [Figs 3(A) and (B); eight retinas], 50 [Figs 3(C) and (D): six retinas] and 90 days after lesion. The dendrites of ganglion cells which survived 30 and 50 days after cutting of the optic nerve had characteristic morphologies (Thanos, 1988b) which could be best determined when the cells were retrogradely labelled 2 days prior to death of the respective animal and dissection of the retina [Figs 3(B) and (D)]. The ganglion cell density was different

SPECIFIC

TRANSCELLULAR

STAINING

OF

MICROGLIA

FIG. 6. Double labelling of microglia. A, Cryostat section through a retina 4 weeks after transection of the optic ne‘rve and transcellular labelling of the microglia with the dye 4Di-1OASP which was used to label the axotomized ganglion cells. Almost all microglial cells are located within the CCL and IPL (arrowhead at bottom of the section). B, The same retinal section ed cells processedfor immunohistochemistrywith the microglia specific antibody 0X-42. Most of the 4Di-lOASP-label1

(arrowheads) 20

and some additional

microglial

cells within

deeper layers of the retina (arrows)

stain for the antibody.

SCale bar :

,um.

when considering retinal eccentricities (Fig. 5). It ranged between 7 f 3 cells mm-” (30 days post-lesion, three retinas, 90 measured fields) in the central zone and 18 k 4 cells mm-” (three retinas, 90 fields) in the periphery of the retina (Fig. 5). The data plotted in Fig. 4 represent the averaged densitiesfrom all three zones of eccentricity. As Figure 5 also shows, 50 days after transection and labelling, the density had further

decreased to 4+ 2 cells mm-* in the central retina (three retinas, 90 fields) and to 9 + 3 cells mm- * in the retinal periphery (three retinas, 90 measured fields). The different ganglion cell densities on the retinal surface presumably indicate that ganglion (:ell degeneration affects the cells in the central retina which have shorter axonal stumps, rather than tho: sewith longer axonal stumps in more peripheral retinal

108

STHANOS

ET AL.

FIG. 7. A, B, Fluorescence photomicrographsfrom whole-mountedretinasshowingthat GFAP-stainedastrocytes[arrows in (A)] do not contain 4Di-1OASP[arrows in (B)]. The photographsrepresentthe sameretinal region. Double-labelled cellswere never seen,indicating that GFAP-positiveneurogliaare not transcellularlylabelledwith 4Di-1OASP.C, D, Sectionthrough the retina 4 weeksafter transection of the optic nerve and transcellular Iabelling of the microglia [arrowheadsin (D)]. The microglialcellsdid not stain for GFAPwhen the samesectionwas processed for immunohistochemistry.Astrocytes within the inner layers of the retina [bottom of sectionin (C)] and cellsof Miiller [arrows in (C)] stainedintensely for GFAP, but they did not contain 4Di-1OASP[compare(C) and (D) which show the samesection].Scalebars: 50 itm.

difference was regions. This central-peripheral confirmed in two animals labelled from the SC and operated at the optic nerves 1 week later. In their retinas examined 30 days after transection, surviving ganglion cells had the following densities: 20 f 5 cells rnrnm2in the central, 28 f 6 cell mm-” in the middle, and 38 + 6 cells rnrnm2in the peripheral retinal zone (two retinas, 90 fields for each retinal eccentricity). The fluorescent dye persisted within the surviving ganglion cells during the entire period of observation

(Figs 2 and 3). Morphologically, the cells which were labelled during axotomy and survived until the time of fixation had brightly fluorescent perikarya and sometimes fluorescent dendrites [Fig. 3(A)]. When the fluorescent dye was deposited into the optic nerve stump 2 days before the animal’s death, only surviving ganglion cell axons stained, as anticipated ; neither debris from degenerated cellsnor microglia (seebelow) were outlined in the retina at the 30th or 50th day after lesion [Figs 3(B) and (D)]. This finding indicates that the axonal stumps of surviving ganglion cells

SPECIFIC

TRANSCELLULAR

STAINING

r-

-L

OF

MICROGLIA

109

-L

300

s

N ‘E E 200. =VI :

100:

A

V

a”

A

MG (GCL)

T

T

V

av

MG (IPL)

FIG. 8. Relationship of labelled microglia to the retinal vasculature at the 30th day after optic nerve cut. A, Distribution microglia within the central retina and along the major vessels which extend radially from the optic papila (OP). A, Artery: V, vein. Some ganglion cells are indicated with arrows. Scale bar: 0.5 mm. B, Large magnification from the vicinity of a blood vessel indicates the alignment of microglia along vessels of different calibers. Scale bar : 100 /Lrn. C, Densities of microglia cells in the neighbourhood of arteries (A), veins (V) and in less vascularized areas (av) of the retina. The graphs represent averaged densities from all eccentricities of the retina. There is no significant difference in the microglial density.

which survived degradation retained some vital functions, in particular the ability to take up and transport substances towards the cell bodies. Characterization

of Labelled Retinal Microglia

From the eighth day after optic nerve transection,

a

population of non-ganglionic cells appeared to contain the fluorescent dye within the OFL and GCL [Figs 2(B) (C) and 4(A) (C)l. These cells had small, irregularly shaped perikarya and dendritic extensions within the GCL [Figs 2(B) and 3(A) (C)l. Their dendritic territories were much smaller than those of ganglion cells. Morphologically, these cells were reminiscent of

110

S.THANOS

E:T AL.

PR

ONL

OPL

INL

IPL

GCL OFL

FIG. 9. Fluorescence photomicrographs from the same area in the retinal whole-mount and Nomarski images from serr [i-thin sections ()f photoconverted material showing microglia within the ganglion cell layer (A, B) and within the inner plexiforrr Llayer (C, D) 50 days after optic nerve transection and labelling with 4Di-1OASP. Microglia within the GCL appear out of focus when the photc Igraph was taken within the IPL and are therefore indicated with asterisks in (C). whereas cells which appear in focus within th te IPL are out of focus in the GCL. Individual cells were photoconverted and are indicated with arrows in (B) an Id (D). The micr ,oglial cells within the IPL are located within the ‘territorial gaps’ of the GCL microglia and vice versa. Seal e bar: 50 ,um. microgli ial cells which have been reported to appear in the devr :loping and in the adult rat retina (Murabe and Sano, 1.981, 1982). The microglial nature of the labelled cells was confirmed by TPPase histochemistry

(Murabe and Sano, 1981. 1982: Schnitzer, 198’ 91,by morphological criteria (Boycott and Hopkins, 1.981) and by vimentin immunohistochemistry, since activated microglia stains for vimentin (Graeber, Streit

SPECIFIC

TRANSCELLULAR

STAINING

OF

MICROGLIA

* 0

+

+G h

.t

*

l *

+

.

*

0 *

*

FIG. 10. Typical examples of retinal areas which illustrate the relationship between surviving ganglioncellsand microglia 3 months after optic nerve transection and labelling with 4Di-1OASP. A, B, Photographs taken from whole-mountswithin the GCL and IPL, showing the ganglion cell body [large arrow in (A)] and the dendrite in (B). Microglial cells occur within the GCL (A) but not within the area covered by the dendrltic branches (B). C, Drawing prepared from the prints of (A) and (B) showing the relationship between ganglion cell, dendritic field, GCL microglia (stars) and IPL-microglia (0). D, Drawing showing a second example of a surviving ganglion cell and its topological relationship to microglia. Conventions as in (C). bv. Blood vessel. Scale bar: 50 pm.

and Kreutzberg, 1988). The nature of transcellularly labelled cells was definitely proved by immunohistochemistry with the monoclonal antibody OX-42 (Hickey and Kimura, 1988) which recognizes microglial cells in the rat retina. As Figs 6 and 7 illustrate, the 4Di-1OASP filled cells also stain positive for the specific antibody indicating that they are microglial cells. The fact that only OX-42-stained microglial cells were also observed within deeper layers of the retina

[Fig. 6(B)] indicates that these cells did obviously not participate in internalization of the fluorescent dye. Labelled microglial cells were not seen in 12 retinas of six control rats in which the fluorescent dye was not placed at the optic nerve stump immediately after transection but rather 12, 28 and 48 days later (for each stage four retinas), i.e. always 2 days prior to the animal’s death. In these cases, in which the axotomyresistant ganglion cells were brightly

stained, no

112

labelled microglia were observed in the retina [Figs 3(B) and (D)]. These control experiments demonstrated that labelling of microglia depends on degradation of fluorescent ganglion cells and on uptake (= endocytosis) of the debris produced by the surrounding microglial cells. These controls, and the fact that longterm labelling of unlesioned ganglion cells with 4DilOASP from the superior colliculus [see Fig. l(B)] does not result in transcellular staining of microglia, led to the following conclusion: the ganglion cells became retrogradely labelled 2 days after lesion, degenerated thereafter and produced the membrane-associated fluorescent debris during progressive degradation. Subsequent distribution of fluorescent ganglion cell debris in the OFL and GCL resulted, therefore, in selective outlining of these retinal cells, which remove the debris. The response to axotomy by other types of retinal glia (astrocytes and Miiller’s cells) has been well described in the rat retina by means of axotomy- or penetrating wound-induced neuronal degeneration (Bignami and Dahl, 1979). According to this report, the normal retinal astrocytes and Miiller’s cells do not stain for GFAP. However, glia in the injured retina accumulate GFAP, indicating a response to ganglion cell degradation. This glial reaction (Fulcrand and Privat, 19 77 : Bignami and Dahl, 19 79 ; Miller and Oberdorfer, 1981) was confirmed in retinas used in the present study both in retinal whole-mounts [Fig. 7(A)] and in sections [Fig. 7(C)], which were processed for double-immunofluorescence with corresponding antibodies. Neither astrocytes, Miiller’s cells, nor retinal interneurons contacting ganglion cells were labelled with 4Di-1OASP (Fig. 7), indicating that the dye does not just leak from the degraded neurons to label other cells, but only becomes internalized by the microglial cells which are obviously specialized to phagocytosis. Topo-chronological Relationship between Ganglion Cell Degeneration and Staining of Microglia In conjunction with the course of ganglion cell degeneration, labelled microglia first appeared in the OFL and GCL at the eighth day after lesion, and their density increased in the GCL to a maximum of 440 f 45 cells mrnd2 14 days after lesion and declined to 400 + 3 5 cells mrnm2 at 30 and 50 days after optic nerve transection (Fig. 4). Three months after axotomy, the density of microglia in the GCL was 380+ 20 cells mm-‘, which is not significantly different from the density at 30 and 50 days after lesion. Parallel to the acceleration and subsequent decrease of labelled microglia in the GCL, a second layer of labelled microglia appeared within the IPL (Figs 4 and 8) at the 14th day after lesion. Its density was 50 + 8 cells mm-” at 14 days, and increased to 80 f 12 cells mm-’ at 30 and 50 days after lesion (Figs 4 and 8). Increase of the IPL microglia was con-

S. THANOS TABLE

ET AL.

I

Relationship between ganglion cell survival and lP1, Microglia MC gaps in IPL l&p. no. 1 2 3 4 5* 6*

Sum %

With GC

Without GC

180 170 185 115 222 189

159 164 166 98 198 169

1061

954 89.9

100

Surviving GC (with dendrites)

Surviving CC (within MG gaps)

28 27 20 22 24

21 6 19 17 24 20

131 100

107 81.6

10

Regions devoid of fluorescent microglia were determined on the retinal surface by focusing the microscope into the IPL. Only regions with diameters larger than 180 pm were defined as devoid of microglia, since the average normal nearest-neighbour distance measured from cell body to cell body was about 60 pm within each layer containing labelled microglia. The data of column 2 represent all regions devoid of microglia scattered throughout the IPL. Column 3 contains the numbers of microglial gaps associated with surviving ganglion cells. Most surviving GC (8 I.h%l were located within regions devoid of microglia. The retinas of animals 1 to 4 were labelled from the ON during transections, whereas those of animals 5 and 6 were retrogradely labelled from the SC and their nerves were transected 1 week later.

comitant with the slight decrease of GCL microglia (Figs 4 and 8), thus indicating that some microglial cells might migrate from the GCL into the IPL. Bilamination of microglia was observed up to 3 months after axotomy, when microglia densities were similar to those determined at the 30th and 50th days after lesion (Fig. 4). The maximum microglial cell density was determined in two control rats whose ganglion cells (1440 f 300 cells mm-“) were retrogradely labelled prior to optic nerve transection. Thirty days after transection, the sum of GCL and IPL microglial cell density was 480 f 45 cells mm-“. This density was comparable with the microglial densities determined after labelling of about 30% of the ganglion cells from the optic nerve (see Fig. 4), which apparently resulted in labelling of the entire microglia. The sizes of labelled cells within each layer of the retina were similar, as the fluorescent images (Fig. 9) indicate. The average distance between microglial cells within the GCL was 68 f 15 pm (four retinas, 120 measured fields). Within the less populated IPL, the average distance was 88 f 18 ihrn. The territorial distribution of labelled microglia displayed a typical staggered pattern within the two layers. Superimposition of the microglial territories of GCL and IPLcells revealed that IPL cells were regularly located in the ‘territorial gaps ’ formed between the GCL cells (Fig. 9). This regular arrangement was partially interrupted in the vicinity of large blood vessels and sometimes around surviving ganglion cells, whose dendritic branches were not degraded (Fig. 10. Table

SPECIFIC

TRANSCELLULAR

STAINING

OF

MICROGLIA

113

(A)

OPL

INL

) IPL

GCL

OFL Axotomy lobelling

Microglio lobelling

ond B-ASP

FIG. 11. Schematicpresentationof the experimentalmodelwhich summarizes the major findingsof the presentstudy. The ganglion (gc) are axotomizedand filled retrogradely with the fluorescentdye designatedas 4Di-1OASP(A). After retrograde degenerationof mostof the ganglion cellsthe fluorescentdebrisis taken up by microglialcells(mg) within the GCLand IPL. Miiller’s cells(mc) and astrocytes(as)respondto degenerationwith accumulationof GFAP [seenasdarker shadowingin (S)]. Remainingganglioncellsare long-termLabelledwith the dye and showmultiple beadsalongthe axonal stump.b, Bipolarcell; gc. ganglioncell; GCL,ganglioncell layer ; h, horizontal cell; IPL. inner plexiform layer ; mc, Miiller’s cell; mg. microglia; OFL. optic fiber layer ; OPL,outer plexiform layer; RPE,retinal pigmentepithelium; PR, photoreceptors.

I). Typically, IPL areas devoid of microglia, called microglial gaps (Table I), were observed in the territories of 107 out of 13 1 (81.6 %) ganglion cells which had completely filled dendrites after 3 months. In the remaining 24 ganglion cells, the microglial densities were reduced within the dendritic territories of the IPL. These observations document the high incidence of ganglion cell survival in regions with reduced densities or absenceof microglial cells. On the other hand, microglial gaps within the IPL were also observed to occur independently of the ganglion cell survival (Table I, 89.9% without ganglion cells) indicating that they are indeed absent in some retinal regions. Similar gaps were not observed within the ganglion cell layer, except for the retinal regions containing large-calibre vessels. 4. Discussion The present study used a new approach to investigate the mechanism of wound healing in the axotomized retina. We applied for the first time a fluorescent carbocyanine dye to retrogradely stain the cells destined to die and to achieve specific ‘ transcellular ’ labelling of the cells responsiblefor removal of cell debris. Figure 11 summarizes the experimental strategy used to label the neurophagic microglia and illustrates the major finding of the present work,

namely that only retinal microglia within the GCL and IPL internalize the fluorescent dye after ganglion cell degeneration. The results proved that only microglial cells are labelled with the fluorescent dye and suggest that only these cells are responsible for removing ganglion cell debris in the adult retina. Methodological Aspects Among the different fluorescent carbocyanine dyes approved so far to label anterogradely and retrogradely neurons, DiI (Honig and Hume, 1986; Thanos and Bonhoeffer. 198 7 ; Magrassi, Purves and Lichtman, 198 7 ; Lichtman, Magrassi and Purves, 198 7 ; Thanos, 1988a. b; Vidal-Sam et al., 1988), DiO (Honig and Hume. 1986 ; Magrassi, Purves and Lichtman, 198 7 ; Lichtman, Magrassi and Purves, 1987: Thanos, 1988b). DiA (Magrassi, Purves and Lichtman, 1987: Lichtman, Magrassi and Purves, 198 7 : BerniceGordon, pers. comm.) and non-carbocyanine dyes (Lichtman, Magrassi and Purves, 1987) have been found to persist over long periods of time in labelled neurons. The lipophilic styryl dye 4Di-1OASP inserts into the membranes with its two fatty tails parallel to the fatty acids and the chromophore at the surface. The probe is not fluorescent except in the cell membrane (information provided by the supplier). Its limited solubility in body fluids makes the dye ideal for

114

a convenient application at either the optic nerve stump or in small slits of the superior colliculus. A further advantage of the dye is the fast labelling of the corresponding neurons, which occurs within 2 days of deposition into the optic nerve and within 3 days of injection into the superior colliculus, which is about 15 mm distant from the cell bodies. Compared to DiI (Thanos, 1988a, b; Vidal-Sanz et al., 1988), 4DilOASP shows a similar pattern of labelling in ganglion and in their dendrites. Introduction of the membranophilic dye 4Di-1OASP as a transcellular marker for activated microglia is an additional tool which can be used to resolve some problems of the classical impregnation techniques (de1 Rio Hortega, 1932), enzyme histochemistry (Murabe and Sano, 1981, 1982; Schnitzer, 1985, 1989). immunohistochemistry with antibodies to monocytes and macrophages (see Perry and Gordon, 1988 for a review), and electron microscopy (Boycott and Hopkins, 1981). The major advantage of 4Di-1OASP is that it remains associated with degradation products of dying cells and is also internalized in microglial cells. Further advantages are that the dye does not leak from non-lesioned labelled neurons, as the experiments for long-term labelling from the superior colliculus showed, and that the molecule is transported within the proximal stumps of transected axons, as the experiments of delayed deposition of the dye for selective labelling of surviving ganglion cells demonstrated. Ganglion Cell Degeneration and Macroglia Reaction Transection of the adult optic nerve results in progressive retrograde degeneration of ganglion cells within the days and weeks following the axotomy (James, 1933 ; Eayrs, 1952 ; Richardson, Issa and Shemie, 1982 ; Misantone, Gerschenbaum and Murray, 1984 ; Grafstein and Ingoglia, 1982 ; Barron et al., 1986; Thanos, 1988b, Vidal-Sanz et al., 1988). The course of ganglion cell degeneration has been confirmed in the present study, which additionally demonstrated a central-peripheral gradient in ganglion cell disappearance. Although normal ganglion cell density in the central retina is about three-fold higher than in the periphery (Perry, 1979), central areas in axotomized retinas were less populated with ganglion cells than the peripheral regions, which is explained by a more massive cell death in the centre of the retina. After optic nerve transection beyond the eyecup, central ganglion cells may have shorter stumps than peripheral cells and may therefore be more vulnerable to axotomy than peripheral neurons with longer axonal stumps. This is in agreement with earlier observations which revealed that the shorter the distance between the site of axotomy and the cell body, the more severe the retrograde effect on the neuron (see Lieberman, 1971; Watson, 1974 ; Grafstein, 1986 for reviews).

S THANOS

ET AL.

The factors which coregulate traumatic degeneration of ganglion cells remain unknown. It has been proposed that signals produced at the site of axotomy are retrogradely transported to the cell body (see Grafstein, 1986 for a review). Alternatively, the local perineuronal environment may produce inductive factors after breakdown of synaptic boutons because of ’ ineffective neuronal transmission ’ (see Watson, 19 74 for a review). Such factors may be initial parts of a cascade of reactions which result in degradation of neurons. As we expected from former studies, the process of lesion-induced structural and functional neuronal alterations in various systems (Purves, 19 75 : Grafstein, 1986) is accompanied by a massive, protracted macroglial response (Fulcrand and Privat, 19 77 ; Bignami and Dahl, 19 79 : Perry and Gordon, 1988: Streit and Kreutzberg, 1988; Graeber, Streit and Kreutzberg, 1988). In the facial nucleus of rats, both microglia and astrocytes responded to toxic ricininduced degeneration (Streit and Kreutzberg, 1988). Microglia phagocytosis did not, however, occur after crush-lesion of the facial nucleus, although the microglia proliferated (Streit and Kreutzberg, 1988). In the optic nerve, macrophages which express Ia antigens are responsible for clearance of degenerating myelin after nerve lesion (Stoll, Trapp and Griffin, 1989). The present study proved directly that the cells which are labelled during degeneration of retinal neurons belong to the microglia and not to the macroglial cells as was assumed by Kalman (1989) since macroglia also appear near the dying ganglion cells and also respond to ganglion cell degeneration by accumulating GFAP (Bignami and Dahl, 19 79 ; Miller and Oberdorfer, 1981; present study). What is the Mechanism of Ganglion Cell Degradation? It is noteworthy that retinal microglia are transcellularly labelled from degraded ganglion cells in a specific chrono-topological sequence which parallels the pattern of ganglion cell degeneration. This sequence of labelling can be best explained by assuming that the time, location and amount of neuronal material that is produced determine the stimulation of microglia to remove this debris. As considerable ganglion cell axon degeneration commences during the so-called traumatic axon reaction within the optic fiber layer at the end of the first week after axotomy (Eayrs, 1952: Watson, 1974; Barron et al., 1986; Schnitzer and Scherer, 1990), microglia are first labelled in the OFL. Glia activation is in agreement with observation in the rabbit retina which showed an enhanced NDPase activity in the microglial membranes at the end of the first week after axotomy (Schnitzer and Scherer, 1990). Increase of microglia density within the GCL and later within the IPL is consistent with the progress of ganglion cell degeneration, which proceeds towards the IPL. Numerical increase of microglia both in the GCL and later

SPECIFIC

TRANSCELLULAR

STAINING

OF

115

MICROGLIA

in the IPL are in conformity with observations that some glial mitosis and migration occurs (Barron et al.. 1986; Schnitzer and Scherer, 1990). Do the ganglion cells die and become phagocytosed thereafter, or does the microglia become activated to degrade the non-functioning, altered neurons ? The initial step in the cascade of debris removal is certainly the axotomy. Then either ganglion cell degeneration is induced which secondarily mediates the phagocytosis, or the axotomy activates the microglia to ‘destroy ’ the cells (= neurophagy) whose axons are interrupted and thus altered. Both possibilities, however, raise the question why axotomy does not affect all ganglion cells. The finding that 4Di-lOASP-labelled IPL microglia are virtually absent in territories of most ganglion cells with intact dendrites might be seen as first speculative evidence that these areas either do not contain microglia or that the microglial cells cannot be activated to phagocytose the axotomized ganglion cells. Alternatively, certain ganglion cells may resist degradation, either by producing substances which suppress or neutralize enzymes involved in cell degradation, or by failing to express the altered antigens which are recognized by the microglia. The interaction between intact or lesioned cells and microglia seems to be the key to understanding the mechanism of phagocytosis. More recent work documents that early mediators in this interaction are microglial proteases (Giulian, 1990 for a review), whose blockade with protease inhibitors can rescue a significant portion of ganglion cells from axotomyinduced degradation (Thanos, 1991). Interestingly, there is a partial translocation of microglia from the GCL to the IPL during the second month after optic nerve lesion, which results in a strong bilaminated arrangement of labelled microglial cells at the second and third month after lesion. There are two pieces of evidence in favour of translocation: first, the sum of microglia within GCL and IPL forms a plateau from the 14th day after lesion onwards, which roughly corresponds to the maximal density of microglia found at the 14th day after injury (see Fig. 4) ; secondly, dye leakage is not the reason for decrease of microglia density within the GCL, since microglia remain labelled up to 3 months after axotomy, indicating that after degradation and endocytosis the dye persists in the microglial membranes. The restriction of labelled microglia to the retinal layers associated with ganglion cells is in line with the view that microglial activation is a highly specialized response of the nervous system, dedicated to the elimination of dead neurons (Perry and Gordon 1988). since ‘resting microglia ’ occur in deeper layers of the retina as well (Murabe and Sano, 1981, 1982; Boycott and Hopkins, 1981). The geometrical arrangement of microglia in the GCL and IPL displayed a striking pattern of alternating cell distribution. The IPL microglia seemed to be shifted into ‘ gaps ’ formed by the regular arrangement

of the GCL microglia. This alternating pattern can be explained by ‘contact inhibition’ exerted between microglial cells, a mechanism which accounts for a regular spreading of cells across the retinal surface. Such contact inhibition may involve microglial proteolytic enzymes (Giulian et al., 1986, 1989; Giulian. 1990) which, once activated by the ganglion cell degradation, can also limit the dendritic territories of neighbouring microglial cells. The concept derived from such a mechanism is that degeneration-dependent thinning of the distance between GCL and IPL results in shortening of the distance between GCL and IPL microglia. Thus, arrangement of the IPL cells into the interspaces of the GCL cells may represent the optimal geometrical arrangement because of inhibition that permits maximum possible distances between the GCL cells and their neighbours within the IPL. In terms of optimized function, such an arrangement would most ensure an adequate, uniform phagocytosis across the degenerating retina. The origin of retinal microglia still remains obscure (Shelper and Adrian, 1986; Perry and Gordon, 1988). although perivascular microglia are bone-marrowderived (Hickey and Kimura, 1988). Immigration of phagocytes occurs in other experimental models like stab lesions in the brain (Perry and Gordon, 1988 for a review). The difference between stab lesions and the retinal axotomy paradigm is that in the second case the blood vessels surrounding the dying cells were not lesioned, since the fifth branch of the ophthalmic artery which vascularizes the retina was not severed during surgery. Immigration of microglia also occurs during post-natal retinal development (Schnitzer, 1985, 1989) when natural neuronal cell death takes place. Whether the same microglial cells which eliminate neurons in the postnatal retina persist until adulthood and then become reactivated by traumatic violence, or whether immigrated microglial cells phagocytose injured neurons, is the object of parallel studies employing double-labelling techniques during postnatal life and in adulthood. These studies revealed that the same microglial cells which are responsible for eliminating neurons during the natural postnatal cell death do also remove cell debris during injuryinduced degeneration (Thanos, 1991). Acknowledgements

The technical contributions of D. Stemmer, M. Wild and S. von Kannen are gratefully acknowledged. We also thank T. Rice for improving the English and Dr H. Schwarz for help with preparing the EM specimens. The work was supported by the Deutsche Forschungsgemeinschaft (research grants : Th 386 2-l and 3-1 to S.T.). References

Barnstable. J. C. and Drtiger, U. C. (1984). Thy-l antigen: a ganglion cell specific marker in rodent retina. Neurascience11, 847-55. R-2

116

S. THANOS

Barron, K. D., Dentinger, M. P., Krohel. G., Easton, S. K. and Mankes, R. (1986). Qualitative and quantitative ultrastructural observations on retinal ganglion cell layer of rat after intraorbital nerve crush. J. Neurocytol. 15. 345-62.

Bignami, A. and Dahl, D. (19 79). The radial glia of Miiller and their response to injury. An immunofluorescence study with antibodies to the glial flbrillary acidic (GFA) protein. Exp. Eye Res. 28, 63-9. Boya, J., Carbonell, A. L., Calvo, J. and Boregon, A. (198 7). Ultrastructural study on the origin of rat microglia cells. Acta Anat. 134, 329-35. Boycott, B. B. and Hopkins, J. M. (1981). Microglia in the retina of monkey and other mammals: Its distinction from other types of glia and horizontal cells. Neuroscience 6. 679-88. Cajal, R. y. S. (1928). Degeneration and Regeneration of the Nervous System. (R. M. May. Trans.) University Press: London and New York. Cammermeyer, J. (19 70). The life history of the microglia cells: A light microscopic study. In Neurosciences Research, Vol. 3. (Eds Ehrenpreis, S. and Solnitzky, 0. C.). Pp. 44-129. Academic Press: New York. de1 Rio-Hortega, P. (1932). Microglia. In Cytology and Cellular Pathology of the Nervous System (Ed. Penfield. W.). Pp. 482-34. Paul B. Hoeber: New York. Eayrs, J. T. (1952). Relationship between the ganglion cell layer of the retinal and the optic nerve in the rat. Br. J. Ophthalmol. 36, 453-459. Fulcrand, J. and Privat, A. (1977). Neuroglial reactions secondary to Wallerian degeneration in the optic nerve of the postnatal rat: Ultrastructural and quantitative study. 1. Comp. Neural. 176, 189-224. Giulian, D. (1990). Microglia and tissue damage in the central nervous system. In: Di;fferentiation and Function of GliaI Cells (Ed. Levi, G.). Pp. 379-89. Alan R. Liss: New York. Giulian, D. R., Allen, R. L., Baker, T. J. and Tomozawa. Y. (1986). Brain peptides and glial growth. I. Gliapromoting factors as regulators of gliogenesis in the developing and injured nervous system. 1. Cell Biol. 101, 803-11. Giulian, D.. Chen, J., Ingeman, J. E., George, J. K. and Noponen. M. (1989). The role of mononuclear phagocytes in wound-healing after traumatic injury to adult mammalian brain. I. Neurosci. 9, 4416-29. Graeber, M. B., Streit. W. J. and Kreutzberg. G. W. (1988). The microglial cytoskeleton : vimentin is localized within activated cells in situ. 1. Neurocytol. 17, 573-80. Grafstein, B. (1986). The retina as a regenerating organ. In The Retina. Part II. (Eds Adler, R. and Faber, D.). Pp. 2 7 5-3 3 5, Academic Press : London. Grafstein, B. and Ingoglia, N. (1982). Intracranial transection of the optic nerve in adult mice: Preliminary observations. Exp. Neural. 76, 318-30. Hickey, W. F. and Kimura. H. (1988). Perivascular microglial cells of the CNS are bone-derived and present antigen in vivo. Science 239. 290-2. Honig, M. G. and Hume, R. I. (1986). Fluorescent carbocyanine dyes allow living neurons of identified origin to be studied in long-term cultures. 1. Cell Biol. 103. 171-87. James, G. R. (19 3 3). Degeneration of ganglion cells following axonal injury. Arch. Ophthalmol. 9, 338-43. Kalman, M. (1989). Dead cells can be phagocytosed by any neighbouring cell in early developing rat brain. Znt. 1. Neurosci. 46, 139-45. Lichtman, J. W., Magrassi, L. and Purves, D. (1987). Visualization of neuromuscular junctions over periods of several months in living mice. 1. Neurosci. 7, 1215-22.

ET AL.

Lieberman. A. R. ( 19 7 1). The axon reaction : A review of the principal features of perikaryal responses to axon injury. Int. Rev. Neurobiol. 14. 49-124. Ling, E. A. (1982). A light microscopic demonstration of amoeboid microglia and microglial cells in the retina of rats of various ages. Arch. Histol. lap. 45, 3744. Magrassi, L.. Purves, D. and Lichtman, J. W. (1987). Fluorescent probes that stain living nerve terminals. I. Neurosci. 7, 1207-14. Miller, N. M. and Oberdorfer, M. (1981). Neuronal and neuroglial responses following retinal lesions in the neonatal rats. 1. Comp. Neural. 202, 493-504. Misantone, L. J., Gerschenbaum, M. and Murray, M. (1984). Viability of retinal ganglion cells after optic nerve crush in adult rats. I. NeurocytoI. 13, 449-65. Mori, S. and Leblond, C. P. ( 1969). Identification of microglia in light and electron microscopy. 7. Comp. Neural. 135, 57-80.

Murabe, Y. and Sano. Y. (1981). Thiaminepyrophosphatase activity in the plasma membrane of microglia. Histothem. 71, 45-52. Murabe, Y. and Sane, Y. (1982). Morphological studies on neuroglia. VI. Postnatal development of microglial cells. Cell Tiss. Res. 225, 469-85. Oehmichen. M.. Wiethiilter, H. and Greaves, M. F. (1979). Immunological analysis of human microglia: Lack of monocytic and lymphoid membrane differentiation antigens. J. Neuropathol. Exp. Neural. 38, 99-103. Perry, V. H. (1979). The ganglion cell layer in the retina of the rat. Proc. R. Sot. Lond. B 204, 363-75. Perry, V. H. and Gordon, S. (1988). Macrophages and microglia in the nervous system. TINS 11. 273-7. 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, 313-26. Purves. D. (1975). Functional and structural changes in mammalian sympathetic neurons following interruption of their axons. I. Physiol. 252, 429-63. Richardson, P. M., Issa, V. K. M. and Shemie, S. (1982). Regeneration and retrograde degeneration of axons in the rat optic nerve. J. Neurocytol. 11, 949-66. Schnitzer, J. (198 5). Distribution and immunoreactivity of glia in the retina of the rabbit. I. Comp. Neural. 240, 128-42. Schnitzer. J. (1989). Enzyme-histochemical demonstration of microglial cells in the adult and postnatal rabbit retina. 1. Camp. Neural. 282, 249-63. Schnitzer, J. and Scherer, J. (1990). Microglial cell responses in the rabbit retina following transection of the optic nerve. 1. Comp. Neural. 302, 779-91. Shelper. R. L. and Adrian, E. K. A. (1986). Monocytes become macrophages : they do not become microglia : A light and electron microscopic autoradiographic study using 12 5-iododeoxyuridine. 1. Neuropathol. and Exp. Nerrrol. 45, l-19. Stall. G,. Trapp, B. D. and Griffin, J. W. (1989). Macrophage function during Wallerian degeneration of the rat optic nerve: Clearance of degenerating myelin and Ia expression. J. Neurosci. 9, 2327-35. Streit. W. J.. Graeber. M. B. and Kreutzberg, G. W. (1990). Functional plasticity of microglia : a review. GLIA 1, 301-7. Streit, W. J. and Kreutzberg, G. W. (1988). Lectin binding by resting and reactive microglia. 1. Neurocytol. 16, 249-70. Thanos. S. ( 1988a). Morphology of ganglion cell dendrites in the albino rat retina: An analysis with fluorescent carbocyanine dyes. 1. Hirnforsch. 6. 6 17-3 1. Thanos, S. (1988b). Alterations in the morphology of ganglion cell dendrites in the adult rat retina after optic

SPECIFIC

TRANSCELLULAR

STAINING

OF

MICROGLIA

nerve transection and grafting of peripheral nerve segments. Cell Zeiss.Res. 259, 599-609. Thanos, S. (1991). The relationship of microglial cells to dying neurons during natural neuronal cell death and axotomy-induced degeneration of the rat retina. E. J. Nuerosci. 3. 1189-207. Thanos. S. and Bonhoeffer, F. (1987). Axonal arborization in the developing chick retinotectal system. 1. Camp. Neural. 261~ 15 5-64. Vaughn, J. E. and Peters, A. (1968). A third neuroglial cell

117

type. An electron microscopic study. I. Camp. Neural. 127, 219-39. Vidal-Sanz, M., Villegas-Perez, M. P., Bray, G. M. and Aguayo, A. J. (1988). Persistent retrograde labeling of adult rat retinal ganglion cells with the carbocyanine dye DiI. Elcp. Neurul. 102. 92-101. Vrabec, F. (1970). Microglia in the monkey and rabbit retina. 1. Neuropath. Exp. Neurol. 29, 241-52. Watson, W. E. (1974). Cellular responses to axotomy and to related procedures. Br. Med. Bull. 30. 112-S.