Individual microglia move rapidly and directly to nerve lesions in the leech central nervous system

Transcription

1 Proc. Nati. Acad. Sci. USA Vol. 86, pp , February 1989 Neurobiology Individual microglia move rapidly and directly to nerve lesions in the leech central nervous system (nerve iinjury/ceil migration/video microscopy) ELLEN MCGLADE-MCCULLOH, ALICE M. MORRISSEY, FERNANDO NORONA*, AND KENNETH J. Department of Physiology and Biophysics, University of Miami School of Medicine, Miami, FL 3311 Communicated by Gunther S. Stent, November 18, 1988 MULLER ABSTRACT Small cells called microglia, which collect at nerve lesions, were tracked as they moved within the leech nerve cord to crushes made minutes or hours before. The aim of this study was to determine whether microglia respond as a group and move en masse or instead move individually, at different rates, and whether they move along axons directly to the lesion or take another route, such as along the edges of the nerve cord. Cell nuclei in living nerve cords were stained with Hoechst 338 dye and observed under dim ultraviolet illumination using fluorescence optics, a low-light video camera, and computer-assisted signal enhancement. Muscular movements of the cord were selectively reduced by bathing in 23 mm MgCI2. Regions of nerve cord within 3,um of the crush were observed for 2-6 hr. Only a fraction of microglia, typically <5%, moved at any time, traveling toward the lesion at speeds up to 7,Im/min. Cells were moving as soon as observation began, within 15 min of crushing, and traveled directly toward the lesion along axons or axon tracts. Movements and roles of leech microglia are compared with their vertebrate counterparts, which are also active and respond to nerve injury. Nerve injury triggers a characteristic sequence of cellular reactions. The early events occur whether or not regeneration follows. Thus, severed axons reseal and retract from the site of lesion, which is rapidly populated by microglia and perhaps other phagocytic cells. There are several possible sources and roles for microglia in regeneration (see, e.g., refs. 1-5). Microglia may arise from blood or from nervous tissue itself, and while microglia are phagocytic, they may play additional roles. The goal of the present study has been to determine whether all microglial cells in the leech travel toward the lesion and by what route they reach it. The leech nervous system has been useful for examining neurons and glia during synapse regeneration (6). Leech microglia, like leech neurons and large glia, resemble their mammalian counterparts in their morphology, physiology, and histochemistry (7-9). Moreover, resting and reactive microglia in the rat brain are selectively stained on their surfaces by a lectin from Griffonia simplicifolia seeds (1) that also in the leech stains microglia but not large glia and neurons (S. Hockfield, R. McKay, and K.J.M., unpublished data). Silver carbonate staining of cell bodies and Feulgen staining of nuclei reveal that leech microglia normally lie scattered among axons as an apparently homogeneous population of cells within the nerve cord and collect at the site of lesion within 24 hr of nerve injury (9). The accumulation peaks within 24 hr and declines slowly, never returning to preinjury levels, even after axons successfully regenerate. Experiments in which nerve cords have been isolated in tissue culture medium and then crushed have shown that The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C solely to indicate this fact. most, if not all, of the microglia at the lesion migrate there from within the nervous system (9). Although microglia occupy <5% ofthe volume ofthe nervous system, the ganglia alone are estimated each to contain 1,-15, microglia (8), compared to only 4 neurons and 8 large glia per ganglion (11). The axon bundles, or connectives, that link ganglia contain thousands of microglia yet have only 2 large glial cells that each ensheathe -3 axons. Furthermore, in the connectives nearly all microglia are capable of moving to the lesion, although usually they do not. In contrast, there is no apparent shift of cells of the perineurial sheath, which are distinctive for their flattened rather than spindle-shaped nuclei. An important question is how microglia within the nerve cord reach the lesion. Do all microglia migrate toward the site of injury, moving en masse, or do only certain cells within the connectives move? Is microglial migration chiefly at the edges of the connective along its outer surface, or do cells anywhere within the connective move? METHODS Preparation and Lesions. Segments of nerve cord were removed from the animal and pinned in Sylgard-coated dishes containing leech saline (12). Cords were stained 2- min with Hoechst 338 dye [.1% (wt/vol) in leech saline] to permit tracking of microglial cells in the living connective. Hoechst dye stained similarly but was not used in the experiments reported here. To optimize Hoechst staining of the cells and reduce background fluorescence, the tissue was incubated overnight at 15C in Leibowitz-15 (L-15; GIBCO) tissue culture medium supplemented with 2% fetal calf serum,.6% glucose, and gentamycin (1 mg/ml) (13). In other studies (14), it has been shown that nerve cords cultured in L-15 will grow for weeks and can regenerate severed axons, which form new appropriate synapses. Connective segments 2-3 mm long were cut from the stained cords between ganglia. Just before observation, the connective segments were crushed with a pair of forceps (Dumont no. 5) ground to a width of 3,m, which severed the axons (15, 16). Except as noted, an isosmotic solution of MgCl2 was added to the L-15 medium to bring [Mg2+] to 23 mm, thereby blocking spontaneous movements of the muscular connectives. Video Microscopy. In most experiments, a single region of paired connectives 22 Am long and 5-2,um from the crush was viewed with low-intensity epifluorescence illumination (Zeiss filter set 48772) through a x4 water immersion objective at room temperature (2 C-23 C) for 2-6 hr after the crush was made. The plane of focus passed through the centers of the connectives. Because bright UV light arrested movement of stained cells (data not shown), the excitation beam from the 1 W dc mercury arc lamp was *Present address: School of Medicine, Universidad Central del Caribe, Cayey, Puerto Rico. 193

2 194 Neurobiology: McGlade-McCulloh et al. Po.Nt.Aa.Si Proc. Natl. Acad. Sci. USAS 86(99 (1989) attenuated by up to 3 log units and a low-light video camera (model 66/ISIT; Dage-MTI, Michigan City, IN) used to view the preparation. Under these conditions, stained cells moved in response to injury. At these low light levels, the fluorescent cells were not visible to the unaided eye. With each experimental preparation, a stained and crushed control nerve cord, not illuminated, was processed to confirm that low light levels did not affect microglia accumulation at the crush. To improve the contrast and signal/noise ratio of the video image, signals were processed with an Image I/AT image processor (Universal Imaging Corporation, Media, PA). Processed and direct images were recorded with videotape recorders (Panasonic NV-85, Panasonic NV-835, or JVC HR-S7U). Histology. After recording, the tissue was immediately fixed in Carnoy's fixative and processed for Feulgen staining (9). Stained crushed cords in the same recording dish, but not illuminated, were also processed. In addition to the unilluminated controls, other controls were run to test the effects of the Hoescht dye and elevated [Mg"~] on the accumulation of microglia, at the crush. The muscular nerve cords lengthened upon fixation in Carnoy's fixative, shrinking their lateral dimensions. In experiments and controls, many fewer microglia accumulated at cuts, such as those made to isolate the nerve cords for tissue culture, than at crushes. Other nerve cords were stained by the del Rio Hortega silver carbonate technique (see, e.g., ref. 17), which stains microglia cell bodies, with results similar to those of Morgese et al. (9). Thus, nerve cords were pinned with glass pins to Sylgard 184 (Dow-Corning), fixed 4-28 days in 4% paraformaldehyde, and then postfixed overnight in formol/ammonium bromide. After washing, the tissue was impregnated 1-2 min in 1% aqueous silver carbonate, reduced 1 min in.4% paraformaldehyde, toned 3 sec in.2% gold chloride, fixed 3 min in 5% sodium thiosulfate, dehydrated, and mounted whole. RESULTS Stained Microglia Accumulate at Lesions in Magnesium. One step in tracking microglia was first to test whether the Hoechst dye used to stain nuclei of living cells and Mg2l used to block contractions interfered with microglia, accumulation at lesions. Some nerve cords were stained with Hoechst dye and connective segments were cut from the cords between ganglia. These Hoechst-stained segments of the connectives were then crushed, placed into culture medium containing 23 MM Mg2l for ':-18 hr to allow microglia to accumulate at lesions, and fixed for Feulgen staining (as in Fig. 1). Other nerve cords were prepared as for the tracking experiments (see Methods). These stained segments were placed in tissue culture medium for up to 24 hr to optimize Hoechst staining, then crushed and incubated in culture medium containing 23 MM Mg2l for an additional 18 hr. Other nerve cords were either stained with Hoechst dye and incubated in culture medium without added Mg2+ or were not stained but were incubated in elevated Mg2. Controls, not stained with Hoechst dye, were crushed and maintained in culture medium without added Mg2+. At least two cords were prepared for each treatment. Microglia accumulation was similar in the five treatment groups described above. Fig. 1 shows that microglia accumulated at the crush, reaching similar levels whether they had been stained with dye and treated with elevated Mg2+ or not, and that these accumulations resembled those seen by Morgese et al. (9). The high density of cells at the crush and variability in crush width made precise counts of cells difficult, but similar crushes had indistinguisihhable-accumuilations. In addition, qualitatively similar accumulations at the crush were observed whether or not the nerve cord had been Hoechst - Mg++ AV" Control crush crush FIG. 1. In nerve cords stained with Hoechst 338 dye and bathed in Mg2+, microglia accumulated at lesions. (Upper) Connectives were stained min with Hoechst dye, crushed, cultured 24 hr in L-15 medium containing 23 mm Mg2l, fixed, and Feulgen-stained. (Lower) Control connectives were not stained with Hoechst dye but were crushed, cultured 24 hr in L-15 with no additional Mg2+, fixed, and Feulgen-stained. (Bar = 2 j.&m.) exposed to the viewing light used during the tracking experiments described below. Tracking Microglia in Living Nerve Cords. Microglia began moving toward the crush within minutes after a lesion was made. Accumulation of microglia was seen 2 hr after crushing and increased, reaching peak levels within 24 hr. An example of sequential views of the connectives, photographed at 5-mmn intervals, is shown in Fig. 2. To assist in detecting moving cells, the image processor was used to subtract each image from the next recorded 5 min later, as shown in the bottom six panels of Fig. 2. Those portions of the image that changed appear black (before) and white (after). Continuous recordings on videotape confirmed the identity of particular cells. Only a fraction of cells moved at any time, and those cells moved at various speeds. For example, cell b moved toward the crush at a nearly steady speed from 15 min to 35 min, as indicated by the lengths of the arrows between the black and white images in the middle four panels of Fig. 2. After 35 min (panel 4' -35'), cell b paused but then began to move again (panel 45'-4'). In contrast, cell a moved at different speeds during the -mmn period of observation shown (note the varied lengths of arrows between the before and after images for cell a in Fig. 2). Some of the moving cells were moving within the viewing window at the start of the 2- to 6-hr observation period (cells a and b), others moved into the field of view and on toward the crush (cell c), and some that had been initially stationary began moving while under observation. Some microglia moved continuously; others stopped and started or occasionally reversed direction. Movement away from the crush was temporary, never exceeded the average dimensions of the microglia cell (see below and Fig. 3C), and in some cases accompanied a rare movement laterally within the connective. Few cells disappeared from the shallow plane of focus,

3 Neurobiology: McGlade-McCulloh et al. Proc. Natl. Acad. Sci. USA 86 (1989) 195 2r3-r -w J ;-~.a -:. - * w ' 6jr * c...j.- >* 4- a--.,nf 357*3w _r *.--a~~~~~~~~~to. A;. -ls Or..W.J' E >~ ~. ;!;tav> X - stt r k aj6)r<ze. t',x4.ft,_w " C --,. <e IIs V.5 -. V 4 * r %W ¼a -v.rmt. * FIG. 2. Migration of microglia within the living nerve cord. Panels are views at 5-min intervals of a.2-mm length of connectives crushed -.2 mm to the right ofthe region viewed. In two top panels, photographed 15 and 2 min after crushing, microglial and perineurial sheath nuclei are visible as bright fluorescent spots. Most of those microglia that moved traveled to the right, toward the crush (e.g., cells a and b). To highlight moving cells, the image recorded at one time (e.g., 15 min) was subtracted from that recorded 5 min later (2 min), creating a difference image (2' - 15'; third panel). In the difference image, the earlier positions of cells appear as dark spots and the later positions are white. Cells that did not move do not appear in the difference image. The six bottom panels show difference images taken at successive 5-min intervals. Three of the moving microglia are labeled a, b, and c, and their movements are marked with arrows. Although most cells of the perineurial sheath did not move, slight drift orjitter of connectives was in some cases detectable, as in the case of the cell marked with an asterisk. which would in effect have required lateral movement within the connectives. Nuclei of cells ofthe perineurial sheath lined the edges of the connectives and moved little. Shifting of perineurial sheath cells thus served to indicate any gross

4 196 Neurobiology: McGlade-McCulloh et al. movement of the nerve cord. In some experiments, the intensity of the excitation light was briefly increased 3-fold every 2 min to improve image quality and to determine whether additional stained cells were present but undetected with dim illumination. When such cells were seen, they were out of the plane of focus. In nerve cords that had not been crushed, microglia did not move. These controls were prepared as experimentals: the day before observation they were stained with Hoechst dye and connective segments were pinned straight. Two control segments were each observed for 2 hr under the usual dim illumination. No nucleus moved perceptibly relative to others in the connective, which included stationary cells of the perineurial sheath. The limit of detection in these experiments was a relative movement of 5,Am in 2 hr. Microglia that did move within crushed connectives were clocked at speeds up to 7,um/min measured over a 1-min interval (Fig. 3A). The average speed during periods of movement was 1.9 Aum/min with a range up to 4.9,um/min (Fig. 3B). Measurements were made at 1-min intervals for 56 cells in 1 preparations. The percentage of microglia moving at any time during the observation period varied as cells entered and left the field of view (see, e.g., cells c and a in Fig. 2) and varied from preparation to preparation, but typical values ranged from 15% to 4% of the microglia. The rest of the microglia did not move during observations up to 6 hr; movements >5 pam (i.e., <1 pum/hr) would have been detected. Cell movement en masse at <1 pum/hr could not have produced the accumulation seen in this and earlier studies (9). A speed more than an order of magnitude faster would have been required. Movement away from the crush, indicated by a negative number, did not exceed the average dimensions of the microglial cell (Fig. 3C; see below). The range of distances traveled by individual nuclei during periods of movement was -41 pum to >2 pam. Some microglial cells entered, traversed, and left the field of view during the period of observation. These cells traveled farther than the 22-,um length of connective under observation. In one case, no MgCI2 was added and the preparation was followed for 2 hr. Only a fraction of the microglia moved, traveling at speeds that were entirely within the range of those in elevated [Mg2"]. Muscular contractions of the connectives made analysis of the shifting nerve cord laborious but confirmed that the addition of Mg2" did not measurably affect microglial migration within the nerve cord, a conclusion first based on Feulgen-stained preparations. Microglial Dimensions and Microglial Movements. Because Hoechst dyes stain only nuclei and thus whole cells cannot be seen, any nuclear movements that are smaller than the dimensions of the cell could be nuclear movements within a stationary cell. It was therefore important to determine whether whole cells were moving. Measurements were made of 18 randomly selected whole microglia in preparations stained with silver carbonate to mark microglial cytoplasm. Their average length was pum (mean ± SD) and their width was <5 pum. Microglial nuclei were situated 4% ± 7% (mean ± SD) of the distance from an end of the cell, indicating that almost all nuclei lay within the central one-third of the cell. Cells that traveled backward all fell within this range; the maximum distance was -41 pum (Fig. 3C). Of course, those nuclei that traveled forward <41 pum could be similarly accounted for as nuclear movement within a stationary cell, but more than two-thirds of moving nuclei traveled farther than this toward the crush and most moved more than the typical cell's dimensions. The same results can be used in considering a converse proposition, that the cell could move within the limits of its dimensions without shifting of the nucleus. This could in theory occur for a few cells, but it would not account for the large accumulation of microglia at the crush. 2 t-j 15 1 c C (at ~% a 215 1, 15 % Peak velocity (pm/min) C. Proc. Natl. Acad Sci. USA 86 (1989) Average velocity (pim/min) 8 * cd I Io ~ ~ ~ s --.- o CP op Average velocity (um/min) FIG. 3. Distributions of velocities and distances traveled by moving cells. Fifty-six moving microglia were tracked at 1-min intervals in 1 preparations. Velocity was computed from the distance traveled during each interval and an average velocity of each cell was determined for all intervals during which the cell moved. (A) The peak velocity was binned at intervals of 1 Aum/min. (B) Distribution of average velocities of the cells, averaged over 1-min intervals in which they moved, not necessarily the total observation period. (C) The distance traveled is shown for each cell, represented by a circle. The length of connective under observation was always 22 Am; thus, only those cells that traversed the field of view were recorded as traveling as much as 22 t.m. Cells initially in the field that exited toward the crush may have traveled farther than indicated, but no cells moving away from the crush entered or left the field of view. DISCUSSION Morgese et al. (9) showed that microglia migrate within the leech nerve cord to crushes and do so within 24 hr of injury.

5 Neurobiology: McGlade-McCufloh et al. Their work raised questions about the route, velocities, and population of cells moving. The present study using low-light video microscopy confims the earlier results and answers several of these questions. Microglia near the lesion begin moving almost immediately toward it, but only a fraction (15-4%o) of the microglia move at any time, and they move at different speeds. Thus, the cells do not move en masse. The microglia travel directly down the connectives, apparently along axons, rather than moving laterally to the edge of the connectives, beneath the perineurial sheath, and then along the nerve cord to the lesion. Lateral movements, which would also have been detected as the spontaneous appearance or disappearance of cells from the focal plane, were rare. It is interesting that not all microglia moved at any time, because with suitable multiple crushes essentially all the microglia within a segment of connective can move to lesions within 1 day (9). Why only some cells respond at any one time to the stimulus and what determines their speeds remain unknown. In mammals and other vertebrates, there may be several sources for microglia, including those that arise endogenously within the brain and not from blood macrophages (1). It has not been possible to track microglial migration within the nervous system in vivo, but in cultures of dissociated brain cells, microglia are motile (3). While vertebrate microglia may divide, there is no evidence that the accumulation of microglia in the leech is due to cell division (K.J.M., unpublished results). Vertebrate and leech microglia are phagocytic and remove cellular debris after nerve injury (16, 18), but their functions during regeneration might extend farther. For example, similar macrophages have been proposed to provide regenerating axons with lipid and perhaps other molecules (19). When isolated leech neurons are plated in tissue culture, it is common for microglia to accompany the neurons and migrate onto the culture dish. Chiquet and Nicholls (2) observed that microglia seem to block neurite outgrowth and that growing neurites avoid the microglia, circumnavigating them. The apparent inhibition of axon growth may relate to an ability of the microglia to channel or direct the growth of axons through the lesion. To judge from their elongated Proc. Natl. Acad. Sci. USA 86 (1989) 197 nuclei, microglia within the lesion are oriented along the axis of the nerve, and they move into the lesion long before the regenerating axons. Perhaps microglia can also move to make way for axons traversing the lesion. We thank Xiaonan Gu, David McCulloh, and Steve Young for valuable discussions and Bob Keane for critical reading of the manuscript. This work was supported in part by National Institutes of Health Grants RO1-NS267 to K.J.M., F32-NS7488 to E.M.- M., and 5T32-NS Ling, E. A. (1981) in Advances in Cellular Neurobiology, eds. Fedoroff, S. & Hertz, L. (Academic, New York), Vol. 2, pp Giulian, D. & Baker, T. J. (1985) J. Cell Biol. 11, Giulian, D. & Baker, T. J. (1986) J. Neurosci. 6, Smith, P. J. S., Howes, E. A. & Treherne, J. E. (1987) J. Exp. Biol. 132, Hickey, W. F. & Kimura, H. (1988) Science 239, Nicholls, J. G. (1987) The Search for Connections: Studies of Regeneration in the Nervous System of the Leech (Sinauer, Sunderland, MA). 7. Coggeshall, R. E. & Fawcett, D. W. (1964) J. Neurophysiol. 27, Kai-Kai, M. A. & Pentreath, V. W. (1981) J. Comp. Neurol. 22, Morgese, V. J., Elliott, E. J. & Muller, K. J. (1983) Brain Res. 272, Streit, W. J. & Kreutzberg, G. W. (1987) J. Neurocytol. 16, Macagno, E. R. (198) J. Comp. Neurol. 19, Nicholls, J. G. & Baylor, D. A. (1968) J. Neurophysiol. 31, Ready, D. F. & Nicholls, J. (1979) Nature (London) 281, Wallace, B. G., Adal, M. N. & Nicholls, J. G. (1977) Proc. R. Soc. London Ser. B 199, Muller, K. J. & Carbonetto, S. (1979) J. Comp. Neurol. 185, Elliott, E. J. & Muller, K. J. (1983) J. Neurosci. 3, Humason, G. L. (1962) Animal Tissue Techniques (Freeman, San Francisco), p Elliott, E. J. & Muller, K. J. (1981) Brain Res. 218, Ignatius, M. J., Shooter, E. M., Pitas, R. E. & Mahley, R. W. (1987) Science 236, Chiquet, M. & Nicholls, J. G. (1987) J. Exp. Biol. 132,