Pergamon 0306-4522(94)00519-2

Neuroscience Vol. 65, No. 4, pp. 973 981, 1995 Elsevier Science Ltd

Copyright ~ 1995 IBRO Printed in Great Britain. All rights reserved

0306-4522/95 $9.50 + 0.00

EFFECTS OF ASTROCYTE IMPLANTATION INTO THE HEMISECTED ADULT RAT SPINAL CORD

J. J. W A N G , * M. I. C H U A H , t + + D. T. W. YEW,* P. C. L E U N G § and D. S. C. TSANGH

Departments of *Anatomy, §Orthopaedics and Traumatology, plBiochemistry, Chinese University of Hong Kong, Shatin, N.T., Hong Kong

tDepartment of Anatomy, University of Tasmania, Hobart, Tasmania 7001, Australia

Abstract--Morphological and biochemical methods were applied to assess the effects of implanting cultured astrocytes into the hemisected adult rat spinal cord. Astrocytes were purified from neonatal rat cortex and introduced into the lesioned spinal cord either in suspension injection or cultured on gelfoam first. The control groups were rats which had hemisection with injection of culture media or with gelfoam grafted alone. At various time points after surgery (two weeks to two months), the spinal cord was removed and processed for routine light microscopy, immunofluorescence, gel electrophoresis and immunoblotting. As early as two weeks after surgery, a significantly smaller volume of scar tissue was consistently found in the experimental groups. This reduced scarring was also confirmed by immuno- fluorescence staining and immunoblotting for glial fibrillary acidic protein in the specimens two months after hemisection. Compared to the control groups, the experimental groups also had more intense staining for neurofilaments, which was confirmed by immunoblotting. However, labelling of the astrocytes with Phaseolus t~ulgar& leucoagglutinin conjugated with fluorescein showed that the astrocytes migrated at a rate of 0.6 mm/day from the original implanted site.

The results therefore suggested that the cultured astrocytes probably exerted their effects over a short time period (less than two weeks) around the lesion site. They could have altered the microenvironment and as a result less scar tissue was formed. Hence, there was less barrier to the regrowth of nerve fibres.

h is often reported that lesions of tracts in the adult spinal cord (SC) result in limited axonal sprouting or slow atrophy and death of the affected neurons. :3"~4"25'2~ However, a few studies, j'~5"z8 i nclud- ing a recent one from our laboratory, 2° indicate that motor fibres in the lesioned adult SC in the rat, squirrel and monkey do undergo appreciable sprouting during recovery.

In recent years, cellular transplants have been increasingly used in the repair of injury in the CNS. j6'19'22'29 Reier and co-workers 26 pointed out that

fetal SC implants formed a bridge that restored some anatomical continuity between the isolated rostral and caudal stumps of the injured SC. The fetal implants also generally survive better than those of adult tissues, probably because of their inherently

;~To whom correspondence should be addressed. Abbreviations: EDTA, ethylenediaminetetra-acetate;

GFAP, glial fibrillary acidic protein; HA, hemisection with cultured astrocyte suspension injected at the time of lesion; HAG, hemisection with immediate implantation

• of gelfoam infiltrated with astrocytes; HBSS, Hank's balanced salt solution; HG, hemisection with implan- tation of gelfoam alone; HH, hemisection with injection of Hank's balanced salt solution only; NF, neuro- filament; PHAL, Phaseolus vulgaris leucoagglutinin; PHAL FITC, Phaseolus vulgaris leucoagglutinin conjugated with fluorescein; SC, spinal cord.

greater capacity for cell division, growth and differen- tiation. However, fetal tissue implants are hetero- geneous in their cellular composition and hence their effect on injured neural tissue cannot be easily resolved in terms of the functional role of each cell type in the implant.

With the advance of research techniques, it is now possible to obtain highly enriched populations of specific cell types. Hence, recent studies have explored the effects of implanting specific cells such as cultured astrocytes in lesion sites. Astrocytes purified from neonatal rat cortex have several advantages: they divide and grow well in tissue culture and can be obtained in large quantities easily. They synthesize and secrete growth factors that might prevent further neuronal degeneration after damage of the central nervous tissue. 2~ In view of this possibility, the objec- tive of this study is to compare the histological recovery of hemisected adult rat SC in the absence or presence of cultured neocortical astrocytes. Cultured astrocytes are introduced into the lesion site either as a cell suspension or as a pellet of gelfoam infiltrated with astrocytes. Following fixed periods of survival, light microscopic observations are made to determine the extent of scarring. Immunofluorescence staining and immunoblott ing for neurofilament proteins are performed to demonstrate neurite growth in the experimental and control spinal cords.

973

974 J. J. Wang et al.

EXPERIMENTAL PROCEDURES

Spinal cord hemisection of rat

Adult female Sprague Dawley rats weighing about 300g were anaesthetized with 2% Nembutal (50mg/kg) intraperitoneally and laminectomy was performed to expose the dorsal surface of the third segment of the lumbar SC. As described in our earlier paper, 2° the meninges of this area were removed and the SC was hemisected on the right side using a surgical scalpel blade (No. 11). Rats were divided into four groups: (i) hemisection with cultured astrocyte suspension injected at the time of lesion (HA); (ii) hemi- section with injection of Hank's balanced salt solution (HBSS; Sigma, St Louis, MO) only (HH); (iii) hemisection with immediate implantation of gelfoam infiltrated with astrocytes (HAG); (iv) hemisection with implantation of gelfoam alone (HG). HA and HAG were the experimental groups, while HH and HG served as controls.

Following SC hemisection, size 4-0 surgical silk was used to suture muscles and adjoining skin. Rats were allowed to survive either two weeks, one month or two months.

Preparation of purified cortical astrocytes

Astrocytes were purified using a modification of the McCarthy and deVellis 23 method, which has been described earlier. 5 Cells dissociated from newborn rat cerebral cortex were plated on to poly-L-lysine-coated tissue culture flasks. After nine to 10 days, the flasks were shaken overnight at 37°C to remove the top cells. The attached cells were then removed by trypsinization and replated at one-third the original density. They were then treated with 10-2mM cytosine arabinoside for seven days, after which they were usually 95 99% pure as indicated by immunostaining with anti-glial fibrillary acidic protein (GFAP) and Nuclear Yellow (0.001%; Hoechst).

To prepare a suspension for injection into the SC, astro- cytes were treated with 0.006% trypsin and centrifuged. The resuspended cells were transferred to a pipette to be introduced into the lesion site in 20 #1 of HBSS at 1.25 × l 0 4

cells//~l. Preparation of astrocyte cultures in gelfoam was

performed according to a modification of the method developed by Kesslak et al. 17 Pieces of sterile gelfoam (3 mm × 3 mm x 2 mm) were soaked in culture medium, which was Dulbecco's modified Eagle's medium (Gibco, Grand Island, NY) supplemented with 10% fetal calf serum (Gibco, Grand Island, NY), 1% penicillin streptomycin (Sigma, St Louis, MO) and 1% modified Eagle's medium-vitamin solution (Gibco, Grand Island, NY), and semi-dried on filter paper. Astrocytes were plated in a volume of 40 Ftl on to each piece of gelfoam at a concen- tration of 750 cells/#l. They were allowed to attach for 2 h, after which more culture medium was added to completely saturate the gelfoam. After three days culture, the gelfoam with astrocytes was implanted into the lesioned SC.

Scanning electron microscopy

Scanning electron microscopy was conducted to verify the presence of viable astrocytes on gelfoam prior to implan- tation into the hemisected SC. The gelfoam infiltrated with cultured astrocytes was fixed with 2.5% glutaraldehyde in 0.1 M phosphate buffer at 4°C for 30 min, and rinsed with two changes of distilled water for 10 min each. It was then postfixed with 1% osmium tetroxide at room temperature for 1 h, washed and dehydrated through a sequential 10-min washing in graded alcohols from 50% to 100%. Next the specimen was treated to sequential 10-min washes in mix- tures of 100% alcohol-Freon 113 in the ratios 4:1, 3 : 2, 2 : 3 and 1 : 4. After that, it was critical point dried in Freon 23, coated with gold-palladium and viewed under the JEOL JSM-35CF scanning electron microscope at 15 kV.

Phaseolus vulgaris leucoagglutinin labelling of astrocytes

In I0 rats, astrocytes tagged with a fluorescent label were implanted into the hemisected SC as a suspension or by culturing first on gelfoam. In this way, we could trace any migration path taken by the implanted astrocytes. The plant lectin Phaseolus vulgaris leucoagglutinin conjugated with fluorescein isothiocyanate (PHAL-FITC; Sigma, St Louis, MO) was used to label the cultured astrocytes. Phaseolus vulgaris leucoagglutinin (PHAL) is commonly used as a marker for graft-derived cells in previous studies and has been shown to be stable for up to three months post- implantationfl '3't2 In order to label the astrocytes, they were trypsinized from the flask and incubated at 37°C for 4 h by gentle agitation with PHAL FITC (8 pg/ml). Cell counts on the fluorescence microscope showed that about 90% of the cells could be labelled by this method.

Light microscopy, measurement of volume of scar tissue

Following a postoperative recovery period of two months, rats were anaesthetized intraperitoneally with 2% Nembutal (50 mg/kg) and perfused with 0.9% saline followed by 10% buffered formalin. A 2.0-cm-long segment of lumbar SC containing the lesion site was dissected and fixed overnight before dehydration in an ascending series of ethanols. The specimen was cleared in xylene, embedded in paraffin and serial sectioned at 7/~m thickness. After de- paraffinization, the sections were stained with haematoxylin and eosin.

The volume of scar tissue at the lesion site was measured by the following steps. Beginning from the first section that contained scar tissue and taking every 20th section there- after, the area of the scarred region in these sections was determined. This was done by camera lucida tracing of the relevant region (e.g., Fig. 1) and then determining the area with a sonic digitizer (Science Accessories Corporation, Southport, CT) attached to an IBM personal computer. The total volume of scar tissue was calculated using the following formula: volume = total area measured in a specimen × 7 x 20.

lmmunofluorescence staining

Rats which had undergone two weeks to two months recovery following hemisection were anaesthetized with 2% Nembutal and perfused with 0.9% saline, followed by 4% paraformaldehyde. A 2-cm segment of the lumbar SC which contained the lesion area was dissected and stored in 4% paraformaldehyde at 4°C overnight. Prior to sectioning, the segment of SC was frozen in liquid nitrogen and 10-/~m- thick longitudinal sections were cut on a cryomicrotome (Reichert-Jung 1800) at -22'~C. They were then mounted on dry gelatin coated slides and allowed to dry.

Polyclonal rabbit anti-GFAP (dilution 1:100; Sigma, St Louis, MO) and monoclonal mouse anti-200,000 tool. wt neurofilament (NF) proteins (dilution 1:40; Sigma, St Louis, MO) were the primary antibodies used. The secondary antibodies were fluorescein conjugated goat anti-rabbit immunoglobulin (dilution 1:100; Jackson Laboratories, West Grove, PA) and goat anti-mouse immunoglobulin (dilution 1:100; Jackson Laboratories, West Grove, PA), respectively. The method for immunofluorescence staining has been described previously. 4 Control sections were stained in the absence of primary antibodies.

Gel electrophoresis and immunoblotting

Gel electrophoresis and protein transfer to a nitro- cellulose membrane were performed according to the instruction manual of the Bio-Rad Modular Mini Electro- phoresis System. Rats from the HA and HH groups which were allowed to survive for two months after hemisection and unoperated rats were used. A 0.5-cm segment of the lumbar SC was dissected out and homogenized in Triton X-100 lysis buffer (10mM NaHzPO4, l l0mM ..NaC1, 50mM EDTA, 0.5% Triton X-100, pH6.5) in the

Astrocyte implantation in

presence of protease inhibitors (20 IU/I aprotinin, 5 mM benzamidine, 5 mM iodoacetamide and l mM phenyl- methanesulphonyl fluoride). Following centrifugation at 12,000g for 15 rain, the supernatant was mixed with the sample buffer (62.5 mM Tris HC1, pH 6.8, containing 10% glycerol, 2% sodium dodecyl sulphate, 5% fl-mercapto- ethanol and 0.05% Bromophenol Blue) in the same volume and heated at 95°C for 10 min. Samples were loaded at 15 lLl to each well at a concentration of 0.26mg/#l. Sodium dodecyl sulphate--polyacrylamide gel electrophoresis stan- dards ranging in molecular weight from 49,000 to 205,000 (Bio-Rad, Richmond, CA) were also applied to the 10% acrylamide gels.

After etectrophoresis, proteins were transferred from gel to nitrocellulose membrane in transblot buffer (25 mM Tris-base, 192 mM glycine and 20% methanol). 31 The membrane was cut into strips corresponding to the different lanes. The strips were incubated overnight at 4 'C with monoclonal mouse anti-GFAP (diluted 1:400), anti-68,000mol, wt NF subunit (diluted 1:100) or anti- 160,000 mol. wt NF subunit (diluted I:100). All primary

hemisected rat spinal cord 975

antibodies were purchased from Sigma, St Louis, MO. Following this, the strips were rinsed with three changes of 0.1% Tween 20 in Tris-buffered saline (100 mM Tris, 0.9% sodium chloride, pH 7.5), and transferred to the solution of the biotinylated secondary antibody (Vector, Burlingame, CA). Subsequent steps were performed according to the instruction manual of the Vectastain ABC kit. Controls consisted of incubating the nitrocellulose strips in 0.1% Tween 20-Tris-buffered saline alone, in the absence of primary antibodies. Immunoblots were scanned by a Personal Densitometer (Molecular Dynamics, CA).

RESULTS

Scanning electron microscopy

Scanning electron microscopy verified that at the time of implan ta t ion into the hemisected SC, the pieces of gelfoam conta ined viable astrocytes. Some were spherical, while others were flat and polygonal

Fig. 1. Hemisected spinal cord, implanted with gelfoam alone. The dotted line demarcates the boundary of the scar tissue. Scale bar = 12 #m.

Fig. 2. Scanning electron micrograph showing the presence of cultured astrocytes on gelfoam four days after plating. Some are spherical (arrowheads), while others are flat (arrow) in shape. Scale bar = 5 #m.

976 J . J . Wang et al.

90

80

70 o

60 e~ .o g

50

40

~, 30

2o o

to

[ ] HA

• HH

[ ] HAG

[ ] HG

n=12 • =P < 0.05

2 weeks 4 weeks 8 weeks

Time after operation Fig. 3. Change in the volume of scar tissue in the hemisected spinal cord in the experimental and control groups from two

weeks to two months after surgery.

in shape (Fig. 2). If the astrocytes had not survived on the gelfoam, they would have detached and would be washed away during processing.

Determination o f volume of scar tissue

The volume of scar tissue in the various groups at different times postsurgery is shown in Fig. 3. In both the experimental (HA, H A G ) and control (HH, HG) groups the volume of scar tissue decreased progress- ively from two weeks to two months postlesion. There is a 20-fold decrease in the H A and H H groups, while a 14-fold decrease is observed in the H A G and H G groups. This means that the corresponding experimental and control groups demonstrated simi- lar rates of recovery in terms of extent of scarring. It should be noted that as early as two weeks posthemisection, the volume of scar tissue in H A and H A G groups was already significantly reduced, com- pared with their respective controls. Therefore, the evidence indicates that the implanted astrocytes had little influence on the rate of reduction in scar tissue from two weeks to two months postlesion. However,

they were crucial in limiting tissue scarring within the first two weeks.

Immunofluorescence staining

Immunoreactivity for G F A P is normally found in astrocytes and the degree of staining often reflects the extent of astrogliosis in the tissue. In the control group (HH) of our study, intense staining for GFAP, particularly around cavities, was found at two weeks postsurgery (Fig. 4). Elsewhere diffuse staining was present. In comparison, the experimental group (HA) appeared to show less intense staining for GFAP. In the gelfoam-implanted groups (HG and HAG), a similar pattern of comparatively reduced staining for G F A P in the experimental group was also obtained (Figs 5, 6). This was somewhat surprising because the H A G group not only contained cultured astrocytes, which were GFAP-posi t ive, but also had more ingrowth of fibrous looking tissues into the gelfoam spaces after two weeks. The staining pattern suggests that much of the tissue that had grown into the gelfoam was not astrocytic gliosis.

Two months after surgery, staining intensity for G F A P around the lesion site was reduced in both HA and H H groups, and it was difficult to distinguish one group from the other. The same can also be said for the H G and H A G groups, in which G F A P became localized within the remnants of the gelfoam and a narrow rim of SC tissue around it.

Immunoreactivi ty for N F in SC segments two weeks after hemisection showed that the experimental groups (HA and H A G ) generally had more NF-posit ive fibres than the control groups (HH and HG). The fibres were irregular and appeared as a meshwork in the region close to the lesion (Fig. 7). More distantly, especially caudal to the lesion, the fibres were more regularly arranged as fascicles. The presence of NF-posit ive fibres around the original lesion site could be observed one month and two months after surgery in all groups. The nerve fibres were usually more numerous in the experimental groups (HA, HAG).

Phaseolus vulgaris leucoagglutinin-labelled astroo'tes

Observations on the location of PHAL-label led astrocytes were made one and two weeks following

Fig. 4. Control hemisected spinal cord injected only with HBSS, two weeks after lesion. Immuno- fluorescent staining for GFAP at the lesion site is most prominent around the cavities (arrowheads).

Scale bar = 30 pm.

Fig. 5. Hemisected spinal cord implanted with astrocytes cultured on gelfoam. Two weeks after lesion, there is abundant growth of tissue into the gelfoam spaces, while immunofluorescent staining for GFAP shows only moderate intensity. The dotted line marks the boundary between the gelfoam on the right and

the intact spinal cord tissue on the left. Scale bar = 30/tm.

Fig. 6. Hemisected spinal cord two weeks following hemisection and implantation with gelfoam alone. Compared to the specimen shown in Fig. 5, there appears to be less ingrowth of tissue into the gelfoam and staining for GFAP is more intense. The dotted line marks the boundary between the gelfoam on the

right and the intact spinal cord tissue on the left. Scale bar = 30 #m.

Astrocyte implantation in hemisected rat spinal cord 977

Figs 4 6.

Fig. 7. Hemisected spinal cord injected with suspension of astrocytes. Two weeks after surgical manipulation, growth of irregularly arranged nerve fibres (arrowheads) into the scar tissue (S) can be observed. Brightly staining NF-positive nerve bundles are located on the intact contralateral side (C).

Scale bar = 30 #m.

Fig. 8. GFAP immunoblotting two n~onths after surgical manipulation. (A) Hemisection with injection of astrocyte suspension. (B) Hemisection with injection of HBSS alone.

Fig. 9. Immunoblotting for neurofilament subuuits, two months after surgical manipulation. (a) NF 160,000 in hemisected spinal cord with injection of astrocyte suspension. (b) NF 68,000 in hemisected spinal cord with injection of astrocyte suspension. (c, d) Control staining performed with mouse ascites fluid in place of specific primary antibodies. (e) NF 68,000 in hemisected spinal cord with injection of

HBSS alone. (f) NF 160,000 in hemisected spinal cord with injection of HBSS alone.

978

Astrocyte implantation in hemisected rat spinal cord 979

implantation. In the group in which PHAL-labelled astrocytes were cultured on gelfoam and then implanted, the astrocytes migrated rostrally and caudally out of the gelfoam and were 4.0 mm into the SC one week after introduction. Usually, more PHAL-labelled astrocytes were present in the region of the SC caudal to the lesion. In subsequent weeks, they migrated at about the same rate (0.6 mm/day) further away from the site of implantation. The astrocytes did not migrate across to the contralateral unlesioned side. In specimens where PHAL-labelled astrocytes were introduced as a cell suspension, by one week the cells had migrated more than 4.0 mm away from the injection site. Hence, the results suggest that migration of astrocytes in SC tissue is retarded to some extent if they are attached to a gelfoam substrate prior to implantation.

Gel electrophoresis and immunoblotting

Distribution of the protein bands from the HA and HH groups was similar to that from the normal SC, suggesting that there was no change in the expression of various proteins following hemisection, regardless of the presence or absence of cultured astrocytes. However, immunoblotting and densito- metry measurements showed that there was a comparatively lower quantity of GFAP in the HA group, i.e. 37% of control value (Fig. 8). Since GFAP is expressed by astrocytes, the lower value in the HA group suggests that there is comparatively less reactive gliosis. Immunoblotting against NF 68,000 and 160,000 showed that they were more abundant in the HA than the HH group (Fig. 9). Densitometry measurements confirmed that the experimental group had 1.55 and 1.25 times as much NF 68,000 and NF 160,000 respectively as the control group. These results thus indicate that there was more nerve fibre growth in the lesioned SC implanted with cultured astrocytes than in those without.

DISCUSSION

This morphological and biochemical study describes the response of the adult rat SC following hemisection and implantation of cultured astrocytes. Rats in the experimental groups had cultured astrocytes introduced into the lesion site either in the form of suspension injection (HA) or they were first allowed to attach on gelfoam (HAG). Volume measurements showed that the amount of scar tissue in the experimental groups (HA, HAG) was always less than that of the control (HH, HG) between two weeks and two months postsurgery. Immuno- fluorescence staining of SC tissue around the lesion site showed that the experimental groups had more NF-positive fibres with comparatively less intense staining for GFAP. This observation was confirmed by immunoblotting.

One of the findings in this study is that tissue scarring in experimental animals was consistently less

than that of control and the discrepancy was already apparent two weeks after hemisection. The similar rate of scar reduction in all groups suggests that the implanted cultured astrocytes probably had little direct effect in promoting recovery of injured tissue over a prolonged period. This notion is also supported by the finding that the implanted astro- cytes migrated quite extensively in the SC, at a rate of about 0.6mm/day. This rate is close to that obtained by Goldberg and Bernstein, 12 who showed that PHAL-labelled astrocytes migrated at a rate of 0.72 mm/day in the adult rat thoracic spinal cord. The results thus suggest that the effect of the implanted astrocytes probably occurs over a short time course, perhaps within a few days, soon after introduction into the lesion. Its influence over this short period of time, however, is sufficient to result in the decreased tissue scarring observed at two weeks posthemisection.

Evidence from several previous studies appears to support this notion. It has been shown that astrocytes and microglia, which are stimulated to become phagocytes during injury, release factors which act on each other as well as neighbouring neurons. 9'~ It is within the first week of spinal cord injury that there is maximal infiltration of phagocytes and removal of cell debris. 6 Astrocytes produce neurotrophic factors such as basic fibroblast growth factor and ciliary neurotrophic factor, 7"27 while microglia produce a variety of cytokines and degradative enzymes. 24 In addition to producing substances with opposing actions, they also secrete mitogens to stimulate each other's division.~° However, in spite of these divergent effects, a recent study has shown that astrocyte- released factors, if present at a sufficiently high concentration, can counteract the neurotoxic effects of microglia. 32 It is possible that in this study the implanted astrocytes played a positive role in counteracting the deleterious effects of microglia and thus reduced the amount of degeneration and scarring.

Immunofluorescence staining for NFs demon- strated that most of the neurite growth was found in a region of the SC caudal to the lesion. It is likely that these neurites are ascending sensory fibres which are attempting to regenerate. Support for this notion can be found in the results of a study by Martin et al. 22 They demonstrated that cultured Schwann cells implanted into a com- pression lesion of the adult rat SC promoted axonal regrowth mostly from dorsal root ganglion cells, not from descending fibres, it could be that sensory fibres intrinsically have a greater potential to regenerate than descending motor fibres. On the other hand, one cannot rule out the possibility that the preferred caudal migratory route of introduced astrocytes can alter the microenvironment such that the region of the SC immediately caudal to the lesion becomes more permissive to neurite growth.

980 J . J . Wang et al.

The secretion of neuro t rophic factors f rom cultured astrocytes and their ability to p romote neuri te growth has been well documented in previous in vitro studies. 7'2j'27 Fur the rmore , Geisert and

Stewart ~ found tha t astrocytes purified from neonata l rat cortex elicited longer neurites f rom neurons than those purified from the adult rat cortex. These in vitro

studies show a direct neuro t rophic effect of the cultured astrocytes. There is also some evidence of the growth p romot ing effect of astrocytes on axons in vivo. Smith and co-workers 3° t ransplan ted astrocyte-coated Mill ipore filters into the brains of acallosal mice and found tha t the t ransplan ts reduced glial scarring and p romoted axonal regenerat ion.

A l though our study has utilized cultured astrocytes from neonata l rat cortex, it is uncer ta in whether the relatively increased neurite growth in the exper-

imental rats, as evidenced by immunofluorescence and immunoblo t t ing , can be directly a t t r ibuted to the neuro t rophic effects of the implanted astrocytes. One major reason for this doub t is the finding that the implanted astrocytes migrated from the original site of injection and also the gelfoam. Perhaps a more convincing explanat ion is tha t the experimental rats had less tissue scarring, evident as early as two weeks postlesion, and this represents less o f ' a barr ier to neurite growth. Hence, an impor t an t subject for fur ther invest igation would be the effect of cultured astrocytes in the lesion site within the first two weeks following implanta t ion .

Acknowledgements--The authors thank Ms C. Au and Mr S. Wong for technical and photographic assistance, respectively. This project was supported by UPGC Grant 221400120.

REFERENCES

1. Aoki M., Fujito Y., Satoni H., Kurosawa Y. and Kasaba T. (1986) The possible role of collateral sprouting in the functional restitution of corticospinal connections after spinal hemisection. Neurosci. Res. 3, 617 627.

2. Bernstein J. J. and Goldberg W. J. (1989) Rapid migration of grafted cortical astrocytes from suspension grafts placed in host thoracic spinal cord. Brain Res. 491, 205 211.

3. Bernstein J. J. and Goldberg W. J. (1991) Grafted fetal astrocyte migration can prevent host neuronal atrophy: comparison of astrocytes from cultures and whole piece donors. Rest. Neurol. Neurosci. 2, 261-270.

4. Chuah M. I. and Au C. (1992) Neural cell adhesion molecules are present in the fetal human primary olfactory pathway. Devl Neurosci. 14, 357 361.

5. Chuah M. I., David S. and Blaschuk O. (1991) Differentiation and survival of rat olfactory epithelial neurons in dissociated cell culture. Devl Brain Res. 60, 123 132.

6. Curtis R., Green D., Lindsay R. M. and Wilkin G. P. (1993) Up-regulation of GAP-43 and growth of axons in rat spinal cord after compression injury. J. Neurocytol. 22, 51-64.

7. Ferrara N., Ousley F. and Gospodarowicz D. (1988) Bovine brain astrocytes express basic fibroblast growth factor, a neurotropic and angiogenic mitogen. Brain Res. 462, 223 232.

8. Geisert E. E. Jr and Stewart A. M. (1991) Changing interaction between astrocytes and neurons during CNS maturation. Devl Biol. 143, 335 345.

9. Giulian D. (1987) Ameboid microglia as effectors of inflammation in the central nervous system. J. Neurosci. Res. 18, 155 171.

10. Giulian D. and lngeman J. M. (1988) Colony-stimulating factors as promoters of ameboid microglia. J. Neurosci. 8, 4707~,717.

11. Giulian D., Vaca K. and Corpuz M. (1993) Brain glia release factors with opposing actions upon neuronal survival. J. Neurosci. 13, 29 37.

12. Goldberg W. J. and Bernstein J. J. (1988) Migration of cultured fetal spinal cord astrocytes into adult host cervical cord and medulla following transplantation into thoracic spinal cord. J. Neurosci. Res. 19, 3442.

13. Goldberger M. E. and Murray M. (1985) Recovery of function and anatomical plasticity after damage to the adult and neonatal spinal cord. In Synaptic Plasticity (ed. Cotman C. W.), pp. 77-110. Guilford Press, New York.

14. Goldberger M. E. and Murray M. (1988) Patterns of sprouting and implications for recovery of function. In Functional Recovery in Neurological Disease (ed. Waxman S. G.), pp. 361 385. Raven Press, New York.

15. Guth L., Barrett C. P., Donati E. J., Deshpande S. S. and Albuquerque E. X. (1981) Histopathological reactions and axonal regeneration in the transected spinal cord of hibernating squirrels. J. Neurol. 203, 297 308.

16. Houle J. (1992) The structural integrity of glial scar tissue associated with a chronic spinal cord lesion can be altered by transplanted fetal spinal cord tissue. J. Neurosci. Res. 31, 120 130.

17. Kesslak J. P., Nieto-Sampedro M., Globus J. and Cotman C. W. (1986) Transplantation of purified astrocytes promote behavioral recovery after frontal cortex ablation. Expl Neurol. 92, 377 390.

18. Kucera P. and Weisendanger M. (1985) Do ipsilateral corticospinal fibers participate in the functional recovery following unilateral pyramidal lesions in monkeys? Brain Res. 348, 297 303.

19. Kuhlengel K. R., Bunge M. B., Bunge R. P. and Burton H. (1990) Implantation of cultured sensory neurons and Schwann cells into lesioned neonatal rat spinal cord. II. Implant characteristics and examination of corticospinal tract growth. J. comp. Neurol. 293, 74-91.

20. Li W. W. Y., Yew D. T. W., Chuah M. 1., Leung P. C. and Tsang D. S. C. (1994) Axonal sprouting in the hemisected adult rat spinal cord. Neuroscience 61, 133 139.

21. Martin D. L. (1992) Synthesis and release of neuroaetive substance by glial cells. Glia 5, 81 94. 22. Martin D., Schoenen J., Delree P., Leprince P., Rogister B. and Moonen G. (1991) Grafts of syngenic cultured adult

dorsal root ganglion-derived Schwann cells to the injured spinal cord of adult rats: preliminary morphological studies. Neurosci. Lett. 124, 44~,8.

23. McCarthy K. D. and deVellis J. (1980) Preparation of separate astroglial and oligodendroglial cell cultures from rat cerebral tissue. J. Cell Biol. 85, 890 902.

24. North R. J. (1978) Concept of activated macrophage. J. lmmun. 121, 806 809.

Astrocyte implantation in hemisected rat spinal cord 981

25. Prendergast J. and Misantone L. F. (1980) Sprouting by tracts descending from the midbrain to the spinal cord: the result of thoracic funiculotomy in the newborn, 21 day old, and adult rat. Expl Neurol. 69, 458~480.

26. Reier P. J., Bregman B. S. and Wujek J. R. (1986) Intraspinal transplantation of embryonic spinal cord tissue in neonatal and adult rats. J. comp. Neurol. 247, 275 296.

27. Rudge J. S., Manthorpe, M. and Varon S. (1985) The output of neuronotrophic and neurite-promoting agents from rat brain astroglial cells: a microculture method for screening potential regulatory molecules. Det4 Brabt Res. 19, 161-172.

28. Shamboul K. M. (1979) Retrograde axon reaction in the lateral vestibular nucleus of neonatal and adult rats. J. Anat. 129, 225 233.

29. Smith G. M. and Miller R. H. (1991) Immature type-I astrocytes suppress glial scar formation, are motile and interact with blood vessels. Brain Res. 543, li1 122.

30. Smith G. M., Miller R. H. and Silver J. (1986) Changing role of forebrain astrocytes during development, regenerative failure, and induced regeneration upon transplantation. J. comp. Neurol. 251, 23 43.

31. Towbin H., Staehlin T. and Gordon J. (1979) Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some applications. Proc. natn. Acad. Sci. U.S.A. 76, 43504354.

32. Vaca K. and Wendt E. (1992) Divergent effects of astroglial and microglial secretions on neuron growth and survival. Expl ~eurol. 118, 62 72.

(Accepted 7 October 1994)