Fibronectin is expressed by astrocytes cultured from embryonic and early postnatal rat brain

Experimental

Cell Research

163 (1986) 175-185

Fibronectin is Expressed by Astrocytes Cultured Embryonic and Early Postnatal Rat Brain PAIVI

LIESI,‘,

** THOMAS

KIRKWOOD

and ANTTI

from

VAHERI’

‘Department of Virology, University of Helsinki, SF-00290 Helsinki, Finland ‘Computing Laboratory, National Institute for Medical Research, Mill Hill, London NW7 IAA, UK

and

In early primary cultures from newborn rat brain, few glial tibrillary acidic protein (GFAP)-positive gliaf cells expressed intracytoplasmic immunoreactivity for fibronectin. After the second week in culture, however, fibronectin was expressed by a distinct population of GFAP-positive flat astrocytes, irrespective of which brain region was studied. In cerebellar cultures, these cells were more abundant than in cortical or neostriatal cultures and often formed a major population of the GFAP-positive cells. The difference in tibronectin expression between cerebellum and the other areas studied was statistically significant. When cultures were started from 9-day-old postnatal rat brain, fibronectinpositive astrocytes appeared earlier than in those from newborn animals, in all areas studied. Further, especially in the case of cerebellum, the number of tibronectin-positive astrocytes increased as a function of time in culture. In cultures started from whole brains of 12-day-old rat embryos, tibronectin was expressed within 24 h in culture by all the cells with morphology of flat astrocytes, positive for vimentin but negative for GFAP. These results indicate that astrocytes cultured from newborn and early postnatal rat brain are a heterogeneous population of cells: depending on the brain region studied and also depending on the age of brain tissue or the time in culture,
Fibronectin is one of the major glycoproteins of the interstitial matrix and is produced by a variety of cell types in vivo and in vitro H-31. It has been associated with malignant transformation [l-3], with regenerative processes outside the central nervous system [4], and with the attachment of a variety of cell types to culture surfaces [5]. Matrix glycoproteins are of interest in studies on nervous tissue as possible candidates in mediating cell-to-cell interactions and celI migration. Indeed, neural crest cells migrate in a matrix containing fibronectin [6] and under culture conditions they preferentially chose fibronectin for their attachment and migration [7]. Primary sensory neurons show enhanced outgrowth of neurites on fibronectin-coated or laminin-coated surfaces [8]. Recent results on brain cells in vitro emphasize a specific role for laminin in neurite outgrowth and in survival of neurons [9-111. Furthermore, laminin has been * To whom offprint requests should be sent. ** Present address: Laboratory of Gene Structure and Expression, National Institute for Medical Research, The Ridgeway, Mill Hill, London NW7 lAA, UK. 12-868333

Copyright @ 1986 by Academic Press, Inc. All rights of reproduction in any form reserved 0014-4827/86 $03.00

176

Liesi, Kirkwood

and Vaheri

shown to be produced in astrocytes in vitro [ 121 and in vivo after brain injury [ 131. No tibronectin has been demonstrated in neuroectodermal cells in embryonic or newborn rat brain [14]. Nor is it expressed by glial cells in normal or injured rat brain [13]. Fibronectin has not been found in short-term cultures of glial cells from whole rat brain [ 15, 161, although fibronectin has been shown to be expressed by cells of certain glial cell strains and by glioma cells in vitro [17, 181. In the present study we show that tibronectin expression in primary astrocytes is a function of the age of the brain tissue used to initiate cultures and of the time in vitro and that it also varies between different brain regions. Further, we show that fibronectin is expressed by early embryonic rat brain cells in vitro. These results further emphasize that astrocytes are a heterogeneous population of cells and suggest that fibronectin may be involved in brain development. MATERIALS

AND

METHODS

Fibronectin was purified as described [19] and fibronectin-free fetal bovine serum (FN-FBS) prepared using a gelatin-Sepharose column 1201. FBS contains 100 &ml of fibronectin of which the treatment removed ~98 %; accordingly, the remaining concentrations of tibronectin in 20 % FBS-FN is maximally 0.4 &ml.

Preparation

of Cultures

Primary cultures from newborn or 9-day-old postnatal rat brain (neostriatum, cerebellum and cerebral cortex) were prepared as described [12]. Meninges were carefully removed to avoid contamination of brain cell cultures. The cells were fed with Eagle’s minimum essential medium (MEM) supplemented with 20% FBS with or without tibronectin. Cultures from 9-day-old embryonic rat brain were prepared as described [12] but no attempt was made to separate meninges. Meningeal cells were prepared by dissociating the meninges through a 200-urn nylon mesh and were cultured in the above medium without tibronectin. Medium of confluent meningeal cell cultures, used as such, after a 24 h conditioning with the cells and extracts of the cells (1% w/v01 in the medium), after repeated freezing and thawing, were used in some of the experiments.

Immunocytochemistry

for Fibronectin,

GFAP, Neurofiaments,

and Vimentin

The cultures were fixed with 3.5 % paraformaldehyde in phosphate-buffered saline (PBS) permeabilized, washed in PBS and double-stained for fibronectin and GFAP in the following sequence: (1) rabbit anti-human fibronectin (for specificity see ref. [21]) diluted 1: 200 for 1 h at room temperature; (2) sheep anti-rabbit immunoglobulins conjugated with fluorescein isothiocyanate (FITC; Wellcome, Beckenham, Kent, England) diluted 1: 20 for 30 min at room temperature; (3) mouse anti-GFAP (a kind gift from Dr D. Dahl; ref. [22]) diluted 1: 50, for 1 h at room temperature; [4] sheep anti-mouse immunoglobulins coupled to tetramethylrhodamine isothiocyanate (TBITC; Cappel Laboratories, Cochranville, Pa.). The cultures were viewed with a Leitz Dialux 20 EB microscope with appropriate filter combinations. The immunocytochemial specificity controls were as described [12]. In some experiments the double staining for libronectin and GFAP was performed using two rabbit primary antisera as described [12]. The immunostaining for vimentin was performed using rabbit antibodies diluted 1: 20 for 1 h at room temperature. The rabbit anti-neurofilament triplet protein and antivimentin were kind gifts from Drs D. Dahl and Richard Hynes, respectively, and their specificities confirmed [23, 241.

Preparation

of Pericellular

Fibronectin

Matrices

Fibronectin matrices of cultured brain cells were prepared as described [25]. In short, the cell monolayers were washed with PBS, incubated in 10 mM Bis-HCl, pH 8.0, containing 0.5% deoxycholate and 1 mM phenylmethylsulfonyl fluoride (PMSF) with three changes for 10 mm each on a Exp Cell

Res 163 (1986)

Fibronectin

in rat astrocytes

177

slowly moving four-way shaker. Then they were incubated in 2 mM Tris buffer containing 1 mM PMSF with three changes and fixed with 3.5% paraformaldehyde in PBS and immunostained for fibronectin.

Immunocytochemistry Rat Brain In Vivo

for Fibronectin

of Embryonic

and Early Postnatal

The whole brains of 12-day-old rat embryos were immersion-fixed with 3.5% pamformaldehyde in PBS for 4 h. Brains of newborn, 9-day-old, and 11-day-old postnatal rats were cut into small pieces and immersion tixed in the same fixative for 6-8 h. After fixation, the tissue was rinsed in PBS for 2W8 h and cut in 6 pm cryostat sections which were then stained for tibronectin. For tissue sections the immunostaining was performed using overnight incubation with rabbit antiserum against human fibronectin diluted 1 : 2 000. The specificity of the immunostaining was controlled by using antibodies preabsorbed with human fibronectin (200 ug/ml).

Quantitation

of Fibronectin-positive

Astrocytes

Cultures double-stained for fibronectin (FITC) and GFAP (TRITC) were used. The number of GFAP-positive glial cells per a microscopic field were counted using high magnification. Then the filter combination was changed and the fibronectin-positive, GFAP-positive glial cells counted. Per each coverslip ten different microscopic fields were counted, selected using TRITC-filter system so that each field contained a good number of GFAP-positive flat astrocytes.

Methods

of Statistical

Analysis

The data analysed were the percentages of GFAP+ glial cells which were also FN+. These were compared among groups of replicate experiments where the factors which differed were (8 region of the brain; (it) age of the animal; (iin number of days in vitro (DIV); (iv) presence or absence of FN in the culture medium. Initially, the data from replicate experiments with the same combinations of the above factors were examined by X*-test to assess whether they might be pooled to give a single percentage figure for each group. These tests showed significant heterogeneity of the percentages between replicate experiments, however, so the data were not pooled. To compare groups of experiments the percentages (P%) of the GFAP+ glial cells which were also FN+ were first transformed using the angular transformation y=sin -’ P/l00 [26]. The reason for this was to correct for the special problems of percentage data (i.e., non-normality of distribution, nonuniformity of variance) in the subsequent analyses. The transformed (y) values and the means and standard deviations for each group of experiments were calculated together with the equivalent values transformed back into percentages. Tests of the statistical significance of differences between groups were made by r-test or one-way analysis of variance, depending on whether two or three groups, respectively, were compared.

RESULTS Neurons ments and GFAP. In 30-50% of

in the cultures were identified by their immunoreactive neurofilacharacteristic morphology and astrocytes by their immunoreactive early cultures independent of the brain region neurons comprised all cells and 80% of the non-neuronal cells were positive for GFAP.

Cerebellum During the first 3 days, fibronectin was expressed (table 1). At the same time form. By 11 days in vitro

in primary cultures from newborn rat cerebellum, by a minor population of GFAP-positive glial cells fibronectin was also deposited in pericellular matrix the number of fibronectin-positive astrocytes had Exp CellRes

163 (1986)

178

Liesi, Kirkwood

Table 1. Fibronectin

and Vaheri in GFAP-positive DIV

Age

glial cells Medium FN+/-

Cerebellum A Newborn

3

-

B Newborn

11

-

C 9-day

postnatal

D Pday

postnatal

+

E Newborn

Cerebral cortex F Newborn

G 9-day

postnatal

H Newborn

3

3

-

11

postnatal

K Newborn

L Newborn

+

of also FN+

l/232 (0.4) l/188 (0.5) l/181 (0.6) 7/210 (3.3) 7/174 (4.0) 13/185 (7.0) 1 l/134 (8.2) 49/l 15 (43) 91/167 (54) 150/246 (61) 86/124 (69) 18/232 (7.8) 161169 (9.5) 33/184 (18) 49/126 (39) 63/148 (43) 53/120 (44) 92/138 (67) l/91 (1.1) 8/178 (1.7) 13/185 (7.0) 6162 (9.7) 8/153 81147 1 l/201 131203

Neostriatum Z Newborn

J 9-day

Fraction GFAP+

(5.2) (5.4) (5.5) (6.4)

lo/125 (8.0) 13008 (12) 20/147 (14) 20/111 (18) O/127 (0.0) 51167 (3.0) 6/157 (3.8) 8153 (15)

l/132 (0.8) 5/204 (2.5) 41103 (3.9) 18/356 (5.1) 23/160 (14) 46f228 (20) 431190 (23) 5/213 (2.3) 7/131 (5.3) 91134 (6.7) 17/142 (12) o/64 (0.0) 2B9 (2.0) 15llO5 (14)

(%)

Fibronectin

in rat astrocytes

179

increased and on some occasions these cells formed more than 60% of the total glial cell population (table 1). Fibronectin was expressed by astrocytes having flat morphology in the form of granular intracytoplasmic immunoreactivity confined to the perinuclear region (fig. 1). Simultaneously, an extensive fibronectin matrix was formed (fig. 2). This matrix was mainly confined to the top of the nonneuronal cells as confirmed by matrix isolation experiments; when the cells had been detached, only occasional fibrillar matrix deposits were seen facing the glass coverslip (fig. 3). In cultures started from 9-day-old postnatal rat cerebellum, abundant expression of fibronectin in GFAP-positive glial cells was demonstrated earlier than when the cultures were started from newborn rat cerebellum; even after 3 days in vitro (table 1). Neostriatum In early neostriatal cultures fibronectin-positive glial cells were few in number, but fine extracellular matrix deposits could be seen in connection with cell explants. After 11 days in vitro there were a few fibronectin-positive astrocytes (fig. 4) which, however, formed less than 20% of the GFAP-positive cell population (table 1). In these older cultures, tibronectin was also deposited in matrix form. This matrix was fairly extensive, as in the case of cerebellum, and was confined to the top of the non-neuronal cells. Cerebral Cortex In early cortical cultures started from newborn rat brain there were few Iibronectin-positive glial cells. In older cultures or in cultures started from 9-dayold rat brain the number of fibronectin-positive astrocytes slightly increased, but these cells formed a minor population of GFAP-positive glial cells (table 1). EfSect of Fibronectin Containing by Cultured Astrocytes

FBS on the Expression of Fibronectin

The expression of fibronectin in primary cultures varied slightly between cultures but the presence of fibronectin in the culture medium did not consistently affect this situation (cf table 1, B vs E and Z vs L). Meningeal cell culture medium or the cells themselves had no effect on the expression of fibronectin by the astrocytes. Results of the Statistical

Analysis

The percentage of FN+ cells was significantly (ZWO.01) higher in cerebellum than in either cerebral cortex or neostriatum for cells from newborns after 11 DIV The number of tibronectin-positive (FN+) astrocytes expressing GFAP immunoreactivity (GFAP+) per ten fields was correlated with the age of the tissue used to initiate the cultures, with the time in culture (DIV, days in vitro) in culture medium containing 20% fetal bovine serum with (FBS+FN) or without tibronectin (FBS-FN). The statistical signiticances of differences between groups, calculated as described in Materials and Methods, were as follows: ZWO.001: A vs B. P-cO.01: A vs C, B vs H vs I, F vs G, Z vs J. ZWO.05: C vs D. The following differences were not significant (Z30.05): A vs F vs K, A vs E, C vs G, E vs L, F vs H, F vs K, H vs I, Z vs K, Z vs L. Exp Cell

Res 163 (1986)

180

Liesi, Kirkwood

and Vaheri

Fig !. 1. Primary culture from newborn rat cerebellum after 15 DIV. Double immunostaining f br (a) iibl *onectin; (b) GFAP. (a) All the flat cells in this picture express perinuclear immunoreactivi ty for tibl -onectin. (b) Double immunostaining for GFAP shows that all these cells are astrocytes exprc the GFAP. x560.

FN- medium. There was no apparent difference, however, after only 3 I 1IV. Lere was no significant difference in either case between cerebral cortex and ostriatum (table 1). l’he percentage of FN+ cells was significantly (P
Cell

Res

163 (1986)

Fibronectin

in rat astrocytes

181

Fig. 2. Primary culture from newborn rat cerebellum after 11 DIV. Immunocytochemistry for fibronectin. Fibronectin is present also in extensive pericellular matrix deposits. x560. Fig. 3. Primary culture from newborn rat cerebellum after 15 DIV. The matrix was isolated as described in the Materials and Methods. Only sparse short tibrils were seen facing the coverslip. No tibrillar network could be demonstrated. x560.

compared with 3 DIV and, at a lower level (P
in Cultures Started from lo-day-old

Embryonic

Rat Brain

Cells from early embryonic rat brain attached to glass coverslips and spread during the first culture days. Most of these cells were flat in morphology and all of them expressed immunoreactivity for vimentin (fig. 5), the intermediate filament protein present in immature brain cells [27, 281. At this stage it was impossible to distinguish cells giving rise to neurons and glial cells. All the cells in these cultures expressed immunoreactivity for fibronectin (fig. 6), even after one day in Exp

Cell

Res 163 (1986)

182

Liesi, Kirkwood

and Vaheri

4. Primary culture from newborn rat cerebellum after 11 DIV. Double immunocytochemistry for (a) tibronectin; (b) GFAP using two rabbit primary antisera as described. (a) A few tibronectinpositive cells are seen (arrows). Fibronectin is also present in matrix form (arrowhead). (b) Double staining for GFAP reveals that some of the astrocytes (arrows) were also positive for tibronectin. Note that due to the use of two rabbit primary antisera the tibronectin immunoreactivity is also detected by the second antibody (TRITC) marking GFAP immunoreactivity. x560.

Fig.

vitro. An extensive fibronectin matrix was present and was now deposited between the glass coverslip and the cell monolayer (fig. 6). Fibronectin

in Embryonic

and Early Postnatal

Rat Brain In Vivo

In tissue sections from H-day-old embryonic rat brain, from newborn, 9-12day-old postnatal rat brain fibronectin was present in capillary and meningeal structures (not shown), just as in adult rat brain. No tibronectin was seen in connection with neuroectodermal cells. DISCUSSION The present study shows that GFAP-positive glial cells in vitro express fibronectin antigen as detected using immunocytochemistry. This is of interest with respect to the possible role of fibronectin for the development of neuronal cells and also with respect to the heterogeneity suggested for glial cells [29]. The present results may also clarify the existing controversy for [17, 301 and against [15, 161 expression of fibronectin in cultured glial cells. We showed here that the expression of immunocytochemically demonstrable fibronectin by a subpopulation of flat astrocytes is dependent on the brain region used (cerebellum vs neostriatum or cerebral cortex) and is expressed earlier by astrocytes in cultures started from 1 l-day-old postnatal rat brain when compared with cultures started from newborn rat brain. This led us to study the expression of fibronectin in late postnatal (9-lZdayExp

Cell

Res 163 (1986)

Fibronectin

in rat astrocytes

183

Fig. 5. Primary culture from lZday-old embryonic rat brain after 3 DIV. Immunocytochemistry for vimentin. All the flat cells attached to glass coverslips express the vimentin intermediate filaments. x340. Fig. 6. Primary culture from lZday-old embryonic rat brain after 3 DIV. Immunocytochemistry for fibronectin. All the flat cells are positive for fibronectin (arrows). An extensive libronectin matrix is formed and deposited as deduced by focusing between the cells and the glass coverslip (arrowheads). X440.

old) rat brain in vivo. However, fibronectin was seen only in connection with capillary and meningeal structures, as reported for adult and newborn rat brain [14]. Astrocytes in primary culture have been shown to express laminin [12], which is also induced in adult brain astrocytes by injury [13]. No regional differences in laminin expression, however, have been reported for in vivoExp

Cell

Res 163 (1986)

184

Liesi, Kirkwood

and Vaheri

derived cultures [12]. Promotion of neurite outgrowth by laminin suggests that it may be specifically needed for neural migration in the brain. In the peripheral nervous system fibronectin has been found to be important for early cellular migration [6, 311. The fact that fibronectin is expressed by cells cultured from 12-day-old rat embryos and also by cells cultured from early postnatal cerebellum could imply that fibronectin is involved during cellular migration, as seems to be the case for neural crest cells and to some extent for sensory neurons [6-9, 311. The fact that no fibronectin was seen in connection with embryonic or late postnatal rat brain in vivo would seem to be in contradiction with its expression by astrocytes in vitro. However, it is possible that fibronectin is not detected intracellularly in vivo because of its rapid secretion. This would be consistent with the report by Hatten et al. [32], who detected fibronectin in developing mouse cerebellar tissue along the route of granule cell migration. The factors responsible for the regional differences in tibronectin expression are not clear. It is evident, however, that glial cells from newborn rat brain form a heterogenous population of cells; glial cells from different brain regions have different capacities to produce tibronectin. Distinct subpopulations of astrocytes with characteristic biochemical properties have been described [29]. In the present study we found that astrocytes, even within the same brain region differ so that some of them are capable in producing laminin [33], fibronectin or both (flat cells) and some of them (star-shaped cells) have the capacity to express only the glial tibrillary acidic protein, but not these matrix glycoproteins. We thank Dr Doris Dahl and Dr Richard Hynes for antibodies. This work was supported by grants from the Lake Oy Research Foundation, Turku, Finland, The Finnish Cancer Society and the Finnish Medical Research Council.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.

Ruoslahti, E, Engvall, E & Hayman, E, Co11 res 1 (1981) 95. Hynes, R & Yamada, K, J cell bio195 (1982) 369. Vartio, T & Vaheri, A, ‘Bends in biochem sci 8 (1983) 442. Vaheri, A, Salonen, E-M, Vartio, T, Hedman, K & Stenman, S, Biology and pathology of the vessel wall (ed N Woolf) p. 161. Praeger, Eastboume (1983). Yamada, K, Ann rev biochem 52 (1983) 761. Duband, J & Thiery, J-P, Dev bio193 (1982) 308. Rovasio, R, Delouvee, A, Yamada, K, Timpl, R & Thiery, J-P, J cell biol 96 (1983) 426. Baron-Van Evercooren, A, Kleiman, H, Ohno, S, Marangos, P, Schwartz, J & Dubois-Dalcq, M, J neurosci res 8 (1982) 179. Rogers, S, Letoumeau, P, Palm, S, McCarthy, J & Furcht, L, Dev bio198 (1983) 212. Manthorpe, M, Engvall, E, Ruoslahti, E, Longo, F, Davis, G & Vat-on, S, J cell biol 97 (1983) 1882. Liesi, P, Dahl, D & Vaheri, A, J neurosci res 11 (1984) 241. - J cell biol % (1983) 920. Liesi, P, Kaakkola, S, Dahl, D & Vaheri, A, EMBO j 3 (1984) 683. Schachner, M, Schoonmaker, G & Hynes, R 0, Brain res 158 (1978) 149. Paetau, A, Mellstrom, K, Westermark, B, Dahl, D, Haltia, M & Vaheri, A, Exp cell res 129 (1980) 337.

Exp Cell

Res 163 (1986)

Fibronectin 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33.

in rat astrocytes

185

Stieg, P, Kimelberg, H, Mazurkiewicz, J & Banker, G, Brain res 199 (1980) 493. Vaheri, A, Ruoslahti, E, Westermark, B & Pontkn, J, J exp med 143 (1976) 64. Jones, T, Ruoslahti, E, Schold, S & Bigner, D, Cancer res 42 (1982) 168. Vuento, M & Vaheri, A, Biochem j 183 (1979) 331. Engvall, E & Ruoslahti, E, Int j cancer 20 (1977) 1. Hedman, K, Vaheri, A & Wartiovaara, J, J cell bio176 (1978) 748. Raju, T, Bignami, A & Dahl, D, Dev biol85 (1981) 344. Dahl, D & Bignami, A, J camp neural 76 (1977) 645. Hynes, R & Destree, A, Cell 13 (1978) 151. Hedman, K, Kurkinen, M, Alitalo, K, Vaheri, A, Johansson, S & Hook, M, J cell biol81 (1979) 83. Armitage, P, Statistical methods in medical research, p. 356. Blackwell, Oxford (1971). Dahl, D, Rueger, D, Bignami, A, Weber, K & Osbom, M, Eur j cell bio124 (1981) 191. Schnitzer, J, Franke, W & Schachner, M, J cell biol90 (1981) 435. Raff, M, Miller, M & Noble, M, Nature 303 (1983) 390. Kavinsky, C & Garber, B, J supramol struct 11 (1979) 269. LeDouarin, N, The neural crest. Cambridge University Press, London (1982). Hatten, M E, Furie, M B & Ritkin, D B, J neurosci 2 (1982) 1195. Liesi, P, EMBO j 4 (1985) 1163.

Received June 5, 1985 Revised version received September 30, 1985

Printed

in Sweden

Exp

Cell Res

163 (1986)