Basic fibroblast growth factor promotes the survival of embryonic ventral mesencephalic dopaminergic neurons—i. Effectsin vitro

0306-4522/93 $6.M)+ 0.00 Pergamon Press Ltd

Neuroscience Vol. 56, No. 2, pp. 379-388, 1993 Printed in Great Britain

IBRO

BASIC FIBROBLAST GROWTH FACTOR PROMOTES THE SURVIVAL OF EMBRYONIC VENTRAL MESENCEPHALIC DOPAMINERGIC NEURONS-I. EFFECTS IN VITRO E. MAYER,*? S. B. DWNETT,*~

R. ~ELLITTERI*$

and J. W. FAW~EIT*§/

*MRC Cambridge Centre for Brain Repair, t~par~en~

of Ex~~mental Psychology and §Physiolo~ical Laboratory, University of Cambridge, Downing Street, Cambridge, U.K. SIstituto di Fisiologia Umana, Universita di Catania, Italia

Abstract-We have studied the effects of basic fibroblast growth factor on rat embryonic mesencephalic neurons in vitro. Basic fibroblast growth factor promotes the survival of dopaminergic neurons in vitro, the effect increasing with dose and reaching a maximum at 10 ng/ml. In the absence of basic fibroblast growth factor the number of tyrosine hydroxylase-stained (tyrosine hydroxylase positive) neurons declines to almost zero within 14 days, whereas in the presence of basicfibroblast growth factor numbers remain almost constant from three to 28 days in vitro. This effect of basic fibroblast growth factor is abolished by preventing non-neuronal cells from appearing in the cultures, apart from a basic fibroblast growth factor-mediated increase in the numbers of tyrosine hydroxylase-positive cells during the first two days in vitro.The presence or absence of non-neuronal cells also iniluences dopaminergic neuronal morphology, the neurons having more, longer, and more varicose processes in the absence of astrocytes. Survival of dopaminergic neurons in vitro in the absence of basic fibroblast growth factor is very dependent on plating cell density, but in the presence of basic fibroblast growth factor this dependency vanishes. It is also possible to make survival independent of plating density by growing the cultures on inverted coverslips, which have the effect of concentrating secreted molecules in the thin layer of medium between coverslip and dish. Our conclusions from these experiments on plating density are that astrocytes probably ~onstitutively secrete a small amount of a trophie factor which promotes survival of dopaminergic neurons, and that the rate of pr~uction of this factor is greatly increased by basic fibroblast growth factor. If basic fibroblast growth factor is withdrawn from cultures after two or seven days the dopaminergic neurons soon die. However, if basic fibroblast growth factor is withdrawn after 14 days, after the period of naturally occurring cell death of these neurons, there is no increase in dopaminergic neuronal death compared to controls in which basic fibroblast growth factor treatment is maintained. If basic fibroblast growth factor is used to improve the survival of dopaminergic neurons grafted in uiuo, it should therefore be sufficient to treat the grafts for 14 days.

One of the most promising techniques for the treatment of Parkinson’s disease is the transplantation of embryonic mesencephalic dopaminergic neurons to replace those which have died in the substantia nigra. While this technique has been shown to be of benefit in human patients, its curative effects are only partial.“.‘3.‘8 Two factors limit the effectiveness of such grafts at present: only 5% or less of transplanted dopaminergic neurons survive to populate the mature graft,s and transplanted neurons cannot receive a normal synaptic input since the transplant has to be placed in the target area in the striatum.3s*

IiTo whom correspondence should be addressed at: Physiolo~~l labofatory, Downing Street, Cambridge CB2 3EG, U.K. Abbreviations:bFGF, basic fibroblast growth factor; BSA, bovine serum albumin; DMEM, Dulbecco’s modification of Eagle’s medium; E, embryonic day; EDTA, ethylenediaminetetra acetate; FITC, fluorescein isothiocyanate; GFAP, glial fibrillary acidic protein; HBSS, Hanks’ balanced salt solution; NGS, normal goat serum; PBS, phosphate-buffered saline; TH, tyrosine hydroxylase.

The poor survival of grafted neurons could be due to anoxic or excitotoxic stress at the time of transplantation, to lack of trophic support at the critical developmental phase of naturally occurring cell death, or to a combination of these processes. If lack of trophic support is a major cause of poor graft survival, treatment with one or several might provide a rational strategy for enhancing graft viability. A number of trophic molecules have been shown to enhance the survival of mesencephalic dopaminergic neurons.‘6.17 In addition a number of trophic factors have been shown to increase the ability of neurons, including dopaminergic neurons from the substantia nigra, to resist a variety of toxic and anoxic insults.24-26One such trophic molecule with widespread effects on many types of neuronal and glial cell in the CNS, including those in the substantia nigra is basic fibroblast growth factor (bFGF).2~‘o~‘s~‘7~20~22~27~‘o~3i Initial studies with bFGF indicate that it can exert a trophic action on dopaminergic neurons in vitro by stimulating astrocytes to release a second factor, but little is known about the optimal conditions for delivery or time-course of effective action. If bFGF is 379

to be used in viva to improve the survival of grafted cells. it is crucial to know how best to use it. In thts paper we examine a number of parameters for bFGF treatment of neurons from rat embryonic day I4 (E 14) mesencephalon, the same tissue that is used for grafting in rat models of Parkinson’s disease. Specifically. we have used standard tissue culture techniques to ask the following questions: (1) How much bFGF must be applied? (2) Must treatment be continued indefinitely, or is it just necessary to treat cells as they go through the period of naturally occurring cell death? (3) Is the effect of bFGF directly on neurons, or via astrocytes? (4) How important is trophic support from other neurons and glia. and to what extent is it affected by bFGF‘? EXPERIMENTAL

PROCEDURES

Preparation of cell suspensions Cell suspensions were prepared from CFHB rat embryos (crown-rump length 10-I 1 mm). Pregnant female rats (breeding colony, Department of Physiology, University of Cambridge) were anaesthetized in ether and given an overdose of equithesin. The embryos were removed by caesarian section and placed in cold ( + 4’ C) Leibowitz L- 15medium, in which all dissection processes were carried out. The ventral mesencephalon was dissected out and the pieces incubated for 6 min in 0.1% porcine trypsin (Sigma T0134) in a solution of phosphate-buffered saline (0.1 M PBS. nH 7.4). containing 0.02% EDTA at 37’C. Deoxvribonullease was then added and the cells lightly agitated, then centrifuged at 100 g for I min, the supernatant removed and the cells resuspended in 1 ml of triturating solution (3 mg bovine serum albumin, BSA, 10 mg deoxyribonuclease, 0.5 mg soybean trypsin inhibitor per ml PBS). The cells were then gently triturated, centrifuged and resuspended in warm (37°C) Dulbecco’s modification of Eagle’s medium (DMEM) with 10% fetal calf serum, or DMEM/FlZ with B27 supplements,“ counted and diluted before plating at determined cell densities on 22-mm diameter poly-L-lysinecoated glass coverslips. Cell counts represent vital cells and values were obtained using Trypan Blue exclusion or Acridine Orange and ethidium bromide.6 To allow for the variability between cell suspensions, all suspensions in any one experiment were divided equally between all conditions. Some wells were treated with bFGF (human recombinant, Boehringer Mannheim, 10 ng/ml). Cultures were fed twice a week, except for cultures in serum-free medium, which were fed on alternate days, with IO-‘M cytosine arabinoside added to alternate feeds. Cultures on inverted coverslips were prepared by allowing the cell suspension to adhere to the coverslip for an hour before inverting and then adding medium. Immunocyrochemistry Individual populations of cells were identified by immunocytochemistry. After washing coverslips twice in 37°C Hanks’ balanced salt solution (HBSS) the cells were fixed for 30 min in ice-cold 4% paraformaldehyde dissolved in phosphate buffer (0.1 M, pH 7.4) rinsed twice in PBS, then

Incubated in a 5% normal goat serum (NGS) in PBS containing 0. I % Triton X-100 (PBS Triton) at room tern peraturc for 2Omin. Cultures on inverted coverslips wcrc inverted in the presence of fixative after 3 r 5-min v.asIic~ in warm HBSS. This was followed by incubation an PBSTriton containing 1% NGS and primary antibodies. for 6 h at +4’ C. Coverslips were then washed three times in PBS and incubated in secondary antibodies [Huroscein isothiocyanate (FITC)-labelled anti-rabbit antibodies (Caltag) 1: 100 to visualize the polyclonal primary antibodies and biotinylated goat anti-mouse I& (Claltag) I: 100 to visualize monoclonal antibodiesj in PBS Triton with 1% NGS, at f4’C for I h. washed three times in PBS and incubated for a further 30 min in rhodamine isothiocyanate-labelled streptavidin (Serotec) I : 100 in PBS, to visualize the monoclonal antibodies. After washing, coverslips were mounted in PBS: glycerol (50: SO) and placed on glass microscope slides, The antibodies used included: for tyrosine hydroxylase (TH) a monoclonal antibody (Boehringer Mannheim) used at a concentration of 1: 200 and a polyclonal antibody (Laboratoires Jacques Boy, France) used at a concentration of I : 2000; a polyclonal antibody for glial fibrillary acidic protein (GFAP) (Dakopatts, U.K.) used at a I :200 dilution; a monoclonal antibody for vimentin (Boehringer Mannheim) used at a I : 200 dilution. As a general neuronal marker we used tetanus toxin which binds to GM1 ganglioside, present on almost all neurons and oligodendrocyte precursors.’ Live cultures were incubated with I : 50 toxin (to give 10 mgjml) in PBS for 30 min at 37 c’. The cultures were then fixed for 20 min (see above) and then stained with rabbit anti-toxin and secondary antibody. Tetanus toxin and rabbit anti-serum (used at a I : 100 dilution) were kindly provided by Dr R. 0. Thompson (Wellcome Laboratories, Beckenham, 1.J.K.l.

Coverslips were mounted on microscope slides in PBS-glycerol and anaiysed on a Leitz dialux fluorescent microscope. Cells were counted over the entire coverslip in the case of low densities of TH-positive neurons. GFAPpositive astrocytes and tetanus toxin-positive neurons and higher densities of TH-containing neurons were counted with the aid of a 500~pm square graticule. The graticule was moved across the coverslip from side to side, then top to bottom, counts being taken every 2 mm, giving 22 positions in all, and counts for the whole coverslip were then estimated. Data were analysed using a statistical analysis of variance (ANOVA. Genstat. Rothamsted Experimental Station), ‘which also supplied standard errors of the difference between means for multiple post hoc comparisons as appropriate.” RESULTS

Time-course ,$broblast

of‘ neuronal

survival in control and basic

growth factor-treated

cultures

Cultures were plated at a density of IO5 cells per coverslip. Typical examples of TH and tetanus toxin staining are shown in Fig. 1 a, b. The survival of TH-immunoreactive neurons is shown in Fig. 2a, and that of tetanus toxin-immunoreactive cells in Fig. 2b.

Fig. 1. (a, b) Same view of a lo-day-old culture stained with both TH (a) and tetanus toxin (b). A dopaminergic neuron in the centre of the field is stained with both antibodies, but the numerous smaller neurons in the culture are only visualized by tetanus toxin. (c, d) Culture grown on an inverted coverslip after 10 days in vitro; c is stained for vimentin, to show the small number of non-neuronal cells, which were GFAP negative; d shows the same field stained with tetanus toxin, to show the many neurons, most of which are associated with the non-neuronal cells. Scale bar = 100 brn.

Effects of bFGF on dopaminergic neurons in vitro

381

38’

E. MAYIIK P/ cil

neurons per mm2 (Fig. 3b) this elIect was not significant (F3,,2 = 1.16, ns.). Plating density

Not surprisingly the higher the density of cells plated per coverslip, the higher the number of TH-positive neurons (Fig. 4a) found to survive after 11 days in vitro. In addition bFGF treatment resulted in a higher rate of survival at all plating densities. All (density &, = 24.25, effects were significant P < 0.001, bFGF treatment F,,50 = 86.69, P < 0.001, interaction F4,%= 19.35, P
(

-n

E

40”

I [

increase in survival proportional to the density of surviving cells in cultures, the data were rc-analysed following log transformation. The density .Y treatment interaction remained highly significant (F4.40= 2.97, P < 0.005). Plotting the data in terms of plating efficiency shows that bFGF has a large and roughly similar effect on TH-immunoreactive neuronal survival at all plating densities (Fig. 4b), the number of surviving neurons being between 1 and 4% of the neurons plated, with no systematic variation with plating density. Comparison of‘ cultures grown on normal and inverted coverslips

Fig. 2. The effects of 10 ng/ml bFGF on the survival of dopaminergic neurons (a) and all neurons (b). Error bars are S.E.M.

If coverslips are inverted in their wells, the cultures are sandwiched between the glass and the bottom of the well, together with a thin layer of medium.

the numbers of TH-positive neurons declined virtually to zero after two to four weeks in control cultures, they were well maintained in cultures treated with bFGF. All effects were significant (time F,,, = 9.99, P < 0.001, treatment F,,*,,= 100.63, P < 0.001, interaction F3,*,,= 3.35, P < 0.05). Counts of tetanus toxin binding neurons showed a similar pattern, with bFGF having a similar marked effect on promoting the survival of all neurons. Approximately 30% of neurons remained after four weeks even in the control cultures, while numbers did not decline at all in bFGF-treated cultures. All effects were highly significant (time F,,32 = 9.46, P < 0.001. treatment F,,,, = 88.21, P < 0.001 and interaction F,,,, = 6.21, P < 0.005).

Whereas

Fibroblast growth factor dose-response

Cells for this experiment were plated at 10’ cells per coverslip, and assayed after 11 days in vitro. The bFGF-associated increase in the survival of TH-positive neurons was dose-dependent (Fig. 3a) and reached asymptote at long/ml of medium. Above this dose no further bFGF-associated increase was it obtained (F,,, = 57.67, P < 0.001). Although appears that bFGF also produced a dose-dependent increase in the number of tetanus toxin-positive

r5 I.-.

0 Fig. 3. The effects of bFGF levels on the survival of dopaminergic neurons (a) and tetanus toxin-stained neurons (b) after I1 days in culture. Error bars are S.E.M.

Effects of bFGF on ~o~inergic 15000 r a) D m

L? E 6 uo

Control bFGF long/ml 1

L 10000 a fi z P

J

0 e

5000

B I b :: 0 5x10S

lo*

383

neurons in &xx

treatment P < 0.001, F,,42 = 33.50, coverslips F ,,,,*= 146.69, P < 0.001, interaction F,,42 = 17.92, P < 0.001). The numbers of GFAP-positive astrocytes’ (Fig. 5b) were markedly reduced by inversion of coverslips (F,,*, = 19.33, P < 0.001). Although bFGF increased the numbers of these cells in both coverslip orientations (Fg.24= 16.56, P < O.OOl), the bFGFassociated increase in their numbers was greatly (interaction inverted coverslips reduced on F 1,24= 14.57, P < 0.001). Post hoc Newman-Keuls comparison revealed that in inverted coverslips there was no significant difference in the number of GFAPpositive astrocytes in the bFGF-treated and control groups (& = 0.26, n.s.). In most of these cultures there was a virtual absence of GFAP-positive astrocytes in the centre of upturned coverslips. Cells tended to form a network of neurons which were generally in close apposition to the rather sparse vimentin-positive glial-type cells (Fig. lc, d). These glia were spindly and unlike the flattened GFAPpositive astrocytes that were observed in large numbers on the very peripheral regions of the coverslips. Time-course of neuronal survival on inverted coverslips

The number of TH-positive neurons (Fig. 6) was low after one day in culture, but then increased, the a

Fig. 4. The influence of plating density on the survival of dopaminergic neurons after 11days in vitro. In a the figures

3000 27 II zi 8 2500

a)

[?

Control

Ea

bFGF

are expressed as absokte numbers of surviving neurons, and in b as surviving neurons as a proportion of the total number of cells plated. Error bars are S.E.M.

Because of the geometry of this arrangement, any molecules secreted by cells in the culture will be much more concentrated in the thin layer of medium under the coverslip than would be the case normally, and conversely access of molecules from the main pool of medium in the well is greatly slowed. As a consequence, the concentrations of rapidly used molecules such as oxygen should be lower than normal. In practical terms, this probably has the same effect as greatly increasing the apparent plating density of a culture, so that cultures which rely on interactions between cells by means of secreted molecules can be grown at lower densities. Under these conditions, the growth and differentiation of astrocytes is greatly decreased.4 A comparison of the same cell suspension cultured on inverted and non-inverted coverslips after 10 days in culture is illustrated in Fig. 5a, b. In the absence of bFGF inverted coverslips had substanti~ly more TH-positive neurons than non-inverted cultures, the survival on inverted coverslips being nearly as good as in non-inverted coverslips in the presence of bFGF. However, bFGF still increased the survival of T&positive neurons in both sets of cultures. All comparisons were significant (orientation of

Normal :

b)

c_ 1000 * 2 z L 0

800

0

0

Control

Es4

WC3

Inverted

i

ii

600

f

s e

$

j 400

I b 9 E 200 E .^ % 8,I

0 NIXTIEd

Fig. 5. The numbers

Inverted

of dopaminergic neurons (a) and GFAP-stained astrocytes (b) in cultures on inverted and non-inverted coverslips after 11 days in vitro,in the presence and absence of long/ml bFGF. Error bars are S.E.M.

k. MAYER (‘I ul.

Fig. 6. The time-course of dopaminergic neuronal survival on inverted coverslips in the presence and absence of long/ml bFGF. Error bars are S.E.M. numbers being very similar in control and bFGFtreated cultures. This large increase must have been due to the appearance of TH in pre-existing neurons rather than the birth of new ones, since division amongst El4 dopaminergic neuronal precursors during the first day in culture is very limited.” Numbers peaked at one week, and there was then a steady decline almost to zero by four weeks. At seven and 10 days the bFGF-treated cultures had more TH-immunoreactive neurons than control, but the differences were much less than in non-inverted cultures. These changes were significant (treatment F,,62 = 7.34, P < 0.01, time F5.62= 27.48, P < 0.001). Although it appears that the increase in TH-positive neuron survival associated with bFGF was different at different time-points, the interaction failed to reach significance (F5,62= 2.15, n.s.). At four weeks, when most of the neurons were dead, there was a dense ring of astrocytes round the edges of the coverslips which were also adherent to the plastic of the dish, and may have sealed the culture surface off completely from access to the culture medium. The e&t of‘ plating density in neuronal survival on inverted coverslips

The numbers of TH-positive neurons per inverted coverslip after 11 days in culture of a cell suspension plated at different densities are shown in Fig. 7a, and the plating efficiency in Fig. 7b. Not surprisingly at higher plating densities more TH.,positive neurons are found (F,,53 = 5 1.75, P < O.OOl), and overall bFGF increases in the sur-

Fig. 7. The survival of dopaminergic neurons on inverted coverslips at different plating densities after 11days in vitro. In a the cell counts are expressed as absolute numbers, and in b they are expressed as a proportion of the total number of cells plated. Error bars are S.E.M.

viva1 of TH-positive neurons (F,,53= 6.98, P < 0.05). There was no significant difference in the bFGF effect at different plating densities (interaction F,,53 = 1.53, n.s.). The observed plating efficiencies in the presence of bFGF are not very different between inverted and non-inverted culture techniques, but the plating efhciency in the absence of bFGF is enormously higher on inverted coverslips. We would suggest that cells in non-bFGF-treated cultures secrete small amounts of a dopaminergic neuronal survival factor, which can be concentrated in the small volume of medium under a coverslip. In the presence of bFGF secretion is upregulated to the extent that the whole medium volume of a tissue culture well can be brought up to the minimum concentration needed to promote neuronal survival.

Fig. 8. (a, b) Same area of a culture grown for 10 days in B27 serum-free medium in a low oxygen atmosphere with long/ml bFGF. Cytosine arabinoside was given on a two-days-on/two-days-off rota. (a) TH stain, showing several dopaminergic neurons with long processes studded with many varicosities. (b) GFAP stain, showing that there are no astrocytes in these cultures. (c, d) views of a culture grown for 10 days in the same medium and conditions, but without treatment with cytosine arabinoside. (d) GFAP stain, showing that many astrocytes develop in such cultures. (c) TH stain. Comparison with a shows that the cells, although more numerous in c have fewer, shorter, finer processes without the marked

varicosities

seen in a. Scale bar = 100 urn.

Effects of bFGF on dopaminergic neurons

in vitro

385

Culture of cells in conditions development of astrocytes

which

prrzw~t

ik

Dissociated El4 nigra grown in DMEM/FlZ supplemented with B27 additives4 in an ambient oxygen concentration of 7% will grow for at least two weeks with very few astrocytes, and good survival of neurons. However, if bFGF is added many astrocytes with fine branching GFAP-positive processes are soon seen, and within a week are almost confluent (Fig. 8 b, d). In order to prevent the astrocytes appearing it is necessary in addition to treat cultures with cytosine arabinoside 10m5M on a two-days-on/two-days-off rota. In cultures grown under these conditions bFGF did not significantly increase the survival of THimmunoreactive neurons at seven or 14 days (Fig. 9). There were on average more TH-containing neurons after two days in bFGF-treated cultures (P < 0.05 by t-test), as there were in cultures grown in serum-containing medium. The morphology of TH-immunoreactive neurons was very different when grown in the absence of astrocytes. Cells growing on astrocytes had fine processes of moderate length with few varicosities, while neurons in astrocyte-free cultures had much longer processes with very marked varicosities along their length (Fig. 8a, c) The effects on neuronal survival of withdrawal of basic ,fibroblast growth ,factor at diJ&erenttimes

This experiment investigated whether cultures must be supplemented with bFGF continuously, or whether a shorter duration of administration is sufficient to produce lasting survival of dopaminergic neurons. After two, seven, or 14 days in the presence of bFGF-supplemented culture medium the cells were rinsed three times in warm (37°C) HBSS and then switched to non-supplemented medium. Thus although it was not possible to remove any bFGF that was attached to or found within cells, no more

0 m

:

>”

Control bFGF

i

8

;

2

3

c

20000

c 3 2 L 8 10000 $ ._ c

0 2 days

7 days

0

I/

0

wlhdi~h,.

bFGF

v I

aL 2 days

bFGF withdrawn at 7 days 0 bFGF withdrawn at 14 days

10 Days

20 after

30 piatmg

40

50

cells

Fig 10. The effects of withdrawing bFGF after various times in vitro. The counts of dopaminergic neurons after bFGF withdrawal are calculated as a percentage of counts obtained from sister cultures in bFGF treatment had been continued. Error bars are S.E.M.

could be obtained from the culture medium. Survival without bFGF was compared with cultures from the same cell suspension that received continuing bFGF supplementation. The time-course of survival of TH-positive neurons after withdrawal of bFGF from the culture medium is illustrated in Fig. 10. When bFGF was withdrawn after two or seven days, there was a rapid decline in the number of TH-immunoreactive neurons compared to control cultures in which bFGF was not withdrawn. However, when bFGF treatment was continued for 14 days, past the period of naturally occurring cell death in the substantia nigra, there was no decline in the number of neurons relative to control over the ensuing two weeks. Pairwise comparison of the overall neuron counts for continuous administration of bFGF and withdrawal at 14 days show that these two groups do not differ overall (t2,48 = 0.40, ns.).

DISCUSSION

40000

2

iii1

14 days

Fig. 9. The survival of dopaminergic neurons cultured in the presence and absence of long/ml bFGF, but under conditions which prevent the appearance of astrocytes. Error bars are S.E.M.

Our results demonstrate that bFGF has a powerful effect on the survival of mesencephalic dopaminergic neurons in culture, on the numbers of other mesencephalic neurons, and on the proliferation of astrocytes. There is a clear dose response, with the maximum survival effect being at 10 ng/ml; this is roughly in line with observations on other types of neuron. We can be confident that the main effect of bFGF on dopaminergic neurons of the age used in this study is one of survival rather than proliferation. In a previous study in which bFGF treated cultures were labelled with 13H]thymidine so as to mark dividing cells, and then stained to visualize TH in dopwe showed that neuronal aminergic neurons, precursors only divide in response to bFGF

Effects of bFGF on dopaminergic neurons in vitro if plated two days earlier than in the present study.” We cannot be. similarly certain about the effects of bFGF on other neuronal cells, and it is possible that some of the increase in neuronal cell number resulting from bFGF treatment could be due to the promotion of division of neuronal precursors. At least part of the effect of bFGF on dopaminergic neurons is via astrocytes or astrocyte precursors, since we find that elimination of these from the culture eliminates the effect of bFGF on long-term survival, in agreement with previous studies.” bFGF induces the secretion of a number of other trophic factors from astrocytes, for instance nerve growth factor and insulin-like growth factor.7.‘2.23 However, neither of these can be entirely responsible for all the effects of bFGF on dopaminergic neurons, since nerve growth factor by itself is without effect,” and the high concentration of insulin in the serumfree medium which we used did not substitute for bFGF in the absence of astrocytes. It is also likely that there is a direct effect of bFGF on dopaminergic neurons or their precursors during the first day or so in culture. We have previously observed a mitogenic effect of bFGF on these cells in culture before any recognizable glial cells have appeared to mediate it, and a direct effect has been reported in single cell cultures.21*2sIn addition, in the present study, bFGF significantly increased the number of TH-immunoreactive neurons present after two days in culture whether glia were present or had been killed. In practical terms, grafts of dissociated El4 mesencephalic tissue placed in vivo soon come to contain large numbers of astrocytes of both graft and host origin, which closely invest the dopaminergic neurons; any trophic effect modulated via astrocytes will therefore be effective almost immediately after transplantation. Our culture conditions were chosen to mimic this progression by allowing rapid division of astroglia. In addition, bFGF is secreted by cells surrounding lesions in the brain, such as are made by the implantation of cells.9J4,19 It is clearly not practicable to administer a trophic factor indefinitely to grafted neurons. It is therefore important to know whether dopaminergic neurons need continuous treatment, or whether it is sufficient to get them through the period of naturally occurring cell death, when neurons are temporarily extremely sensitive to the lack of trophic support, and into adulthood where trophic support is less essential. We have therefore looked at the effects of administering bFGF to cultures for a limited period, then withdrawing it and maintaining cultures in the absence of bFGF and observing the level of dopaminergic neuronal survival. The period of naturally occurring cell death in the substantia nigra is reported to begin just before birth and continue for about four days thereafter. Our cultures were made from El4 animals, so treatment for 14 days should have been more than sufficient, assuming that our cultures behave like other neurons in vitro and maintain their

387

developmental programme so as to become sensitive to lack of trophic support at the normal time.29,33Our results support the hypothesis that it is sufficient to administer bFGF until the end of the period of naturally occurring cell death. bFGF administered for the first two or seven days of culture had little long-term effect on dopaminergic neuronal survival, while if bFGF was administered for 14 days before withdrawal, the survival of neurons over the ensuing four weeks was as good as in cultures in which bFGF treatment was maintained. We cannot definitely say from our results whether the dopaminergic neurons no longer need trophic support after two weeks in culture, or whether there are by this time sufficient astrocytes to provide that support even when not stimulated by bFGF. In practical terms, however, grafts implanted in vivo will contain many astrocytes, and the present results suggest that infusion of bFGF for the first two weeks after grafting should be sufficient. The survival of dopaminergic neurons is very dependent on factors received from other cells. Thus in low density cultures neuronal ceil death in the absence of bFGF is rapid and complete. bFGF treatment removes much of this sensitivity to low cell density, probably by greatly increasing secretion of trophic factors by astrocytes and thereby increasing their local concentrations. We made a series of cultures in which the coverslips were inverted, the cells and a thin layer of medium now being in the narrow gap between coverslip and culture dish. This has the effect of slowing diffusion of molecules from the medium to the cells, and also from the cells back into the medium. Diffusion can occur, since tetanus toxin or Methylene Blue put into the medium will stain the cells, but molecules secreted by cells will be concentrated under the coverslip; the end result is probably rather similar to culturing cells at extremely high density. Under such conditions astrocytic development and proliferation is very much delayed, but survival of dopaminergic and other neurons, particularly at low densities is very much better. Additionally, cell survival in the absence of bFGF is almost as good as that seen in the presence of bFGF in non-inverted cultures over the first two weeks. This result implies that cells in the culture, probably glia, secrete low amounts of a dopaminergic neuronal survival factor even in the absence of bFGF, but that the level of secretion is greatly increased by bFGF. A second implication is that the factor must be produced by relatively undifferentiated glia as well as by fully differentiated cells, since hardly any differentiated glia were present in our inverted cultures. As judged from its properties in tissue culture bFGF is clearly a good candidate for improving the very poor survival of embryonic mesencephalic neurons grafted in vivo. In the following paper we report on the effects of bFGF on such grafts, following the treatment regimen suggested by the results of the experiments described in this report.

REFERENCES

1. Abney E. R., Bartlett P. P. and Raff M. C. (198 I) Astrocytes, ependymal cells, and oligodendrocytes develop on schedule in dissociated cell cultures of embryonic rat brain. &ol Bioi. 83, 3OI-310. 2. B6hr M., Vanselow J. and Thanos S. (1989) Abihty of adult rat ganglion cells to regrow axons in vitro can be influenced by tibroblast growth factor and gangliosides. Neurosci. Left. %, 197.-201. 3. Bjiirklund A. and Stenevi U. (1984) Intracerebral neural implants: neuronal replacement and reconstruction of damaged circuitries. A. Rer. Narrosci. 7, 279-308. 4. Brewer G. J. and Cotman C. W. (1989) Survival and growth of hippocampal neurons in defined medium at low density: advantages of a sandwich culture technique or low oxygen. Brain. Res. 494, 65-74. 5. Brundin P., Barbin G., Streaker R. E., Isacson 0.. Prochiantz A. and Bjorklund A. (1988) Survival and limction of dissociated rat dopamine neurons rated at different developmental stages or after being cultured in t$tuc~.Becl Brain Res. 39, 233-243. 6. Brundin P., Isacson 0. and Bjorklund A. (1985) Monitoring cell viability in suspensions of embryonic CNS tissue and its use as a criterion for intracerebral graft survival. Brain Res. 331, 251-259. 7. Drago J., Murphy M., Carroll S. M.. Harvey R. P. and Bartlett P. F. (199I) Fibroblast growth factor-mediated proliferation of central nervous system precursors depends on endogenous production of insulin-like growth factor I. Proc. natn. Acad. Sci. U.S.A. 88, 2199-2203. 8. Dunnett S. B., Bjijrklund A., Gage F. H. and Stenevi U. (1985) Neural Grafing in the Mamma&an CNS (eds Bjorklund A. and Stenevi U.), pp. 451-469. Elsevier, Amsterdam. 9. Eckenstein F. P., Shipley G. D. and Nishi R. (1991) Acidic and basic fibroblast growth factors in the nervous system: Distribution and differential aIteration of levels after injury of central versus peripheral nerve. J. Neurosci. 11,412.-419. 10. Engele J. and Bohn M. C. (1991) The neurotrophic effects of fibroblast growth factors on dopaminergic neurons in vitro are mediated by mesencephalic glia. J. Neurosci. 11, 3070-3078. 11. Freed C. R., Breeze R. E., Rosenburg N. L., Schneck S. A., Krick E., Qi J.-X., Lone T.. Zhang Y.-B., Snyder J. A., Wells T. H., Ramig L. O., Thompson L., Nazziotta J. C., Huang S. C., Crafton S. T., Brooks D., Sawle G., Schroter G. and Ausari A. A. (1992) Survival of implanted fetal dopamine cells and neuronlogic improvement 12 to 46 months after transplantation for Parkinson’s disease. New Engl. J. Med. 327, 1549-1555. 12. Fukumoto H., Kakihana M. and Suno M. (1991) R~ombinant human basic ffbroblast growth factor (rhbFGF) induces secretion of nerve growth factor (NGF) in cultured rat astroghal cells. Neurusci. Lerr. 122, 221.-224. 13. Garcia I., Martinou I., Tsujimoto Y. and Martinou J.-C. (1992) Prevention of programmed cell death of sympathetic neurons by the bcl-2 proto-oncogene. S’cicnre 258, 302--304. 14. Gomez-Pinilla F., Lee J. W.-K. and Cotman C. W. (1992) Basic FGF in adult rat brain: cellular distribution and response to entorhinal lesion and fimbria-fornix transection. J. Neurosci. 12, 345-355. IS. Grothe C., Wewetzer K., Lagrange A. and Unsicker K. (1991) Etrects of basic tibroblast growth factor on survival and choline acetyltransferase development of spinal cord neurons. Den/ Brain Res. 62, 257.-261. 16. Hyman C., Hofer M., Barde Y.-A., Juhass M., Yancopoulos G. D., Squint0 S. P. and Lindsay R. M. (1991) BDNF is a neurotrophic factor for dopaminergic neurons of the substantia nigra. Naiure 350, 230-232. 17. Knusef B., Michel P. P., Schwaber J. S. and Hefti F. (1990) Selective and nonselective stimulation of central cholinergic and dopaminergic development in oirro by nerve growth factor, basic fibroblast growth factor, epidetmal growth factor, insulin and the insulin-like growth factors I and II. J. Neurosci. 10, 558-570. 18. Lindvall O., Brundin P., Widner H., Rehncrona S., Gustavii B., Frackowiak R., Leenders K. L., Sawle G., Rothwell J. C., Marsden C. D. and Bjiirklund A. (1990) Grafts of fetal dopamine neurons survive and improve motor function in Parkinson’s disease. Science 247, 574 -577. 19. Logan A., Frautschy S. A., Gonzalez A.-M. and Baird A. (1992) A time course for the focal elevation of synthesis of basic fibroblast growth factor and one of its high-affinity receptors (fiR) following a localized cortical brain injury. J. Neurosci. 12, 3828-3837. 20. Matsuda S., Saito H. and Nishiyama N. (1990) Effect of basic fibroblast growth factor on neurons cultured from various regions of postnatal rat brain. Brain Res. 520, 310.-316. 21. Mayer E., Dunnett S. B. and Fawcett J. W. (1993) Mitogenic effects of bFGF on embryonic mesencephalic dopaminergic neurons. Devl Brain Res. 72, 253-258. 22. Murphy M., Drago J. and Bartiett P. F. (f990) Fibroblast growth factor stimulates the proliferation and differentiation of neural precursor cells in vitro. J. Neurosci. Res. 25, 463-475. 23. Ono T., Saito H., Kishimoto T., Okumoto T. and Miyamoto K. (1991) Stimulation of biosynthesis of nerve growth factor by acidic fibroblast growth factor in cultured mouse astrocytes. Neurosci. LAM. 126. 18-20. 24. Otto D. and Unsicker K. (1990) Basic FGF reverses chemical and morphological deficits in the nigrostriatal system of MPTP-treated mice. J. Neurosci. 10, 1912-1921. 2s. Otto D. and Unsicker K. (1990) Basic FGF reverses chemical and mo~hofo~cal deficits in the nigrostria~l system of MPTP-treated mice. J. Neurosci. 10, 1912-1921. 26. Spina M. B., Squint0 S. P., Miller J., Lindsay R. M. and Hyman C. (1992) Brain-derived ne~o~ophi~ factor protects dopamine neurons against 6-hydroxydopamine and .~-methyl-4-phenyfpy~dinium ion toxicity: involvement of the glutathione system. J. Neurochem. 59, 99-106. 27. Unsicker K., Reichert-Preibsch H. and Wewetzer K. (1992) Stimulation of neuron survival by basic FGF and CNTF is a direct effect and not mediated by non-neuronal cells: evidence from single cell cultures. Deal Brain Res. 65,285-288. 28. Unsicker K., Reichert-Preibsch H. and Wewetzer K. (1992) Stimulation of neuron survival by basic FGF and CNTF is a direct effect and not mediated by non-neuronal cells: evidence from single cell cultures. Deal Brain Res. 65,285-288. 29. Vogel K. S. and Davies A. M. (1991) The duration of neurotrophic factor independence in earty sensory neurons is matched to the time course of target fiefd innervation. Neuron 7, 819-830. 30. Wahcke P., Cowan W. M., Ueno N.. Baird A. and Guillemin R. (1986) Fibroblast growth factor nromotes survival of dissociated hippocampal neurons and enhances neurite extension. P&c. natn. Acid. Sci. 33, 3Of2-3016. 31. Westermann R., Grothe C. and Unsicker K. (1990) Basic fibroblast nrowth factor (bFGFj.,. a multifunctional growth factor for neuroectodermal cells. J. Cell Sci. h, &ppl. 13, 97-l 17. 32. Winer B. J. (1971) Statistical Principles in Experimental Design. McGraw-Hill, New York. 33. Wright E. M., Vogel K. S. and Davies A. M. (1992) Neurotrophic factors promote the matu~tion of developing sensory neurons before they become dependent on these factors for survival. Neuron 9, 139-150. (,4ccepted

10 May f993)