Basic fibroblast growth factor (bFGF) promotes the survival of a subpopulation of non-neuronal cells developing from trunk neural crest (Kalcheim, Devi Biol. 134, 1-10, 1989). It was therefore important to determine whether this factor is present in the nervous system at early developmental stages. Immunocytochemistry using specific polyclonal and monoclonal antibodies was combined with three highly sensitive assays: bFGF-induced proliferation of bovine adrenal cortex-derived capillary endothelial cells (ACE), a radioimmunoassay for bFGF (RIA) and Western blot analysis. bFGF immunoreactivity was localized to the cytoplasm of neuroepithelial cells derived from embryonic day 2 (E2) quail neural tubes and cultured for one day in a chemically defined medium. Specific staining was observed in young sensory neurons in cultures of neural crest clusters as well as in a subpopulation of non-neuronal cells. In cultured E7 dorsal root ganglia, immunostaining was confined to neuronal cell bodies and fibers. In situ, staining of spinal cord and ganglionic neurons appeared on E6 and increased In intensity towards E10. Various mesoderm-derived structures such as the limb buds, the mesenchyme dorsal to the neural tube, the vertebral muscles and cartilage showed specific staining patterns in addition to neural tissue. In agreement with the results of immunocytochemical studies, 1.4 ng bFGF per mg protein was detected in spinal cord extracts by RIA as early as E3, its concentration increased to 8.0ng mg−1 on ES and then to a maximum of 18.0 ng mg−1 protein on E10, this was followed by a subsequent decrease in concentration in older embryos. On the other hand, high levels of bFGF were present in vertebral tissues from E10 onwards. Extracts of immunopositive tissues were subjected to heparin-Sepharose affinity chromatography and eluted in a stepwise salt gradient. Fractions that eluted from the columns at 2 M NaCl contained a bFGF-like protein as revealed by their ability to stimulate the proliferation of ACE cells and by Western blot analysis. These data demonstrate that bFGF is expressed during early nervous system development in both central and peripheral neurons.
Neurons and non-neuronal cells of the dorsal root ganglia (DRG) develop from the neural crest (NC), a transient embryonic structure of vertebrate embryos that individualizes from the closing neural folds that give rise to the neural tube (NT) (Le Douarin, 1982; Weston, 1970). Growing evidence suggests that specific interactions occurring between NT and NC during the first stages of ganglion ontogeny promote survival and/or differentiation of the various cell types constituting the DRG. For instance, mechanical separation of early migrated NC cells of the DRG anlage from the NT is followed by death of only those cells that remain distal to the barrier (Kalcheim and Le Douarin, 1986). Using this experimental system, brain-derived neuro-trophic factor, a CNS-derived protein endowed with survival-promoting action on sensory neurons (Davies et al. 1986; Hofer and Barde, 1988), was found to rescue temporarily a significant amount of NC cells separated from the NT (Kalcheim et al. 1987). In NC cultures, this factor promoted survival of a subpopulation of sub-stance P immunoreactive neurons (Kalcheim and Gen-dreau, 1988). Recently, another CNS-derived mol-ecule, basic fibroblast growth factor (bFGF) was reported to stimulate, both in vivo and in vitro, the survival of a substantial number of HNK-1 immunoreactive non-neuronal cells differentiating from the NC (Kalcheim, 1989), without affecting their initial attachment to the substrate (unpublished observations).
bFGF, a known mitogen and growth factor for mesoderm and neuroectoderm-derived cells (for review see Gospodarowicz et al. 1987; Rifkin and Moscatelli, 1989), has been implicated in the regulation of a variety of embryonic processes such as mesoderm induction (Kimelman and Kirschner, 1987; Kimelman et al. 1988; Slack et al. 1987), eye development (Mascarelli et al. 1987), angiogenesis (Gospodarowicz et al. 1979; Folk-man and Klagsbrun, 1987; Risau, 1986) and muscle differentiation (Clegg et al. 1987; Seed and Hauschka, 1988). In agreement with the proposed biological activities, bFGF has been localized to the yolk and white of unfertilized chick eggs (Seed et al. 1988), to striated muscle cells and their precursors (Joseph-Silverstein et al. 1989), and to developing limb buds (Munaim et al. 1988; Seed et al. 1988).
In addition, the identification of bFGF in extracts of nervous tissue such as embryonic brain (Risau et al. 1988) is suggestive of a role for this growth factor in the development of the nervous system. Evidence for such a role is provided by the findings that bFGF has trophic effects on rat embryonic neurons derived from various CNS regions (Morrison et al. 1986; Walicke, 1988; Walicke et al. 1986), on chick spinal cord and ciliary ganglion neurons (Unsicker et al. 1987) and on rat PC12 pheochromocytoma cells (Wagner and D’Amore, 1986; Neufeld et al. 1987b; Togari et al. 1985).
The acidic form of FGF has similar effects to those reported for bFGF on neuronal survival and neuritic outgrowth of a variety of embryonic neurons, but the concentration required for half-maximal activity in the different systems is significantly higher than that required with the basic form of the molecule (Neufeld et al. 1987b). In addition, the FGF-related proto-onco-gene int-2 is expressed in particular regions of the developing CNS of mouse embryos such as cerebellar Purkinje cells, retina and sensory regions of the inner ear (Wilkinson et al. 1989), suggesting that additional members of the FGF family may be implicated in development of nervous structures.
In the present study, we report the immunocytochemical localization of bFGF in young neurons of the NT and NC, and its continued expression by neurons of the spinal cord and DRG. Confirmation of these results is obtained using a highly sensitive RIA that detects low levels of bFGF in E3 quail NT. The specific concentration of this molecule increases towards E10, and then progressively decreases to hatching time. In addition, heparin-Sepharose (HS)-purified material from embryonic spinal cords is shown to be mitogenic for cultured ACE and baby hamster kidney (BHK-21) cells, and the active fractions comigrate with standard bFGF on Western blots. Taken together, these results support the hypothesis that bFGF, localized in early neurons of the CNS and DRG, may act as a survival factor for non-neuronal precursors in developing ganglia.
Materials and methods
Quail embryos (Coturnix coturnix Japonica) were used for this study and kept in a humidified incubator at 38°C. Embryos were killed at different ages ranging from E2 to El 4.
Cultures of neural tube
The trunk region corresponding to somitic levels 5–15 of 20-to 25-somite-stage embryos was microsurgically excised. Tissues were then incubated in 20 % pancreatin (Grand Island Biological Co., GIBCO) in Ca2+/ Mg2+-free Tyrode buffer for 5–10 min at room temperature until the NT detached from the adjacent mesoderm. Enzymatic digestion was stopped by addition of Dulbecco’s modified Eagle’s medium (GIBCO) containing 10 % heat-inactivated newborn calf serum (GIBCO). Isolated tubes were subsequently washed in Tyrode buffer and transferred to serum-free Basic Brazeau Medium (BBM) (Brazeau et al. 1981). Pooled NT were mechanically dissociated and the equivalent of two tubes was plated in a volume of 2001t1 BBM in the center of 35 mm laminin-coated dishes (NUNC), as previously described (Kal-cheim and Gendreau, 1988).
Cultures of neural crest clusters and DRG cells
Neural crest clusters were isolated from explanted neural tubes at 35–40 h, as described by Loring et al. (1981). Clusters were pooled in BBM and dissociated mechanically. Three thousand cells were cultured in a final volume of 200 μl of BBM in the center of 35 mm laminin-coated wells. DRG were excised from 7-day-old embryos and cultures were prepared as previously described (Dulac et al. 1988). Dissociated cells were cultured in BBM at a density of 5×105 cells m1−1
Cultures of ACE and BH K-21 cells
ACE cells were isolated from bovine adrenal cortex as previously described (Gospodarowicz et al. 1986). BHK-21 cells were cultured and assayed for their responsiveness to bFGF as previously described (Neufeld et al. 1986).
Cultures were fixed with 4 % formaldehyde in phosphate-buffered saline (PBS) pH 7.4 for 30 min at room temperature followed by washing in PBS.
Embryos were killed at different stages, trunk areas were fixed in Bouin’s fluid, embedded in paraffin wax, and 5 μm serial sections were cut and fixed on gelatin-coated glass slides. Before staining, all cultures and sections were preincubated for 15–30 min in a solution of PBS containing 5 % newborn calf serum.
To stain for bFGF, the antibodies used were as follows. (1) Two neutralizing polyclonal antibodies raised in rabbits against human recombinant bFGF; these antibodies were purified on Protein A-Sepharose, and tested for specificity as previously described (Neufeld et al. 1987a; Schweigerer et al. 1987). (2) A monoclonal antibody raised against bovine pituitary bFGF. This antibody was purified on Protein A-Sepharose and tested for specificity using Western blots as described elsewhere (Massoglia et al. 1987). The polyclonal antibodies were used for localization studies both in culture and in situ, the monoclonal was used only on tissue sections. Very similar staining patterns were obtained with the poly-clonal or monoclonal antibodies. Most of the results presented in this work were obtained with the polyclonal antibodies.
Double stainings were performed combining bFGF antibodies with: (1) the HNK-1 monoclonal antibody (Abo and Balch, 1981) that recognizes most neural crest cells and their derivatives, as well as cells in the spinal cord from E3.5 onwards (Lipinski et al. 1983); (2) a monoclonal antibody against the 200 × la3 subunit of neurofilament proteins (Amer-sham, England) to visualize neuronal cells and fibers.
All antibodies were diluted in serum-containing PBS and incubated with cultures or sections for 1 h at room temperature followed by overnight incubation at 4°C. Second antibodies were incubated for 1 h at room temperature. In double-stained preparations, the polyclonal anti-bFGF antibodies were applied first, followed by incubation with goat anti-rabbit lg conjugated with fluorescein isothiocyanate (Nordic, Tilburg, The Netherlands). After washing with PBS containing serum, the HNK-1 antibodies were directly ap-plied. In contrast, if double stained with neurofilament antibodies, cultures were first treated for 10-15 min with 0.25 % Triton in PBS before the antibody was applied. HNK-1 and neurofilament antibodies were visualized by a goat antimouse lg coupled to tetramethyl rhodamine (Sigma, St. Louis).
Appropriate controls, e.g. substituting nonimmune serum for the first antibody, or absorbing the antibody with excess antigen, were performed for the bFGF antibodies on parallel preparations. Sections and cultures were photographed using a Zeiss Axioscope microscope equipped with epi-illumination on TMax film (Kodak) at 400 ASA.
Extraction of embryonic tissues and Heparin-sepharose (HS) chromatography
Neural tubes were isolated from 3-day-old quail embryos by treatment with pancreatin (as described above) to avoid contamination by neighbouring tissues. All the other tissues (spinal cord, cartilage, muscle, limb buds) were excised and collected in ice-cold buffer containing 20 mM Tris-HCI pH 7.0, 0.3 M NaCl, and 0.5 mM PMSF. Pooled tissues were transferred to a dounce homogenizer to which two volumes of the same buffer containing 1 % Triton X-100 were added. Homogenized suspensions were centrifuged at 40.000g for 30 min at 4°C. Supernatants were collected and either used directly for RIA determinations, or subjected to affinity purification on HS columns (2 ml bed volume g-1 wet tissue) (Pharmacia, Sweden). The columns were eluted stepwise with 20 mM Tris pH 7.0 containing increasing salt concentrations, e.g. 0.6, 1.0, and 2.0 M NaCl, as previously described (Neufeld et al. 1987a). Fractions eluting from the HS columns were further analysed using biological and Western blot assays. Protein concentration was determined according to Peterson (19n).
Iodination of bFGF and RIA
Pure bovine-derived pituitary bFGF (Neufeld and Gospodar-owicz, 1985) was iodinated by the chloramine-T method. bFGF (4μg/10μ1) was added to 60μ1 of 0.2M sodium phos-phate buffer pH 7.2, and water to give a final volume of 120 μI.
Na 1251 (0.7 mCi) (Kamag) was subsequently added. The reaction was started by the addition of 8 μI of a freshly prepared 1 mgml-1 chloramine-T solution at room temperature. After 45 s, the reaction was stopped by the addition of 50 μI of 10 mM KI and 50 μI of 1 mg ml-sodium metabisulfite.
Free Na1251 was separated from the 1251-bFGF as previously described (Neufeld and Gospodarowicz, 1985). Specific activity was 170.000-230.000 cts min-1 ng-1 bFGF. RIA for bFGF was performed in triplicate samples according to a previously published protocol (Neufeld et al. 1987a). Each type of extract was tested two times with similar results.
Fractions eluted from the HS column were diluted in PBS containing 0.2 % gelatin. Aliquots (10 μl) were added in duplicate every other day to 1 × 104 BHK-21 cells seeded into 24-cluster wells in a final volume of 1 ml of serum-free medium (Neufeld et al. 1986). Cells were counted in a Coulter counter on day 4. Alternatively, the aliquots were added in duplicate every other day to 5 × la3 ACE cells. ACE cells were counted in a Coulter counter on day 5 (Neufeld et al. 1987a).
Western blot analysis
For the detection of bFGF by Western blot, aliquots of fractions eluted from the HS column were chromatographed on 13 % polyacrylamide NaDodSO4 slab gel under reducing conditions. Proteins were then electroblotted onto nitrocellu-lose at 0.4 amp for 1 h. Transferred proteins were cross-linked to the nitrocellulose using glutaraldehyde (Karey and Sir-basku, 1989), stained with Poinceau red to reveal size markers and exposed to anti-bFGF polyclonal antibodies as previously described (Neufeltl et al. 1987a). Bound antibody was detected using goat anti-rabbit IgG antibodies coupled to alkaline phosphatase (Sigma) (Neufeld et al. 1987a). l ng of bFGF could be detected using this technique.
Immunocytochemical localization of bFGF in the avian nervous system
Neural tubes from 2-day-old embryos were grown in BBM for 1-to-4 days. As early as one day after seeding, neuroepithelial cells of the NT displayed a cytoplasmic pattern of staining with the anti-bFGF antibodies that colocalized with neurofilament protein staining in double-immunolabeled cultures (Fig. 1 A,B). In contrast, bFGF-immunoreactivity on NT cells preceded the expression of the HNK-1 epitope (Fig. 1 C,D) that was detected on these cells only from culture day 3 onwards (not shown). The few HNK-1-positive cells observed in the periphery of the NT cell aggregates were therefore most likely to be NC cells migrating away from the main explant (Fig. 1 D) . In cultures established from NC clusters that contain only NC cells without contamination by somitic cells or NT cells (Loring et al. 1988), bFGF immunoreactivity was observed in the cytoplasm of virtually all HNK-1 positive neurons in BBM, with faint staining also present in the neurites (Fig. 2 A,8). In a previous study, it was demonstrated that these neurons express substance P immunoreactivity as well, and derive from a subpopulation of early committed sensory precursors within the NC (Ziller et al. 1987). Moreover, a significant subpopulation ofHNK-1 immunoreactive cells with flat and polygonal morphology alsostained for bFGF in these cultures (Figs 2A, and 3). Six hundred such cells were counted in seven double-labeled cultures and it was found that 53.4% were positive for both markers, whereas 32% were HNK-1 positive and bFGF negative, and surprisingly, the remaining 14.5 % were bFGF positive and HNK-1 negative. As shown in Fig. 3E. bFGF staining on these non-neuronal cells was unevenly distributed in the cytoplasm, which remained in many cases largely unstained. In contrast to NC cultures, staining of cultured DRG with bFGF antibodies appeared restricted to the neurons (Fig. 2C).
Staining of tissue sections with anti-bFGF antibodies revealed a positive reaction product in the mesenchyme dorsally localized with respect to the NT from E4 onward (Fig. 4,A) but no specific staining at this age could be seen in the NT itself. Staining of the mesenchyme appeared fibrillar, suggesting that the extracellular material rather than the cells was immunoreactive. Specific staining of the nervous system became evident on E6 and increased in intensity towards E10. Neurons of the dorsal and ventral regions of the spinal cord stained cytoplasmically, in contrast to the membranous pattern of staining obtained with the HNK-1 antibody in double-labeled sections (Fig. 4 C,D and E,F). A similar pattern was detected in neurons of the DRG (Fig. 48), in agreement with the results obtained in DRG cultures. Moreover, neurons of the sympathetic ganglia were also stained (data not shown). Replacement of the first antibody by nonimmune serum (data not shown) or preincubation of the anti-human antibodies with a 103.fold-excess human recombinant bFGF completely eliminated reactivity in cultures (Fig. IC), and in tissue sections (not shown).
Immunocytochemical localization of bFGF immunoreactivity in tissues outside the nervous system
bFGF immunoreactivity was found in the limb buds of E4 quail embryos (data not shown). At this age, the myotomes developing into vertebral muscles were also slightly positive. Staining intensity of vertebral muscles increased with embryonic age while the pattern of staining changed. For example, at E7 bFGF immunofluorescence was confined to the cell cytoplasm, whereas at E13 strong immunostaining was observed in extracellular spaces, associated with the membranes ensheathing the muscle fibers (Fig. 5 A,8). In contrast, vertebral cartilage was unstained on E7 (Figs 4B and 5D), but on E13 many cells were cytoplasmically labeled (Fig. SC).
Specific patterns of bFGF immunoreactivity were observed in additional structures of the E7-13 embryos: in blood vessels such as the endothelium and smooth muscle of the dorsal aorta (Fig. 5D) and the endothelium of smaller vessels, smooth muscles in the gut, the basement membrane underlying the surface ectoderm and the epithelial lining of mesonephric tubules and ducts (data not shown).
RIA of bFGF in extracts of spinal cord and vertebral tissues
The amount of bFGF-like immunoreactive material was measured in spinal cord extracts of different embrlonic ages. Small concentrations of bFGF (1.4 ng mg-protein) were already detected in E3 NT (Fig. 6). This measurement was inexact as the concentration of protein in the sample was very low. The RIA signal was low as well, but was within the linear range of the assay. In the rest of the samples, protein concentrations were considerably higher, and the protein quantification is therefore more reliable. The concentration of bFGF-like immunoreactive material increased with age and reached a peak at E10 (l8ngmg-1 protein), followed by a subsequent decrease to low levels close to hatching time (Fig. 6).
On the other hand, the concentration of bFGF-like immunoreactive material in vertebral cartilage was low on E7 (1.8 ng mg-1 protein), and increased progressively to 11.8 ng mg-protein on E10, and to 35 and 33.5 ng mg-1 protein, on E12 and El 4, respectively. A concentration of 4 ng mg-1 protein was present in E4 limb bud extracts. High levels of immunoreactive material, comparable to the amounts detected in vertebral cartilage, were measured in muscle during the last third of the embryonic period. These ranged between 23 and 30ngmg-1 protein.
The amounts of bFGF measured by RIA in tissue extracts showed a temporal pattern consistent with the results of immunocytochemical studies. Thus, it appears that values of 4ngmg-1 protein like those measured in E4 limb buds, and higher levels of bFGF were detectable by immunostaining on sections. In contrast, as was the case for spinal cord and cartilage amounts of factor ranging between 1 and 2 ng mg-1protein were below the threshold of detection by antibody staining.
Effect of embryonic spinal cord extracts on cell proliferation
The RIA determinations and immunocytochemical localization studies described above strongly suggest that bFGF is present in the developing nervous system. However, all of this experimental evidence is based upon detection of bFGF by antibodies. To corroborate this evidence by methods that are not dependent upon immunological recognition, we have extracted E10 spinal cords and subjected them to HS-affinity chromatography (Gospodarowicz et al. 1984; Neufeld et al. 1987a). Following elution with 0.6 Mand 1 M NaCl, the column was eluted with buffer containing 2 M NaCl. Fractions (0.7 ml) that eluted from the column at 2 M NaCl were tested for their ability to stimulate the proliferation of ACE cells (Fig. 7), and BHK-21 cells (not shown). A peak of bioactivity eluted from the column at this stage, whose chromatographic behavior was similar to that of bFGF, and unlike that of aFGF which is known to elute with 1 M NaCl (Gospodarowicz et al. 1984).
Similarly, fractions with bFGF-like activity were detected in the 2 M salt eluates of E12 brain extracts chromatographed upon a HS column (data not shown).
Identification of bFGF by Western blots
The biologically active brain fractions recovered from HS columns upon elution with 2 M salt were subjected to Western blot analysis using an antibody directed against full-length human recombinant bFGF. A single band with a relative molecular mass of 16x 103 was labeled by the antibody in each of the two biologically active fractions (Fig. 8, lanes 1 and 2). This molecular mass corresponds in size to the N-terrninal truncated form of bFGF which is found in the bovine pituitary (Gospod-arowicz et al. 1984), and is lower than that of full-length human bFGF which migrates as a band of 18 × 103 (Fig. 8, lane 3). When the bFGF content of the biologically active spinal cord fractions recovered from the 2 M salt elution of HS columns was analyzed by Western blot, a different profile was observed. All of the biologically active fractions (29-31) contained an im-munoreactive band which migrated at a relative molecular mass of 18 x 103 (Fig. 9, lanes 2-4). This band corresponds in size to that of full-length human 155 amino acid-long bFGF (lane 1) (Abraham et al. 1986). In addition, the fraction that contained the highest bioactivity (fraction 39 in Fig. 7, lane 3 in Fig. 9), revealed an additional band with a relative molecular mass of 16x 103, resembling the band observed in blots of brain extracts.
The present study provides evidence using several complimentary techniques, demonstrating that bFGF is expressed during development in the CNS and peripheral nervous system of avian embryos. bFGF-like immunoreactivity was localized to young neurons developing from the NT and the NC and later in neurons of the spinal cord and peripheral ganglia (Figs 1, 2 and 4). RIA measurements revealed the presence of small quantities of bFGF in the E3 NT and showed that its concentration increased with embryonic age until E10 (Fig. 6). Further evidence for the identification of the factor as bFGF was provided by a one-step purification of E10 spinal cord extracts on a HS column that yielded biologically active fractions in the 2 M salt eluate (Fig. 7). These fractions contained bFGF-like immunoreactive material when analyzed by Western blots (Fig. 9). The biochemical results, however, do not completely exclude the possibility that additional, heparin-binding proteins exist in the embryonic CNS, and might be in part responsible for the mitogenic activity measured in cultures of ACE cells.
The presence of bFGF-like immunoreactivity was detected at an age equivalent to E3 both in NT cells in vitro and in NT extracts by RIA. In contrast, immuno-localization in situ was only possible from E6 onwards. Several reasons may account for the delay observed. On one hand, it is possible that the levels of factor expressed by the cells in vivo are below the threshold of detection by immunocytochemical means. Indeed, RIA measurements showed low concentrations of bFGF-like material at E3. On the other hand, the conditions used for fixation and embedding may partly reduce the antigenicity of cytoplasmic bFGF.
The finding that a bFGF-like molecule is present in cells of the young NT supports the hypothesis that CNS-derived factors like bFGF can stimulate the survival of non-neuronal cells during the early stages of DRG formation (Kalcheim, 1989). Since ganglionic precursors migrate and organize in close proximity to the NT, a short-range action of bFGF could account for this survival effect. Whether bFGF is secreted together with heparan sulfate proteoglycan from NT cells and deposited in the surrounding ECM, as proposed for other systems (Vlodavsky et al. 1987), or whether the factor leaks out of the producer cells (Schweigerer et al. 1987), or is secreted in a yet unknown manner remains to be elucidated. It is possible that additional factors produced by cells of the developing CNS, like brainderived neurotrophic factor, exert their action on NC-derived neurons via specific interactions with ECM proteins.
A progressive increase in the concentration of bFGF towards E10 followed by a decrease towards E13, was determined by RIA of spinal cord extracts. This temporal pattern of bFGF expression might be correlated with the formation or stabilization of neuromuscular contacts taking place around E10. This however, might not be the only function of bFGF at this stage in development, since the factor is found in neurons of both the dorsal as well as the ventral horns of the spinal cord (Fig. 4).
bFGF, localized in embryonic spinal cord neurons (Figs 1 and 4), was implicated in promoting survival of E6 chick spinal cord motoneurons (Unsicker et al. 1987). This suggests the possibility of an autocrine mechanism of action of bFGF in the CNS. Such a mechanism has been previously demonstrated for BHK cells transfected with a bFGF expression vector (Neu-feld et al. 1988).
We have detected a bFGF-like mitogenic activity in embryonic spinal cord and brain extracts confirming the results obtained by immunological methods. Recently, both the acidic and basic forms of FGF have been found in E14 and adult chick brains (Risau et al. 1988). No attempt has been made in this work to study the ontogeny of bFGF expression in the brain. Yet, other investigators have reported that the factor is localized in neurons of different regions of 18-to 20-day-old rat brains, although the antibodies used in that study crossreacted with acidic FGF (Pettmann et al. 1986). In the brain, bFGF may serve as a neurotrophic factor for multiple types of neurons (Walicke, 1988) as well as glial cells (Weibel et al. 1985).
Peripheral neurons differentiating from the NC in serum-free conditions also express bFGF-like immunoreactivity (Fig. 2). Thus, upon differentiation from E4 onwards, DRG neurons may be an additional or alternative source of bFGF for the non-neuronal cells in the sensory ganglia.
In addition to neurons, a significant subpopulation of HNK-1 positive cells with non-neuronal morphology (53.4%) is also FGF immunoreactive, as well as 14.5 % of similar cells that are HNK-1 negative. Because in DRG cultures staining is confined to neurons (Fig. 2C), the possibility exists that at least partof these positively stained non-neuronal cells in NC cultures represent neuronal precursors.
Outside the nervous system, bFGF was localized by RIA to the limb buds (data not shown), in agreement with previously reported findings (Munaim et al. 1988; Seed et al. 1988). In a recent report, antibodies raised against human bFGF stained the cytoplasm of striated muscle cells and their myotomal precursors (Joseph-Silverstein et al. 1989). In agreement with this report, we have localized bFGF immunoreactivity to the cytoplasm of E7 vertebral muscles (Fig. 5). In contrast, staining of E13 sections revealed that most of the staining was associated with the muscle ECM (Fig. 5). To our knowledge, bFGF has not been found previously in embryonic muscle ECM in vivo, although it was localized to the endomysium of skeletal muscles in normal as well as dystrophic adult mice (DiMario et al. 1989). bFGF staining product appeared associated with the ECM in various other locations, for instance, in the subectodermal basement membrane of E7-13 embryos and in the mesenchyme dorsally localized with respect to the NT in E4 and ES embryos (Fig. 4 A). It is important to determine whether, in these instances, the ECM contains heparan sulfate proteoglycans, known for their ability to bind bFGF, stabilize it against proteolytic degradation (Gospodarowicz and Cheng, 1986; Moscatelli, 1988; Saksela et al. 1988; Vlodavsky et al. 1987) and modulate its activity (Neufeld et al. 1987b).
From the distribution patterns of bFGF observed by us and by others, it is tempting to hypothesize that this growth factor plays a role in the development of multiple embryonic tissues including the nervous systern. In the axis comprising the developing spinal cord and peripheral ganglia, neuronally derived bFGF may participate in modulation of survival and differentiation of NC-derived non-neuronal cells.
We thank Ms Chana Carmeli for excellent technical assistance, and Dr Ron Goldstein for critical reading of the manuscript. Human recombinant bFGF was kindly provided by California Biotechnology Corporation. This work was supported by grants from the Muscular Dystrophy Association of America, the Israel Academy of Sciences and Humanities (to C.K) and from the Israel-United States Binational Science Foundation and the Israel Cancer Research Fund (to G.N).
adrenal cortex-derived capillary endothelial cells
Basic Brazeau medium
basic fibroblast growth factor
baby hamster kidney cells
central nervous system
dorsal root ganglia