PDGF receptors on cells of the oligodendrocyte-type-2 astrocyte (O-2A) cell lineage

The rat optic nerve contains three types of macroglial cells, which develop on a predictable schedule. Type-1 astrocytes first appear at embryonic day 16 (E16), oligodendrocytes at the time of birth and type-2 astrocytes in the second postnatal week (Miller et al. 1985). Studies in culture suggest that oligodendrocytes and type-2 astrocytes develop from a common, bipotential O-2A progenitor cell, while type-1 astrocytes develop from a separate precursor cell (Raff et al. 1983, 1984a).

A clue as to how O-2A progenitor cell differentiation is controlled is provided by comparing the behaviour of these cells in vitro and in vivo. Whereas O-2A progenitor cells in vivo proliferate extensively before differentiating into postmitotic oligodendrocytes (Skoff et al. 1976a,b; Noble & Murray, 1984) or type-2 astrocytes (Miller et al. 1985), when dissociated from the developing optic nerve and placed in culture, the progenitor cells prematurely stop dividing and differentiate within 2 days: if cultured in 10% fetal calf serum (FCS) they become type-2 astrocytes; if cultured without FCS they become oligodendrocytes (Raff et al. 1983, 1984b). These findings indicate that both the timing and direction of O-2A progenitor cell differentiation can be dramatically influenced by environmental signals. The oligodendrocyte pathway of differentiation seems to be the constitutive pathway, which is triggered automatically when a progenitor cell is deprived of signals from other cells: if a single progenitor cell is cultured alone in a microwell in <1% FCS, it prematurely stops dividing and differentiates into an oligodendrocyte (Temple & Raff, 1985). Type-2 astrocyte differentiation in vitro, by contrast, requires the presence of inducing factors (Temple & Raff, 1985; Hughes & Raff, 1987; Hughes et al. 1988).

The normal timing of oligodendrocyte differentiation can be reconstituted in vitro by coculturing embryonic O-2A progenitor cells with purified type-1 astrocytes, or by adding medium conditioned by type-1 astrocytes (ACM) (Noble & Murray, 1984; Raff et al. 1985). Now the progenitor cells continue to divide, oligodendrocytes first differentiate on the equivalent of the day of birth, and new postmitotic oligodendrocytes continue to develop from dividing progenitor cells for at least two weeks, just as in vivo (Raff et al. 1985). How do type-1 astrocytes stimulate O-2A progenitor cells to divide and how do they influence the timing of oligodendrocyte development in culture? Four lines of evidence suggest that these effects of type-1 astrocytes are mediated by PDGF. (1) Purified PDGF can replace type-1 astrocytes or ACM, both in stimulating the proliferation of O-2A progenitor cells (Noble et al. 1988; Richardson et al. 1988) and in reconstituting the normal timing of oligodendrocyte development in culture (Raff et al. 1988). (2) Cultures of purified type-1 astrocytes secrete PDGF and contain messenger RNA encoding PDGF A chains (Richardson et al. 1988). (3) If ACM is fractionated by gel filtration, the mitogenic activity for O-2A progenitor cells is found in the same fractions as radiolabelled PDGF (Richardson et al. 1988). (4) Anti-PDGF antibodies inhibit the ability of ACM, both to stimulate O-2A progenitor cell proliferation (Richardson et al. 1988; Noble et al. 1988) and to reconstitute the normal timing of oligodendrocyte development in vitro (Raff et al. 1988). Although it remains to be demonstrated that PDGF functions in these ways in vivo, it is apparently present in the developing optic nerve: extracts of developing optic nerve stimulate O-2A progenitor cells to divide in culture and most of this activity is inhibited by anti-PDGF antibodies (Raff et al. 1988).

How does PDGF contribute to the normal timing of oligodendrocyte differentiation? Experiments in vitro suggest that PDGF keeps O-2A progenitor cells dividing until an intrinsic clock in these cells initiates the process leading to oligodendrocyte differentiation (Raff et al. 1985). For example, clonal analyses show that the descendants of a single progenitor cell usually differentiate more or less synchronously after the same number of divisions when stimulated to divide by type-1 astrocytes or PDGF (Temple & Raff, 1986; Raff et al. 1988). The nature of the clock is unknown, but it apparently controls when O-2A progenitor cells become unresponsive to PDGF: even if an excess of PDGF is added to cultures of embryonic optic nerve, progenitor cells stop dividing and differentiate into oligodendrocytes on schedule (Raff et al. 1988).

These observations raise a number of questions about the response of O-2A progenitor cells to PDGF. Does PDGF act directly on progenitor cells or via an intermediary cell? What is the nature of the PDGF receptor expressed by the PDGF-responsive cells in the optic nerve? Two types of PDGF receptor have been identified on cells of the human foreskin fibroblast line AG 1523 (Heldin et al. 1988): type A receptors bind all three forms of dimeric PDGF (AA, BB and AB) with high affinity, while type B receptors bind BB with high affinity, AB with lower affinity and AA with very low affinity (Heldin et al. 1988; Hart et al. 1988; Escobedo et al. 1988; Gronwald et al. 1988; Claesson-Welsh et al. 1988). Does the initial loss of mitotic responsiveness to PDGF reflect a loss of PDGF receptors on O-2A progenitor cells, or is the block in the response pathway distal to the receptors?

In the present study, we have attempted to answer these questions by analysing the nature and distribution of PDGF receptors on optic nerve cells in vitro, using ¹²⁵I-PDGF. We show that O-2A progenitor cells express PDGF receptors (which resemble the type A receptors on human fibroblasts in their ability to bind all three forms of PDGF), suggesting that PDGF acts directly on these cells. We also show that many newly formed oligodendrocytes express PDGF receptors although they do not divide in response to PDGF; this makes it unlikely that receptor loss is the reason that progenitor cells initially lose mitotic responsiveness to PDGF.

Optic nerves were dissected from postnatal day 7 (P7) or newborn (PO) Sprague-Dawley rats and dissociated into single cells using trypsin, EDTA and collagenase as previously described (Miller et al. 1985). Approximately 20000-40000 cells were cultured on poly-D-lysine-coated glass coverslips in 0-5 ml Dulbecco’s modified Eagle’s medium (DMEM), containing 0-5% FCS and additives modified from Bottenstein & Sato (1979), as previously described (Miller et al. 1985). In some cases the cultures were treated with human PDGF (1 ng ml⁻¹). To study fresh suspensions of newborn optic nerve cells, the nerves were dissociated with EDTA and collagenase, in the absence of trypsin.

The major component of human PDGF (hPDGF) is a heterodimer of one A and one B chain with variable proportions of the homodimers (Hammacher et al. 1988a). The hPDGF used in our binding studies (both labelled and unlabelled) was purified from human platelets and contained ∼ 70% PDGF-AB and ∼30% PDGF-BB. Recombinant PDGF-AA and PDGF-BB were purified from supernatants of yeast cells transfected with recombinant DNA vectors containing the coding sequence for the human PDGF A chain or B chain, as previously described (Ostman et al. 1989). The hPDGF used to treat cultures was of unknown dimeric composition and was obtained from R and D Systems Inc. Basic fibroblast growth factor (FGF) and epidermal growth factor (EGF) were obtained from Collaborative Research Inc. hPDGF and PDGF-BB were labelled with ¹²⁵I by the Bolton-Hunter method (Bolton & Hunter, 1973) to a specific activity of ∼37 000 cts min⁻¹µ g∼¹ and ∼57 000 cts min⁻¹ ng⁻¹, respectively. PDGF-AA was labelled with ¹²⁵I by the chlora-mine-T (Hunter & Greenwood, 1962) method to a specific activity of ∼64000 cts min⁻¹ ng⁻¹. Some experiments were carried out using human ¹²⁵I-PDGF of unknown dimeric composition (specific activity ∼40 000 cts min⁻¹ ng⁻¹) obtained from PDGF Inc.

Cultures of optic nerve cells on coverslips were washed once in DMEM containing 10% FCS, 0·1% bovine serum albumin (BSA), and made up to pH7-4 (binding buffer). The cells were then incubated for 1 h at room temperature in 30 µl binding buffer containing ¹²⁵I-PDGF at a concentration of 33 ng ml⁻¹, unless otherwise stated. This concentration of ¹²⁵I-PDGF was at least 20-fold higher than the concentration of PDGF in 10% FCS. FCS was used in the binding buffer because it helped the cells retain their original morphology during the radiolabelling procedure; in this way morphology as well as antigenic phenotype could be used to distinguish the various cell types in autoradiographs. In competition studies, 100-fold excess of unlabelled hPDGF, EGF or FGF were added with the ¹²⁵I-PDGF.

For receptor down-regulation experiments, the cells were preincubated for Ih at 37 °C with 100 ng ml⁻¹ of unlabelled PDGF-AA, PDGF-BB, or with the vehicle used to reconstitute the lyophilized PDGF (10mm-acetic acid with 1mg ml⁻¹ BSA). Before being exposed to ¹²⁵I-PDGF as above, the cells were washed once in binding buffer containing 20mm-acetic acid, pH 3·75 (in order to dissociate receptor-bound unlabelled PDGF at the cell surface) and then in binding buffer.

After radiolabelling, the cells were fixed with 4% paraformaldehyde, which was added during the last 4 min of incubation with ¹²⁵I-PDGF. The cells were then washed three times in minimal Eagle’s medium buffered with 0-02M-Hepes (MEM-Hepes) and containing 10% FCS and 0·1% sodium azide, and once in MEM-Hepes. Washing before fixation gave much poorer cell preservation and did not significantly reduce non-specific binding. The cells were then immunolabelled, as described below, and the coverslips were mounted (with cells uppermost) on glass slides using Fluorospar (BDH). One day later, the slides were dipped in K2 emulsion (Ilford), fan-dried at room temperature, and exposed at 4°C for 6 days before they were developed in Contrast-FF (Ilford) and fixed in Hypam (Ilford). A second coverslip was mounted in Citifluor (Citifluor) on top of the cells and was sealed with nail varnish. The cells were examined in a Zeiss Universal fluorescence microscope using a x63 objective and photographed using Kodak Tri-X film (ASA 400). The macroglial cell types were identified by their characteristic morphologies and antigenic phenotypes: astrocytes were labelled by antibodies against glial fibrillary acidic protein (GFAP) (Bignami et al. 1972), oligodendrocytes by anti-galactocerebroside (GC) antibody (Raff et al. 1978), and O-2A progenitor cells by the A2B5 monoclonal antibody (Raff et al. 1983). Unless stated otherwise, each experiment was carried out in triplicate and at least 100 cells were counted in each experiment.

After fixation with 4% paraformaldehyde, but before dipping in emulsion, cells were surface-stained, either with A2B5 monoclonal antibody (Eisenbarth et al. 1979; ascites fluid diluted 1:100) followed by rhodamine-coupled goat antimouse Ig (G anti-MIg-Rd, Cappel, diluted 1:100) or with monoclonal anti-GC antibody (Ranscht et al. 1982) followed by fluorescein-coupled, class-specific, goat anti-mouse IgG3 (G anti-MIgG3-Fl, Nordic, diluted 1:100). The cells were then postfixed in methanol at −20°C for 2 min. In some experiments, cells were stained with anti-GFAP antiserum (Pruss, 1979). In these cases the cells were not fixed with paraformaldehyde: after washing, they were fixed in methanol at − 20°C for 2min, and then stained with rabbit anti-GFAP serum (diluted 1:1000) followed by fluorescein-coupled sheep anti-rabbit Ig (S anti-RIg-Fl, Wellcome, diluted 1:100).

For cross-competition experiments, 1-day-old cultures of P0 optic nerve cells were incubated with ¹²⁵I-PDGF-AA or ¹²⁵I-PDGF-BB, with or without a 100-fold excess of unlabelled PDGF-AA or PDGF-BB, in 30 µl phosphate-buffered saline containing 1 mg ml ⁻¹ BSA, 0·01 mg ml ⁻¹ CaCl₂·2H₂O and 0.01mg ml⁻¹ MgSO₄·7H₂O, pH7·4 (PBS-BSA). Binding was terminated by washing three times in PBS-BSA at 4°C. The cells were then lysed with 30 µ of an aqueous solution of 20mm-Hepes, pH 7·4, containing 1% Triton X-100, 10% (vol/vol) glycerol, and BSA (0·1 mg ml⁻¹) for 30min at room temperature. The radioactivity of the lysate was then counted in a γ-counter (Nuclear Enterprises).

For Scatchard analysis, 1-day-old cultures of P7 optic nerve cells were incubated with ¹²⁵I-hPDGF, as for autoradiography, but the cells were not fixed; they were washed three times in MEM-Hepes with 10% FCS and 0·1% sodium azide, once in MEM-Hepes, and then lysed as above. Residual bound ¹²⁵I-PDGF was solubilized from the coverslips with 0·3m-sodium hydroxide and the lysates were counted in a γ-counter.

P0 optic nerve cells were cultured for 1 day with hPDGF (2 ng ml⁻¹) and bromodeoxyuridine (BrdU, Boehringer) was added to a final concentration of 5X10⁻⁵M for the entire period, or for the last 6−12 h. BrdU is incorporated into replicating DNA (Gratzner, 1982). After fixation with 4% paraformaldehyde for 2min, the cells were stained with anti-GC antibody followed by G anti-IgG3-Fl, and then postfixed in acid-alcohol (5% glacial acetic acid: 95% ethanol) at·20°C for 10 min. They were then treated with 2N-HC1 to denature the nuclear DNA (Yong & Kim, 1987), followed by 0·1 M-Na₂B₄O₂, pH8·5, both for lOmin at room temperature, and then labelled with monoclonal anti-BrdU antibody (Magaud & Mason, 1988) (culture supernatant diluted 1:5 in PBS containing 1% Triton X-100 and 2% FCS) followed by G anti-MIg-Rd (diluted 1:100 in the same solution). The cells were mounted and examined as described above.

P0 optic nerve cells were cultured for 1 day, incubated with ¹²⁵I-hPDGF (33 ng ml⁻¹), and then stained for A2B5, GC or GFAP expression. Cells were judged to be radiolabelled if there were more than 10 silver grains over them. About 60% of O-2A progenitor cells and about 50% of oligodendrocytes, but no type-1 astrocytes, were radiolabelled (Fig. 1). Less than 5% of the non-macroglial cells (mainly meningeal cells, macrophages and endothelial cells) were radiolabelled. Therefore, most of the cells that bound ¹²⁵l-PDGF belonged to the O-2A lineage (Table 1). The percentage of radiolabelled O-2A progenitor cells increased with increasing concentrations of ¹²⁵I-hPDGF, reaching a plateau at around 30 ng ml⁻¹ (Fig. 2). To demonstrate the specificity of binding, the cells were incubated with 33 ng ml⁻¹ of ¹²⁵I-hPDGF in the presence of a 100-fold excess of unlabelled hPDGF, FGF or EGF. While FGF and EGF had no detectable effect on the binding, hPDGF greatly reduced the percentage of radiolabelled cells (Table 1) as well as the average number of silver grains over the radiolabelled cells (not shown).

To determine whether PDGF receptors are present on O-2A progenitor cells in vivo as well as in vitro, cells were dissociated from PO optic nerves without trypsin and immediately incubated with ¹²⁵I-hPDGF (33 ng ml⁻¹), with or without a 100-fold excess of unlabelled hPDGF. A mean of 66 ± 1% of the A2B5⁺ cells were radiolabelled in the absence of unlabelled PDGF, while 11 ± 2% were radiolabelled and the average number of silver grains overlying the cells was lower in its presence (n = 3). When PO optic nerves were dissociated with trypsin in the normal way, and then immediately incubated with ¹²⁵l-hPDGF, no radiolabelled cells were seen, indicating that the PDGF receptors on optic nerve cells are trypsin sensitive, as previously shown for other cell types (Bowen-Pope & Ross, 1985).

To quantify the interaction between PDGF and its receptors, Scatchard analysis of binding data obtained on optic nerve cultures by γ -counting was carried out. There are at least two problems with such studies on optic nerve cultures: the cultures contain a number of different cell types and it is difficult to obtain large numbers of cells. In order to maximize the number of O-2A progenitor cells for analysis, we used P7 optic nerve cells cultured for one day with hPDGF (1 ng ml⁻¹). This low concentration of PDGF would not be expected to significantly down-regulate PDGF receptors (Heldin et al. 1982), but is sufficient to keep O-2A progenitor cells dividing and to prevent their premature differentiation into oligodendrocytes (Noble et al. 1988; Richardson et al. 1988). Each of these cultures contained about 4000 O-2A progenitor cells, about 1500 oligodendrocytes, and no type-2 astrocytes. When the cultures were incubated in ¹²⁵I-hPDGF and studied by autoradiography, 84 ± 3% of the O-2A progenitors and 43 ± 4% of the oligodendrocytes were radiolabelled; non O-2A lineage cells accounted for less than 7% of the radiolabelled cells (n = 3). Therefore, more than 90% of the binding of ¹²⁵I-hPDGF assessed by γ-counting probably represented binding to O-2A progenitors and oligodendrocytes.

When cultures of P7 optic nerve cells were incubated with increasing concentrations of ¹²⁵I-hPDGF and then assayed by γ-counting, the binding tended to saturate at 25-40X10⁻¹°M and was half maximal at 3−4X10⁻¹⁰M (Fig. 3A). The Scatchard plot of the binding data (Scatchard, 1949) was linear (Fig. 3B), suggesting a single class of receptor. The apparent K_d was about 3·45×10⁻¹⁰M and the average number of receptors per cell was about 2·2×10⁵ (Fig. 3B). To assess the influence of the 10% FCS on the binding of ¹²⁵1-PDGF, a 3-point assay was done in the absence of FCS: the binding increased about 1·4-fold in the absence of FCS (not shown).

The results obtained with ¹²⁵I-PDGF-AA and ¹²⁵I-PDGF-BB in autoradiography experiments on 1-day-old cultures of P0 optic nerve cells (Table 2) were very similar to those obtained with ¹²⁵I-hPDGF (see Table 1): with all 3 ligands 60−65% of the O-2A progenitor cells were radiolabelled and most of the radiolabelling was displaceable by an excess of hPDGF.

Moreover, in γ-counting experiments on 1-day-old cultures of P0 optic nerve cells, the binding of ¹²⁵I-PDGF-AA and the binding of ¹²⁵I-PDGF-BB were similar and both ligands were displaced about equally by both unlabelled PDGF-AA and unlabelled PDGF-BB (Table 3). These results suggest that the PDGF receptors on rat O-2A lineage cells resemble the type A PDGF receptors described previously on human fibroblasts that bind the three PDGF dimers about equally; type B receptors on fibroblasts bind PDGF-BB better than PDGF-AB, and PDGF-AA hardly at all (Heldin et al. 1988; Hart et al. 1988; Escobedo et al. 1988; Gronwald et al. 1988; Claesson-Welsh et al. 1988).

To further characterize the binding specificity of the PDGF receptors on O-2A progenitor cells, we carried out receptor down-regulation experiments, which have been used previously to identify the type of PDGF receptor present on fibroblasts (Heldin et al. 1988; Hart et al. 1988). Preincubation of P0 optic nerve cultures with either unlabelled PDGF-AA or unlabelled PDGF-BB for 1 h at 37 °C (followed by acid washing to remove any receptor-bound unlabelled PDGF exposed on the cell surface) reduced the proportion of O-2A progenitor cells that bound ¹²⁵I-PDGF-AA or ^l25I-PDGF-BB by 60−80%, as assessed by autoradiography (not shown). PDGF-AA was just as efficient as PDGF-BB in down-regulating the receptors that bound ¹²⁵I-PDGF-BB, confirming that the PDGF receptors on O-2A progenitor cells resemble the type A PDGF receptor on human fibroblasts, at least in terms of their ligand specificity.

To determine if the loss of O-2A progenitor cell mitotic responsiveness to PDGF results from a loss of PDGF receptors, P0 optic nerve cells were cultured in a low but mitogenic concentration (1 ng ml⁻¹) of unlabelled hPDGF. After 1 day, the binding of ¹²⁵I-hPDGF to oligodendrocytes was determined by autoradiography. Presumably all of the oligodendrocytes in these cultures will have differentiated on schedule, having lost responsiveness to PDGF (Raff et al. 1985, 1988; and see below). Yet in 3 experiments, 35 out of 73 oligodendrocytes (47 ± 4%) were radiolabelled, suggesting that the loss of PDGF receptors is not a prerequisite for oligodendrocyte differentiation. When P0 optic nerve cells were studied after 3 days in culture without PDGF, only 12 ± 1% of the oligodendrocytes were radiolabelled, suggesting that oligodendrocytes lose PDGF receptors with time.

In order to study PDGF receptor expression on type-2 astrocytes, P7 optic nerve cells were cultured for 3 days in 10% FCS to induce most of the O-2A progenitor cells to differentiate into type-2 astrocytes (Raff et al. 1983, 1984b). The cells were then incubated with ¹²⁵I-hPDGF and analysed by autoradiography. In three experiments, 74 ± 6% of the type-2 astrocytes were radiolabelled. Thus, like oligodendrocytes, most type-2 astrocytes retain PDGF receptors, at least initially, after they differentiate in culture.

In order to determine whether PDGF can stimulate newly developed oligodendrocytes to synthesize DNA, we cultured P0 optic nerve cells with hPDGF (2 ng ml⁻¹) for 24 h and added BrdU for the last 12 h of the culture period. The cells were then immunolabelled for GC expression and BrdU incorporation. Although more than 15% of the cells in the cultures were BrdU⁺, all of the oligodendrocytes were BrdU⁻ (239 oligodendrocytes. were examined in 6 cultures). These results suggest that PDGF is not mitogenic for newly developed oligodendrocytes, even though many of them have PDGF receptors on their surface.

When BrdU was added to cultures for 24 h, 5% of the oligodendrocytes were BrdU⁺; these presumably developed from O-2A progenitor cells that had incorporated BrdU before they differentiated into oligodendrocytes.

We have demonstrated that most O-2A progenitor cells in the developing rat optic nerve have PDGF receptors on their surface. Moreover, most of the cells in cultures of newborn rat optic nerve that have detectable PDGF receptors are O-2A lineage cells: type-1 astrocytes and the great majority of non-macroglial cells in these cultures did not bind detectable amounts of ¹²⁵I-PDGF. These findings suggest that PDGF acts directly on O-2A progenitor cells to stimulate their proliferation and prevent their premature differentiation into oligodendrocytes (Richardson et al. 1988; Noble et al. 1988; Raff et al. 1988).

Scatchard analysis of PDGF binding to mainly O-2A lineage cells in P7 optic nerve cultures suggests that these cells have a single class of high affinity receptors with a K_d of at most 3·5×10⁻¹⁰M. While this value falls within the reported range of K_ds (10⁻⁹−10⁻¹¹ M) for PDGF receptors on other cell types (Bowen-Pope & Ross, 1985), it is about 20-fold greater than the concentration of PDGF required for half-maximal stimulation of O-2A progenitor cell proliferation in culture (about 0·2×10¹¹M) (Richardson et al. 1988). While at least some of this discrepancy is probably caused by technical problems in the binding assays, such as the presence of 10% FCS and perhaps the loss of labelled ligand during washing, it raises the possibility that O-2A progenitor cells, like human fibroblasts (Heldin et al. 1981), have ‘spare’ PDGF receptors, so that only a small fraction of the receptors on a cell need be occupied to produce a maximal proliferative response.

Dissociation of bound ¹²⁵I-PDGF during the multiple washes required for immunostaining may account, at least in part, for the finding that only 60−65% of the O-2A progenitor cells in newborn optic nerve were labelled by the radioactive PDGF in autoradiographs; about 80% of progenitor cells in the newborn nerve can be stimulated by PDGF in culture to synthesize DNA (unpublished observations). Scatchard analysis of the binding data generated by γ-counting without immunolabelling indicated that each O-2A lineage cell, on average, has about 2·2×10⁵ receptors (which is similar to the estimate of 3X10⁵ PDGF receptors on human foreskin fibroblasts in culture, Heldin et al. 1981), while with immunolabelling the maximum number of silver grains detected by autoradiography over a radiolabelled cell was only about 100. Thus the combination of immunostaining and autoradiography was relatively inefficient at detecting PDGF receptors, so that the proportions of radiolabelled cells determined in these studies should be taken as minimal estimates.

Two types of PDGF polypeptides, A chains and B chains, have been described (Johnsson et al. 1982), each encoded by a separate gene (Dalla Favera et al. 1982; Betsholtz et al. 1986). At least three forms of PDGF occur naturally, each consisting of two polypeptide chains bound together by disulphide bonds (Johnsson et al. 1982): human platelets, for example, contain mainly PDGF-AB (Hammacher et al. 1988a), some human osteosarcoma (Heldin et al. 1986), melanoma (Westermark et al. 1986) and glioma (Hammacher et al. 1988b), cell lines make PDGF-AA, while PDGF isolated from porcine platelets resembles PDGF-BB (Stroobant & Waterfield, 1984). In addition, there are at least two types of PDGF receptors that have been distinguished on human fibroblasts by their size, ligand specificity, and by reactivity with monoclonal antibodies (Heldin et al. 1988; Hart et al. 1988). The type A receptor binds all three PDGF dimers, whereas the type B receptor binds PDGF-BB with high affinity, PDGF-AB with lower affinity, and PDGF-AA hardly at all (Heldin et al. 1988; Hart et al. 1988; Escobedo et al. 1988; Claesson-Welsh et al. 1988). DNA sequencing studies indicate that the type B receptor, like several other growth factor receptors (Yarden & Ullrich, 1988), is a transmembrane tyrosine kinase (Yarden et al. 1986; Claesson-Welsh et al. 1988; Gronwald et al. 1988). The type A receptor has not been structurally or functionally characterized.

Two findings in the present study indicate that, in terms of ligand specificity, the PDGF receptors on O-2A progenitor cells in the newborn rat optic nerve resemble fibroblast type A PDGF receptors. First, iodinated PDGF-AA, PDGF-BB and hPDGF all bound about equally to O-2A progenitor cells, and the binding could be displaced about equally by an excess of unlabelled PDGF-AA, PDGF-BB or hPDGF. Second, preincubation with either unlabelled PDGF-AA or PDGF-BB were equally effective at down-regulating most of the PDGF receptors on O-2A progenitor cells, as assessed by the binding of either ¹²⁵I-PDGF-AA or ¹²⁵I-PDGF-BB in autoradiographs.

Separate functional studies indicate that type-A-like receptors are mainly, and perhaps exclusively, responsible for mediating the proliferative response of O-2A progenitor cells to PDGF: PDGF-AA is at least as potent as PDGF-AB in stimulating progenitor cell proliferation in cultures of P7 optic nerve, whereas PDGF-BB is 5- to 10-fold less potent (Pringle et al. 1989). These findings, taken together with our present binding studies, and the observations that type-1 astrocytes in culture (Richardson et al. 1988) and probably in vivo (Pringle et al. 1989) make PDGF A-chain mRNA but little, if any, B-chain mRNA, make a persuasive case for a crucial role for PDGF-AA, acting via type-Alike PDGF receptors, in the stimulation of O-2A progenitor cell proliferation in the developing rat optic nerve. By contrast, only the type B receptors seem to mediate the proliferative response of human fibroblasts to PDGF (Heldin et al. 1988; Hart et al. 1988).

Although O-2A progenitor cells are stimulated to proliferate by PDGF in cultures of perinatal optic nerve, they do not proliferate indefinitely: even in the continuing presence of an excess of PDGF, the progenitor cells eventually stop dividing and (if the concentration of FCS is <1%) differentiate into postmitotic oligodendrocytes (Raff et al. 1988). The molecular mechanism responsible for this loss of mitotic responsiveness to PDGF is not known, but our present results suggest that it is not the loss of PDGF receptors: at least half of the oligodendrocytes that developed in cultures of newborn optic nerve, either in the presence or absence of PDGF, had readily detectable PDGF receptors, even though these cells no longer divided in response to PDGF. The block in mitotic responsiveness to PDGF must therefore lie downstream from the receptors. On the other hand, the proportion of oligodendrocytes that had PDGF receptors decreased about 4-fold over three days in culture, suggesting that these cells eventually lose their PDGF receptors. When O-2A progenitor cells were induced by 10% FCS to differentiate into type-2 astrocytes (Raff et al. 1983), most of these newly formed astrocytes also retained their PDGF receptors.

In summary, we have provided evidence that O-2A progenitor cells respond directly to PDGF by means of type-A-like PDGF receptors and that receptor loss is not the reason that these cells eventually lose mitogenic responsiveness to PDGF.

We thank Julia Burne, Elian Collarini, David Colquhoun, Laura Lillien and Anne Mudge for advice, and David Mason for the anti-BrdU antibody. I.K.H. is supported by a MRC Training Fellowship.