Heparan sulfate proteoglycan and FGF receptor target basic FGF to different intracellular destinations

Basic fibroblast growth factor (bFGF) is a prototype of the heparin binding growth factor family, which is composed of aFGF (FGF-1), bFGF (FGF-2), int-2 (FGF-3), K-FGF (FGF-4), FGF-5, FGF-6 and KGF (FGF-7) (Burgess and Maciag, 1989; Finch et al., 1989; Marics et al., 1989). Basic FGF promotes the proliferation and differentiation of a wide range of cells of mesenchymal and neuroectodermal origin and is an important regulator of angiogenesis and wound healing (Burgess and Maciag, 1989). Basic FGF binds to a family of tyrosine kinase-containing FGF receptors (FGFRs) that includes at least four members, designated FGFR-1 through FGFR-4 (Lee et al., 1989; Dionne et al., 1990; Houssaint et al., 1990; Partanen et al., 1991; Pasquale, 1990; Reid et al., 1990; Keegan et al., 1991; Stark et al., 1991). In addition, splicing variants have been found for these receptors that have altered affinities for members of the FGF family (Johnson et al., 1991; Miki et al., 1992; Werner et al., 1992).

Cells also bind bFGF through the heparan sulfate chains of cell surface proteoglycans, such as syndecan-1, a well characterized heparan sulfate proteoglycan, which has been shown to bind bFGF with a dissociation constant in the nmolar range (Kiefer et al., 1991). Syndecan-1 is a member of a family of related proteoglycans (syndecans 1-4) that are defined by a highly conserved cytoplasmic domain (reviewed by Bernfield et al., 1992). Basic FGF also binds to other cell surface heparan sulfate proteoglycans, including a lipid linked proteoglycan (Brunner et al., 1991). Heparan sulfate proteoglycans are obligate partners in binding of FGFs to their FGFRs; bFGF, aFGF and K-FGF do not bind to FGFRs unless heparan sulfate, or its analog heparin, is present (Rapraeger et al., 1991; Yayon et al., 1991; Olwin and Rapraeger, 1992; Ornitz et al., 1992; Kan et al., 1993), and heparan sulfate proteoglycans are required for FGF induction of fibroblast mitogenesis and negative regulation of myoblast differentiation (Rapraeger et al., 1991; Olwin and Rapraeger, 1992). These results suggest that a complex of bFGF, FGFR and heparan sulfate proteoglycan is required for bFGF binding and biological response.

Intracellular targeting of bFGF after binding and endocytosis by surface receptors is not well defined but may play a role in bFGF-mediated responses. Unlike some other growth factors that are rapidly degraded once internalized, bFGF is long-lived (Moscatelli, 1988) and may play a regulatory role after internalization. A fraction of the exogenously added bFGF escapes degradation in the endosomal pathways and is targeted to the nucleus (Bouche et al., 1987; Amalric et al., 1991). Targeting of the internalized bFGF could therefore play a role in the divergent responses mediated by this growth factor.

It is not known if the intracellular fate of bFGF is determined by the receptor that binds bFGF at the cell surface. The multiple bFGF receptors have different rates of internalization and intracellular fates. The FGFRs are rapidly down-regulated in response to binding bFGF (Burgess and Maciag, 1989). In contrast, the proteoglycans are constitutively internalized by several pathways, one being a fast pathway in which the proteoglycans are completely degraded with a half-life of 30 minutes and a second pathway in which the cells accumulate heparan sulfate fragments cleaved from the internalized proteoglycans (Yanagishita and Hascall, 1984). Thus, bFGF could have altered fates depending which proteoglycan bound bFGF.

In this study, we used bFGF conjugated to saporin (bFGF-saporin) as a facile means of determining if binding of bFGF by different cell surface receptors results in bFGF targeting to different sites. Saporin is a ribosomal inactivating protein that is cytotoxic once it enters the cytoplasm. However, saporin lacks cell binding capabilities. Binding and uptake of saporin require that it be conjugated to another protein, in this case bFGF, for which cell surface receptors are available. Saporin conjugated to bFGF (bFGFsaporin) is cytotoxic to cells that are bFGF responsive (Lappi et al., 1989, 1991; Beattie et al., 1990). We find that Swiss 3T3 cells that express heparan sulfate proteoglycan and FGFR are killed by exposure to bFGF-saporin. However, NMuMG cells, which express little or no FGFRs and have abundant proteoglycan, internalize bFGF-saporin but are resistant to its cytotoxicity. Transfection of these cells with one particular member of the FGFR family, FGFR-1, results in the transfected cells being killed by bFGF-saporin. In addition, blocking synthesis of heparan sulfate chains in cells expressing both FGFRs and proteoglycans conferred resistance to bFGF-saporin. This suggests that the intracellular fate is dependent upon the receptor that internalizes the bFGF. Basic FGF-saporin endocytosed on heparan sulfate proteoglycan enters a pathway in which it cannot exert its cytotoxic effect, possibly leading to the lysosome where it would be degraded. Alternatively, FGFR-1, perhaps in a complex with heparan sulfate, targets bFGF-saporin to a different destination. This latter destination ultimately allows the saporin to gain access to the cytoplasm.

The NMuMG (normal murine mammary gland) epithelial cells, the resulting transfected cell lines, and the Swiss 3T3 cells were routinely maintained in 10% fetal bovine serum (FBS; Tissue Culture Biologics, Tulane, CA) in Dulbecco’s modified Eagle’s medium (DME; Gibco, Grand Island, NY), with 4.5 g/l glucose, penicillin/streptomycin (P/S), and 10 µg/ml bovine pancreatic insulin (Sigma, St. Louis, MO). Cells were not allowed to exceed 70% confluence before passage.

NMuMG cells were transfected using Lipofectin Reagent (Gibco BRL, Gaithersburg, MD) with the FGFR-1 gene in a Bluescript vector containing the maloney virus promoter (Mo/mFR1/SV), a gift from Dr David Ornitz (Harvard Medical School, Boston, MA), and PKO-neo plasmid that confers neomycin resistance (Ornitz et al., 1992). Lipofectin-DNA solution was prepared according to the manufacturer’s directions. Briefly, 19 µg Mo/mFR1/SV and 1 µg PKO-neo in 50 µl water were incubated with 50 µl of 1 mg/ml Lipofectin Reagent 15 min, then 3 ml DME with P/S was added.

NMuMG cells at 50% confluence were washed twice with DME. The Lipofectin/FGFR-1 DNA solution was incubated with the cells for 5.5 h at 37°C in a 5% CO ₂ incubator. Then 3 ml 20% FBS/DME with P/S was added. After 2 d cells were transferred to selection medium: 10% FBS/DME with P/S and 2 mg/ml geneticin (Gibco, Grand Island, NY). Two weeks later, clones were selected, grown to sufficient density and screened for FGFR binding sites. Several clones were subcloned in 96-well plates and resulting clones were again screened for FGFR binding sites.

NMuMG cells transfected with the PKO-neo plasmid alone were prepared as above with the exception that the cells were transfected with a total of 50 µg PKO-neo DNA alone. Clones were selected for resistance to geneticin.

Poly-adenylated RNA was prepared from cells essentially as described by Bradley et al. (1988). Briefly, confluent monolayers of cells were rinsed, suspended by treatment with trypsin, washed three times with PBS containing 25 µM aurin tricarboxylic acid, pelleted and frozen for storage. The pellets were resuspended in lysis buffer (0.2 M NaCl, O.2 M Tris-HCl, pH 7.5, with 0.15 mM MgCl₂, 2% SDS, 200 µg/ml Proteinase K and 20 µM aurin tricarboxylic acid) at 1×10⁷ cells/ml and homogenized by passage through a 18 gauge needle 4 times and then through a 22 gauge needle an additional 4 times. The cell lysate was incubated for 2 h at 45°C with intermittent mixing, adjusted to 0.5 M NaCl and incubated with oligo(dT)-cellulose for 60 min at room temperature. The oligo(dT)-cellulose was pelleted and washed four times with 0.5 M NaCl, 0.1 M Tris-HCl, pH 7.5, and then eluted with 5 washes of 0.1 M Tris-HCl, pH 7.5.

For northern blot analysis, RNA samples (3 µg) were electrophoresed in a formaldehyde-agarose gel and transferred to Nytran membrane (Schleicher and Schuell, Keene, NH) by capillary action. The RNA on the filter was hybridized with FGFR-1 cDNA probe labeled with ³²P using Prime-a-gene labeling system (Promega, Madison, WI: Feinberg and Vogelstein, 1984). Hybridization was carried out at 42°C for 16 h in 50% formamide, 2.5× Denhardt’s, 0.1% SDS, 100 µg/ml sssDNA (sonicated salmon sperm DNA) and 5× SSPE. The filter was washed twice in 6× SSPE at room temperature and then 1× SSPE at 65°C.

Human recombinant bFGF, a gift from Dr Brad Olwin (University of Wisconsin, Madison, Wisconsin), was iodinated by the chloramine T method to a specific activity of approximately 200 µCi/fmole (Burrus and Olwin, 1989). Specific activity was determined with a mitogenic assay on Swiss 3T3 cells in comparison to unlabeled bFGF. Basic FGF-saporin, a gift from Dr Andrew Baird (The Whittier Institute for Diabetes and Endocrinology, La Jolla, CA), was iodinated by the same procedure with the exception that the iodination reaction was terminated with a saturated tyrosine solution instead of dithiothreitol. Specific activity of the bFGF-saporin was determined with a cytotoxic assay on Swiss 3T3 cells in comparison to unlabeled bFGF-saporin.

For bFGF binding, cells were plated in 24-well plates at approximately 150,000 cells/well and incubated with 50 pM ¹²⁵I-bFGF for 2 h at 4°C in DME with 20 mM Hepes (pH 7.4) and 0.1% bovine serum albumin (BSA). The labeling solution was removed and the cells were washed 3 times with 150 mM NaCl, 10 mM Tris-HCl, pH 7.4. Basic FGF binding to heparan sulfate sites was determined as the bFGF removed during 3 washes of 2 M NaCl, 0.1% BSA, 10 mM Tris-HCl, pH 7.4 (Moscatelli, 1987). Subsequently, the FGFR sites were determined as the bFGF removed during an additional 3 washes with 2 M NaCl, 0.1% BSA, 10 mM sodium acetate, pH 4. Nonspecific binding in the pH 4 washes was assessed by competition with 5 nM unlabeled bFGF to equivalent wells during binding.

For crosslinking, cells were plated in 6-well plates at 500,000 cells/well and incubated with 500 pM ¹²⁵I-bFGF 4°C for 2.5 h. The labeling solution was replaced with 0.1 mg/ml disuccinimidyl suberate, DSS (Pierce Chemical Company, Rockford, IL), in PBS, pH 7.7 and incubated 30 min at 4°C. Cells were washed as described for the binding assay to reduce background and then twice with PBS. Cells were then extracted in 200 µl sample buffer and cell extract equivalent to 200,000 cells was separated by 7.5% SDS-PAGE. The gel was dried and exposed to Kodak X-Omat AR film (Eastman Kodak Co., Rochester, NY).

Cells at 50-80% confluence in 24-well plates were washed twice with DME and incubated 24 h with DME containing 1 mg/ml BSA and P/S. The medium was removed and replaced with DME containing 1 mg/ml BSA, 10 µg/ml insulin and P/S with 0-270 pM bFGF and incubated for 18 h. The insulin does not stimulate growth in the cells, however it potentiates the mitogenic response to bFGF. DNA synthesis was measured by adding 1 µCi/ml [³H]thymidine (NEN Research Products, Boston, MA) and incubating at 37°C for 4-6 h. The cells were then incubated at room temperature for 10 min with 5% trichloroacetic acid, washed and solubilized in 0.1% NaOH. Incorporated [³H]thymidine was counted in a LS 5800 liquid scintillation counter (Beckman Instruments, Inc., Irvine, CA).

Approximately 500,000 cells in 60 mm dishes were incubated with 5 ml 100-200 pM ¹²⁵I-FGF in 10% FBS/DME and P/S for 8 h at 37°C. Cells were washed as in the binding assay to remove surface bound ¹²⁵I-bFGF-saporin, then washed once with TES (160 mM NaCl, 5 mM EDTA and 20 mM Tris-HCl, pH 7.6). Residual surface bound ¹²⁵I-bFGF was removed with 0.1% trypsin in TES. The cells were pelleted and counted in a LKB Gamma counter (Stockholm, Sweden). Incubation of cells at 4°C followed by treatment with trypsin contain less than 10% of the c.p.m. measured in cells incubated at 37°C.

Cells were plated in 10% FBS/DME at 5000 cells/well in 24-well plates and incubated overnight. Basic FGF-saporin or unconjugated bFGF and saporin together were then added to the final concentration indicated in the figures. Cells were incubated an additional 24 h and DNA synthesis was measured as for mitogenic assays. In addition cells cultured in bFGF-saporin do not increase in cell number and after 3-5 days the cells round up and detach from the dish.

Cells were plated in 24-well plates at 5% confluence in sulfate-free DME with 10% FBS, 30 mM chlorate, P/S, 10 µg/ml insulin and 100 pM EGF (Collaborative Research Inc. Lexington, MA) with the proper adjustments in salt concentration (Baeuerle and Huttner, 1986; Rapraeger et al., 1991). After 1-2 days pretreatment, the medium was removed and replaced with chlorate medium or chlorate medium with 10 mM sulfate added. Basic FGF-saporin or bFGF and non-conjugated saporin were added to a final concentration of 100 pM each. After a 24 h incubation DNA synthesis was measured as for mitogenesis.

Both proteoglycans and FGFRs bind bFGF at the cell surface and are capable of internalizing the ligand. In order to determine if the intracellular fate of bFGF depends upon the receptor by which it is internalized, the ability of the cell to target FGF into the cytoplasm was studied using bFGFsaporin. These studies utilized two cell types: Swiss 3T3 fibroblasts, which express FGFRs and FGF-binding heparan sulfate proteoglycan and which respond to bFGF, and NMuMG epithelial cells, which express heparan sulfate proteoglycans, but have little or no expression of FGFRs and show no mitogenic response to the growth factor.

Swiss 3T3 cells were incubated with various concentrations of bFGF-saporin for 24 hours and DNA synthesis was measured as a convenient assay for saporin-mediated cytotoxicity. Cells were incubated in serum-containing medium to promote full growth potential. Therefore, addition of bFGF to these cultures had no stimulatory or inhibitory effect on cell growth (data not shown). However, bFGFsaporin conjugate at 10-100 pM inhibits growth by 40-80% as measured by inhibition of DNA synthesis (Fig. 1). Cells left in bFGF-saporin for 3-5 days round up and detach from the dish. At equivalent concentrations, nonconjugated saporin has no effect. Basic FGF-saporin was cytotoxic in the same concentration range as bFGF exerts its mitogenic effects, suggesting that bFGF-saporin-mediated cytotoxicity and bFGF-mediated mitogenesis employ the same receptors (data not shown).

NMuMG cells, which do not respond mitogenically to bFGF, are refractory to bFGF-saporin. At concentrations where 80% of the Swiss 3T3 cells are killed (100 pM), NMuMG cells are unaffected by the bFGF-saporin (Fig. 1). One explanation for this difference may be differences in the receptors mediating bFGF binding to these cell types. Therefore, binding of bFGF to NMuMG cells was compared to that of Swiss 3T3 cells. Binding occurs to two categories of receptors, heparan sulfate proteoglycans and FGFRs. Basic FGF bound to heparan sulfate proteoglycan is released by a 2 M salt wash at neutral pH, whereas bFGF bound to the FGFRs is resistant to this wash and requires a subsequent acid wash (Moscatelli, 1987). Basic FGF binds heparan sulfate proteoglycan on both NMuMG cells and Swiss 3T3 cells (Fig. 2A). It also binds to FGFRs on the Swiss 3T3 cells; however, these sites are few or absent on the NMuMG cells. Therefore, a major difference between these two cell types is the presence of the FGFRs.

Heparan sulfate proteoglycans are constitutively endocytosed by cells, providing a means of bFGF uptake (Yanagishita and Hascall, 1984). NMuMG cells are thus equipped to internalize bFGF or bFGF-saporin. The Swiss 3T3 cells also internalize bFGF or bFGF-saporin via this mechanism as well as via endocytosis of bFGF bound to FGFRs. The Swiss 3T3 cells express several members of the FGFR family. One specific member of the FGFR family, FGFR-1, is present on Swiss 3T3 cells as shown by Northern analysis of poly(A)⁺ RNA (Fig. 2B). These cells express an abundant message of 4.4 kb. In contrast, the NMuMG cells do not express any detectable FGFR-1 mRNA (Fig. 2B).

The potential role of FGFR-1 in bFGF targeting was explored by expressing this receptor in the NMuMG cells. Cells were co-transfected with the plasmid Mo/mFR1/SV containing the FGFR-1 cDNA and the PKO-neo plasmid that confers neomycin resistance. Of approximately 20 clones generated, two clones expressing receptor (R1-1 and R1-2) and two clones expressing neomycin resistance alone (Neo-10 and Neo-5) were used in these studies. FGFR-1 expression was confirmed in three ways. First, the cells were examined for expression of FGFR-1 mRNA (Fig. 3). Poly(A)⁺ RNA was isolated from the cells and screened by northern analysis. Clone R1-1 expresses a 4.4 kDa mRNA that hybridizes with radiolabeled FGFR-1 probe; the mRNA is the same size as that of the Swiss 3T3 cells. FGFR-1 mRNA is not detectable in clones Neo-10 and Neo-5.

Secondly, expression of the FGFR-1 protein at the cell surface was examined by binding of iodinated bFGF. Clones R1-2 and R1-1 bound bFGF to FGFRs (Fig. 4A). The amount of binding observed was similar to that of Swiss 3T3 cells. Binding to heparan sulfate proteoglycan is not significantly changed by the transfection (data not shown). The Neo-10 and Neo-5 clones retained the negative bFGF binding characteristics of the parental NMuMG cell line.

Thirdly, the size of the cell surface receptor was examined. Iodinated bFGF was cross-linked to the surface of the transfected cells using DSS, a protein-protein cross-linker, and the size of the resulting complexes was analyzed on SDS-PAGE. A complex with apparent molecular mass of 170 kDa was seen in cells transfected with FGFR-1, which corresponds to a similar complex found in Swiss 3T3 cells (Fig. 4B). The Swiss 3T3 cells also express another band of approximately 150 kDa, which may be another FGFR or alternatively spliced form of FGFR-1 (Fig. 4B). No complex was seen in the Neo-10 and Neo-5 clones. Heparan sulfate binding of bFGF is not detected as DSS does not cross-link the bFGF to the heparan sulfate chains. Taken together, these results demonstrate that FGFR-1 is expressed and binds bFGF in the transfected cells.

To determine if the cell surface FGFR-1 is functional in the transfected cells, the growth response of the cells to bFGF was measured. Cells were serum starved and then stimulated with various concentrations of bFGF for an additional 18 hours before DNA synthesis was measured by incorporation of [³H]thymidine in a 6 hour pulse. The R1-1 cell line responds to bFGF with half maximal stimulation at approximately 10 pM (Fig. 5). This concentration of bFGF is similar to that required for half-maximal stimulation of other bFGF-responsive cell lines. The Neo-10 cells fail to respond to bFGF at the concentrations tested (Fig. 5). Thus, FGFR-1 is functional in the FGFR-1 transfected NMuMG cells and capable of mediating bFGFinduced growth. This implies also that parental NMuMG cells must contain the signal transduction machinery required for the bFGF response and lack only sufficient expression of FGFRs.

The NMuMG clones expressing or not expressing FGFR-1 were now used for comparison of bFGF-saporin intracellular pathways. However, it was important to ensure that equivalent amounts of the toxin were internalized and thus available for targeting by both cell types.

Uptake of iodinated bFGF-saporin was measured at 37°C in the culture medium used for the cytotoxicity experiments. The integrity of the iodinated compound was monitored by examining the iodinated complex by SDS-PAGE, which demonstrated that 90% of the radiolabel was in the bFGFsaporin conjugate (data not shown). The concentration of iodinated bFGF-saporin was determined by comparison to unlabeled bFGF-saporin in a cytotoxicity concentration curve on Swiss 3T3 cells (data not shown).

A range of bFGF-saporin concentrations was used to measure its accumulation in the Neo-10 and the R1-1 cells (data not shown). Accumulation, of course, is a function of uptake and loss from the cells. ¹²⁵I-bFGF accumulation in these cells is a direct measure of uptake, however, as no loss of intracellular iodinated material is seen during chases of up to 6 hours. This demonstrated that the R1-1 cells takeup more bFGF-saporin than equivalent numbers of the Neo-10 cells. This is to be expected as both cell types internalize bFGF on heparan sulfate proteoglycan and the R1-1 cells also have additional uptake due to the FGFR-1. Uptake data are shown at two concentrations where the R1-1 and the Neo-10 transfects accumulate equivalent amounts of bFGFsaporin. The R1-1 cells internalized 0.42 fmoles of ¹²⁵IbFGF-saporin/500,000 cells over an 8-hour period when incubated with 100 pM ¹²⁵I-bFGF-saporin; the Neo-10 cells take up at least this amount (0.44 fmoles/500,000 cells) when incubated with 200 pM ¹²⁵I-bFGF-saporin over an 8-hour period (Fig. 6a). These relatively long time points were employed to determine the total amount of bFGF-saporin the cells would be exposed to during the cytotoxicity assay. Uptake for shorter (4-hour) periods showed a similar relationship, with 0.31 fmole uptake/500,000 cells by the Neo-10 clone and 0.25 fmole uptake/500,000 cells for the R1-1 clone. Evaluation of uptake for periods of time longer than 8 hours was not deemed feasible, as bFGF-saporin began to be cytotoxic to the R1-1 clone. The ¹²⁵I-bFGF-saporin is not degraded to iodinated products that are released to the culture medium because chloroquine, a lysosomal inhibitor, had no affect on the total cellular accumulation of ¹²⁵I-bFGFsaporin (data not shown). Also cells chased in the absence of bFGF following a 3-hour uptake show no loss of label during ensuing 6-hour chases (data not shown).

Basic FGF-saporin had different cytotoxic effects at the concentrations where the two clones internalize equivalent amounts of bFGF-saporin; bFGF-saporin killed the R1-1 cells that express the FGFR-1 and had no effect on the Neo-10 clone (Fig. 6b). Non-conjugated bFGF together with free saporin had no cytotoxic effect on either clone. Because the Neo-10 cells were resistant to bFGF-saporin despite its internalization by these cells, the two cell types must target the bFGF-saporin differently once it is internalized.

Basic FGF-saporin is cytotoxic to FGFR-1 clones at concentrations that are at least 5-fold lower than concentrations at which cells without FGFR-1 are resistant. Cytotoxicity is exhibited at concentrations as low as 30 pM on R1-1 and R1-2 clones (Fig. 7), but the Neo-10 cells and Neo-5 cells are resistant to at least 150 pM bFGF-saporin. Basic FGF together with unconjugated saporin at concentrations equivalent to the bFGF-saporin had no effect on any of the clones (data not shown). An additional three FGFR-1 clones and three neomycin clones were studied with similar results; clones that express FGFR-1 at levels within a 2-fold range of Swiss 3T3 cells are susceptible to bFGF-saporin, while clones expressing little or no FGFR-1 are resistant (data not shown). Thus FGFR-1 negative cells are resistant to bFGFsaporin despite internalizing as much if not more bFGFsaporin than is cytotoxic to the cells expressing FGFR-1.

Cell surface heparan sulfate is required for bFGF binding to FGFRs and for bFGF-induced mitogenesis of Swiss 3T3 cells. To verify that heparan sulfate proteoglycan must be part of the complex that endocytoses bFGF-saporin, we blocked the sulfation of newly synthesized heparan sulfate by the addition of chlorate to the culture medium of growing cells, thus inhibiting bFGF binding to the glycosaminoglycan (Baeuerle and Huttner, 1986; Rapraeger et al., 1991). Addition of excess sulfate restores the sulfation of the chains and the bFGF binding characteristics. Treatment for 24 h with 30 mM chlorate reduces radiolabeled sulfate incorporation into heparan sulfate of NMuMG cells by greater than 90% (data not shown) and reduces ¹²⁵I-bFGF binding by 85%. Treatment of the R1-1 clone with 30 mM chlorate completely blocked the cytotoxic effect of the FGF-saporin conjugate on these cells, as compared to untreated controls (Fig. 8). Treatment of the cells with non-conjugated saporin together with bFGF also was not cytotoxic (Fig. 8). As expected, however, restoring functional heparan sulfate proteoglycan by the addition of 10 mM sulfate in the continued presence of 30 mM chlorate restores the cytotoxic effect of FGF-saporin (Fig. 8).

Signaling of growth and differentiation by bFGF involves multiple pathways, and internalization of bFGF may be involved in one or more of these pathways. We have therefore used bFGF-saporin as a means of detecting differences in the intracellular targeting of bFGF when bound and internalized by different receptors. NMuMG cells bind bFGF and bFGF-saporin via cell surface heparan sulfate proteoglycans; however, they express little or no FGFRs and are not responsive to bFGF. Nevertheless, bFGF and bFGFsaporin are internalized by these cells, but the cells apparently do not target the bFGF-saporin to a compartment from which it can gain access to cytoplasmic ribosomes. In contrast, NMuMG clones expressing a specific tyrosine kinase-containing receptor, FGFR-1, bind bFGF via a proteoglycan/FGFR-1 complex, respond to bFGF mitogenically at concentrations similar to other bFGF-responsive cell lines and are killed by bFGF-saporin. Thus, endocytosis of the FGF-saporin conjugate on FGFR/heparan sulfate proteoglycan complexes results in its targeting to a compartment from which the bFGF-saporin gains access to the cytoplasm, leading to inactivation of the ribosomes.

There are at least four FGFRs, designated FGFR-1 to FGFR-4, that bind members of the FGF family. NMuMG cells were transfected with the FGFR-1 as a representative member of the FGFR family. The transfectants express FGFR-1, as chemical crosslinking of ¹²⁵I-bFGF cell surface proteins reveals a protein-¹²⁵I-bFGF complex of 170 kDa, an appropriate size for FGFR-1 cross-linked to bFGF. The binding and crosslinking characteristics are similar to those of Swiss 3T3 cells, a cell line which expresses endogenous FGFR-1. The FGFR-1 transfected into the NMuMG cells is active as these cells now respond mitogenically to bFGF at concentrations that stimulate other bFGF-responsive cell lines.

Intracellular sorting of bFGF-saporin is dependent upon the receptor by which the ligand is internalized. We have shown here that bFGF-saporin is internalized by both HSPG and FGFR. This supports other work demonstrating that bFGF is internalized by both types of receptors (Rusnati et al., 1993; Roghani and Moscatelli, 1992). Basic FGFsaporin internalized after binding heparan sulfate proteoglycan at the cell surface is not targeted to a compartment from which the saporin has cytoplasmic access. It is possible that it is targeted to lysosomes where it is degraded. The susceptibility of cells expressing FGFR-1 to bFGF-saporin cytotoxicity suggests that the FGFR provides sorting information that results in the bFGF-saporin being transported to a different compartment. This sorting may occur at the cell surface, perhaps by localization to specific sites for endocytosis. There are several different pathways for endocytosis, including clathrin-coated pits and clathrin-independent pathways (VanDeuers et al., 1989). The most abundant proteoglycan on NMuMG cells is syndecan-1, a member of a family of proteoglycans (syndecans 1-4) characterized by their well-conserved cytoplasmic domains (Bernfield et al., 1992). The role of this conserved region is not known, although it contains several conserved tyrosines, which could mediate endocytosis through coated pits as is the case with several other receptors such as the mannose 6-phosphate receptor and the LDL receptor (Pearse, 1988; Glickman et al., 1989; Pearse and Robinson, 1990). This domain could therefore play an important role in the endocytosis of the bFGF-receptor complex. Alternatively, the sorting may occur once in the endosomal pathway, perhaps by retrieving the bFGF/heparan sulfate complex from a route leading to lysosomes and targeting it to another pathway. Such a retrieval could possibly occur in the endosomes where FGFR would be captured and localized in discrete vesicles. Alternatively, the bFGF-saporin may be targeted to a single site whether bound either to receptor or to proteoglycan alone, but might gain access to the cytoplasm from this site only if bound to receptor. Other members of the FGF receptor family, which are highly homologous (Kornbluth et al., 1988; Pasquale, 1990; Keegan et al., 1991; Partanen et al., 1991; Stark et al., 1991), may also direct intracellular targeting of FGFs in the same manner.

Intracellular sorting of bFGF out of the endocytic pathway may be significant in light of the proposed signal transduction pathway in which bFGF is targeted directly to the nucleus. Basic FGF has been found in the nucleus of a number of bFGF-responsive cells (Bouche et al., 1987; Baldin et al., 1990; Renko et al., 1990; Speir et al., 1991; Woodward et al., 1992). Studies on exogenously added bFGF indicate that endocytosed bFGF can be targeted to the nucleus despite lacking a nuclear localization sequence (Bouche et al., 1987; Bugler et al., 1991). Such a pathway would require bFGF to leave the vesicular endocytic system and traverse the cytoplasm to reach the nucleus. Basic FGF-saporin may be detecting such a pathway by its cytotoxic effect on cells. Several studies indicate that nuclear targeting of bFGF may signal physiological responses. Expression of bFGFs containing nuclear localization sequences is reported to regulate growth and cellular transformation (Couderc et al., 1991; Quarto et al., 1991), suggesting that they act by an internal autocrine mechanism. Also, deleting a putative nuclear localization sequence of aFGF eliminates nuclear localization of the growth factor and abolishes its ability to stimulate cell growth (Imamura et al., 1990; Imamura et al., 1992). However, the mutant still binds and triggers cell surface receptors as shown by stimulation of receptor phosphorylation and induction of c-fos. Nuclear localization of the deletion mutant and acquisition of mitogenic activity can be restored by adding the nuclear localization sequence of Histone 2B to the mutant growth factor (Imamura et al., 1990, 1992). Thus, the biological action of FGFs may require intracellular sorting of the ligand in addition to receptor activation.

Several criteria demonstrate that the sorting of bFGFsaporin faithfully monitors intracellular targeting of bFGF. First, unconjugated saporin is not cytotoxic to cells. Second, endocytosis of the bFGF-saporin does not necessarily lead to cell killing, but is dependent on the receptor that carries out internalization. Third, the amount of bFGF-saporin necessary for killing is similar to concentrations of bFGF that promote growth. Thus, bFGF-saporin appears to be using the same physiological mechanism as bFGF. Taken together these results indicate that bFGF-saporin is binding to the bFGF receptors for internalization and is specifically sorted depending on receptor type.

The targeting of bFGF-saporin requires heparan sulfate as well as FGFR-1. This is not surprising, as it has been established that bFGF requires this glycosaminoglycan in order to recognize FGFR (Rapraeger et al., 1991; Yayon et al., 1991; Ornitz et al., 1992). However, heparan sulfate proteoglycan may also play an important role in the uptake and sorting of FGF. A hallmark feature of heparan sulfate proteoglycan uptake is cleavage of the heparan sulfate chains into discrete fragments in a prelysosomal compartment (Yanagishita and Hascall, 1984). Several different pathways are utilized, however. One is a fast pathway in which heparan sulfate of the proteoglycan is completely degraded with a half-life of 30 min. The heparan sulfate of this fast pathway is derived from phosphoinositol-linked proteoglycan such as glypican (Yanagishita, 1992). An additional pathway results in much degradation of the heparan sulfate fragments in a prelysosomal compartment with a much slower a half-life of ∼4 hours. Recent work by Yanagishita, 1992) has demonstrated that endocytosis of membrane-spanning proteoglycans, presumably syndecans, are the source of heparan sulfate in the slow pathway that accumulates heparan sulfate fragments. The generation of such fragments would allow the FGFR/FGF/heparan sulfate fragment complex to be divorced from the core protein of the proteoglycan and be targeted separately. Thus, intracellular heparan sulfate fragments may also sustain binding of bFGF to the FGFRs, and protect bFGF from acid and proteolytic denaturation as the bFGF moves though the endocytic pathway (Gospodarowicz et al., 1987; Saksela and Rifkin, 1990). Heparan sulfate is found in the nucleus in a sequence-specific and cell cycle-specific manner (Fedarko and Conrad, 1986; Fedarko et al., 1989), suggesting that the heparan sulfate fragments accompany bFGF to the nucleus.

The authors thank Dr Brad Olwin for providing bFGF, Dr Andrew Baird for providing bFGF-saporin, and Dr David Ornitz for providing the Mo/mFR1/SV plasmid. This work was supported by grants from the National Institutes of Health (HD21881). Jane Reiland was supported by Developmental Biology Training grant (HD017118) and is currently supported by American Heart Association Grant/Wisconsin Affiliate, Inc.