Regulation of Secreted Frizzled-related Protein-1 by Heparin*

Secreted Frizzled-related protein-1 (sFRP-1) belongs to a class of extracellular antagonists that modulate Wnt signaling pathways by preventing ligand-receptor interactions among Wnts and Frizzled membrane receptor complexes. sFRP-1 and Wnts are heparin-binding proteins, and their interaction can be stabilized by heparin in vitro. Here we report that heparin can specifically enhance recombinant sFRP-1 accumulation in a cell type-specific manner. The effect requires O-sulfation in heparin, and involves fibroblast growth factor-2 as well as fibroblast growth factor receptor-1. Interestingly, further investigation uncovers that heparin can also affect the post-translational modification of sFRP-1. We demonstrate that sFRP-1 is post-translationally modified by tyrosine sulfation at tyrosines 34 and 36, which is inhibited by the treatment of heparin. The results suggest that accumulation of sFRP-1 induced by heparin is in part due to the relative stabilization of unsulfated sFRP-1 and the direct stabilization by heparin. The study has revealed a multifaceted regulation on sFRP-1 protein by heparin.

Wnt proteins are a large family of structurally related secreted glycoproteins that mediate fundamental biological processes such as cell polarity and proliferation, tissue patterning, and tumorigenesis (1)(2)(3)(4). The Wnt signaling event is initiated by the binding of Wnt proteins to a membrane receptor complex consisting of a seven-pass transmembrane molecule of the Frizzled (Fz) 2 family (5) and members of the low density lipoprotein receptor-related family (LRP5 or LRP6) (6 -8). The binding of Wnts to the receptors activates various intracellular signaling pathways according to the Wnt, Fz, and cell type involved (3,4). In the canonical or Wnt/ ␤-catenin pathway, Wnt binding activates disheveled protein and leads to the inhibition of glycogen synthase kinase-3␤ and subsequent stabilization of ␤-catenin. ␤-Catenin translocates to the nucleus, interacts with DNA-binding proteins of the T-cell factor/ lymphoid enhancer-binding factor family, and activates transcription of target genes (1)(2)(3)(4).
Wnts bind to Fz proteins through a cysteine-rich domain, which contains 10 cysteines conserved in all members of the Fz family (5,9). The cysteine-rich domain is not unique to Fz and is also the N-terminal domain of secreted Frizzled-related proteins (sFRPs), a family of glycoproteins that are ϳ300 amino acids in length (10 -16). sFRPs are capable of binding to Wnts and Fz receptors, and are also modulators of Wnt signaling (17). sFRP-1 is a 35-kDa prototypical member of the sFRP family (14, 18 -20). It has been shown that sFRP-1 acts as a biphasic modulator of Wnt signaling, counteracting Wnt-induced effects at high concentrations and promoting them at lower concentrations (18). Deletion of sFRP-1 in mice leads to decreased osteoblast and osteocyte apoptosis and elevated bone mineral density (21).
Heparin and heparan sulfate (HS) are mammalian glycosaminoglycans with the highest negative charge density of known biological macromolecules, and are known to bind by ionic interactions with a variety of proteins such as fibroblast growth factors (FGFs) (22). The interactions of HS with heparin-binding proteins can have a direct effect on important cellular processes such as FGF cell signaling events (23,24). sFRP-1 as well as Wnts are heparin-binding proteins (13,16,18) and, in fact, sFRP-1 was originally isolated and identified from the heparin-binding fraction of human embryonic lung fibroblast-conditioned medium (13). The heparin-binding domain of sFRP-1 is in the C-terminal region of the protein and has weak homology with netrins (10,11). The complex of sFRP-1 and Wg (Drosophila Wnt1 homologue) can be stabilized by heparin in vitro, suggesting that heparin or endogenous heparan-sulfate proteoglycan (HSPG) may promote sFRP-1/Wg binding by serving as a scaffold to facilitate interaction between sFRP-1 and Wg (18). Lowering tissue HSPG levels has been shown to impair Wnt signaling in vivo, supporting the notion that HSPG plays an important role in the Wnt signaling regulation (25)(26)(27)(28).
In this study, we report two unexpected effects of heparin on sFRP-1. One is that heparin can stimulate sFRP-1 accumulation in a cell type-specific manner. The effect involves FGF-2 and FGF receptor-1. Surprisingly we also find that heparin can affect the post-translational modification of sFRP-1, which induces a mobility shift of the protein on SDS-PAGE. Further study shows that sFRP-1 is tyrosine-sulfated at tyrosines 34 and 36, and heparin treatment inhibits the modification. The data suggests that direct stabilization by heparin and the relative stabilization of unsulfated sFRP-1 contribute partly to the accumulation of sFRP-1 induced by heparin.
Transient and Stable Expression-Transient expression was performed in either 50-ml spinners or 1-liter spinners. For the 50-ml culture volume, 25 g of plasmid DNA was mixed with 400 g of polyethylenimine (25 kDa, linear, neutralized to pH 7.0 by HCl, 1 mg/ml, Polysciences (Warrington, PA)) in 2.5 ml of serum-free 293 media. For the 1-liter volume, 500 g of DNA was mixed with 4 mg of polyethylenimine in 50 ml of serumfree 293 media. Then the mixtures were mixed with either 50 ml or 1 liter of HEK293 cells in 293 media with 5% FBS at a cell density of 0.5 ϫ 10 6 cells/ml. The spinners were incubated at 37°C with a rotation rate of 170 rpm on a P2005 Stirrer (Bellco) for 72-144 h before harvest.
For the establishment of CHO-DUKX sFRP-1 stable lines, construct pWZ1028 was transfected into CHO-DUKX cells and the transfectomas were selected against 50, 100, or 200 nM MTX for 3 weeks. After screening 72 colonies, three 200 nM MTX-resistant clones (200-10, -11, -12) with the highest expression of sFRP-1 were isolated. In this study, HEK293 cells were also found to be sensitive to MTX at concentrations of 100 nM and above, even though they have two copies of the dihydrofolate reductase gene. To construct a HEK293 stable line for sFRP-1, pWZ1028 was transfected into HEK293-EBNA and transfectomas were selected against 100 or 250 nM MTX for 3 weeks. The two best clones, 100-5 and 100-20, were then isolated. To establish Lec3.2.8.1 stable lines for sFRP-1, pWZ1097 was transfected into Lec3.2.8.1 and the transfectomas were selected against 10 or 25 M methionine sulfoximine for 3 weeks. Several positive clones were isolated.
Protein Purification-The conditioned media containing sFRP-1-(His 6 ) from HEK293 cells or from Lec3.2.8.1 cells was supplemented with NaCl to 1 M final concentration and equilibrated with nickel-NTA (Qiagen) resin at 4°C for about 1 h. The nickel-NTA resin was collected by centrifugation at 3,000 ϫ g (Sorvall H-6000A/HBB-6), suspended in 1 M NaCl, 25 mM Tris⅐HCl, pH 7.5, and packed into a column (GE Healthcare). After extensive washing with 1 M NaCl, 25 mM Tris⅐HCl, pH 7.5 (buffer A), and buffer A plus 15 mM imidazole, sFRP-1-(His 6 ) protein was eluted with buffer A plus 200 mM imidazole. The partially purified sFRP-1 was concentrated to about 2 ml using 10K MWCO concentrators (Vivascience) and applied to a Superdex TM 200 size exclusion column (SEC) (GE Healthcare) pre-equilibrated in Buffer A. The protein concentration of the SEC fractions containing sFRP-1 was determined by absorbance at 280 nm using the calculated extinction coefficient.
Metabolic Labeling-Stable Lec3.2.8.1 cells for sFRP-1-(His 6 ) (25-6) were grown to confluence and switched to a low sulfate RPMI medium (specialty medium), supplemented with 10% FBS. Before sulfate incorporation experiments, cells were pretreated or mock-treated with sodium chlorate for 3 h. 50 Ci/ml of [ 35 S]sulfate (PerkinElmer Life Sciences) was incubated with cells in the presence or absence of heparin (50 g/ml) for 18 h. For labeling in transient HEK293-EBNA expression, cells at 24 h post-transfection were resuspended into labeling medium with [ 35 S]sulfate for 18 h as above. sFRP-1-(His 6 ) in the conditioned media were affinity purified by passing through a small nickel-NTA column.

RESULTS
Heparin Specifically Stimulates Recombinant sFRP-1 Accumulation in HEK293 Cells-A DNA construct expressing a C-terminal His 6 -tagged sFRP-1 was transiently transfected into HEK293 cells, and between 72 and 120 h, sFRP-1 was detected in the media, although at relatively low levels (Fig. 1A, lanes 1). As part of an effort to increase the expression of sFRP-1 for planned structure/function studies, heparin was tested as an additive to the cell culture media at various times during transfection. Adding heparin at the time of transfection reduced sFRP-1 expression to undetectable levels, most likely a result of the negatively charged heparin binding strongly to the polyanionic polyethylenimine thus interfering with the DNA uptake process (data not shown). Nevertheless, adding heparin 24 or 48 h after transfection significantly enhanced sFRP-1 accumulation in the media (Fig. 1A, lanes 2 and 3). When the cell pellets were analyzed, intracellular sFRP-1 was also accumulated with the treatment of heparin (lane 4 versus 5 and 6). We then tested various concentrations of heparin to determine the optimum concentration for sFRP-1 accumulation. As can be seen in Fig.  1B, heparin at concentrations above 5 g/ml gave the highest sFRP-1 induction levels, whereas higher levels of heparin resulted in slightly lower sFRP-1 accumulation. Interestingly, under similar expression conditions, the unrelated secreted human proteins aggrecanase I and II (30) did not show heparin inducible accumulation (Fig. 1C). Collectively, the results indicate that heparin specifically stimulates both extracellular and intracellular sFRP-1 accumulation in HEK293 cells.
Purified sFRP-1 Is Biologically Active-To determine whether the enhancement of sFRP-1 by heparin affects biological activity, sFRP-1 produced by HEK293 cells was purified. As described under "Experimental Procedures," 1 liter of conditioned medium from heparin-induced transient expression of sFRP-1 was prepared. NaCl was added to the conditioned medium at a final concentration of 1 M and sFRP-1 was purified as described under "Experimental Procedures." During the purification process, 1 M NaCl was included in all washing or elution buffers. As shown in Fig. 2, A and B, sFRP-1 was substantially pure after nickel-NTA (Fig. 2, A, left panel, and B, lanes 1 and 2), and was nearly homogeneous after the Superdex 200 SEC column (Fig. 2, A, right panel, and B, lanes 3 and 4). The sFRP-1 final yield was about 1 mg/liter and runs as an apparent monomer by analytical SEC analysis ( Fig. 2A, right panel). N-terminal sequencing confirmed the identity of sFRP-1 and ESI-MS analysis revealed a heterogeneous and higher than expected mass that was assumed to be a result of N-linked glycosylation at two potential sites, Asn 172 and Asn 262 (data not shown).
To determine whether the purified sFRP-1 protein was biologically active, the Wnt3 antagonistic activity of sFRP-1 was measured in a functional cell-based assay (31). In this assay, U2OS cells were co-transfected with a ␤-galactosidase-expressing plasmid, a Wnt3-expressing plasmid, as well as a reporter plasmid containing 16 copies of TCF DNA-binding sites placed in a reverse orientation upstream of a minimal thymidine kinase promoter and the luciferase gene. Then the cells were treated with various amounts of protein samples in the medium for 20 h. The cells were lysed and assayed for luciferase and ␤-galactosidase activity. The luciferase activity was normalized for transfection efficiency with ␤-galactosidase activity. As shown in Fig. 2C, Wnt3 increased the TCF luciferase reporter gene expression in the transfected U2OS cells. The addition of either nickel-NTA-purified sFRP-1 or SEC-purified sFRP-1 decreased the Wnt-mediated response in a dose-dependent manner, whereas the buffer had no effect on the Wnt-mediated Conditioned media were harvested at 120 h after DNA transfection. Protein samples were separated by SDS-PAGE and immunoblotted with anti-His 4 antibody. C, heparin effect is target-specific. Plasmid DNAs, encoding Aggrecanase-1 (Agg-1-A520-Flag) or the active-site mutant (Agg-1-E362Q-A520-Flag), or Aggrecanase-2 (Agg-2-H556-His6), were transiently transfected into HEK293FT cells. 50 g/ml of heparin was added to the culture at 48 h after DNA transfection. Conditioned media were harvested at either 120 or 144 h. Protein samples were analyzed by SDS-PAGE and immunoblotting with anti-His 4 or anti-FLAG.
TCF-luciferase reporter activation. These data clearly demonstrate that the purified sFRP-1 is functional.
Heparin Induction of sFRP-1 Expression Is Cell-type Specific and Requires Heparin O-Sulfation-In addition to transient expression of sFRP-1 in HEK293 cells, stable CHO lines expressing sFRP-1 production were also established. Stable clones of CHO-Dukx cells were obtained by methotrexate selection and screened for sFRP-1 expression by immunoblotting, as described under "Experimental Procedures." Three clones (200-10, 200-11, and 200-12) were isolated that expressed sFRP-1 at ϳ50% lower levels than observed in HEK293 transient expression without heparin induction. sFRP-1 produced in CHO cells migrated at a higher apparent M r than sFRP-1 from HEK293 cells (Fig. 3A, lanes 1-6  versus 7), which is likely due to more extensive N-linked carbohydrate modification in CHO cells. To determine whether heparin can also stimulate sFRP-1 accumulation in CHO cell lines, 90% confluent monolayers were treated with 50 g/ml of heparin. Conditioned medium was harvested at 72 h posttreatment and analyzed for sFRP-1 expression by Western blot analysis. Interestingly, the expression of sFRP-1 in CHO cells was not induced by heparin (Fig. 3A, lane 1 versus 2; lane 3 versus 4; lane 5 versus 6). Similar results were also obtained when conditioned media were harvested at 96 h post-treatment with heparin (data not shown). Clearly the response to heparin in CHO cells is different from that in the HEK293 transient expression experiments and suggests a cell type-specific mechanism.
To rule out the possibility that heparin-induced sFRP-1 accumulation was unique to transient expression in HEK293 cells, a stable line of HEK293 expressing sFRP-1 was established as described under "Experimental Procedures." The stable line was treated with heparin and conditioned media were col-lected at different time points and sFRP-1 levels were determined by Western blot analysis. As shown in Fig. 3B, the expression of sFRP-1 in this line responded to heparin at 48 h (lane 4 versus 3) and the effect increased with a longer incubation (lanes 6 and 8). The data clearly indicated that heparin induces sFRP-1 accumulation in both transient and stable HEK293 systems. The enhancement effect by heparin on sFRP-1 appears to be cell-type specific.
The interaction of heparin-binding proteins with HS is determined by the sequence and sulfation level of the sugar moieties of HS (32,33,35). To test whether O-sulfation and N-sulfation are required for heparin activity in sFRP-1 accumulation, sFRP-1-transfected HEK293 cells were incubated with chemically modified heparin that was completely N-desulfated, followed by N-acetylation. Modified heparin lacking 2-O-sulfation was also examined for its ability to stimulate sFRP-1 accumulation. As shown in Fig. 3C Heparin Does Not Affect Protein Release from the Cell Surface Matrix or mRNA Level of sFRP-1-Heparin is known to release or extract extracellular proteins such as Wnts from the cell surface matrix (3). To test whether the heparin effect is simply a competitive binding effect with cell surface proteoglycans, a whole cell sFRP-1 binding study was performed. As shown in Fig. 4A, purified His 6 -tagged sFRP-1 was incubated with HEK293 cells in cultured media in either the presence or absence of heparin. The concentration of sFRP-1 was at 1 g/ml, which is similar to the level during the transient expression. Because untransfected HEK293 cells do not produce His 6tagged sFRP-1, the latter can be chased into the cell cultures under these conditions. Cells mixed with His 6 -tagged sFRP-1 were harvested at different time points. Cells and media were separated by centrifugation and analyzed by immunoblotting. In Fig. 4A, lanes 1-4, the levels of sFRP-1 protein in the conditioned media without added heparin were similar to those in the media with heparin (lanes 9 -12). In lanes 5-8, some sFRP-1 was associated with HEK293 cells in the absence of heparin, whereas no such association was found when heparin was present (lanes 14 -17). The percentage of this cell-bound species in the total sFRP-1 population was, however, very low when compared with the total in the conditioned media (lanes 1-4 versus [5][6][7][8], indicating that the heparin effect is not mainly due to protein extraction from the cell surface matrix. To see if accumulation of sFRP-1 induced by heparin is due to the increase of mRNA, total RNA from heparin-treated and mocktreated cells were isolated and hybridized to digoxigenin-labeled probes against sFRP-1 and glyceraldehyde-3-phosphate dehydrogenase. As shown in Fig. 4B, there is no significant difference in mRNA levels between heparin-treated and mock-  1 and 3) or non-reduced conditions (lanes 2 and 4). C, Wnt-3 antagonistic activity of purified sFRP-1. Wnt3 antagonistic assays with U2OS cells transfected with TCF-luciferase were performed as described in Ref. 31. Heparin Induction on sFRP-1 Accumulation Involves Fibroblast Growth Factor and Its Receptor-It is well known that fibroblast growth factors bind heparin with relatively high affinity and are thus termed "heparin-binding growth factors" (22). Numerous reports have demonstrated the involvement of HSPGs in FGF cell signaling (23,24). To determine whether FGF signaling pathways are involved in sFRP-1 induction by heparin, purified FGF-1 (acidic FGF) and FGF-2 (basic FGF) were tested as additives in cell culture experiments. Prior experiments demonstrating the effects of heparin addition for the increased accumulation of sFRP-1 were carried out in the presence of serum in the medium. If we now test for the accumulation of sFRP-1 in serum-free media, we can see a dependence on the presence of exogenous growth factors. As shown in Fig. 5A, FGF-1 had no effect on sFRP-1 accumulation (lane 2 versus 1), whereas FGF-2 stimulated sFRP-1 accumulation in the presence of heparin (lanes 3 and 4 versus 1). FGF-2 alone, without heparin, had no affect on sFRP-1 accumulation (data not shown). These data indicated that FGF-2 is indeed involved in the heparin effect. To further demonstrate the involvement of FGF signaling pathways in the accumulation of sFRP-1, neutralizing antibodies against FGFR were tested. For this experiment, cells grown in the presence of serum were pretreated with antibodies against either FGFR-1 or FGFR-2 for 6 h followed by the addition of heparin to the cell cultures. The conditioned media were collected and analyzed by SDS-PAGE and immunoblotting. As shown in Fig.  5B, pretreatment with anti-FGFR-2 had little effect on heparin-induced sFRP-1 accumulation (lanes 5 and 6). In contrast, anti-FGFR-1 significantly diminished the heparin effect (lanes 3, 4, 7, and 8), further demonstrating that the FGF signaling pathway is indeed involved in mediating heparin-induced sFRP-1 accumulation.
When Expressed in Lec3.

Cells, a Mobility Difference between sFRP-1 of Heparin-treated Cells and Those of Mock-treated Cells Is Identified, and This Difference Is Not Due to N-Linked Glycosylation-
We established stable cell lines expressing sFRP-1 from the glycosylation deficient CHO cell line Lec3.2.8.1 with the intention of producing protein suitable for structure determination. This cell line carries mutations in four enzymes involved in a sugar modification pathway resulting in carbohydrate structures with severely truncated N-linked and O-linked sites (34). Production of sFRP-1 from this cell line was relatively low, but to our surprise, this mutant CHO line is responsive to heparin treatment (Fig. 6A). This is contrary to the result seen previously when using the CHO-DUKX parental line and further demonstrates a cell type-specific rather than species selective response. Interestingly, there is a mobility difference between sFRP-1 from treated and mock-treated cells (lane 1 versus 2; 3 versus 4; 5 versus 6; 7 versus 8, 9 versus 10, 11 versus 12). sFRP-1 from heparin-treated cells migrates as a doublet, whereas sFRP-1 from mock-treated cells predominantly migrates as the lower form. This difference is not due to the direct binding of heparin to sFRP-1, because adding heparin to the conditioned media from mock-treated cells does not change the migration pattern of sFRP-1 (data not shown).
Because sFRP-1 contains two potential N-linked glycosylation sites (Asn 172 and Asn 262 ), with Asn 172 being utilized in vivo (20), it is possible that the mobility difference caused by heparin treatment is a result of additional N-linked glycosylation. To test this possibility, nickel-NTA-purified sFRP-1 from heparintreated and mock-treated cells were digested with Endo H or PNGase F to remove sugar modification. As shown in Fig. 6B,  (20). The sFRP-1 doublet from heparin-treated cells was also responsive to treatments of Endo H and PNGase F, but both forms appear to migrate faster in SDS-PAGE ( lanes 5 and 6 versus lane  4). The lower band of the doublet migrates similarly to those of heparin mock-treated sFRP-1 (lane 2 versus 5; 3 versus 6), after the enzyme digestion. The upper band also shifts but does not change in relative intensity compared with the lower, indicating the mobility difference is not due to N-linked glycosylation.

sFRP-1 Is Post-translationally Modified by Tyrosine Sulfation at Tyrosines 34 and 36, Which Are Inhibited by Heparin-
Inspection of the primary sequence of sFRP-1 reveals that it contains two possible tyrosine-sulfation motifs (36,37) at Tyr 34 and Tyr 36 . To examine whether the mobility difference of sFRP-1 caused by heparin treatment is due to tyrosine sulfation, sodium chlorate, a selective inhibitor of tyrosine sulfation (38), was used. We treated the stable Lec3.2.8.1 cell line (25-6) with 30 or 60 mM sodium chlorate in the presence or absence of heparin (Fig. 7A). Prior to treatment, the sFRP-1 protein migrates as a doublet in the SDS-PAGE (lane 1). Sodium chlorate treatment causes a mobility shift to the predominantly upper species (lanes 3 and 5). This slower migration pattern of unsulfated protein is also observed in other tyrosine-sulfated proteins (39,40). As observed in lane 2, heparin treatment in the absence of the inhibitor of tyrosine sulfation results in the same apparent gel shift to apparently unsulfated sFRP-1. In addition, the protein signal of sFRP-1 in the chlorate-treated sample with no heparin (lane 3) is significantly stronger than the protein signal in the corresponding chlorate mock-treated sample (lane 1), indicating that unsulfated sFRP-1 appears to be relatively more stable than the sulfated protein.
To provide direct proof of tyrosine sulfation in sFRP-1, a metabolic labeling experiment with [ 35 S]sulfate was performed as described under "Experimental Procedures." The Lec3.2.8.1 stable line for sFRP-1 was incubated in a low sulfate medium. Cells were treated or mock-treated with sodium chlorate before FIGURE 5. Heparin induction of sFRP-1 accumulation requires fibroblast growth factor and its receptor. A, the heparin effect on sFRP-1 accumulation requires FGF-2. The stable HEK293 line for sFRP-1 (100-5) was grown to confluence in a 6-well plate, and cells were replaced with fresh serum-free media containing 50 g/ml of heparin. 50 ng/ml of FGF-1 or FGF-2 were added to the media in the proper wells. Conditioned media were harvested 48 h after media replacement. Protein samples were separated by SDS-PAGE and immunoblotted with anti-His 4 antibody. B, fibroblast growth factor receptor-1 (FGFR-1) is required for the heparin effect on sFRP-1 accumulation. The stable HEK293 cell line for sFRP-1 (100-5) grown to confluence in a 6-well plate was pre-treated or mock-treated with rabbit polyclonal anti-FGFR-1 or anti-FGFR-2 antibodies for 6 h. Then the cells were treated with 50 g/ml of heparin. Conditioned media were harvested 48 h after heparin treatment. Protein samples were separated by SDS-PAGE and immunoblotted with anti-His 4 antibody. N-linked glycosylation. A, sFRP-1 migrates differently in Lec3.2.8.1 stable lines when treated or mock-treated with heparin. Stable lines of sFRP-1-(His 6 ) in Lec3.2.8.1 cells were generated as described under "Experimental Procedures." Cells of different stable lines were treated or mock-treated with heparin, and conditioned media were collected at 72 h. Protein samples were analyzed by immunoblotting with anti-His 4 antibody. B, the migration difference is not due to N-linked glycosylation. Conditioned media from the stable Lec3.2.8.1 cell (25-6) mock-treated or treated with heparin were passed through a nickel-NTA column as described under "Experimental Procedures." The protein samples after nickel-NTA were treated with Endo-H or PNGase F according to the manufacturer's protocols for 20 h. Protein samples were analyzed for immunoblotting with anti-His 4 antibody.

FIGURE 6. A mobility difference between sFRP-1 from the heparintreated cells and those from mock-treated cells is identified and this difference is not due to
overnight labeling with [ 35 S]sulfate in the presence or absence of heparin. sFRP-1 protein in the conditioned media was affinity purified by nickel-NTA columns. The protein samples separated by SDS-PAGE were either immunoblotted with anti-His 4 antibody or exposed to phosphorimager for radioactive detection. As shown in Fig. 7B, sFRP-1 can indeed be labeled with [ 35 S]sulfate (lane 7). In the heparintreated sample (lane 8), [ 35 S]sulfatelabeled sFRP-1 is also present with a stronger signal. This is consistent with previous data that heparin treatment enhances the accumulation of both the sulfated and unsulfated proteins resulting in an apparent doublet in SDS-PAGE (Fig. 6). As predicted, treatment of sodium chlorate drastically abolishes the labeling of [ 35 S]sulfate (Fig. 7B, lanes 7 versus 9 and 11; 8 versus 10 and 12). This data supports the conclusion that the upper band of the doublet is the unsulfated sFRP-1, because the signal of this protein band remains strong in the immunoblotting with the chlorate-treated samples (lanes 2-6), whereas 35 S labeling signals in these samples are abolished drastically (lanes 9 -12).
To provide direct evidence that sulfation detected in sFRP-1 is indeed due to tyrosine sulfation, three sFRP-1 mutants substituting Tyr 34 and Tyr 36 individually or simultaneously with Phe were generated. These constructs along with wild type sFRP-1 were expressed transiently in Lec3.2.8.1 cells and conditioned medium examined following 120 h in the presence or absence of heparin. As shown in Fig.  7C, mutants Y36F and Y34F/Y36F were expressed in a similar level to wild type (lanes 1-4, 7, and 8). Interestingly, Y36F and Y34F/Y36F migrate as an upper band of the sFRP-1 doublet (lane 10 and 12 versus 9), which is similar to the heparin-treated (Fig. 7A, lane 2) and sodium chlorate-treated sFRP-1 (Fig. 7A, lanes 3 and 5). The expression of Y34F is barely detectable (Fig. 7C, lanes 5 and 6), indicating that this mutant is not stable. To   was metabolically labeled with [ 35 S]sulfate in the presence or absence of sodium chlorate as described under "Experimental Procedures." Conditioned media were passed through nickel-NTA columns as described under "Experimental Procedures." The protein samples separated by SDS-PAGE were exposed to a phosphorimager overnight or immunoblotted with anti-His 4 antibody. C, mutant sFRP-1 Y36F and Y34F/Y36F migrate as the upper band of the doublet of sFRP-1, which is similar to the heparin-treated and sodium chlorate-treated sFRP-1. Lec.3.2.8.1 cells were transiently transfected with pWZ1049 (wild type (WT) sFRP-1 His 6 ), pWZ1134 (Y36F), pWZ1143 (Y34F), and pWZ1148 (Y34F/Y36F) for 120 h in the presence or absence of heparin (50 g/ml). The conditioned media were analyzed by immunoblotting with anti-His 4 antibody. Lanes 9 -12 were samples of lanes 1, 3, 5, and 7 in a separated gel. D, sFRP-1 is tyrosine sulfated at Tyr 34 and Tyr 36 . HEK293 cells were transiently transfected with pWZ1049, pWZ1134, pWZ1143, and pWZ1148 for 24 h, cells were metabolically labeled with [ 35 S]sulfate in the presence or absence of heparin (50 g/ml) as described under "Experimental Procedures." Conditioned media were passed through nickel-NTA columns as described under "Experimental Procedures." The protein samples separated by SDS-PAGE were exposed to a phosphorimager overnight or immunoblotted with anti-His 4 antibody.
determine whether these sFRP-1 mutants can be labeled with [ 35 S]sulfate, construct DNAs were transfected into HEK293 cells and the cells were incubated with [ 35 S]sulfate as described under "Experimental Procedures." sFRP-1 protein in the conditioned media was affinity purified by nickel-NTA resin. The protein samples were either immunoblotted with anti-His 4 antibody or exposed to phosphorimager for radiodetection. As shown in Fig. 7D Fig. 7B (lane 8 versus 7). For the Y34F and Y36F single mutant samples the labeling signal decreases compared with the wild type (Fig.  7D, lane 12 and 14 versus 10), and for the double mutant sample labeling is not detected (lane 16). Taken together, the results indicate that both Tyr 34 and Tyr 36 undergo sulfation in wild type sFRP-1, with Tyr 36 predominating.
The Relative Stabilization of Unsulfated sFRP-1 and the Direct Stabilization by Heparin Contribute in Part to the Accumulation of sFRP-1 Induced by Heparin-To test whether the tyrosine sulfation state of sFRP-1 has an effect on the stability of the protein, we carried out two sets of whole cell binding experiments similar to that outlined in Fig. 4A. First, sFRP-1 protein in the conditioned media of both heparin-treated and mock-treated stable Lec3.2.8.1 cells were nickel-NTA purified and used for the assay. Consistent with previous data (Fig.  4A), sFRP-1 from heparin-treated cells (mainly unsulfated sFRP-1) is fairly stable within 1 h regardless of the presence of heparin (Fig. 8A, lanes 6 versus 7-10; 16 versus 17-20). In contrast, sFRP-1 from the conditioned media of the mocktreated cells (mainly sulfated sFRP-1) is less stable without heparin than with heparin (Fig. 8B, lanes  16 versus 19 and 20; lanes 6 versus 9 and 10), suggesting that tyrosinesulfated sFRP-1 is relatively less stable than unsulfated sFRP-1 in the absence of heparin. To provide further evidence for the stability effect of tyrosine sulfation, the tyrosine sulfation-free mutant protein (Y34F/ Y36F) was used for the whole cell binding assay. Wild type and mutant sFRP-1 protein were nickel-NTA purified from the conditioned media of mock-treated HEK293 cells transiently transfected with the construct DNAs and used for the assay. As shown in Fig. 8C, mock-treated wild type sFRP-1 protein (mainly sulfated sFRP-1) is very stable up to 16 h in the culture  Fig. 4A were carried out. In panels A and B, nickel-purified sFRP-1 either from heparintreated conditioned media (A) or from mock-treated conditioned media (B) of the stable Lec3.2.8.1 line were incubated with cells grown at a cell density of 0.5 ϫ 10 6 /ml at a concentration of 0.5 g/ml. In panels C and D, wild type and mutant (Y34F/Y36F) sFRP-1 were nickel purified from the conditioned media of 120-h transient transfected HEK293 cells and incubated with cells grown with a cell density of 0.5 ϫ 10 6 /ml at a concentration of 0.5 g/ml. The mixtures contained either no heparin or 50 g/ml of heparin. Cell cultures were collected at different time points. Cells and media were separated by centrifugation. The samples were analyzed by immunoblotting with anti-His 4 antibody. media with the presence of heparin (lanes 5-8). In contrast, sFRP-1 becomes less stable in the culture at 5 (lanes 3 versus 1) and 16 h (lanes 4 versus 1), indicating that heparin can directly stabilize sFRP-1. For Y34F/Y36F tyrosine sulfation free mutant sFRP-1, the protein is stable in the presence of heparin (Fig. 8D,  lanes 5-8). In the absence of heparin, the mutant sFRP-1 is also less stable than in the presence of heparin (lanes 1-4), but it is relatively more stable than wild type sFRP-1 without heparin (Fig. 8, C, lanes 1-4, versus D, lanes 1-4). These data together suggests that the unsulfated sFRP-1 is relatively more stable than the sulfated protein, which along with the direct stabilization effect of heparin contributes partly to the accumulation of sFRP-1 induced by heparin.

DISCUSSION
In this study, we have discovered that heparin, a highly negatively charged macromolecule, can exert two unexpected effects on sFRP-1 protein. One is that it can specifically stimulate protein accumulation of recombinant sFRP-1 in a cell typespecific manner. The enhancement effect requires O-sulfation in heparin and involves the FGF signaling pathway. Another surprising effect is that heparin can affect the post-translational modification of sFRP-1. We have provided evidence for the first time that sFRP-1 is tyrosine-sulfated at two N-terminal tyrosines and the modification can be inhibited by the treatment of heparin. Subsequent sequence analysis for the current sFRP protein family members reveals that tyrosine-sulfation sites are highly conserved in sFRP-1 and sFRP-5 ( Fig. 9), but are not present in sFRP-2, sFRP-3, and sFRP-4 (41), suggesting that the heparin regulating tyrosine-sulfation modification may be significant in differentiating the function among different sFRP proteins.
sFRP-1 is known to be a heparin-binding protein (13,18,19) and it has been suggested in previous studies that heparin might stabilize sFRP-1 and help mediate the release of sFRP-1 from the extracellular matrix into the media (13,19). Our data indicates that the accumulation of sFRP-1 induced by heparin is a complicated event. Although we did observe that some sFRP-1 was bound to the cell surface matrix, the quantity was only a very small fraction of total protein. Our results suggest that the relative stabilization of unsulfated sFRP-1 might also contribute partly to the accumulation. Mutating Tyr 34 to Phe destabilizes sFRP-1 (Fig. 7C, lanes 5 and 6), implying this region is related to protein stability. The tyrosine sulfation-free double mutant Y34F/Y36F is much more stable than Y34F alone, suggesting it could be due to the stabilization effect of de-sulfation (Fig. 7C, lane 11 versus 12). Our data confirm the notion that heparin can directly stabilize the sFRP-1 protein. Both sulfated and unsulfated sFRP-1 are stable in the culture with the presence of heparin. Nonetheless, with all these factors, it still cannot explain the overall accumulation of both intracellular and extracellular sFRP-1 upon the addition of heparin. The involvement of the FGF pathway suggests that additional factor(s) may participate. Our data indicate that heparin and FGF pathways may control the post-mRNA level of protein synthesis, as Northern analysis shows that the mRNA level of sFRP-1 is not affected by the presence of heparin (Fig. 4B). FGF and heparin may activate genes whose products can up-regulate the translational process, or facilitate protein trafficking along the secretory pathway.
Protein tyrosine sulfation is a post-translational modification that occurs in the trans-Golgi network and affects a number of membrane or secreted proteins (36,37). In mouse and human, one of two tyrosyl-protein sulfotransferases, TPST1 and TPST2, mediates the process of tyrosine sulfation (42)(43)(44). They catalyze the transfer of sulfate from 3Ј-phosphoadenosine 5Ј-phosphosulfate to tyrosine residues in proteins (45). Functional significance for tyrosine sulfation has been documented in a handful of cases, including protein-protein interaction between P-selectin glycoprotein ligand-1 and P-selectin (46 -48), optimal viral entry for CCR5 coreceptor function (40), ligand recognition of hormone receptors (39), and binding of platelet glycoprotein Ib␣ to von Willebrand factor (49). Here we show for the first time that sFRP-1 is also a tyrosine-sulfated protein, in addition to its N-linked glycosylation modification. We provide evidence that tyrosine sulfation could partially destabilize the sFRP-1 protein, which is consistent with the biosynthetic studies (13) in which sFRP-1 is susceptible to degradation in the absence of heparin. Because sFRP-1 functions under the extracellular matrix environment where active enzyme is subjected to tight regulation, the relative stability of the protein could be physiologically important.
The finding that the extracellular heparin can inhibit intracellular post-translational modification of sFRP-1 is unanticipated. This indicates that heparin may inhibit the process of tyrosine sulfation such as the tyrosyl-protein sulfotransferases enzymes or sulfate donor pathways. Given that heparin is a highly charged molecule that cannot penetrate membranes, it must activate a signal transduction pathway to carry out the effect. Our finding of the involvement of the FGF pathway is consistent with this hypothesis. The FGF family encompasses over 20 factors, with two prototypical members: acidic FGF (FGF-1) and basic FGF (FGF-2) (22,24,50,51). The FGF ligands bind to four related FGFRs expressed on most types of cells in tissue culture. Additional structural variants of FGFRs resulting from alternative splicing also exist (24, 50 -52). Our study has shown a specificity of FGFs and FGFRs on sFRP-1 accumulation, demonstrating that FGF-2 and FGFR-1 are involved. It has been long suspected that the FGF signaling pathway interacts with the Wnt signaling system for normal limb development (53). Our findings here seem to support the speculation that these pathways may be interconnected.
Uncovering the multifaceted regulation of sFRP-1 by heparin may be significant for understanding Wnt signaling regulation. It will be interesting to know whether tyrosine-sulfated and unsulfated sFRP-1 may function differently in interacting with Wnt. sFRP-1 is expressed in multiple organs and cell types (13), thus HS and FGF may provide an additional level of regulation of the accumulation of sFRP-1 in tissues where it modulates Wnt function. Changing the amount of cell surface HS in different tissue cells may alter the intercellular distribution of sFRP-1. An additional level of complexity may be that sFRP-1 has a biphasic effect on Wnt signaling, neutralizing Wnt-induced effects at high concentrations and stabilizing them at lower concentrations (18). sFRPs have also been suggested to facilitate boundary definition in the developing organism by limiting the range of Wnt activity (17). Thus an uneven gradient of sFRP-1 might produce an uneven gradient of active Wnt protein in regions where a Wnt is expressed uniformly.
In conclusion, this study has discovered two unexpected roles of heparin in regulating sFRP-1 protein. The findings of heparin-regulated sFRP-1 accumulation and tyrosine sulfation might provide a better understanding of Wnt signaling regulation.