We wanted to know if Sulf-2, as a 6-O endosulfatase, could modulate the interaction of known heparin-binding growth factors to heparin. We prepared and purified amino-terminal FLAG-tagged HSulf-2 (rHSulf-2) from stably transfected kidney 293 cells (Figs. 1A and 1B). The major rHSulf-2 component detected by the H2.3 antibody migrated with an apparent molecular weight of 72 kDa. Immunoblotting of rHSulf-2 with a FLAG antibody revealed a protein band of the same molecular weight (Fig. 1B). We conclude that the 72 kDa component represents an N-terminal subunit with a C-terminal truncation. It should be noted that there are several potential furin cleavage sites that are C-terminal to the position of the peptide immunogen (aa 484–504) 15. It is not known at present whether protein processing is necessary for enzymatic activity against heparan sulfate substrates. We also prepared an enzymatically-inactive mutant of HSulf-2 15 in which two essential cysteines were mutated to alanines (rΔCCHSulf-2). As we had previously shown for CHO cell-produced material 15, rHSulf-2 exhibited endosulfatase activity against heparin in a concentration- and time- dependent manner (Figs. 1C and 1D). The assay employs HPLC to examine to the disaccharide composition of treated and untreated heparin (see Methods). After 24 h of enzyme treatment, 80% of the trisulfated disaccharides were converted to the disaccharide products. The mutant protein was inactive in this assay (data not shown). We next developed solid phase ELISAs to measure the binding of protein ligands to immobilized heparin-BSA. Antibodies directed to the ligands were used for detection. For "pre-binding" effects, we first treated heparin-BSA with rHSulf-2 and then tested for its ability to support ligand binding (Fig. 2A). For "post-binding" effects, we determined the ability of rHSulf-2 to release the ligand from a preformed complex with heparin-BSA (Fig. 2A).

We initially focused our attention on three growth factors: VEGF₁₆₅, FGF-1 and FGF-2. Sulfate groups in heparin/HS, including glucosamine-6-sulfates, are required for the interaction of VEGF₁₆₅ and FGF-1 with heparin/HS 283136. In contrast, FGF-2 binding to heparin/HS depends on N-sulfation and iduronate-2-sulfation but not on 6-O-sulfation 283637. We observed saturable binding of VEGF₁₆₅, FGF-1 and FGF-2 to heparin-BSA (Figs. 2B and 4B). We estimated K_d's for these interactions and the binding of three chemokines (CXCL12, CCL21, and CXCL8) to heparin-BSA (Table 1). Negligible binding was seen when the wells were coated with BSA alone (data not shown). As a positive control to eliminate heparin-dependent binding, we pre-digested heparin-BSA with a mixture of heparinases (0.2 mU of heparinase I, 0.1 mU of heparinase II and 0.04 mU of heparinase III) for 4 h and found that this digestion reduced the binding of VEGF by >92%. Treatment of heparin with rHSulf-2 reduced the binding of varying concentrations of VEGF to heparin-BSA almost to background levels (Fig. 2B). rHSulf-2 treatment diminished the binding of FGF-1 by >60%. In contrast, rHSulf-2 had only a negligible effect on the binding of FGF-2 to heparin (Fig. 2B). To determine the efficacy of the enzyme, we fixed the concentration of ligand at approximately its K_d and tested varying concentrations of rHSulf-2 or the inactive mutant, rΔCCHSulf-2. For FGF-2, only minimal effects of rHSulf-2 were seen up to 100 ng of the enzyme, whether tested in the pre-binding (Fig. 2C) or post-binding assay (Fig. 2D). In contrast, treatment of heparin-BSA with 50 ng or more of rHSulf-2 substantially reduced its binding to VEGF in a dose-dependent manner (Fig. 2C), whereas the mutant enzyme was largely inactive. In the post-binding assay, rHSulf-2 was also efficacious in releasing VEGF from preformed complexes with heparin-BSA (Fig. 2D, Table 2). The slight inhibitory effects of rΔCCHSulf-2 (Figs. 2C and 2D) in both the pre-binding and post-binding assays may be attributable to a competing heparin-binding activity of Sulf-2 protein due to its highly basic hydrophilic domain 15.

Chemokines are also heparin/HS-binding proteins 6. However, limited information is available about the specific sulfation requirements for these interactions. We therefore took advantage of rHSulf-2 to determine the contributions of glucosamine-6-sulfation to these interactions. rHSulf-2 was very effective in reducing the interaction of SDF-1 (CXCL12) with heparin-BSA in both pre-binding and post-binding assays (Fig. 3A and 3B, Table 2). In contrast, rHSulf-2 had only partial inhibitory effects on the binding of SLC (CCL21) to heparin-BSA (Fig. 3A). IL-8 (CXCL8) bound less well to heparin-BSA, but rHSulf-2 treatment still attenuated the interaction (Fig. 3A).

SULF2 transcripts are upregulated in breast, central nervous system and colon carcinomas 2425. Moreover, transcripts are detected in some cultured human breast carcinoma cell lines and the mRNA-positive cells release Sulf-2 protein in their conditioned medium 25. Sulf-2 expression is highest in the conditioned of MCF-7 cells and we have demonstrated that the secreted protein has associated arylsulfatase activity using 4-methylumbelliferyl sulfate as the substrate 25. As shown in Fig. 4A, the CM from these cells exhibited concentration- and time-dependent endoglucosamine-6-sulfatase activity against heparin. In pilot experiments, the CM also showed activity against bovine intestinal heparan sulfate (HS) producing a substantial reduction in the amount of the trisulfated disaccharide with a corresponding increase in the amount of the disulfated disaccharide product (data not shown). Using CM as a source of the native enzyme, we analyzed the "pre-binding" and "post-binding" effects on the binding of VEGF or FGF-2 to heparin. Consistent with the rHSulf-2 findings, pretreatment of heparin-BSA with a fixed concentration of MCF-7 CM strongly reduced the subsequent binding of VEGF₁₆₅ or FGF-1 but had no effect on the binding of FGF-2 (Fig. 4B). In a dose-dependent manner, CM prevented (Fig. 4C) and reversed (Fig. 4D) the interaction of VEGF₁₆₅ or FGF-1 with heparin-BSA. Similar effects were observed with SDF-1 (data not shown). In contrast, the same range of CM concentrations had no appreciable effect on the binding of FGF-2 to heparin-BSA in either the pre-binding or post-binding assays (Figs. 4C and 4D). To confirm that the reductions in heparin binding were, in fact, due to Sulf-2, we employed anti-HSulf-2 antibodies 25 for immunodepletion of Sulf-2 from MCF CM. Pre-clearing was carried out by incubating CM with protein A-Sepharose complexed to a mixture of two specific antibodies and then centrifuging the mixture. As shown by Western blotting, the specific antibodies removed the majority of HSulf-2 protein from the CM, whereas control IgG removed only a relatively small amount of the protein (Fig. 4E). In parallel with the depletion of HSulf-2 protein, the "pre-binding" activity of the precleared CM, measured with respect to VEGF₁₆₅ binding, was greatly diminished as compared to that of the control IgG-treated CM (Fig. 4F).

Heparin is a structural analogue of the S-domains of heparan sulfate. We wanted to determine whether the effects of Sulf-2 on heparin interactions would be repeated with bona fide heparan sulfate. For this purpose, we conjugated BSA with heparan sulfate chains (HS-BSA) that were isolated from porcine intestinal mucosa. FGF-1 bound to HS-BSA, although with a 3-fold lower affinity than to heparin-BSA (Fig. 5, Table 1). This binding was abolished by pre-digestion of the HS chains with heparinases. Pretreatment of the HS-BSA with MCF-7 CM or rHSulf-2 substantially diminished FGF-1 binding (55–70% reduction), whereas rΔCCHSulf-2 did not exert this effect (Fig. 5).

The following materials were obtained commercially from the source indicated. Heparin conjugated with bovine serum albumin (Heparin-BSA), BSA, heparinases (I, II and III) were from Sigma (St Louis, Mo., USA); heparan sulphate (porcine intestinal mucosal) from Organon (Oss, The Netherlands); DEAE-Sephacel from Sigma (Poole, UK); Phenyl-Sepharose CL-4B from Amersham Biosciences (Little Chalfont, UK); Protein A Sepharose beads from RepliGen (Needham, MA); recombinant human VEGF₁₆₅, human CXCL12 (SDF-1), anti-human VEGF antibody (Ab), anti-human FGF-1 Ab, anti-human FGF-2 Ab, anti-human CXCL12 Ab, anti-CXCL8 Ab and anti-human CCL21 Ab from R&D Systems. All are polyclonal antibodies produced in goats. Recombinant human FGF-1 (acidic FGF), human FGF-2 (basic FGF), human CCL21 (SLC) and CXCL8 (IL-8) were from Leinco Inc. A biotinylated swine anti-goat IgG (H+L) antibody and a streptavidin conjugated with alkaline phosphatase were from Caltag. FLAG-tagged versions of HSulf-2 and the inactive mutant designated ΔCCHSulf-2 were produced in 293 cells and purified as previously described 25. Polyclonal rabbit anti-bodies against HSulf-2 (H2.1 and H2.3) were generated and prepared as previously described 25.

Porcine intestinal mucosal heparan sulfate was coupled, via its reducing end, to BSA based on the procedure of Najjam et al. 48. Briefly, 2 mg of sodium cyanoborohydride was added to a mixture of 50 mg of HS and 2 mg of BSA in 2 ml of 0.2 M potassium phosphate buffer pH 8. The mixture was then incubated for 2 days at 37°C. After dilution to 10 ml with 0.15 M NaCl, 20 mM phosphate, pH7.4, it was applied to a 5 ml column of DEAE-Sephacel equilibrated in the same solution. After washing with 0.15 M NaCl solution, followed by buffered 0.4 M NaCl, to remove reagents and any non-conjugated BSA, the HS-containing material was then step-eluted with 0.5 ml aliquots of buffered 1.5 M NaCl. HS-containing fractions were identified by spotting aliquots onto Whatman filter paper and staining with Azure A. The pooled fractions were adjusted to 3 M in (NH₄)₂SO_4, by addition of solid (NH₄)₂SO₄, and then applied to a 2 ml column of Phenyl-Sepharose CL-4B equilibrated in 3 M (NH₄)₂SO₄. After washing with 3 M (NH₄)₂SO₄ to elute free HS, the bound HS-BSA conjugate was step-eluted with 0.5 ml aliquots of 0.15 M NaCl, 20 mM phosphate pH7.4. Fractions containing conjugated BSA were identified by their absorbance at 280 nm. The final product ran as a diffuse band, stainable with both Coomassie Blue and Azure A, close to the top of a 10% polyacrylamide SDS-PAGE gel and with a higher apparent MW than free BSA. The BSA content of the conjugate was assessed by absorbance at 280 nm. The HS content was determined by digestion with a mixture of heparinases I, II and III, and measurement of the resulting increase in absorbance at 232 nm. It was estimated that the average level of conjugation was 1–2 HS chains/BSA molecule.

FLAG-tagged versions of active Sulf-2 (rHSulf-2) and inactive enzyme (rΔCCHSulf-2) were produced in stably transfected kidney 293 cells as described previously 25. The purities of rHSulf-2 and rΔCCHSulf-2 were estimated to be 90% and 80%, respectively based on Coomassie blue staining of SDS-PAGE gels. The amount of Sulf-2 protein was quantified by the BCA assay (Pierce). The amounts of rHSulf-2 and rΔCCHSulf-2 were equalized by an ELISA employing an anti-FLAG M2 antibody (Sigma) as the capture reagent and a mix of H2.1 and H2.3 antibodies for detection. The human breast carcinoma cell line, MCF-7, was grown in RPMI-1640 containing 10% fetal calf serum in a 150 mm-flask. Cells (80% confluent) were rinsed with PBS and OptiMEM (Invitrogen) and then incubated in 25 ml of OptiMEM at 37°C in 5% CO₂ for 72 h. The CM was collected and then 100-fold concentrated on a Centricon-30. Subsequently, the concentrated CM was dialyzed into 50 mM HEPES, pH 8.0.

The experimental design for these ELISA is schematically shown in Fig 2A. To immobilize heparin, 100 ng/ml of heparin-BSA in phosphate-buffered saline (PBS) was added to the wells (100 μl/well) of a 96-well plate (Immulon 2HB, Dynex Labs). The plate was placed at 4°C overnight. The wells were washed 3 times with PBS containing 0.1% Tween-20 (PBS-T) and then blocked with 3% BSA (Sigma) in PBS-T at 25°C for 2 h. To determine the effects of Sulf-2 on the ligand-binding activities of heparin ("pre-binding" effects), the wells were washed as above and then incubated with 100 μl of a reaction mixture containing 5 μmol of HEPES, pH 7.6, 1 μmol of MgCl₂ and enzyme (rHSulf-2, rΔCCHSulf-2 or MCF-7 CM) at 37°C for 4 h. In the cases where we carried out dose response studies with protein ligands, incubation with the enzyme was overnight. The wells were washed 3 times with PBS-T and then incubated with 25 μl of heparin-binding factors (25 nM VEGF, 12.5 nM FGF-2, 12.5 nM FGF-1, and 100 nM CXCL12) in PBS at 25°C for 30 min. The wells were washed 3 times and incubated with 1 μg/ml of each primary antibody in 0.1% BSA/PBS (100 μl/well) at 25°C for 1 h. The wells were washed as above and incubated with 1.2 μg/ml of a biotinylated secondary antibody in 0.1% BSA/PBS (100 μl/well) at 25°C for 30 min. Then, the wells were washed and incubated with 2 μg/ml of an alkali phosphatase-conjugated streptavidin in 0.1% BSA/PBS (100 μl/well) at 25°C for 30 min. Finally, the wells were incubated with PNPP (Pierce) in cacodylate buffer, pH 9.2 (100 μl/well) at 25°C for 5 minutes, and then OD 405 nm was read on a microplate reader. For the analysis of "post-binding" effects (i.e. dissociation of heparin-ligand complexes), 25 μl of each heparin-binding factor was added to plates coated with heparin-BSA after blocking with 3% BSA. The plates were incubated at 25°C for 30 min to form the heparin-factor complex. The wells were washed as above and then incubated with 100 μl of the enzyme reaction mixture as described above. After washing the wells, the amount of bound factor was determined with the antibody detection procedures described above. For the "pre-binding" study of FGF-1 binding to heparan sulfate-BSA, 1 mg/ml of heparan sulfate-BSA was coated on a 96-well plate (100 μl/well). Other procedures are the same as described above for the heparin-BSA assay. Representative data are shown from one of 2–3 independent experiments in all cases.

A purified FLAG-tagged HSulf-2 (10 ng) or concentrated MCF-7 CM (10 μl) was separated by electrophoresis on a reducing SDS-7.5% polyacrylamide gel (Bio-Rad), and blotted onto ProBlott™ membrane (Applied Biosystems, Foster City, CA). The membrane was blocked with 5% skim milk/PBS-T for 1 h and then incubated overnight with an anti-HSulf-2 peptide antibody (see below) at a concentration of 1 μg/ml in 5% skim milk/PBS-T at 4°C. The membrane was washed and incubated with horseradish peroxidase-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch Laboratories Inc, West Grove, PA) (0.2 μg/ml) for 1 h. Bound antibodies were visualized with SuperSignal West Pico Chemiluminescent reagent (Pierce).

A mixture of anti-HSulf-2 peptide antibodies, H2.1 (5 μg) and H2.3 (5 μg) 25, or a rabbit IgG (10 μg), 10 μl of a 50% (v/v) suspension of protein A-Sepharose (IPA-400HC, RepliGen) and 100 μl of 100-fold concentrated CM of MCF-7 cells were mixed in a total volume of 1 ml with 150 mM NaCl and 50 mM Tris-HCl, pH 7.4. The mixture was rocked overnight at 4°C. The immunocomplexes bound to the protein A beads were removed by centrifugation. The "pre-binding" and "post-binding" ELISA activities remaining in the supernatant were assayed as described above.

Endoglucosamine-6-sulfatase activity of rHSulf-2 and MCF-7 CM was determined by previously established procedures 15. The standard reaction mixture contained 5 μmol of HEPES, pH7.8, 1 μmol of MgCl₂, 5 μg of porcine intestinal heparin (Sigma), and a purified FLAG-tagged HSulf-2 or a 100-fold concentrated MCF-7 CM in a total volume of 100 μl. The mixture was incubated at 37°C. The reaction was stopped by heating at 100°C for 2 min. The heparin was digested into disaccharides by a mix of heparinase I, heparinase II and heparinase III 15. The disaccharides were then analyzed by HPLC on a Partisil-10 SAX column (Whatman, Fairfield, NJ) 15. The enzymatic activity was defined by calculating the increase in the moles of unsaturated disulfated disaccharide, IdoA(2S)-Glc(NS), in the digested heparin.