Domain-specific Modification of Heparan Sulfate by Qsulf1 Modulates the Binding of the Bone Morphogenetic Protein Antagonist Noggin*

We have reported previously that Noggin is a heparin-binding protein and associates with the cell surface through heparan sulfate proteoglycans, where it remains functional for the binding of bone morphogenetic proteins (BMPs). Here we report that the binding of Noggin to the cell surface is highly selective for heparan sulfate and that specific structural features are required for the interaction. Noggin binds most efficiently to heparin sequences composed of 10 or more monosaccharides; N-, 6-O-, and 2-O-sulfates contribute to this interaction. In addition, we have shown that the developmentally regulated endosulfatase Qsulf1 selectively removes sulfate groups from the 6-O position of sugars within the most highly sulfated S domains of heparan sulfate, whereas 6-O-sulfates in the NA/NS domains are not substrates for the enzyme. The activity of Qsulf1 in cells in culture results in the release of Noggin from the cell surface and a restoration of BMP responsiveness to the cells. This shows that Noggin binds to the S domains of heparan sulfate and provides evidence that, in addition to modulating Wnt signaling in vivo by the release of heparan sulfate bound Wnt, Qsulf1 also modulates BMP signaling by the release of surface-bound Noggin.

Heparan sulfate proteoglycans are found ubiquitously both on the surface of cells as well as within the extracellular matrix, where they bind and modify the functions of a diverse array of ligands (1). Loss of function mutations in enzymes of the heparan sulfate biosynthetic pathway have confirmed that this polysaccharide has essential roles during development. Foremost among these is the regulation of cellular responsiveness to a number of growth factors and morphogens that control patterning events (2). Mutations in members of the glypican family of cell-surface heparan sulfate proteoglycans, in both vertebrates and invertebrates, have been specifically associated with loss of bone morphogenetic protein (BMP) 1 activ-ities (3,4). Although the precise molecular mechanisms by which these specific heparan sulfate proteoglycans regulate cellular responsiveness to BMPs in vivo is not known, recent in vitro data does support the conclusion that heparan sulfate can directly augment signaling of BMPs through their receptors (5,6).
Previously we have shown that the BMP antagonist Noggin remains associated with the surface of cells by binding to heparan sulfate, where it remains functional for the inhibition of BMPs (7). We have proposed that the interaction with heparan sulfate in vivo is likely to regulate the range of influence of Noggin by restricting its diffusion, thus providing another mechanism by which cellular responsiveness to BMPs can be regulated by heparan sulfate in vivo (7). This hypothesis supposes that the interaction between Noggin and heparan sulfate is sequence specific and implies that there should be a developmentally regulated mechanism in vivo for the specification of heparan sulfate structures that modulate Noggin binding to the cell surface.
Heparan sulfate sequences in vivo are highly variable between tissues, and the molecular mechanisms that regulate the occurrence of this variability is an issue of significant importance in understanding the functions of heparan sulfate proteoglycans in vivo. One recently appreciated mechanism for the developmentally regulated modification of the sulfation of heparan sulfate is the expression of endo-6-O-sulfatases such as Qsulf1, which are capable of post-synthetically removing 6-O-sulfate groups from glucosamine residues in heparan sulfate and in the chemically related heparin (8 -10).
Here we report an extension of our previous studies that include a determination that Noggin binds to the cell surface selectively through heparan sulfate (HS) in a structurally specific manner. Using heparin as a chemical analogue of HS, we have found that the HS-binding site in Noggin can accommodate up to 10 monosaccharides, and N-, 2-O-, and 6-O-sulfate residues are needed for optimum interaction. Furthermore, we have identified a mechanism by which the occurrence of Noggin-binding sites can be regulated in vivo by the specific action of Qsulf1 on the S domains of HS.

EXPERIMENTAL PROCEDURES
Antibodies-RP57-16 was a gift of Regeneron (Tarrytown, NY). This rat monoclonal antibody was generated by using native human Noggin protein as immunogen. Ascites fluid from SCID mice, affinity-purified by protein G affinity chromatography, was used in Western blotting, immunoprecipitation, and immunofluorescence as indicated below. PhosphoSMAD 1,5,8 was obtained from Cell Signaling Technologies, Inc.
Plasmids-The eukaryotic expression plasmids, pQSulf1 and pMutQSulf1, encoding expression of full-length human Qsulf1 and an active-site mutant, respectively, were a gift from C. P. Emerson, Jr. (8).
Cell Culture and Transfection-Chinese hamster ovary cells (CHOK1) stably expressing Noggin have been described previously (7). Cells were maintained in DMEM/F12 media (BioWhittaker, Inc.) containing 10% fetal bovine serum (HyClone). Liposome-mediated transfection was performed by using Geneporter (Gene Therapy Systems) according to the manufacturer's recommendations. Stable cell lines previously transfected with Noggin were transfected with plasmids encoding Qsulf1 or MutQSulf1 and selected in DMEM/F12 media containing 10% fetal bovine serum with 200 g/ml hygromycin, and individual clones were harvested and subcultured. Western blotting with anti-myc monoclonal antibody (9E10) identified positive clones expressing similar levels of wild-type or mutant Qsulf1 enzyme.
Metabolic Labeling, Pulse-chase, and Immunoprecipitation-For metabolic labeling and assessment of Noggin turnover in the cell layer, cells were incubated in methionine-and cysteine-free DMEM (Invitrogen) for 40 min. Trans 35 S-Label (MP Biomedicals) was then added to each well at 200 Ci/ml, and cells were incubated at 37°C for 30 min. Subsequent to this pulse, cells were washed once with DMEM/F12 media containing 10% fetal bovine serum and chased in the same media. For competition experiments, heparin, heparin fragments, or other glycosaminoglycans were added during the chase period at 1 g/ml.
At the specified time intervals, media was recovered and the cell layers lysed using cold 1% Nonidet P-40, 0.5% deoxycholic acid, 0.1% SDS, 1 mM magnesium chloride, 0.5 mM calcium chloride in phosphate-buffered saline (PBS) containing protease inhibitors of 1 g/ml pepstatin A, 0.25 mg/ml N-ethylmaleimide, and 0.5 mM phenylmethylsulfonyl fluoride. Media was brought to similar conditions by the addition of concentrated buffer. Cell debris was removed from both cell layer extracts and media samples by centrifugation at 14,000 rpm for 1 min. Supernatants were then precleared by the addition of rat IgG (Sigma) at 1 g/ml, incubated for 30 min at 4°C, followed by the addition of protein G-Sepharose and incubation overnight at 4°C. Samples were then centrifuged, and the supernatants were used for immunoprecipitation.
RP57-16 (Regeneron) was added to each sample at 1 g/ml and incubated for 1 h at 4°C. Protein G-Sepharose was subsequently added and incubated for an additional 1 h. Beads were washed twice with the original lysis buffer without protease inhibitors and then twice with PBS containing magnesium and calcium chloride. Immunoprecipitated products were recovered from the beads by boiling in SDS-PAGE sample buffer and analyzed by electrophoresis on 10 -20% gradient SDS-PAGE gels (Bio-Rad); immunoprecipitated products were detected by autoradiography or PhosphorImager.
Western Blot Analysis-Extracts from cells expressing human Noggin and either Qsulf1 or MutQSulf1 were electrophoresed on 10 -20% gradient SDS-PAGE gels (Bio-Rad) and transferred to Immobilon-P (Millipore) by electroblotting. Filters were blocked for 1 h at room temperature using 5% non-fat dry milk in 20 mM Tris-HCl, pH 7.4, 150 mM sodium chloride (TBS), plus 0.1% Tween 20 (TBST). Blocked filters were probed with either RP57-16 at 20 ng/ml or 9E10 at 1:100 dilution of conditioned culture media in 2.5% non-fat dry milk in TBST for 1 h at room temperature. After three washes with TBST, incubation with an anti-rat or anti-mouse horseradish peroxidase-conjugated secondary antibody in TBST for 1 h and three remaining washes with TBST; protein bands were detected by ECL (Amersham Biosciences).
Purification and Analysis of Heparan Sulfate Chains-HS chains were obtained from cultured CHOhNG cells grown in DMEM/F12 media supplemented with 10% fetal calf serum. These cells were transfected with wild-type QSulf1 or the mutated form of QSulf1. Clones of cells containing differential amounts of QSulf1 were obtained. These cells were metabolically radiolabeled with 20 Ci/ml D-[6-3 H]glucosamine hydrochloride (PerkinElmer Life Sciences) and 20 Ci/ml [ 35 S]Na 2 SO 4 (MP Biomedicals) for 20 h at 37°C. The culture media was removed and frozen for future use. The cells were washed twice with PBS, then treated with 1 mg/ml trypsin (Sigma) for 10 mins at 37°C. The trypsin was then neutralized with 2 mg/ml trypsin inhibitor (Sigma). Cells were centrifuged at 1000 rpm for 10 mins at 4°C. The cell pellet was then extracted with 10 mls 20 mM sodium phosphate, pH 7.0, 0.15 M NaCl, and 1% Triton X-100. The pellets were then frozen at Ϫ20°C. The supernatants were treated with 100 g/ml Pronase (Sigma) for 3 h at 37°C. The Pronase was heat-inactivated at 80°C for 15 mins. The samples were then applied to a DEAE HiTrap (Amersham Biosciences) column for purification. The column was washed extensively with 20 mM sodium phosphate, pH 7.0, 0.3 M NaCl, and 1% Triton X-100. The column was eluted with 1 M NaCl in the phosphate Triton buffer. The recovered HS and chondroitin sulfate chains were dialyzed against 50 mM NaCl, 50 mM Tris-HCL, pH 8.0, concentrated 10-fold by reverse osmosis against 20% polyethylene glycol, added 5 mM CaCl 2 , and then digested with 0.1 unit/ml chondroitinase ABC for 4 h at 37°C. The intact HS chains were then recovered by reapplication to a DEAE HiTrap column as described above, but without Triton X-100 in the buffers. The samples were then lyophilized. The HS from the cell pellet was also purified in this manner but with the modification that, after Pronase treatment, the pellets were then digested with 6.25 units/ml (RQ1) DNase (Promega) for 30 mins at 37°C and then heat inactivated at 65°C for 10 mins. The disaccharide composition of HS was determined after depolymerization with heparinases I, II, and III, and separation on SAX-HPLC disaccharides were identified by comparing elution positions with disaccharide standards as described previously (11). HS was hydrolyzed at positions of N-sulfation by low pH (1.5) nitrous acid and the depolymerization products were resolved by gel filtration on Bio-Gel P.10 and SAX-HPLC (12).
Immunofluorescence-For the cellular localization of Noggin, live cells grown on glass coverslips were washed twice with warm serumfree medium, incubated with primary antibodies for Noggin (RP57-16 at 4.4 mg/ml) for 30 min at 37°C, and then washed twice with PBS. The cells were fixed with 4% paraformaldehyde in PBS for 10 min, washed twice with PBS, and then permeabilized in 0.5% Nonidet P-40 for 15 min. The cells were again washed twice with PBS and incubated with secondary antibodies mixed with Oregon Green Phalloidin (Molecular Probes) at 1:200 for 30 min at 22°C. The cells were washed twice in PBS followed by a water rinse, then mounted in ProLong Antifade (Molecular Probes). Immunofluorescence microscopy was performed with a confocal microscope configured with krypton and UV with the appropriate wavelength filters (568 and 488 nm) for CY3 and Phalloidin excitation.
For the assessment of cellular responsiveness to BMP4, live cells grown on glass coverslips were washed twice and then incubated for 3 h at 37°C in warm serum-free medium. They were then subsequently washed twice with PBS and incubated with BMP4 at 250 ng/ml for 30 mins at 37°C. Cells were washed two times with PBS, fixed with 4% paraformaldehyde in PBS for 10 min, washed twice more with PBS, and then permeabilized in 0.2% Triton X-100 for 5 min. After three additional washes with PBS, they were quenched with freshly prepared 0.1% sodium borohydride for 5 mins at 22°C and then given one final wash with PBS. Prior to the addition of primary antibody, cells were blocked by incubating with 10% goat serum and 1% BSA in PBS for 1 h at 22°C. Cells were incubated overnight at 4°C with a 1:100 dilution of anti-PhosphoSMAD 1,5,8 (Cell Signaling Technology, Inc.). The cells were then washed three times with PBS and incubated with CY3 donkey anti-rabbit at 1:200 for 45 min at 22°C. After three washes in PBS and a water rinse, the coverslips were then mounted in ProLong Antifade (Molecular Probes) and evaluated by standard immunofluorescence microscopy.

RESULTS
Noggin Binds Selectively to Heparan Sulfate at the Cell Surface-Our previous studies established that Noggin binds to heparin in vitro and to heparan sulfate on the surface of cultured cells (7). In support of the selectivity of this interaction, we performed pulse-chase experiments in which we found that Noggin was displaced from the cell surface by heparin, but not by chondroitin sulfate. In the present studies our goal was to further identify the specificity of the Noggin-glycosaminoglycan interaction at the cell surface. Because heparin is more highly sulfated than chondroitin sulfate and, therefore, has a higher net negative charge, we wanted to see if heparan sulfate with its lower net negative charge, but with similar structural features as heparin, had the same activity.
CHOK1 cells, stably expressing human Noggin, were pulselabeled with Trans 35 S-Label for 30 min and then chased in unlabeled media in the presence or absence of glycosaminoglycans, as we have described previously (7). Labeled Noggin was detected in the total cell layer at each time point by immunoprecipitation and SDS-PAGE followed by autoradiography. The amount of labeled Noggin was quantified and is shown in Fig.  1; it is plotted as the log Noggin recovered as a function of time.
Heparan sulfate derived from porcine mucosa ( Fig. 1, HS, open diamonds) had equal activity to heparin (Fig. 1, Heparin, open squares) in the displacement of Noggin from the plasma membrane. By contrast, neither chondroitin sulfate (Fig. 1, CS, open triangles), which has a higher net negative charge than does heparan sulfate, nor dermitan sulfate (Fig. 1, DS, open circles), which contains iduronic acid and is therefore structurally more related to heparan sulfate, had any activity. Similarly, the unsulfated glycosaminoglycan hyaluronic acid (Fig. 1, HA, crossed squares) was unable to displace Noggin from the cell surface. Together, these results imply that Noggin interacts with heparan sulfate in a structurally specific manner and suggested that further experimentation was warranted to elucidate the structural features required for this interaction.
Size Dependence of Noggin Glycosaminoglycan Interactions-To establish the minimum size requirement for the interaction between Noggin and heparan sulfate or heparin, we assessed the ability of various sizes of heparin fragments to displace Noggin from the CHO cell surface. As above, CHOK1 cells stably expressing Noggin were pulse-labeled with Trans 35 S-Label for 30 min and then chased in unlabeled media in the presence or absence of equivalent amounts of heparin fragments of defined size, previously isolated by gel filtration chromatography. Labeled Noggin was recovered from the cell layer by immunoprecipitation and analyzed by gel electrophoresis as above.
As shown in Fig. 2, heparin fragments of 10 -12 saccharides in length (Fig. 2, dp 10, dp 12), had equal activity in the displacement of cell-surface Noggin as did intact heparin (Fig.  2, Heparin, open squares). Heparin fragment of either 4 (Fig. 2, dp 4, open diamonds) or 6 (dp 6, open circles) saccharides in length were inactive, whereas heparin fragments 8 saccharides in length (dp 8, open triangles) had intermediate activity. This suggests that the optimum structural requirement for Noggin binding to heparan sulfate requires a sulfated sequence of 10 -12 mono-saccharides in binding site length, similar in size to that required for the interaction between heparan sulfate and the extracellular matrix proteins fibronectin and endostatin (13)(14)(15).
Noggin Binding to Heparan Sulfate Depends upon N-, 6-O-, and 2-O-Sulfation-To determine whether Noggin requires specific structural features of the heparan sulfate chain for binding, we repeated similar pulse-chase immunoprecipitation experiments in which we assessed the ability of chemically modified heparin chains to displace cell surface-bound Noggin. Heparin which has been fully desulfated (Fig. 3, open diamonds) shows no activity in the displacement of cell-surface Noggin, confirming the essential requirement for sulfate groups in the binding of Noggin. Similarly, heparin chains which have been either merely N-desulfated (Fig. 3, crossed  squares), or N-desulfated and reacetylated (Fig. 3, crossed diamonds) are also inactive, indicating that specifically N-sulfate residues are essential for Noggin binding. Both 6-O (Fig. 3, open triangles) and 2-O-desulfated heparins (Fig. 3, open circles) show a similar significant but partial loss of activity. Although not forming as strong a structural feature of the Noggin binding site, as is apparently the case for N-sulfation, these results suggest that sulfation at 6-O and 2-O positions are also critical for Noggin binding. Together these results imply that Noggin typically binds to the most highly modified domains along the polymer chain of HS.
The Enzyme Qsulf1 Selectively Targets the Catalytic Removal of 6-O-Sulfate Groups within the S Domains of Heparan Sulfate-The enzyme known as Qsulf1 is a developmentally regulated cell-surface protein with strong homology to lysosomal heparan sulfate 6-O-sulfatase. To further evaluate the specificity of this enzyme, stable clones of CHO cells were prepared expressing either the native Qsulf1 or a mutant enzyme bearing conversion of two critical cysteines, which are required for catalytic activity, to alanines (C89A,C90A) (8). The trypsin released heparan sulfate from the surface of CHO cells transfected with either the native or mutant Qsulf 1 enzyme was degraded by heparinases, and the disaccharides were analyzed by SAX-HPLC. The HS from cells harboring the mutated enzyme had a typical HS composition that was composed of nearequal amounts of N-acetylated and N-sulfated units, with 6-Osulfates present as ⌬HexA-GlcNAc,6S, ⌬HexA-GlcNS,6S, and ⌬HexA,2S-GlcNS,6S. In total, 9% (based on 3 H-label) of disaccharides were 6-O-sulfated (Fig. 4). The HS isolated from cells transfected with the native Qsulf1 enzyme had a nearly identical content of N-acetylated and N-sulfated units as that from cells containing the mutant enzyme; however, there was a striking difference in relative amounts of the trisulphated species (⌬HexA,2S-GlcNS,6S), which was reduced by about 80%; this reduction was accompanied by a significant increase in the disulphated component ⌬HexA,2S-GlcNS (Fig. 4). The amounts of the other two 6-O-sulfated units were unaffected.
To determine the location of action of Qsulf1 along the heparan sulfate chain, we performed nitrous acid cleavage of the same chains. Low pH (1.5) nitrous acid specifically cleaves N-sulfated disaccharides by deaminative scission; N-acetylated units are unaffected. Analysis of scission products by high resolution gel filtration on Bio-Gel P.10 reflects the domain organization of HS in which most of the GlcNS residues are in either contiguous sequences (S domains) or in regions of alternating N-acetylated and N-sulfated units (mixed sequences or NA/NS domains). S domains are degraded to disaccharides by nitrous acid, NA/NS regions yield tetrasaccharides, and larger oligomers (dp6 and above) are derived from sections of the polymer chain composed of sequences of two or more N-acetylated units interrupted by GlcNS residues.
The HNO 2 profiles from HS produced by cells expressing the native or mutant enzyme were virtually identical, indicating that transfection with the Qsulf1 enzyme did not affect the formation of the domain structure of HS (Fig. 5). Sulfation present in the tetrasaccharide peak is due mainly to 6-Osulfation of amino sugars (GlcNAc and GlcNSO 3 ). The amounts of 35 S-label (evaluated on the basis of the 3 H: 35 S ratio) were similar in the tetrasaccharide fragments, suggesting that the native enzyme was unable to remove 6-O-sulfate groups from the NA/NS domains. The tetrasaccharides were further examined by SAX-HPLC, and the profiles showed similar proportions of non-, mono-, and di-O-sulfated structures (Fig. 6). This finding essentially confirms that the Qsulf1 enzyme does not attack 6-sulfates in the NA/NS regions; its principal target is 6-O-sulfate groups in trisulphated disaccharides (HexA,2S-GlcNS,6S) which occur almost exclusively in the S domains of HS.
Qsulf1 Activity Releases Cell Surface-bound Noggin-Our results above indicated that 6-O-, 2-O-, and N-sulfate residues all participated in the binding of Noggin to heparan sulfate. This suggested that Noggin might actually bind to the S domains of heparan sulfate where the trisulfated disaccharides (HexA2S-GlcNS,6S) reside. Because Qsulf1 seems to selectively target these regions of the heparan sulfate chain, we reasoned that Qsulf1 might modulate cell-surface binding of Noggin. To assess this possibility, the CHO cells stably expressing Noggin, used in the studies above, were subsequently transfected with either Qsulf1 or the mutant enzyme and selected for stable expression.
As described above, pulse-chase analysis of these cells was performed, followed by immunoprecipitation of cell layer-associated Noggin and its quantification after PAGE-gel electrophoresis. Cells expressing active Qsulf1 (Fig. 7, diamonds) had significantly increased clearance of Noggin from the cell layer when compared with control cells (Fig. 7, triangles) not expressing enzyme. Cells expressing MutQSulf1 (Fig. 1, circles) in which the catalytic site of the enzyme has been mutated revealed identical pulse-chase kinetics as control cells. This altered kinetics of Noggin release results in dramatically altered steady-state levels of Noggin on the cell surface, as detected by immunohistochemistry of fixed, non-permeabilized cells.
As we have described previously, CHO cells expressing Noggin display abundant punctate localization of this protein at FIG. 3. Differential activity of chemically modified heparins in the displacement of Noggin from the cell surface. CHOK1 cells stably expressing Noggin were labeled with Trans 35 S-Label in methionine-and cysteine-free media and chased in cold media containing various chemically modified heparins, each at 1 g/ml. Heparin that had been fully desulfated (᭛) was unable to displace cellsurface Noggin, demonstrating the importance of sulfation to binding. Both N-desulfated (µ) and N-desulfated reacetylated (crossed diamonds) heparins were also inactive, demonstrating the N-sulfation is essential for binding. Heparins that were 2-O desulfated (E) and 6-O desulfated (‚) showed reduced activity, demonstrating a relative requirement for these sites for binding. the cell surface, where it is bound by means of heparan sulfate (Fig. 8A). In contrast, cells simultaneously expressing Qsulf1 show a dramatic reduction in cell-surface Noggin (Fig. 8G). A similar reduction in level of cell-surface Noggin was detected (Fig. 8E) even with lower levels of Qsulf1 expression (Fig. 8K). Only at extremely low levels of Qsulf1 expression (Fig. 8K) did cell-surface staining return to near-control levels. As in the case of the pulse-chase analysis, loss of cell-surface localization depended upon intact enzyme activity, as shown by co-expression of mutated Qsulf1 (Fig. 8I).
Release of Cell-surface Noggin by Qsulf1 Activity Results in BMP4 Responsiveness-In our previous studies we have shown that Noggin, bound to the cell surface through an interaction with heparan sulfate, remains functional at that location in the binding of BMP4. In our present study we have shown that Qsulf1 catalytic activity results in the release of this Noggin from the plasma membrane. Therefore, we would predict that the activity of Qsulf1 functions to release the localized BMP inhibitory activity of Noggin near the plasma membrane, thus resulting in an increased accessibility of BMP to its signaling receptor at the plasma membrane and an increased BMP responsiveness of the cell.
To access this, QS3-NG and QMut-NG cells (Fig. 8) were incubated in serum-free medium and treated with BMP4 in culture. After stimulation with BMP4, cells were permeabilized and their responsiveness was accessed by the detection of phos-phoSMAD 1,5,8 using immunofluorescence. QMut-NG cells expressing the catalytic site mutant of Qsulf1 have, as shown above, abundant plasma membrane localization of Noggin; consistent with this, they fail to respond to exogenous BMP4 as indicated by the lack of SMAD phosphorylation (Fig. 9A). In contrast, expression of active Qsulf1 results as shown above in the near-quantitative release of cell-surface Noggin. This results in a release of membrane-associated BMP inhibitory activity that leads to the restoration of BMP4 responsiveness of these cells, as can be seen by the detection of phosphorylated SMAD in the nucleus (Fig. 9C). DISCUSSION Patterning events during development are determined by both the short-and long-range action of signaling molecules that regulate cell fate. One important class of these signaling molecules is the family of BMPs (16,17), which specify cell fate in response to the local level of BMP activity in the immediate environment of the cell (18 -20). Secreted proteins, such as Noggin, act as antagonists of BMP function by binding BMPs and preventing their interactions with BMP receptors at the cell surface (21,22). Thus antagonists like Noggin can function to regulate patterning events by helping to shape activity gradients of BMPs in vivo. We have reported previously that Noggin release and diffusion from the cell surface depends upon heparan sulfate (7). Noggin binds tightly to the cell surface through heparan sulfate proteoglycans and can be directly internalized and degraded at the plasma membrane without significant diffusion. Therefore, we have proposed that heparan sulfate proteoglycans might regulate BMP-dependent patterning events by controlling the range and duration of action of Noggin in vivo (7).
Our current findings highlight the fact that the binding of Noggin to the cell surface is specific to heparan sulfate and that there are certain structural characteristics essential to the Noggin binding site. The minimal Noggin binding site seems to accommodate a length of 10 monosaccharides and has an essential requirement for N-sulfate groups, as well as a relative requirement for 6-O-and 2-O-sulfate, suggesting that Noggin binding sites contain tri-sulfated disaccharides. The identification that Noggin binding to heparan sulfate has specific structural requirements supports the hypothesis that Noggin, like many other heparin-binding proteins, may bind selectively in vivo to relatively specific sequences of heparan sulfate. Although heparan sulfate is ubiquitous in nature, its specific structure is extremely heterogeneous. There is mounting evidence that the structure of heparan sulfate chains found in vivo may be both spatially and temporally regulated to impart tissue-specific preferences for the functional binding of particular ligands (23)(24)(25). How this structural diversity is generated is complex. There are multiple isozymes for many of the key enzymes in the biosynthesis of heparan sulfate. The developmentally regulated expression of these different isozymes is likely to contribute to some of the diversity found within heparan sulfate structures in vivo. In addition, it has been recently appreciated that there are membrane-associated sulfatases capable of editing heparan sulfate sequences in vivo and apparently, therefore, modifying heparan sulfate functions. One such enzyme is Qsulf1 (8 -10).
Qsulf1 was originally cloned as a Sonic hedgehog-responsive gene (Shh) activated during somite formation and was recognized to have homology to the lysosomal N-acetylglucosamine sulfatases and therefore speculated to be a heparan sulfate 6-O-sulfatase (8), an activity recently confirmed (9,10). Blockage of Qsulf1 function in vivo through the use of antisense phosphorothiolated oligonucleotides resulted specifically in the inhibition of Myf 5 expression, a Wnt-induced gene, in the epaxial somite muscle progenitor cells (8). Qsulf1 has been shown in vitro to directly regulate Wnt1 signaling, apparently by modification of heparan sulfate chains to a lower affinity binding state for Wnt1 (9). This is speculated to FIG. 7. Catalytic activity of Qsulf1 is associated with more rapid release of Noggin from the cell surface. CHOK1 cells stably expressing Noggin, as well as either active Qsulf1 or the inactive catalytic site mutant MutQSulf1 were labeled with Trans 35 S-Label in methionine-and cysteine-free media and chased in cold media. Cells expressing Qsulf1 (छ) had a significantly reduced half-life of Noggin at the cell surface, whereas cells expressing MutQsulf1 (E) were indistinguishable from control (‚). result in the release of Wnt1 and its increased availability for signaling.
Our results suggest that Qsulf1 might have a more complex relationship to the control of patterning events during development. Qsulf1 expression in the medial somite is likely to have important in vivo functions in the regulation of Noggin activity as well. Noggin is expressed in the dorsomedial lip of the developing somite. Antagonism of BMP4, arising from the lateral plate mesoderm, is thought to be a function of Noggin, which contributes with Wnt to the specification of epiaxial musculature in the medial somite (26 -28). Coincident expression of Qsulf1 in the medial somite would be expected, based upon our in vitro data, also to result in reduced binding of Noggin to heparan sulfate proteoglycans on the surface of medial somitic tissue. This would be expected to lead to more effective diffusion of Noggin within this tissue, whereas diffusion of Noggin into the adjacent lateral somite would likely be restricted by heparan sulfate on the surface of these cells with higher affinity binding, thus helping to establish a limit on Noggin action and defining the boundary between the lateral and medial somite. Qsulf1 could play a similar role in the release of Noggin from the surface of cells within the notochord, where both of these proteins are also co-expressed and, therefore, have other functions with respect to BMP antagonism in neural patterning.
The discovery that the catalytic activity of Qsulf1 modifies the cell-surface binding of components of at least two developmentally significant signaling pathways suggests that this has evolved as a general mechanism to regulate cellular responsiveness to perhaps many heparin-binding growth factors. Two proteins related to Qsulf1 have been identified in humans: Hsulf-1, which seems to be the ortholog of Qsulf-1, and Hsulf-2. Each had a distinct pattern of expression in adult tissues (10). These enzymes have very similar activities on heparin substrates, but it remains to be determined whether their actions on heparan sulfate can be distinguished in ways that might reflect functional divergence.
Our analysis of the composition of heparan sulfate chains after exposure to Qsulf1 has revealed a detail previously not fully appreciated. The activity of Qsulf1 seems to be almost exclusively targeted toward the highly sulfated S domains of the heparan sulfate chain, while leaving the NA/NS domains virtually unchanged. This is clear from disaccharide analyses of Qsulf1 transfectants which have a marked reduction (ϳ80%) in the level of the trisulphated disaccharide HexA,2S-GlcNS,6S, which is found almost exclusively in the S domains (11). However, we consistently found that there was always a small proportion of this disaccharide that resisted Qsulf1 actions. The 6-O-sulfate in the NA/NS domains, where the HexA-GlcNAc,6S unit is present together with HexA-GlcNS,6S (29,30), were not affected by Qsulf1. Taking into consideration the sequence and sulfation patterns of the S domains in HS, these findings suggest that Qsulf1 recognizes the 6-O-sulfate groups in regions where the substituted GlcNS residue is flanked by iduronate-2-sulfate (i.e. IdoA,2S-GlcNS,6S-IdoA,2S). On the other hand, 6-O-sulfated sequences in the NA/NS domains, which take the form of GlcA/IdoA-GlcNAcϮ6S-GlcA-GlcNSϮ6S are unaffected (Figs. 4 and 6). Thus, Qsulf1 has an "editing" effect on the 6-sulfation pattern of heparan sulfate; this targeted Qsulf1-mediated 6-O-desulfation of HS clearly impairs Noggin and Wnt/Wg binding, indicating that these proteins interact with the S domain in HS. Moreover, we can also predict that it will suppress signaling induced by FGF1 and FGF2, which also requires S domain 6-O-sulfation (31,32). It remains to be seen whether these two growth factors are coexpressed with Qsulf1 in developing tissues. In summary, we have shown that Noggin binding to heparan sulfate depends upon 6-sulfate groups being present in the S domains. Qsulf1 impairs Noggin binding by specifically removing these sulfate groups, without affecting 6-O-sulfates in the NA/NS domains. Thus, Qsulf1 subtly changes the sulfation pattern of heparan sulfate on the cell surface, "tuning" its bioactivities to its key functions in the embryo of regulating the actions of growth factors and morphogens at different stages of development. Further molecular and genetic studies will be helpful in ascertaining the full significance of Qsulf1 to biological processes in vivo.