Sulf Loss Influences N-, 2-O-, and 6-O-Sulfation of Multiple Heparan Sulfate Proteoglycans and Modulates Fibroblast Growth Factor Signaling*

Sulf1 and Sulf2 are two heparan sulfate 6-O-endosulfatases that regulate the activity of multiple growth factors, such as fibroblast growth factor and Wnt, and are essential for mammalian development and survival. In this study, the mammalian Sulfs were functionally characterized using overexpressing cell lines, in vitro enzyme assays, and in vivo Sulf knock-out cell models. Analysis of subcellular Sulf localization revealed significant differences in enzyme secretion and detergent solubility between the human isoforms and their previously characterized quail orthologs. Further, the activity of the Sulfs toward their native heparan sulfate substrates was determined in vitro, demonstrating restricted specificity for S-domain-associated 6S disaccharides and an inability to modify transition zone-associated UA-GlcNAc(6S). Analysis of heparan sulfate composition from different cell surface, shed, glycosylphosphatidylinositol-anchored and extracellular matrix proteoglycan fractions of Sulf knock-out cell lines established differential effects of Sulf1 and/or Sulf2 loss on nonsubstrate N-, 2-O-, and 6-O-sulfate groups. These findings indicate a dynamic influence of Sulf deficiency on the HS biosynthetic machinery. Real time PCR analysis substantiated differential expression of the Hs2st and Hs6st heparan sulfate sulfotransferase enzymes in the Sulf knock-out cell lines. Functionally, the changes in heparan sulfate sulfation resulting from Sulf loss were shown to elicit significant effects on fibroblast growth factor signaling. Taken together, this study implicates that the Sulfs are involved in a potential cellular feed-back mechanism, in which they edit the sulfation of multiple heparan sulfate proteoglycans, thereby regulating cellular signaling and modulating the expression of heparan sulfate biosynthetic enzymes.

Sulf1 and Sulf2 are two heparan sulfate 6-O-endosulfatases that regulate the activity of multiple growth factors, such as fibroblast growth factor and Wnt, and are essential for mammalian development and survival. In this study, the mammalian Sulfs were functionally characterized using overexpressing cell lines, in vitro enzyme assays, and in vivo Sulf knock-out cell models. Analysis of subcellular Sulf localization revealed significant differences in enzyme secretion and detergent solubility between the human isoforms and their previously characterized quail orthologs. Further, the activity of the Sulfs toward their native heparan sulfate substrates was determined in vitro, demonstrating restricted specificity for S-domain-associated 6S disaccharides and an inability to modify transition zone-associated UA-GlcNAc(6S). Analysis of heparan sulfate composition from different cell surface, shed, glycosylphosphatidylinositol-anchored and extracellular matrix proteoglycan fractions of Sulf knock-out cell lines established differential effects of Sulf1 and/or Sulf2 loss on nonsubstrate N-, 2-O-, and 6-O-sulfate groups. These findings indicate a dynamic influence of Sulf deficiency on the HS biosynthetic machinery. Real time PCR analysis substantiated differential expression of the Hs2st and Hs6st heparan sulfate sulfotransferase enzymes in the Sulf knock-out cell lines. Functionally, the changes in heparan sulfate sulfation resulting from Sulf loss were shown to elicit significant effects on fibroblast growth factor signaling. Taken together, this study implicates that the Sulfs are involved in a potential cellular feed-back mechanism, in which they edit the sulfation of multiple heparan sulfate proteoglycans, thereby regulating cellular signaling and modulating the expression of heparan sulfate biosynthetic enzymes.
Heparan sulfate proteoglycans (HSPGs) 2 are essential regulators of cell signaling and development that are ubiquitously present on the cell surface and in the extracellular matrix (ECM) of virtually all animal cells (1). Composed of a dynamically sulfated heparan sulfate (HS) polymer and a protein backbone, HSPGs can be divided into three functionally distinct families: the transmembrane syndecans, the glycosylphosphatidylinositol (GPI)-anchored glypicans, and the ECM-associated proteoglycans perlecan, agrin, and collagen XVIII (2). A major function of the HS component of HSPGs is to act as a regulatory cofactor for a variety of signaling molecules and morphogens (3)(4)(5)(6). In addition to the HS chains, the protein backbone of HSPGs also plays a critical role in determining proteoglycan function, regulating processes such as cellular adhesion, HSPG shedding, and endocytosis (1,7,8).
HSPG biosynthesis is a non-template-driven process, relying on multiple enzyme activities to generate a glycosaminoglycan polymer with distinct sulfation patterns attached to a protein core. Synthesis begins with the attachment of a tetrasaccharide linker sequence onto a target serine residue, followed by polymerization of alternating GlcA and GlcNAc residues up to 200 disaccharides in length. Once synthesized, the diversity of HS structure is generated by a variety of heparan-modifying enzymes in the Golgi, which epimerize a portion of the GlcA residues to iduronic acid and add sulfate to some of the 2-Opositions of the hexuronic acid (UA) and N-and 6-O-positions of GlcNAc (9). Importantly, the cumulative action of these biosynthetic enzymes is incomplete, generating a defined HS domain structure composed of highly (S-domains), partially (transition zones), and nonsulfated regions (10). These patterns of sulfation are cell type-and developmental stage-specific (11), serving as dynamic templates to promote or inhibit specific cellular interactions and signaling events (12,13).
In recent years, the activity of two additional HS-modifying enzymes has been described, Sulf1 and Sulf2 (14,15). These heparan sulfate 6-O-endosulfatases are unique in their ability to postsynthetically edit 6-O-sulfation patterns at the cell surface. * This work was supported by the Deutsche Forschungsgemeinschaft, the Fonds der Chemischen Industrie, and Shire Human Genetic Therapies Inc. (Cambridge, MA). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental Tables S1-S6 and Figs. S1 and S2. 1 To whom correspondence should be addressed: Fakultät fü r Chemie, Biochemie I, Universität Bielefeld, Universitätsstr. 25 (16 -20). In order to understand the biological significance of these enzymes during mammalian development, knock-out mouse models were generated and subsequently characterized. Although loss of Sulf1 or Sulf2 was shown to result in relatively mild phenotypes variably affecting lung development as well as growth and viability, the loss of both Sulfs resulted in severe developmental ablations and early post-natal lethality (21)(22)(23)(24)(25). These phenotypic observations, in conjunction with previous analyses of HS material from Sulf knock-out mouse embryonic fibroblasts (MEFs), support the idea that the Sulfs act cooperatively in vivo to modify HS sulfation patterns and regulate developmental processes (26). Despite the biological importance of the Sulfs, much remains to be learned about the endogenous activity and HSPG substrate specificity of these enzymes. Indeed, primary Sulf knockout MEFs have been shown to exhibit changes in multiple sulfate groups not accounted for by in vitro Sulf activity assays employing heparin substrate analogs (15,26,27). Further, inconsistent results regarding the ability of the Sulfs to modify HSPG substrates within the ECM have been reported (28,29). To address these issues, in vitro activity analysis using native HS substrates was carried out in conjunction with the characterization of cell surface, GPI-anchored, shed, and ECM-associated HSPGs from immortalized WT and Sulf knock-out cell lines. These studies demonstrate a highly restricted 6-O-sulfate substrate specificity for the Sulfs in vitro, which was contrasted by dynamic effects of Sulf loss on N-, 2-O-, and 6-O-sulfated moieties in vivo. This comparative analysis implicates that, in addition to their 6-O-endosulfatase activity, Sulf1 and Sulf2 are able to dynamically modulate the abundance of nonsubstrate N-, 2-O-, and 6-O-sulfate groups on proteoglycans throughout the cell surface and the ECM by influencing the HS biosynthetic machinery. In support of this novel observation, real time PCR expression analysis was able to verify a significant impact of Sulf loss on the abundance of HS sulfotransferase transcripts. Finally, the modulation of HS sulfation patterns resulting from Sulf deficiency observed in this study was shown to elicit significant effects on FGF signaling, underlining the importance of these enzymes in regulating cellular response.
Cell Surface and Shed Proteoglycan Fractionation-Conditioned medium was collected after 48 h, cleared of cellular debris via centrifugation, and pooled as the shed proteoglycan fraction. Cells were subsequently washed twice in PBS and treated for 10 min with trypsin/EDTA solution (Invitrogen). Cells were pelleted and washed twice in HEPES, 0.9% NaCl, 0.5 mM EDTA, and supernatant samples were pooled together as the cell surface proteoglycan fraction.
GPI-anchored Proteoglycan Fractionation-The protocol for GPI-anchored glypican enrichment and purification was based on the protocol of Tumova et al. (32). Briefly, cells were washed three times with PBS and incubated for 45 min at 37°C in serum-free DMEM containing 0.067 units/ml phosphatidylinositol-phospholipase C (PI-PLC). Supernatant was collected, and cells were washed twice with 10 mM HEPES, 0.9% NaCl, 0.5 mM EDTA. These wash fractions were pooled with the supernatant as the glypican proteoglycan fraction.
ECM Proteoglycan Fractionation-MEF cell lines were plated onto 15-cm sterile glass Petri dishes coated with poly-L-lysine (Sigma) and 1 g/ml laminin (Sigma). Lipid extraction was carried out based on previous experiments by Heremans et al. (33). Briefly, cells were washed twice with ice-cold PBS, washed three times for 5 min with 0.5% Triton X-100 in 10 mM Tris/HCl, 150 mM NaCl, 10 mM EDTA, pH 8.0, and washed three times for 10 min with 0.5% sodium deoxycholate in 10 mM Tris, 150 mM NaCl, pH 8.0. The matrix that remained attached to the glass plates was stringently washed three times with 0.2 M Tris, 2 M NaCl, pH 8.0, before finally being washed three times with 0.2 M Tris, pH 8.0. ECM was harvested in PBS containing collagenase II (10 units/ml) (PAA Laboratories, Pasching, Austria) and incubated overnight at 37°C.
Disaccharide Composition Analysis-Purified HS chains were digested with a combination of heparinases I, II, and III, and the resultant disaccharides were identified following strong anion exchange-HPLC as previously described (31).
Cell Surface Proteoglycan Profiling-Cells were grown to 100% confluence on a 10-cm plate and were washed twice with ice-cold PBS, harvested, pelleted, and resuspended in 50 mM HEPES, pH 8.0. Cells were homogenized manually with a Dounce homogenizer. Cellular debris was cleared by centrifuging at 500 ϫ g for 5 min at 4°C, and the supernatant was centrifuged at 100,000 ϫ g for 1 h at 4°C to pellet the membrane fraction. Cell membranes were resuspended in 50 mM HEPES containing 0.5% Triton X-100, pH 8.0, and incubated on ice for 30 min. The sample was centrifuged at 100,000 ϫ g for 30 min, and the Triton-soluble supernatant was collected, diluted to 0.1% Triton with 50 mM sodium acetate, 0.5 mM calcium acetate, pH 7.0, and digested for 2 h at 37°C with 2 mIU heparinase II and 2 mIU heparinase III.
GPI-anchored Proteoglycan Profiling-Cells were grown to 100% confluence on a 10-cm plate, and GPI-anchored glypican was extracted as described above. Following PI-PLC treatment, conditioned medium was bound to a 1-ml HiTrap DEAE FF column (GE Healthcare), washed with 20 ml of PBS, and eluted with PBS containing 1.5 M NaCl. Protein-containing fractions were pooled, dialyzed against 50 mM sodium acetate, 0.5 mM calcium acetate, pH 7.0, and digested for 2 h at 37°C with 2 mIU heparinase II and 2 mIU heparinase III.
Shed Proteoglycan Profiling-Cells were grown to 100% confluence on a 10-cm plate, washed twice with PBS, and incubated for 24 h in serum-free DMEM. Conditioned medium was bound to a 1-ml HiTrap DEAE FF column (GE Healthcare), washed with 50 ml of PBS, and eluted with PBS containing 1.5 M NaCl. Samples were dialyzed and digested with heparinases II and III, as described above.
ECM Proteoglycan Profiling-Confluent cells plated on glass coverslips in a 24-well dish were treated for 2 h with 2 mIU heparinase II and 2 mIU heparinase III in serum-free DMEM followed by ECM extraction, as described above. ECM composition was analyzed via immunofluorescence microscopy according to a standard fixation and staining protocol as previously described (34). Alternatively, the ECM was scraped from the glass coverslips for Western blotting.
Cloning of Human Sulf1 and Sulf2 cDNAs and Construction of Expression Vectors-A 5Ј-truncated Sulf2 cDNA, designated KIAA1247 (35), was obtained from the Kazusa Institute (Kisarazu, Chiba, Japan). Using primer 1247_His, a RGS-His 6 tag-encoding sequence, followed by a HindIII site, was added to the 3Ј-end of the coding region by PCR (for PCR primers, see Table S1). The product was subcloned as a 1.8-kb EcoRI/Hin-dIII fragment into vector pMPSVEH (36). The lacking 5Ј region of the Sulf2 cDNA was obtained through rapid amplification of cDNA ends-PCR using the GeneRacer kit (Invitrogen). As gene-specific noncoding primers, 1247_SP3 and, for nested PCR, 1247_SP4 were used. From the obtained 1.0-kb product (Sulf2-N), the 5Ј-untranslated region was deleted by adding a 5Ј-NcoI site at the position of the start codon (primer 1247_NcoI), which allowed fusion with the optimized Kozak sequence of a pMPSVEH-based arylsulfatase A expression vector described earlier (37). Full-length Sulf2-RGS-His 6 -encoding cDNA was obtained by joining the 5Ј and 3Ј fragments via the internal EcoRI site, which was amplified also with the rapid amplification of cDNA ends product. For construction of the pCI-neo-based expression vector, the 3Ј Sulf2-RGS-His 6 fragment was cloned first as an EcoRI/HindIII fragment (blunted at the HindIII end) into pCI-neo (Promega), which had been opened by EcoRI and SmaI, and then assembled with the 5Ј EcoRI/EcoRI fragment excised from the pMPSV-Kozak-Sulf2-N construct.
Cloning of Sulf1 cDNA also started from a 5Ј-truncated EST fragment (KIAA1077), which was obtained from the Kazusa Institute (38). Using primer 1077_His, an RGS-His 6 tag encoding sequence was added to the 3Ј-end of the coding region, followed by an MscI site. The obtained 2.3-kb product (Sulf1-C) was subcloned as an AccI/MscI fragment. The lacking 5Ј region of Sulf1 cDNA was obtained through rapid amplification of cDNA ends-PCR using the gene-specific noncoding primer 1077_SP3. From the obtained 0.4-kb Sulf1-N product, the 5Ј-untranslated region was deleted by adding a 5Ј NcoI site at the position of the start codon (primer 1077_NcoI), which allowed generation of an optimized Kozak sequence (as above). Full-length Sulf1-RGS-His 6 -encoding cDNA was obtained by joining 5Ј and 3Ј fragments via the internal AccI site. For that purpose, Kozak-Sulf1-N was cloned as an EcoRI/AccI fragment into the MCS of pCI-neo and then assembled with Sulf1-C as an AccI/AccI fragment. Full-length sequencing resulted in the same coding sequences for both Sulf1 and Sulf2 as published by Morimoto-Tomita et al. (15). To generate plasmids with Sulf cDNAs encoding enzymatically inactive Sulf1-C87A/C88A or Sulf2-C88A/C89A mutants, the QuikChange method (Stratagene) was applied, using complementary mutagenesis primers Sulf1CA or Sulf2CA (each forward and reverse), respectively (for PCR primers see Table S2).
Expression and Localization Analysis of Human Sulfs in HT1080 Cells-Human fibrosarcoma HT1080 cells were transfected with pCI-neo plasmids containing wild type or mutant Sulf cDNAs using magnet-assisted transfection according to the manufacturer's instructions (IBA, Göttingen, Germany). Stable clones were selected with 800 g/ml G-418 sulfate (PAA Laboratories), and drug-resistant cells were cloned and expanded. Sulf expression in total lysates was analyzed by lysing cells with gentle sonification in 10 mM HEPES, 0.5 M NaCl, pH 7.4, centrifugation at 100,000 ϫ g for 30 min at 4°C, and Western blotting using anti-RGS-His 6 antibodies. Detergent solubility was tested by lysing cells manually with a cell Dounce homogenizer in 50 mM HEPES, pH 8.0, centrifuging 500 ϫ g for 5 min at 4°C to remove cellular debris and subsequently centrifuging the supernatant at 100,000 ϫ g for 1 h at 4°C to isolate the membrane fraction. Membranes were treated with 1 or 2.5% Triton X-100 or Brij98 on ice for 30 min prior to centrifugation at 100,000 ϫ g for 1 h at 4°C to separate detergentsoluble and detergent-insoluble fractions. For detection of secreted Sulfs in Western blots, confluent Sulf-expressing cell lines were cultivated for 72 h in serum-free DMEM prior to 100ϫ concentration of conditioned medium using Centricon-30 units (Millipore), as previously described (15). Immunofluorescence analysis of permeabilized HT1080 cells expressing Sulf1 or Sulf2 was performed as previously described (34) but without LysoTracker treatment. Alternatively, live cells were stained to specifically label cell surfacelocalized proteins. Sulf-expressing HT1080 cells were incubated for 1 h at 4°C with anti-RGS-His 6 antibodies in serum-free DMEM, washed with PBS, and fixed with 4% paraformaldehyde. After blocking with 2% fetal calf serum, the cells were labeled with an Alexa-488-conjugated goat antimouse IgG secondary antibody (Molecular Probes) and imaged as previously described (34).
In Vitro Sulf Activity Analysis-For in vitro analysis of Sulf activity from cell lysates, Sulf-expressing cells were harvested, pelleted, and lysed with gentle sonification in 20 mM Tris, 0.5 M NaCl, 10 mM imidazole, pH 7.4. Cleared cell lysate was added to 100 l of nickel-Sepharose and incubated at 4°C overnight. Nickel beads were washed three times with 20 mM Tris, 0.5 M NaCl, 40 mM imidazole, pH 7.4, and finally twice in 50 mM Tris, 10 mM MgCl 2 , pH 7.4, activity buffer. Nickel-bound Sulfs were incubated overnight at 37°C in activity buffer containing 200,000 3 H cpm of purified HS isolated from Sulf1/2 double knock-out MEFs. For analysis of secreted Sulf activity, cells were cultivated in DMEM with 10% fetal calf serum for 72 h prior to (NH 4 ) 2 SO 4 precipitation of conditioned medium. Medium precipitate was resuspended and dialyzed against activity buffer. The sample volume of the dialyzed material was reduced to 10% of the starting volume using a SpeedVac and incubated overnight at 37°C with 200,000 3 H cpm of purified HS as above. HS was rebound on 500 l of DEAE resin and washed with 20 ml of PBS to remove excess proteins prior to elution with 1.5 M NaCl. HS samples were desalted on PD10 columns (GE Healthcare), and disaccharide compositions were analyzed as previously described (31).
Real Time PCR-Total RNA was extracted from MEF cell lines using the RNeasy minikit (Qiagen). Synthesis of firststrand cDNA from mRNA transcripts was performed using the iScript cDNA synthesis kit (Bio-Rad). Real time PCR was carried out using the LightCycler FastStart DNA Master (Plus) SYBR Green I Kit and the LightCycler instrument (both from Roche Applied Science). The conditions for denaturation, annealing, and extension were repeated 45 times as follows: denaturation at 95°C for 10 s, annealing at 61°C for 5 s, and extension at 72°C for 10 s. All reactions were performed in triplicates. Real time PCR primers are listed in Table S3. Primer specificities were analyzed by agarose gel electrophoresis to verify amplification of single products of the expected sizes and by melting curve analysis. cDNA dilution series were performed to calculate PCR primer efficiencies. Efficiencies were 90 -100% for all primer sets used. Relative quantification of transcripts was carried out according to Pfaffl (39) with data normalized to the housekeeping gene Rpl13a.
FACS Analysis-Cells were grown to 100% confluence, washed twice with PBS, and incubated for 2 h in serum-free DMEM containing 1 mIU heparinase II and 1 mIU heparinase III. Cells were washed twice in ice-cold PBS, harvested with a rubber policeman, and transferred to FACS tubes. Cells were labeled with the 3G10 antibody (1:200) in DMEM plus 1% fetal calf serum at 4°C, washed twice in ice-cold PBS, and labeled with goat anti-mouse Alexa-488-conjugated secondary antibodies in DMEM plus 1% fetal calf serum. Cells were washed in ice-cold PBS, resuspended in PBS plus 1 mM EDTA, and analyzed using a FACSCalibur TM system (BD Biosciences).
Growth Factor Signaling-Cells were grown to 100% confluence, starved overnight in serum-free DMEM, and induced with no FGF2 (control) or 1 ng/ml FGF2 for 10 min at 25°C prior to lysis in 10 mM HEPES, 1% Triton X-100, and protease inhibitor mixture (1:100) (Sigma) by gentle sonication. Lysates were collected and cleared via centrifugation at 20,000 ϫ g for 30 min at 4°C. Equal protein amounts (Coomassie Plus Bradford assay; Pierce) were analyzed for diphosphosphorylated and nonphosphorylated ERK1/2 signal intensity on Western blot, as previously described (31). Signals were quantified using the AIDA 4.06 software package (Raytest, Straubenhardt, Germany).

Subcellular Localization and Secretion of Human Sulf1 and
Sulf2-To determine the subcellular distribution of the mammalian Sulfs, human Sulf1 and Sulf2 enzymes were expressed in human HT1080 fibrosarcoma cells. Live cell immunolabeling of nonpermeabilized cells revealed a strict cell surface staining (Fig. 1, A and B). Immunofluorescence analysis of permeabilized cells also exhibited a strong cell membrane staining pattern with additional staining localized to the ER (Fig. 1C), as In Vivo and in Vitro Functional Characterization of Sulf1/2 OCTOBER 10, 2008 • VOLUME 283 • NUMBER 41 verified by co-localization with the ER marker protein-disulfide isomerase (data not shown). Similar results as shown for Sulf1 in Fig. 1, A-D, were observed for Sulf2 (data not shown). The membrane localization of the Sulfs was further substantiated by differential centrifugation of cell lysates and Western blot detection of the Sulfs in the total membrane fraction (Fig.  1E). In line with previous reports, association of the Sulfs with cell membranes was sensitive to 0.5 M NaCl (data not shown), indicating a strictly ionic interaction with the cell surface. To further define the cell membrane association of the Sulfs, membrane fractions were treated with 1 or 2.5% Triton X-100 or Brij98 detergent in salt-free HEPES buffer. Interestingly, the Sulfs could not be extracted by this treatment, even at high detergent concentrations, indicating a firm association with detergent-insoluble membrane domains (Fig. 1E).
Analysis of conditioned medium from Sulf-expressing cell lines revealed a significant degree of secretion (Fig. 1E). Interestingly, although little proteolytic processing was observed in cellular fractions of Sulf-expressing cells, proteolytic processing was observed for secreted Sulf2 and, to a very low extent, Sulf1. Indeed, detection of the secreted fraction using antibodies against the C-terminal RGS-His 6 tag revealed bands at ϳ45 and 40 kDa in addition to the full-length 130 kDa band (Fig. 1E). Using antibodies directed against native Sulf1 or Sulf2, a corresponding N-terminal 75-kDa fragment could be detected in the secreted fraction in addition to the full-length protein (data not shown).
In Vitro Sulf Activity toward Native HS Substrate-To determine the in vitro Sulf activity toward native HS substrate, Sulf1 or Sulf2 enriched from the cell lysates or the secreted fractions of Sulf-expressing cell lines was incubated with purified radiolabeled HS from Sulf1/2 double knock-out MEFs. Fig. 2 displays the results from these experiments showing the relative amounts of the three potential 6-O-sulfated disaccharide substrates, UA-GlcNAc(6S), UA-GlcNS(6S), and UA(2S)-GlcNS(6S), before and after Sulf treatment. Incubation of HS with either Sulf1 or Sulf2 from cell lysates ( Fig. 2A) resulted in a ϳ20% decrease in UA-GlcNS(6S) and a ϳ65% decrease in UA(2S)-GlcNS(6S), whereas no activity toward the monosulfated UA-GlcNAc(6S) disaccharide could be detected. The mutant Sulf negative controls Sulf1-C87A/C88A and Sulf2-C88A/C89A were observed to be completely inactive. Similar results were obtained for secreted Sulfs (Fig. 2B) showing a ϳ25% reduction in UA-GlcNS(6S) and a ϳ90% reduction in UA(2S)-GlcNS(6S). Again, although a higher activity was observed for the secreted Sulfs, no activity toward the UA-GlcNAc(6S) disaccharide was detected. Endogenous secreted Sulf activity could be detected in the conditioned medium of untransfected WT MEFs, which also showed preferential substrate specificity toward the tri-sulfated UA(2S)-GlcNS(6S) disaccharide (Fig. 2B). As an endogenous Sulf-deficient control, conditioned medium from Sulf1/2 double knock-out MEFs was tested for secreted Sulf activity and was found to be completely inactive (Fig. 2B). It should be noted that the radiolabel-based analytical procedure implemented in this study to quantify HS disaccharides is not sensitive enough to detect rare disaccharide units such as 3O-sulfated 6S disaccharides or UA(2S)-GlcNAc(6S). Thus, the possibility cannot be excluded that these uncommon 6S disaccharides serve as potential substrates for the Sulfs. Importantly, no Sulf activity toward N-or 2-O-sulfate moieties was detected in either cellular or secreted Sulf fractions (Tables S4 and S5), underlining the high specificity of these enzymes for the indicated 6-O-sulfate groups.
HSPG Analysis of Sulf Knock-out Cells-To characterize the proteoglycan-specific effects of the Sulfs in vivo, cell surface, GPI-anchored, shed, and ECM HSPG pools were extracted from immortalized WT and Sulf knock-out MEFs for HS analysis. In order to determine the proteoglycan composition in each of these different fractions, total HSPGs present in each sample were analyzed via Western blotting using the 3G10 monoclonal antibody. Following heparan lyase treatment, the 3G10 antibody enables visualization of all HSPGs present in a sample via recognition of the ⌬4 -5-UA epitope, an oxidized FIGURE 2. In vitro HS activity assays with human Sulf1 or Sulf2 recovered from cell lysates or conditioned medium. Radiolabeled HS purified from Sulf1/2 double knock-out cells was used as a substrate for the following in vitro assays. A, incubation with nickel-bound Sulf1, Sulf2, or inactive Sulf mutants (mixture of Sulf1-C87A/C88A and Sulf2-C88A/C89A) from the cell lysates of overexpressing cell lines; B, incubation with concentrated conditioned medium from either Sulf1/2 double knock-out MEFs, WT MEFs, or HT1080 Sulf1-or HT1080 Sulf2-overexpressing cell lines, as indicated. The bar graphs show the relative amounts of 6-O-sulfated disaccharides, as determined by disaccharide composition analysis (see "Experimental Procedures"). Error bars, S.D. from three independent heparinase digests of the treated or untreated HS substrate. Data significance was calculated using the two-tailed t test assuming equal variance between samples. *, p Ͻ 0.05; **, p Ͻ 0.01.
form of uronic acid that remains attached to the proteoglycan core (40).
The results from this proteoglycan profiling show distinct HSPG compositions in each of the four fractions (Fig. 3A). The cell surface fraction is composed of five prominent bands representing HSPGs associated with the cell membrane (Fig. 3A). The 65 and 30 kDa bands correspond in size with the full-length and proteolytically processed GPI-anchored proteoglycan glypican (41), whereas bands at ϳ45, 40, and 35 kDa probably represent transmembrane proteoglycans, such as syndecans.
No differences in the cell surface profiles between WT and Sulf knock-out cells were observed, indicating that Sulf loss has no major effects on HSPG expression or processing (data not shown). The GPI-anchored glypican fraction was extracted by treating cells with PI-PLC. As seen for the glypican fraction in Fig. 3A, treatment of cells with PI-PLC released the expected 65 and 30 kDa bands, which could be enriched from the conditioned medium. Interestingly, analysis of cell surface HSPG profiles following PI-PLC treatment showed no significant depletion of the 65 and 30 kDa bands, indicating that only a small fraction of cell surface glypican is sensitive to PI-PLC cleavage (Fig. 3A). Resistance of glypican to PI-PLC cleavage has previously been shown to correlate with acylation of the GPI anchor inositol moiety (42), indicating that the majority of glypicans at the cell surface of these MEFs are probably modified with an acyl chain. Profiling analysis of the shed fraction from WT and Sulf knock-out cells demonstrates that fulllength 65-kDa glypican is the primary HSPG that is constitutively released into the medium along with smaller amounts of high molecular weight HSPGs, most of which are probably secreted HSPG elements of the ECM (Fig. 3A). Again, no differences between WT and Sulf knock-out cell lines were observed, indicating that Sulf loss has no significant effect on constitutive HSPG shedding (Fig. 3A). Last, profiling of the ECM fraction showed HSPG bands larger than 250 kDa, a profile typical for the ECM, which probably corresponds to the high molecular weight ECM HSPGs perlecan and collagen XVIII (33) (Fig. 3A). The lack of low molecular weight cell surface or shed HSPGs in this fraction substantiates the stringency of the applied ECM extraction protocol. ECM composition and extraction efficiency were further verified via immunofluorescence microscopy, which showed complete removal of cell-associated protein-disulfide isomerase staining following lipid extraction, whereas a strong perlecan signal was retained (Fig.  3B). Finally, to determine whether Sulf loss has any quantitative effect on the HSPG population, relative HSPG amounts associated with WT and Sulf knock-out cell lines were determined via FACS, again using the 3G10 antibody. In support of profiling results, Sulf loss had no substantial effect on the overall amount of HSPG present at the cell surface (Fig. 3C), indicating that Sulf knock-out strictly affects HS sulfation and not the proteoglycan landscape of the cell.
Radiolabeled HS material from the above characterized cell surface, glypican, shed, and ECM proteoglycan fractions was isolated from the described MEF cell lines and subjected to detailed disaccharide analyses to determine the effects of Sulf loss on HS structure specifically for each HSPG fraction. Interestingly, characterization of HS from cell surface, shed, GPIanchored, and ECM HSPG fractions of WT and Sulf-deficient MEFs revealed that loss of Sulf1, Sulf2, or both can have different effects on 6-O-sulfated disaccharide composition (Fig. 4,  A-D). Indeed, analysis of cell surface proteoglycan 6-O-sulfation from Sulf1-deficient MEFs revealed a small but significant ϳ10% increase in the monosulfated 6S disaccharide UA-GlcNAc(6S) compared with the WT, whereas Sulf2 and Sulf1/2 double knock-out HS exhibited much larger ϳ50 and ϳ30% increases, respectively (Fig. 4A). Further, although similar increases in the di-and trisulfated 6S disaccharides FIGURE 3. Proteoglycan profiling of Sulf knock-out cells. WT, Sulf1 knockout (1 Ϫ/Ϫ ), Sulf2 knock-out (2 Ϫ/Ϫ ), and Sulf1/Sulf2 double knock-out (1 Ϫ/Ϫ 2 Ϫ/Ϫ ) cells were subjected to a proteoglycan fractionation protocol, as described under "Experimental Procedures." A, HSPGs present in the cell surface, glypican, shed, and ECM fractions were detected with the anti-⌬HS 3G10 monoclonal antibody, which recognizes the ⌬UA epitope generated by heparan lyase treatment of HSPGs. B, immunofluorescence analysis of WT MEFs using anti-protein-disulfide isomerase (␣-PDI; green) and anti-perlecan (red) antibodies before and after lipid extraction for ECM purification. C, FACS quantification of cell surface HSPGs for WT, Sulf1 and Sulf2 single, and Sulf1/2 double knock-out MEFs, again using the anti-⌬HS 3G10 antibody following heparan lyase treatment. Control cells represent MEFs not treated with heparan lyase enzymes but stained with primary and secondary antibodies.
UA-GlcNS(6S) and UA(2S)-GlcNS(6S) were observed for the Sulf1 or Sulf2 knock-outs in the cell surface proteoglycan fraction, the Sulf1/2 double knock-out exhibited an exceptional ϳ60 -80% increase in UA(2S)-GlcNS(6S) compared with either single knock-out, indicating cooperative turnover of the trisulfated 6S substrate (Fig. 4A). Analysis of the shed proteoglycan fraction in WT and Sulf-deficient MEFs revealed similar results as those observed in the cell surface fraction with the exception that no significant increase was observed for the mono-and disulfated UA-GlcNAc(6S) and UA-GlcNS(6S) disaccharides in the Sulf1 knock-out MEFs, compared with the WT (Fig. 4B). Differences in the effect of either Sulf1 or Sulf2 loss on 6-O-sulfation were especially prevalent in the glypican proteoglycan fraction (Fig. 4C). Indeed, analysis of the Sulf1 knock-out HS from this fraction revealed no significant increase in either mono-or di-sulfated 6S disaccharides, whereas the trisulfated 6S disaccharide, UA(2S)-GlcNS(6S), exhibited a ϳ90% increase compared with the WT. In contrast, the Sulf2 knock-out exhibited ϳ50% increases in all three mono-, di-, and trisulfated 6S disaccharides compared with the WT (Fig. 4C). Finally, analysis of WT and Sulf1/2 double knockout ECM proteoglycan fractions substantiated significant activity of the Sulfs within the extracellular matrix, revealing ϳ20 and ϳ100% increases in di-and trisulfated disaccharides compared with the WT, respectively (Fig. 4D). Interestingly, in contrast to cell surface and shed proteoglycan fractions in which Sulf1/2 double knock-out HS compositions were also analyzed, no significant increase in the monosulfated UA-GlcNAc(6S) disaccharide was observed in the ECM proteoglycan fraction.
In addition to differential effects of Sulf1, Sulf2, or Sulf1/2 loss on 6-O-sulfation patterns, dynamic changes in N-and 2-O-sulfation were also observed in the proteoglycan fractions derived from the Sulf knock-out cell lines. Fig. 5A shows the relative percent changes in N-, 2-O-, and 6-O-sulfation observed for the different Sulf knock-out cell lines in the cell surface proteoglycan fraction. It is interesting to note that although loss of Sulf1 had no significant effect on N-or 2-O-sulfation in this proteoglycan fraction, loss of Sulf2 resulted in ϳ10 and ϳ20% decreases in N-and 2-O-sulfate moieties, respectively, compared with the WT. Conversely, the Sulf1/2 double knock-out revealed ϳ10% increases in N-and 2-O-sulfate moieties compared with the WT (Fig. 5A). Importantly, the Sulf1, Sulf2, and Sulf1/2 knock-outs were shown to also elicit changes of N-and 2-O-sulfate moieties in the other shed, glypican, and ECM proteoglycan fractions analyzed in this study, with effects similar to those observed for cell surface HSPGs (Fig. S1). Additional data are provided in Table S6, which exhibits the relative abundance of all HS disaccharides from each proteoglycan fraction.
Real Time PCR Analysis-The effect of the Sulf1, Sulf2, and Sulf1/2 knock-outs on a number of nonsubstrate sulfate moieties implies an influence of these enzymes on the HS biosynthetic machinery. To determine whether Sulf loss impacts the expression of HS biosynthetic enzymes, real time PCR analysis was carried out for Sulf1 and Sulf2 as well as the HS sulfotransferases Hs2st1, Hs6st1, Hs6st2, and Hs6st3 (Fig. 5B). All of the corresponding transcripts were shown to be expressed in the MEF cell lines used in this study (Fig. S2). Expression of Hs6st3 was low compared with the other targets analyzed. No real time PCR analysis could be carried out for the N-deacetylase/N-sulfotransferase isoforms, since these enzymes have been shown to exhibit regulation via internal ribosomal entry sites and not at the transcript level (43). Analysis of Sulf expression in Sulf1-, Sulf2-, and Sulf1/2-deficient cell lines revealed that, in contrast to previous characterizations of primary Sulf knock-out MEFs (26), no compensatory up-regulation of Sulf1 or Sulf2 occurs in these knock-out cell lines (Fig. 5B). Interestingly, expression analysis of the 2-O-sulfotransferase, Hs2st1, revealed dynamic effects of Sulf loss on its transcript abundance (Fig. 5B). Indeed, although loss of Sulf1 was not observed to influence Hs2st1 expression in these cell lines, loss of Sulf2 resulted in a ϳ2.5fold decrease in Hs2st1 expression. Conversely, loss of both Sulfs in the Sulf1/2 double knock-out resulted in a ϳ1.5 increase in Hs2st1 expression. Sulf loss was also shown to have dynamic effects on the expression of each of the three 6-Osulfotransferase isoforms present in mammalian cells, Hs6st1, Hs6st2, and Hs6st3 (Fig. 5B). Explicitly, Hs6st1 exhibited no significant change in Sulf1 knock-out MEFs, a ϳ1.5-fold decrease in the Sulf2 knock-out MEFs, and a ϳ2-fold increase in the Sulf1/2 double knock-out. Hs6st2 was increased ϳ2-fold in the Sulf1 knock-out, decreased ϳ1.5-fold in the Sulf2 knockout, and increased ϳ3-fold in the Sulf1/2 double knock-out. Finally, Hs6st3 was increased ϳ2-fold in the Sulf1 knock-out and ϳ1.3-fold in the Sulf2 knock-out, whereas no significant change was observed in the Sulf1/2 knock-out. As discussed below, these real time PCR expression data correlate with a number of important changes observed in 2S and 6S moieties (see Fig. 5A and Fig. S1), suggesting a functional interaction between Sulf activity and expression of HS biosynthetic enzymes.
FGF Signaling-As shown in Fig.  6, A and B, induction of Sulf1 and Sulf2 single knock-out MEF cell lines with 1 ng/ml FGF2 resulted in a ϳ5and ϳ4-fold increase of phospho-ERK1/2 reporter response as compared with WT MEFs, respectively. In contrast, induction of the Sulf1/2 double knock-out cell line with 1 ng/ml FGF2 elicited an exceptional ϳ11-fold increase in reporter signal compared with the WT. The compound effect of Sulf1 and Sulf2 loss on cell signaling response correlates to a significant degree with the observed changes in 6-O-sulfation (Fig. 4, A-D), underlining the notion of functional cooperativity for these enzymes.

DISCUSSION
The discovery of the Sulfs as postsynthetic editors of HS 6-Osulfation patterns revealed a novel mechanism by which HS sulfation can be regulated. Determining how these enzymes affect different proteoglycan families in vivo is critical for understanding how cells and tissues modulate their HSPG environment to regulate cell signaling and biological response.
The Sulfs Are Detergent-insoluble Membrane-associated Proteins and Can Be Secreted-Initial characterization of the quail sulfatases QSulf1 (14) and later QSulf2 (29) revealed that this family of sulfatases localizes to the cell surface and is able to endolytically modulate HS 6-O-sulfation. To investigate the cellular localization and activity of the mammalian orthologs, human Sulf1 and Sulf2 carrying a C-terminal RGS-His 6 tag were cloned and expressed in human fibrosarcoma HT1080 cells. Importantly, human and murine Sulfs are highly homologous enzymes phylogenetically juxtaposed within the Sulf branch of the sulfatase family (15,29). In agreement with previous QSulf results, immunofluorescence microscopy analysis demonstrated a dual cell surface and ER localization for the human Sulfs (14,29). The association of the Sulfs with the cell surface probably plays an important role in mediating the association of these enzymes with their extracellular HSPG substrates. Whether ER localization results from enzyme overexpression and subsequent folding problems or is relevant to intracellular Sulf function remains to be determined.
In line with previous reports, cellular fractionation under different extraction conditions revealed a strict membrane association of the Sulfs, which is sensitive to 0.5 M salt, indicating that the cell surface localization of these enzymes is due to electrostatic interactions and not to trans-membrane integration into the lipid bilayer (15). This ionic membrane association  In Vivo and in Vitro Functional Characterization of Sulf1/2 OCTOBER 10, 2008 • VOLUME 283 • NUMBER 41 is mediated by the highly charged hydrophilic domain of the Sulfs (29). Interestingly, the Sulfs were found to partition specifically into detergent-insoluble fractions, which may indicate affinity for specific lipid compositions or receptors associated with detergent-insoluble membrane domains. Despite the strong association of the Sulfs with membrane subdomains, analysis of the conditioned medium from Sulf-expressing cell lines revealed significant secretion of both the full-length 130-kDa form and, particularly for Sulf2, a proteolytically processed N-terminal 75-kDa form. Previous reports have demonstrated that proteolytically processed Sulf may also be enzymatically active (20,44). Thus, although active full-length Sulf can be secreted, Sulf processing could represent a regulatory mechanism for modulating the solubility and activity range in the extracellular matrix. Substantiating the relevance of Sulf secretion, endogenous Sulf activity was detected also in the secreted fraction of untransfected WT MEFs.
Interestingly, secretion and detergent solubility data obtained for the mammalian Sulfs differ significantly from those reported for the QSulf orthologs (14,29). QSulf1 and QSulf2 are not secreted and are found to partition more promiscuously into detergent-soluble and -insoluble membrane fractions (29). Thus, although QSulf1 and QSulf2 display 86 and 88% homology with their human orthologs, respectively, important functional differences may exist between species.
Sulf1 and Sulf2 Exhibit Identical Specificity toward Native HS in Vitro-The biological function of HS depends on its sulfate composition and the organization of different sulfate groups into highly sulfated, partially sulfated, and nonsulfated domains. Although characterization of human Sulf activity using structurally homogenous heparin as a model substrate has been reported (15,27), understanding the specificity of these enzymes within their native HS substrates is essential for comprehending Sulf function. To address this question, Sulf1 and Sulf2 activities from cellular and secreted fractions of Sulfexpressing cell lines were analyzed using purified radiolabeled HS from Sulf1/2 double knock-out MEFs as a substrate. The results from this analysis reveal restricted enzyme specificities of both Sulf1 and Sulf2 toward the disulfated and, primarily, trisulfated 6S disaccharide units UA-GlcNS(6S) and UA(2S)-GlcNS(6S) within the HS chain. It is important to note that although these in vitro studies demonstrate redundant substrate specificities for Sulf1 and Sulf2, these analyses implement large amounts of enriched enzyme and cannot rule out subtle differences in substrate specificity that may be present at lower endogenous enzyme levels.
By using native HS from double knock-out MEFs as a substrate that has not yet been exposed to endogenous Sulf activity, these experiments show for the first time the activity of Sulf1 and Sulf2 toward their genuine substrate as biosynthetically delivered to the cell surface. Importantly, both Sulf disaccharide substrates are associated primarily with highly sulfated S-domains within the HS chain. These S-domains may be important for substrate recognition or functional activation of the Sulf enzymes. On the other hand, neither Sulf is able to modify the monosulfated 6S disaccharide UA-GlcNAc(6S), a disaccharide unit exclusively found in HS transition zones (45). The significant 6S content of HS transition zones and their location adjacent to the highly sulfated S-domains in the HS chain are thought to be essential for a number of HS-ligand interactions (12,13). The inability of the Sulfs to modify this domain type may indicate a functional importance for retaining 6S residues within these regions.
Sulf Loss Differentially Affects the HS Sulfation Patterns of Multiple HSPGs-The unique enzymatic activity of Sulfs, in conjunction with their specific subcellular localization and low level secretion poses the important question of whether or not these enzymes elicit their effects in a proteoglycan and/or regionally specific manner. To determine the in vivo proteoglycan specificity of the Sulfs, cell surface, GPI-anchored, shed, and ECM-associated proteoglycan fractions were extracted from immortalized WT, Sulf1, Sulf2, and Sulf1/2 knock-out MEFs. After characterizing the extraction efficiency and proteoglycan composition of each fraction by Western blot profiling using the previously described HSPG-specific 3G10 antibody (40), the effect of Sulf loss on each fraction was determined by HS disaccharide composition analysis.
Comparison of the HS composition analysis data from Sulf knock-out proteoglycan fractions with the in vitro Sulf activity analysis in this study reveals important differences between Sulf in vitro activity and the effect of Sulf loss in vivo. One major difference is the impact of Sulf loss on the nonsubstrate UA-GlcNAc(6S) disaccharide in the Sulf knock-out cell lines, a result that has been observed and characterized in a previous study (26). Since this effect cannot be explained by a loss of enzymatic activity, changes in UA-GlcNAc(6S) abundance as a result of Sulf loss implicate a functional influence of Sulf deficiency on the HS biosynthetic machinery to modulate transition zone 6-O-sulfation. Importantly, significant differences between Sulf1 or Sulf2 loss on HS 6-O-sulfation of different proteoglycan fractions analyzed in this study are also manifest primarily within the nonsubstrate or low affinity substrate disaccharide units, UA-GlcNAc(6S) and UA-GlcNS(6S), indicating that these differential effects result from a distinct impact of Sulf1 or Sulf2 loss on 6-O-sulfotransferase activity. In addition to differential changes in nonsubstrate 6-O-sulfation in Sulf knock-out cell lines, a dynamic influence on N-and 2-O-sulfation was also observed. Interestingly, although loss of Sulf1 had no significant effect, loss of Sulf2 caused a decrease in N-and 2-O-sulfation, and loss of both Sulfs resulted in an increase in these sulfate moieties. These results further suggest a dynamic interplay between Sulf activity and the HS biosynthetic machinery in vivo.
Loss of Sulf1 or Sulf2 was also shown to elicit divergent effects on the different HS proteoglycan fractions. Specifically, although the Sulf1 knock-out resulted in significant increases in all three 6S disaccharides in the cell surface proteoglycan fraction, no significant increases of the mono-and disulfated 6S disaccharides UA-GlcNAc(6S) and UA-GlcNS(6S) were observed for the Sulf1 knock-out in the shed or glypican proteoglycan fractions. Further comparison of the influence of the Sulf1/2 double knock-out on different proteoglycan fractions revealed that although loss of both Sulfs affects the abundance of all three 6S disaccharides in the cell surface and shed proteoglycan fractions, Sulf1/2 loss has no effect on the monosulfated UA-GlcNAc(6S) disaccharide in the ECM proteoglycan frac-tion. These differences in HS sulfation patterns between the distinct proteoglycan fractions analyzed in this study imply small but significantly different affinities of the Sulfs or the biosynthetic enzymes influenced by their loss toward different cell surface, shed, glypican, or ECM proteoglycan substrates.
Finally, this study demonstrates for the first time evidence of endogenous Sulf activity within the ECM. This finding contrasts with previous studies analyzing the impact of Sulf overexpression in tumor cell lines (28). In this earlier study, Sulf overexpression failed to disrupt recombinant FGF⅐FGF receptor complex formation in the ECM, indirectly suggesting an inability of these enzymes to modify the sulfation of matrixassociated HSPGs. These conflicting results could be due to differences in the extracellular mobility of the Sulfs in alternate cell lines. Of note, ECM modulation by the Sulfs is likely to have far reaching consequences for growth factor mobilization and may therefore represent a highly regulated process.
Sulf Loss Leads to Differential Expression of HS Biosynthetic Enzymes-Detailed analysis of HS sulfation patterns from the different proteoglycan fractions of Sulf-deficient cells has revealed dynamic effects of Sulf1, Sulf2, and Sulf1/2 knock-outs on N-, 2-O-, and 6-O-sulfate groups that cannot be explained by loss of Sulf enzymatic activity alone. Modulation of these nonsubstrate sulfate moieties implies dynamic effects of Sulf loss on the HS biosynthetic machinery. To substantiate this hypothesis, real time PCR analysis was carried out for Sulf1, Sulf2, and the 2-O-and 6-O-sulfotransferases Hs2st1, Hs6st1, Hs6st2, and Hs6st3. Importantly, a number of changes in HS biosynthetic enzyme expression could be correlated with specific changes in HS sulfation.
Although compensatory expression of Sulf1 or Sulf2 in Sulf-deficient cells has been observed in previous studies (26), no differential Sulf expression was observed in the immortalized cell lines used in this study. This lack of compensatory expression corresponds to the similar increases of UA(2S)-GlcNS(6S) in the proteoglycan fraction from either Sulf1 or Sulf2 knock-out cell lines. Further, changes in the expression of Hs2st1, the single 2-O-sulfotransferase present in mammalian cells, directly correlate with the changes in 2S observed in the Sulf knock-out MEFs. Indeed, both Hs2st expression and 2S sulfation state show no change in the Sulf1 knock-outs, a significant decrease in the Sulf2 knock-outs, and an increase in the Sulf1/2 double knock-outs.
Changes in the expression of the three 6-O-sulfotransferase isoforms were more complex, exhibiting dynamic positive or negative regulation in the different Sulf knock-out MEFs. The lack of correlation between Hs6st expression in the Sulf1 and Sulf2 single knock-outs and the observed changes in overall 6S substantiate previous assumptions that the Sulfs are in fact the major regulators of 6-O-sulfation in the cell (26). Due to the dominant role of the Sulfs in modulating 6S, the major influence of the Hs6st enzymes is likely to be apparent within the HS transition zones where neither Sulf is active. Indeed, changes in the abundance of the transition zone-associated 6S disaccharide unit UA-GlcNAc(6S), which is not affected by Sulf activity in vitro, could be associated with increases in different Hs6st isoforms in the different Sulf knock-out MEFs.
Sulf Loss Has Dynamic Effects on Cell Signaling-To functionally characterize the biological impact of the multiproteoglycan changes in HS sulfation patterning observed in the Sulf knock-out cell lines, the cell signaling response toward FGF2 was analyzed. The effect of Sulf loss on FGF signaling correlates to a significant degree with the observed changes in 6-O-sulfation, showing moderate increases in the Sulf1 or Sulf2 single knock-outs and a synergistic increase in the Sulf1/2 double knock-out. These findings complement recent data obtained from the analysis of FGF response in 6-O-sulfotransferase-deficient MEFs, showing that despite the dependence of FGF2 receptor activation on multiple N-, 2-O-, and 6-O-moieties (46), 6S appears to play an exceptional role in mediating signaling response (47). Nevertheless, compensatory changes in Nand 2-O-sulfation observed in the Sulf2 single knock-outs are likely to procure some degree of moderation to the growth factor response. Compensatory modulation of HS sulfation patterning has been observed in a number of different HS sulfo- In Vivo and in Vitro Functional Characterization of Sulf1/2 OCTOBER 10, 2008 • VOLUME 283 • NUMBER 41 transferase knock-out models (31,47,48), with changes in nonsubstrate sulfate moieties acting to attenuate the primary effect of the knock-out on overall HS charge distribution and cell signaling.
The influence of Sulf1 or Sulf2 loss on the expression of HS biosynthetic enzymes described in this study implies functional interactions between Sulf activity, gene expression, and the regulation of critical HS modulating factors. Previous analysis has demonstrated that Sulf loss can also influence the expression and activity of one another and elicit significant changes in HS sulfation patterns (26). A working model is presented in Fig. 7, illustrating how changes in the expression or activity of cell surface Sulf1 and Sulf2 may influence cell signaling to modulate the expression of HS biosynthetic enzymes as well as the expression and activity of one another. Importantly, evidence supporting regulatory feedback functions for the Sulfs in response to HS-dependent signaling factors was recently presented by Yue et al. (49), who demonstrated specific up-regulation of Sulf1 expression in response to transforming growth factor-␤1 cell signaling, which served to negatively regulate transforming growth factor-␤1 cellular response.
Conclusions-The mammalian Sulfs are cell surface enzymes that associate with detergent insoluble membrane subdomains and can be secreted as active full-length and proteolytically processed forms. In an MEF model system, endogenous activities of Sulf1 and Sulf2 lead to modification of cell surface, GPIanchored, shed, and ECM-associated proteoglycan sulfation. The loss of Sulf activity can also dynamically influence the expression of HS biosynthetic enzymes, leading to a modulation of nonsubstrate N-, 2-O-, and 6-O-sulfate moieties. Finally, Sulf1/2 double deficiency was shown to have a synergistic impact on FGF signaling, demonstrating the cooperative function of the Sulfs in modulating cell signaling response. Overall, this study changes our understanding of the Sulfs from cell surface 6-O-endosulfatases to global multisulfation modulators. Future analysis will focus on understanding the impact of these dynamic effects on developmental abnormalities observed in Sulf knock-out mice.