SC1/Hevin

SC1, a member of the BM-40 family of extracellular matrix proteins, was recombinantly expressed in a eukaryotic expression system. The full-length protein as well as truncated versions were purified to homogeneity under non-denaturing conditions. Matrix-assisted laser desorption ionization time-of-flight mass spectrometry of full-length SC1 revealed a mass of 87.8 kDa of which 16.8 kDa is contributed by posttranslational modifications. In electron microscopy, after negative staining, SC1 was revealed as a globule attached to a thread-like structure. A calcium dependence of the SC1 conformation could be demonstrated by fluorescence spectroscopy. In the extracellular matrix of cultured osteosarcoma cells SC1 was found associated with collagen I-containing fibrils, and binding of SC1 to reconstituted collagen I fibrils could be demonstrated by immunogold labeling and electron microscopy. SC1 showed a broad expression in a variety of tissues.

SC1, a member of the BM-40 family of extracellular matrix proteins, was recombinantly expressed in a eukaryotic expression system. The full-length protein as well as truncated versions were purified to homogeneity under non-denaturing conditions. Matrix-assisted laser desorption ionization time-of-flight mass spectrometry of full-length SC1 revealed a mass of 87.8 kDa of which 16.8 kDa is contributed by posttranslational modifications. In electron microscopy, after negative staining, SC1 was revealed as a globule attached to a thread-like structure. A calcium dependence of the SC1 conformation could be demonstrated by fluorescence spectroscopy. In the extracellular matrix of cultured osteosarcoma cells SC1 was found associated with collagen I-containing fibrils, and binding of SC1 to reconstituted collagen I fibrils could be demonstrated by immunogold labeling and electron microscopy. SC1 showed a broad expression in a variety of tissues. SC1 was originally cloned from a rat brain expression library by screening with a polyclonal antibody raised against synaptic junction glycoproteins (1). Its mRNA is expressed widely in the brain and could be detected in many types of neurons. In immunoblots of brain proteins SC1 was detected as a doublet band of 116 and 120 kDa, whereas the molecular mass predicted from sequence was 70.6 kDa (1). A homologous cDNA called hevin was also cloned from a library derived from human high endothelial venule cells (2). Human high endothelial venule cells form specialized postcapillary vascular sites (high endothelial venules) that mediate extensive extravasation of circulating lymphocytes from the blood into lymphoid organs and sites of chronic inflammation. The high sequence identity suggests that hevin is the human SC1 ortholog. SC1/hevin is expressed in many tissues as a secreted glycoprotein of the extracellular matrix.
Sequence analysis showed that SC1/hevin belongs to the BM-40/SPARC/osteonectin family of proteins and is composed of a highly acidic domain I without homology to any known protein, a follistatin-like domain (FS) and an extracellular calcium-binding domain (EC) (Fig. 1). The presence of the FS and the EC domain is the hallmark of the BM-40/SPARC/osteonectin family, which consists of BM-40/SPARC/osteonectin, SC1/ hevin, QR1, tsc36/flik, the testicans, and the SMOCs (3).
Functional assays suggest that SC1/hevin may serve as an antagonist of cell adhesion, an effect also studied extensively for SPARC (4). Hevin was shown to inhibit the attachment and spreading of endothelial cells on fibronectin substrates, and it was suggested that hevin modulates the high endothelial venule cell adhesion and phenotype, thus facilitating lymphocyte transendothelial migration (5). The effects of SC1/hevin on cell adhesion are of particular interest due to the fact that SC1/ hevin is down-regulated in many types of cancer cells and may serve as a negative regulator of cell growth and proliferation (6 -8). Furthermore, expression of SC1/hevin is associated with the migration of myotomes during somitogenesis in early mouse embryos and undergoes a rapid down-regulation just before myotome emigration from the somitic environment (9). In another line of work, SC1/hevin was shown to bind to Blymphocyte precursors (10) and to augment B cell lymphopoiesis (11,12), effects that are mediated by the N-terminal acidic domain I. Gene targeting of SC1 in mouse did not show any obvious phenotype (13), a finding that is not too surprising considering the number of members of the BM-40 family that may play a compensatory role.
The major objective of this investigation was to characterize SC1/hevin at the protein level, to investigate the functionality of its predicted calcium-binding EC domain, and to identify mechanisms by which it can be integrated into the extracellular matrix. For this purpose we expressed full-length SC1 as well as fragments with domain-specific truncations. These were analyzed for structural features as well as for the ability to bind calcium and collagen fibrils. The purified SC1 FS-EC domain pair was used to raise an antibody with which we could demonstrate a colocalization of SC1/hevin with collagen I in the extracellular matrix of SAOS-2 cells and determine the distribution of SC1 in many tissues also outside the central nervous system.

Construction of Expression Vectors for Full-length and Truncated
Forms of SC1-The cDNA clone for mouse SC1 was kindly supplied by Dr. Peter J. McKinnon. Four cDNA fragments, representing the fulllength SC1, the pair of I and FS domains, the pair of FS and EC domains, and the EC domain, were generated by PCR using the following primers: forward primer-1, 5Ј-GCCCCGCTAGCCCCGACAAGTAC-AAGGTTC, reverse primer-2, 5Ј-CAATGACTGCGGCCGCTCAAAAGA-GGAGGTTTTCAT, forward primer-3, 5Ј-GCCCCGCTAGCCTCTTGCA-CGAACTTCCAAT, forward primer-4, 5Ј-GCCCCGCTAGCCCCTGCTT-GTACGGACTTT, and reverse primer-5, 5Ј-CAATGACTGCGGCCGCT-CAAATAGATTTGCAAGCTCCG. Primers 1 and 2 were used for fulllength SC1, primers 1 and 5 were used for the pair of I and FS domains, primers 3 and 2 were used for the pair of FS and EC domains, and primers 4 and 2 were used for the EC domain. The primers introduced new restriction sites (NheI, NotI) and a stop codon. The NheI/NotIrestricted PCR products were purified and cloned into the corresponding restriction sites of pCEP-Pu/BM40 (14) to obtain the final expression vectors pCEP-Pu-SC1, pCEP-Pu-SC1-I-FS, pCEP-Pu-SC1-FS-EC, and pCEP-Pu-SC1-EC. The cDNA fragment of I-FS was additionally cloned in the modified expression vector pCEP-Pu with an N-terminal tag consisting of six histidines, a Myc epitope, and an enterokinase recognition site for cleavage of the tag (15). The eukaryotic expression vector pRcCMV (Invitrogen) was used to produce a modified vector pRcCMV/BM40 (5.5 kilobases). Therefore, a HindIII/NotI fragment containing the signal peptide sequence and some of the 5Ј-untranslated region of BM-40 (105 bp) together with the full-length SC1 sequence (1.9 kilobases) was cleaved from pCEP-Pu-SC1, purified, and ligated to the same restriction sites in pRcCMV. The resulting vector was designated pRcCMV-SC1. The correct insertion and sequences of all amplified fragments were verified by cycle sequencing of both strands using an ABI Prism 377 Automated Sequencer (PE Biosystems).
Cell Culture and Transfection-The human embryonic kidney cell line EBNA-293 (Invitrogen), which constitutively expresses the EBNA-1 protein from Epstein-Barr virus was used for transfection with pCEP-Pu-SC1-I-FS, pCEP-Pu-SC1-FS-EC, and pCEP-Pu-SC1-EC and selected with puromycin as described (14). The fibrosarcoma cell line HT1080 (kindly provided by Dr. M. Aumailley) was transfected with pRcCMV-SC1. Two days after transfection the cells were split 1:10, and selection was started by adding 800 g/ml G418 (Invitrogen). Resistant clones were isolated after 4 -5 weeks and then screened for expression of SC1 by immunoblotting with an antibody against the N-terminal portion of SC1 (16). Conditioned serum-free media were stored at Ϫ80°C.
Purification of Full-length SC1-Conditioned serum-free medium was dialyzed against 50 mM BisTris, pH 7.0, containing 0.5 mM phenylmethylsulfonyl fluoride and passed at 4°C over a DEAE-Sepharose (Amersham Biosciences) column equilibrated in the same buffer. Fractions that eluted between 0.26 -0.33 M NaCl showed immunoreactivity with the antibody to SC1. These were dialyzed against 50 mM BisTris, pH 7.0, and applied to a Q-Sepharose (Amersham Biosciences) column equilibrated in the same buffer. The recombinant protein was eluted at 0.35-0.41 M NaCl. Further purification was carried out on a HiTrap heparin column (Amersham Biosciences) equilibrated in 10 mM sodium phosphate, pH 7.0. The protein was eluted at 0.21 M NaCl. After concentration on a Resource Q (Amersham Biosciences) column equilibrated in 20 mM BisTris, pH 6.0, the protein was further purified by gel filtration on Sephadex G-75 (Amersham Biosciences) in 0.3 M NaCl, 50 mM Tris-HCl, pH 7.4. The SC1 pool was dialyzed against 20 mM BisTris, pH 5.9, and final purification was performed on a Resource Q column equilibrated in the same buffer. The pure SC1 eluted at 0.32-0.35 M NaCl (linear gradient 0 -1 M NaCl) and was dialyzed against 5 mM Tris-HCl, pH 7.4.
Purification of the I-FS Domain Pair-Conditioned serum-free medium was dialyzed against 0.1 M NaCl, 50 mM sodium phosphate, pH 8.2, and passed over an affinity column with cobalt Talon Metal Affinity Matrix (Clontech). The protein was eluted at a concentration of 20 -70 mM imidazole. I-FS-containing fractions were further purified by gel filtration on a column of Sephadex G-75 equilibrated in 50 mM Tris-HCl, pH 7.4. The final pool was dialyzed against 5 mM Tris-HCl, pH 7.4.
Purification of the FS-EC Domain Pair-Conditioned serum-free medium was dialyzed against 50 mM Tris-HCl, pH 8.6, and passed over a column of DEAE-Sepharose equilibrated in the same buffer at 4°C. The recombinant protein was found in the flow-through and was, after dialysis against 50 mM sodium acetate, pH 4.9, applied to a column of SP-Sepharose (Amersham Biosciences). The FS-EC protein eluted at 0.35-0.77 M NaCl and was, after concentration by ultrafiltration (YM-10, Amicon), further purified by gel filtration on a column of Sephadex G-75 equilibrated in 0.2 M NaCl, 50 mM Tris-HCl, pH 8.6. Fractions containing the FS-EC-protein were dialyzed against 20 mM MES, 1 pH 6.0, and subjected to chromatography on Resource S matrix, where it eluted at 0.25 M NaCl. The final pool was dialyzed against 5 mM Tris-HCl, pH 7.4.
Purification of the EC Domain-Conditioned serum-free medium was dialyzed against 50 mM Tris-HCl, pH 8.6, and passed through a DEAE-Sepharose column at 4°C. The pH of the flow-through was adjusted to pH 9.0 and concentrated on a HiTrap Q column (Amersham Biosciences) equilibrated in 50 mM Tris-HCl, pH 9.0. The EC domain was eluted at 0.5 M NaCl. Final purification was carried out by gel filtration on a column of Sephadex G-75 equilibrated in 5 mM Tris-HCl, pH 7.4.
Verification of Recombinant Proteins-The identity of all proteins was verified by MALDI-TOF peptide mass fingerprinting, and in the case of the FS-EC domain pair and the EC domain alone, also by Edman degradation (model 473A, PE Biosystems).
N-Glycosidase F and Endoglycosidase H Digestion-Purified proteins (20 pmol) were reduced with 0.5% of mercaptoethanol and denatured by heating at 100°C for 5 min. After denaturation, the proteins were incubated with 0.3 units of enzyme per mg of protein for 24 h at 37°C in the buffers recommended by the supplier (Roche Molecular Biochemicals). Digested and non-digested samples were analyzed by MALDI-TOF mass spectrometry.
Mass Spectrometry-Samples for MALDI-TOF mass spectrometry were prepared as described (17) with ␣-cyano-4-hydroxycinnamic acid (Sigma) as matrix. Detection and analysis of cations was performed utilizing the high or low mass detector in the linear mode of a Bruker Reflex III. The time of flight analyzer was calibrated using the M r of bovine serum albumin dimer (Sigma) of 132,859.0.
Electron Microscopy of Purified SC1-Isolated full-length SC1 (5 g/ml) was visualized by negative staining with 0.75% uranyl formate on carbon-coated grids rendered hydrophilic by glow discharge (18). The samples were examined in a Zeiss E 902 electron microscope.
Circular Dichroism (CD) Spectroscopy-CD spectra were recorded in a Jasco model 715 CD spectropolarimeter at 25°C in thermostatted quartz cells of 1-mm optical path length. The molar ellipticity [] (expressed in deg ⅐ cm 2 ⅐ dmol Ϫ1 ) was calculated on the basis of a mean residue molecular mass of 110 Da. The Ca 2ϩ dependence of the CD spectrum was measured by the addition of 2 mM CaCl 2 . Reversibility of the conformational change was tested by subsequent addition of 4 mM EDTA. To improve the signal to noise ratio, five spectra were accumulated. A base line with buffer (5 mM Tris-HCl, pH 7.4) was recorded separately and subtracted from each spectrum. Protein concentrations were determined from UV spectra using extinction coefficients calculated according to Gill and von Hippel (19). The program SELCON was used to estimate the content of secondary structure elements.
Fluorescence Spectroscopy-Intrinsic fluorescence was measured with a PerkinElmer LS50B spectrofluorometer in 10-mm path length rectangular cells at 25°C with excitation at 280 nm. Emission spectra for the full-length SC1 (0.5 M) and the I-FS and FS-EC domain pairs (each 1 M) were recorded in 5 mM Tris-HCl, pH 7.4, and for the EC domain (1 M) in 0.15 M NaCl, 50 mM Tris-HCl, pH 7.4. All spectra were recorded with 2 mM CaCl 2 and after the addition of 4 mM EDTA. The percent change in fluorescence intensity was calculated as ⌬F 350 ϭ 100 ϫ (F Ca Ϫ F EDTA )/F Ca , with F Ca representing the fluorescence signal at Ca 2ϩ saturation, and F EDTA representing the signal in the presence of excess EDTA.
Preparation of Antibodies against SC1-The purified FS-EC domain pair was used to immunize a rabbit. The antiserum was purified by affinity chromatography on a column with the antigen coupled to CNBractivated Sepharose (Amersham Biosciences).
Binding of SC1 to Reconstituted Collagen I Fibrils-Solutions of highly purified chicken collagen I were prepared in 0.4 M NaCl, 0.1 M Tris-HCl, pH 7.4. Samples were degassed under vacuum and then transferred to microcuvettes (light path, 1 cm) and mixed with an equal volume of distilled water at 4°C. The cuvettes were sealed and placed into a spectrophotometer (Beckman UV 640, equipped with a multicell holder Micro Auto 12) connected to a water bath at 37°C. Aggregation was followed by monitoring turbidity at 313 nm. When a plateau was reached, aliquots were spotted onto Parafilm sheets. Nickel grids coated with Formvar/carbon were floated for 5 min on drops to allow adsorption of fibrils. The grids were washed with PBS and treated for 30 min with 5% bovine serum albumin in PBS. The adsorbed fibrils were allowed to react with full-length SC1 or the FS-EC domain pair in 0.15 M NaCl, 50 mM Tris, pH 7.4, containing 10 mM CaCl 2 for 90 min. After washing 5 times with PBS, the grids were incubated with the affinitypurified SC1 antibody in PBS containing 0.2% (w/v) dried skim milk and, after renewed washing with a suspension of colloidal gold particles (18 nm), coated with goat antibodies to rabbit IgG (Jackson ImmunoResearch) in PBS containing 0.2% (w/v) dried skim milk. Finally, the grids were washed with distilled water and negatively stained with 2% uranyl acetate for 10 min. In control experiments SC1 or the first antibody was omitted. Electron micrographs were taken at 80 kV with a Philips CM 10 electron microscope.
Immunofluorescence Microscopy of SaOS-2 Cell Layers-The human osteosarcoma cell line SaOS-2 (DSMZ, Braunschweig) was cultivated in McCoy's 5A medium supplemented with 10% of fetal calf serum, 2 mM L-glutamate, and optionally, 50 g/ml sodium ascorbate. For immunofluorescence microscopy cells were plated onto plastic chamber slides, cultivated until they reached confluency, fixed in 2% paraformaldehyde in PBS for 10 min, and blocked in 1% bovine serum albumin. Immunolabeling for SC1 was done by consecutive treatment for 1 h with the affinity-purified rabbit antibody to SC1 followed by a diaminotriazinylaminofluorescein (DTAF)-conjugated goat antibody against rabbit IgG and for collagen with a mouse monoclonal antibody (Sigma) followed by Cy™3-conjugated affinity-purified goat antibody against mouse IgG (Jackson ImmunoResearch). Pictures of the cell layers were taken with a fluorescence laser scanning microscope (Leica).
Immunoblotting of Tissue Extracts-Mouse tissues were homogenized and extracted in 5 volumes (ml/g) 0.15 M NaCl, 50 mM Tris-HCl, pH 7.4, containing a protease inhibitor mixture (Complete, Roche Molecular Biochemicals). The homogenates were centrifuged, and aliquots of the supernatants were submitted to SDS-PAGE. For immunoblotting, proteins were transferred electrophoretically to a nitrocellulose filter and developed with the affinity-purified antibody against SC1, followed by peroxidase-conjugated swine anti-rabbit IgG (DAKO) and the enhanced chemiluminescence reagent (Amersham Biosciences).
In inhibition experiments the first antibody was preincubated with 50 g of the FS-EC domain pair for 2 h at room temperature and then applied to an immunoblot of kidney extract and recombinant SC1 protein.

Recombinant Production of Full-length and Truncated SC1
Proteins-Eukaryotic cells were used for protein expression to ensure proper folding and posttranslational modifications. The use of the BM-40 signal peptide gave targeting through the endoplasmic reticulum and Golgi apparatus and secretion into the medium from which recombinant proteins could be purified. The human embryonic kidney cell line EBNA-293 was initially used for expression of all four constructs, corresponding to full-length SC1, the I-FS and FS-EC domain pairs, and the EC domain (Fig. 1). Although the shorter constructs were produced at acceptable levels, expression of full-length SC1 was low in this cell line. As an alternative we transfected the human fibrosarcoma cell line HT1080 with a plasmid coding for full-length SC1 and selected stably transformed clones. One of these showed good SC1 expression and only slight degradation of the protein and was used for recombinant production of the full-length protein. In SDS-PAGE full-length SC1 migrated above the 94-kDa marker, although the calculated peptide mass is only 70.9 kDa (Fig. 2, Table I). The I-FS domain pair was cloned with a His-Myc tag and could be purified by cobalt affinity chromatography. In SDS-PAGE, the protein gave a single band with an apparent molecular mass of 95 kDa (Fig.  2), in contrast to the calculated mass of 57.6 kDa. The purified FS-EC domain pair gave a single SDS-PAGE band at 33 kDa, whereas the EC domain migrated as a 17 kDa protein (Fig. 2), in both cases in reasonable agreement with their predicted molecular masses. Taken together, the results show an unex-pectedly low mobility in SDS-PAGE of the larger proteins, which may be caused by posttranslational modifications within the I and FS domains and/or the highly acidic nature of the I domain.
Molecular Mass and Glycosylation of Recombinant Fulllength and Truncated SC1 Proteins-In an attempt to resolve the cause for these discrepancies all proteins were analyzed by MALDI-TOF mass spectrometry either directly or after digestion with N-glycosidase F or endoglycosidase H (Table I) (Table I). The masses determined for N-glycosidase F-digested full-length SC1 and domain I-FS were about 12 kDa higher than that predicted. Because the mass of the N-glycosidase F-digested FS-EC domain pair was as expected from sequence, these 12-kDa posttranslational modifications of full-length SC1 and domain pair I-FS are located in domain I.
Electron Microscopic Appearance of Full-length SC1-The purified full-length SC1 was visualized by negative staining electron microscopy (Fig. 3). SC1 was seen as a monomeric particle with a length of about 25 nm composed of a globule with a diameter of 9.5 nm attached to a thread-like structure with a length of about 15 nm. Because the x-ray structure of the FC-EC domain pair from the homologous protein BM-40 shows a rather compact fold (20), the globule is likely to correspond to the FS and EC domains that make up the C-terminal part of SC1, whereas the thread-like structure is contributed by the highly acidic N-terminal domain I.
Conformation and Calcium Binding of Full-length and Truncated SC1-The conformations of the recombinant proteins were analyzed by circular dichroism in the far-ultraviolet region. Spectra of full-length SC1 and the I-FS domain pair exhibited a marked negative band at 200 nm, indicating a large content of unordered structures (Fig. 4). In contrast the FS-EC domain pair and the isolated EC domain displayed shoulders at 222 nm typical for ␣-helices. Analysis of the CD spectra (Table  II) suggested that the EC domain has the highest content of ␣-helix followed by the FS-EC domain pair, which also shows a  1 and 5), I-FS tag (lanes 2 and 6), FS-EC (lanes 3 and 7), and EC (lanes 4 and 8) were submitted to SDS-PAGE under non-reducing (lanes 1-4) and reducing (lanes 5-7) conditions. The 12% polyacrylamide gel was stained with Coomassie Brilliant Blue. The molecular mass of marker proteins are given at the left margin.
high content of ␤-sheet. Domain I consists largely of unordered structures, which it contributes to full-length SC1 and the I-FS domain pair. All four proteins gave similar CD spectra when equilibrated in 2 mM CaCl 2 or 4 mM EDTA (Fig. 4).
The potential calcium dependence of SC1 conformation was also studied by intrinsic fluorescence. Upon excitation at 280 nm all proteins showed emission maxima between 339 and 351 nm that were not changed by the addition of 2 mM CaCl 2 (Fig.  5). However, calcium addition led to a decrease in fluorescence intensity for all proteins containing the EC domain, i.e. fulllength SC1 (24%), the pair of FS and EC domains (64%), and the single EC domain (97%) itself. The conformational change was reversible when EDTA was added in excess. In contrast, the fluorescence signal of the I-FS domain pair was not influenced by calcium (results not shown).
Binding of SC1 to Reconstituted Fibrils of Collagen I-Earlier studies performed with the homologous protein BM-40/ SPARC/osteonectin showed a distinct binding of this family member to collagen via its EC domain (21,22) and a loss of BM-40 from the extracellular matrix of collagen I-deficient mice (23). Because the EC domain, which has in BM-40 been shown to carry the collagen binding site, is conserved in SC1, we decided to test for affinity to collagen I. As collagen I is present in tissues in fibrillar form, we used reconstituted fibrils prepared from highly purified collagen I to which SC1 was added. Electron microscopy showed a marked binding of both recombinant full-length SC1 and the FS-EC domain pair to such collagen I fibrils (Fig. 6), whereas control experiments, performed by omitting SC1 or the antibody to SC1 used for detection, were negative. The fibrils showed the characteristic periodic banding pattern and carried SC1 bound to their surface.
Colocalization of SC1 and Collagen I in the Extracellular Matrix of SaOS-2 Osteosarcoma Cells-To further analyze the relevance of the interaction between SC1 and collagen I we studied the deposition of the two proteins in the matrix produced by cultured human osteosarcoma SaOS-2 cells by immunofluorescence microscopy. An antiserum was raised in rabbit against the FS-EC domain pair, and specific antibodies were affinity-purified by binding to the original antigen coupled to CNBr-Sepharose. The purified antibodies against SC1 showed no cross-reactivity to other members of the BM-40 protein family, i.e. BM-40, testican-1, -2, -3, TSC-36, SMOC-1, either in native or in SDS-denatured form (results not shown).
When ascorbate was present in the SaOS-2 cell culture medium, allowing secretion of collagens, a filamentous extracellular network was produced (Fig. 7, A-C). Immunostaining revealed a colocalization of SC1 and collagen I in large portions of this network. In the absence of ascorbate neither collagen I nor SC1 could be detected in the surrounding space, showing that collagen is required for the deposition of SC1 into the extracellular matrix produced in these cultures (Fig. 7, D-E). Under these conditions the intracellular staining for both proteins was increased, which indicates that SC1 is retained in the secretory pathway when collagen secretion is abolished due to the lack of ascorbate.
Distribution of Different Size Forms of SC1 between Mouse Tissues-The tissue distribution of SC1 was studied by immunoblot analysis of extracts of adult mouse tissues (Fig. 8A). The strongest signals were found in extracts of brain, cerebellum, and lung, giving a band migrating at about 116 kDa. Weaker signals with the same apparent molecular mass were detected in samples from heart, eye, and muscle. Another prominent band at about 55 kDa was found in brain, cerebellum, heart, eye, muscle, and kidney. In kidney this was the only form of SC1 detected. To further verify that this lower band represents  a true SC1 fragment or variant, recombinant full-length SC1 protein and kidney tissue extract were tested in an inhibition experiment. A dilution of the affinity-purified antibodies was preincubated with the recombinant FS-EC domain pair against which the antibodies were originally raised. This completely abolished antibody binding to both the full-length recombinant SC1 and to the 55-kDa band from kidney extracts, whereas a control sample of the same antibody dilution without the addition of inhibitor gave strong signals (Fig. 8B). DISCUSSION SC1 has been proposed to perform a number of important functions in development and regeneration of the central nervous system (1, 16, 24 -26) in the adhesion of high endothelial venule cells (2,5), in muscle differentiation (9), and in the maturation of B-lymphocytes (10 -12). However, the structural features of SC1 have not been characterized beyond sequence interpretation (1, 2), and even though it is described as an extracellular matrix protein, its mode of integration into extra-cellular matrices has not been investigated. We therefore undertook the systematic expression of full-length and truncated SC1 proteins to be able to determine some of their important structural features and to provide a basis for future mechanistic studies of SC1 functions. Two important properties of BM-40 are the functionality of the EC domain in calcium binding and the ability to bind fibrillar collagens. Therefore, we investigated if both are conserved in SC1. To determine whether a collagen binding activity can be of physiological relevance, we studied SC1 to determine of it occurs in close association with fibrillar collagens and is expressed in tissues rich in a collagenous matrix.
Full-length SC1 was recombinantly expressed in the human fibrosarcoma cell line HT1080 and the truncated SC1 proteins (Fig. 1) in the human embryonal kidney cell line EBNA-293. Both of these cell types have been used frequently for the expression of other mammalian proteins and have proven to reliably provide a correctly folded product with the right pattern of disulfide bonds. We recently compared the glycosylation pattern of human bone sialoprotein expressed in EBNA-293 cells with that of the same protein purified from human bone and found that, although differences were detected in both Nand O-linked glycans, functional properties such as enhancement of cell attachment and hydroxyapatite binding were retained in both forms (15). Because there is no abundant tissue source for purification, it has not been possible to perform an equally detailed analysis of SC1. Nevertheless, the consistency of the mobilities of the tissue-derived and the recombinant full-length protein in SDS-PAGE (Fig. 8) indicate considerable similarities in processing.
Both full-length SC1 and the I-FS domain pair showed a lower mobility in SDS-PAGE than expected from the molecular mass calculated from the amino acid sequence. A part of the discrepancy could be explained by the presence of N-glycans on   ) and MALDI-TOF mass spectrometry (88 kDa) is that the highly negatively charged domain I (106 acidic residues of 400 residues) causes an atypical binding of SDS to SC1, leading to an anomalous electrophoretic behavior. The predicted domain structure of SC1 could be partially confirmed by electron microscopy of recombinant SC1. The full-length protein is seen as a flexible tail of about 15-nm length connected to a globular structure of 9.5-nm diameter (Fig. 3). The x-ray structure of the FS-EC domain pair from BM-40 shows a comparatively compact fold and a close interaction between these domains, yielding a roughly globular structure of about the same dimensions as the SC1 globule seen by electron microscopy (20). Accordingly, the flexible tail must be the domain I, a conclusion that is also supported by the high content of unordered secondary structure in this domain (see below). This domain has been shown to increase the survival and proliferation of B lineage lymphocytes and their precursors (10 -12). Its extended flexible structure will make it well available to cells also if SC1 is anchored in the extracellular matrix. It may be that its high negative charge density and potential O-linked glycosylation plays a role in its receptor interactions.
Circular dichroism spectroscopy showed that the EC domain is rich in ␣-helix and that the FS domain has a comparatively high content of ␤-structure (Fig. 4, Table II). This is in good agreement with the detailed knowledge of the three-dimensional structure of the homologous domains in BM-40 (28,20) and structure-based sequence alignments. On the other hand, the domain I contains largely unordered secondary structure, in agreement with its thread-like appearance in electron microscopy. The circular dichroism spectra obtained in the presence of saturating amounts of calcium or with excess EDTA were closely similar, pointing to a rearrangement of secondary structure in SC1 upon calcium binding that is too small to be detected by circular dichroism. In contrast, large and reversible changes in intrinsic fluorescence were detected for all SC1 proteins containing the EC domain when calcium was added and removed (Fig. 5). This demonstrates that the SC1 EC domain is active in calcium binding, which is an agreement with the fact that both EF-hands are conserved when compared with BM-40. EF-hand 2 is a classic EF-hand as seen in multiple cytosolic EF-hand proteins in which the acidic side chains of residues at position 1, 3, 5, and 12 within the EF-hand loop mainly contribute to calcium binding. The EF-hand 1 of BM-40 is of a non-classical type as it contains a one-amino acid insertion and a proline in cis conformation in the EF-hand loop. This leads to a structural rearrangement such that the carbonyl oxygen of the proline coordinates the calcium ion instead of an oxygen from the side chain of residue 3 (28). In SC1 the residues of EF-hand 1 including the insertion and the proline are strictly conserved compared with BM-40; therefore, calcium coordination is most probably identical to the one seen in BM-40. The fluorescence change upon calcium addition further shows that the calcium binding induces a conformational change even though this could not be detected by circular dichroism. Calcium-induced conformational changes as well as a specific proteolytic cleavage that may have related structural consequences have been shown to enhance the binding of BM-40 to collagen (21,22,29). Only the I-FS domain pair was unaffected by calcium binding, as determined by intrinsic fluorescence, making it likely that these domains do not bind FIG. 7. Immunofluorescence laser scanning microscopy of SC1 and collagen I in monolayer cultures of SaOS-2 osteosarcoma cells. Cells were cultivated in the presence (A-C) or absence of ascorbate (D-F). In A and D SC1 was detected with affinity-purified rabbit antibodies raised against the purified FS-EC domain pair and a DTAF-conjugated affinity-pure goat anti-rabbit IgG (green). In C and F collagen I (Col. I) was detected with a mouse monoclonal antibody followed by a Cy3™-conjugated goat anti-mouse IgG (red). B and E are merged images. The bar corresponds to 40 m.
FIG. 8. Immunoblot analysis of SC1 in mouse tissue extracts. A, mouse tissues were extracted with 5 volumes (ml/g) of 0.15 M NaCl, 50 mM Tris-HCl containing 10 mM EDTA, and protease inhibitors. Aliquots were applied to 12% SDS-polyacrylamide gels under reducing conditions, and SC1 was detected by immunoblotting using an affinitypurified antiserum to SC1. B, recombinant full-length SC1 and kidney extracts were submitted to SDS-PAGE followed by immunoblotting as in A. Inhibition of specific antibody binding was performed by incubation of a dilution of the affinity-purified antiserum to SC1 with recombinant FS-EC protein, the antigen against which the serum was originally raised. The control lanes were developed with an identical antiserum dilution that was not treated with inhibitor. calcium with high affinity or that calcium binding does not strongly affect their fold.
BM-40 has been shown to bind strongly to several collagens (21,22), and Mov13 mice, which are deficient in collagen I, do not retain BM-40 in their extracellular matrix (23). We therefore determined if the collagen affinity of BM-40 is conserved in SC1 and if this may indeed be a mechanism by which SC1 is anchored in extracellular matrices. Immunogold electron microscopy showed that both full-length SC1 and the FS-EC domain pair bind to the surface of fibrils reconstituted from purified collagen I (Fig. 6). Furthermore, when monolayers of SaOS-2 osteosarcoma cells, which express both SC1 and collagen I, were studied by double immunofluorescence microscopy, a close colocalization between these proteins was seen in filamentous structures in the extracellular matrix (Fig. 7, A-C). When the same cell line was grown in the absence of ascorbate, resulting in intracellular retention of collagen due to decreased activity of the prolyl hydroxylase, such extracellular deposition of SC1 was lost in parallel with that of collagen I (Fig. 7, D-F). Indeed, the intracellular staining for both proteins was increased, creating the impression that SC1 was retained in compartments of the secretory pathway when collagen I could not be processed and externalized. It may be that complexes between collagen I and SC1 are formed already in the endoplasmic reticulum or in the Golgi apparatus.
A screen performed by immunoblot of tissue extracts showed a broad distribution of SC1, although the largest amounts were found in brain (Fig. 8). The only two organs in which no SC1 could not be detected by extraction and immunoblot was liver and adrenal gland, tissues characterized by a high cellularity and a low collagen content. In tissue extracts, SC1 was detected mainly as an SDS-PAGE band at 116 kDa, corresponding to the mature protein. However, some lower bands were also specifically stained, and one of these, at 55 kDa, was the major SC1 reactivity in muscle and the only band in kidney. There is no evidence in the literature for alternative splicing of SC1 mRNA, and it is, therefore, most likely that the 55-kDa band results from a proteolytic cleavage that is particularly prominent in certain tissues. Because the antibodies were raised against the purified FS-EC domain pair, the fragment must contain epitopes from these domains. Considering the flexible and unordered nature of domain I, it is tempting to propose a cleavage in this domain, releasing a large part of it from the rest of the molecule. Because domain I has been implicated in interactions with B-lymphocyte precursor cells (11), it may be that cellular interactions of SC1 may be modulated by such cleavage.