Neuroglycan C, a novel membrane-spanning chondroitin sulfate proteoglycan that is restricted to the brain.

Monoclonal antibodies were raised to membrane-bound proteoglycans derived from rat brain, and four monoclonal antibodies that recognized a 150-kDa chondroitin sulfate proteoglycan with a core glycoprotein of 120 kDa were obtained. Immunohistological study revealed that the proteoglycan was associated with developing neurons. We screened rat brain cDNA libraries using the four monoclonal antibodies and isolated overlapping cDNA clones that encoded the entire core protein of 514 amino acids plus a 30-residue signal peptide. The deduced amino acid sequence suggested an integral membrane protein divided into five structurally different domains: an N-terminal domain to which chondroitin sulfate chains might be attached, a basic amino acid cluster consisting of seven arginine and two lysine residues, a cysteine-containing domain, a membrane-spanning segment, and a C-terminal cytoplasmic domain of 95 amino acids. On Northern blots, the cDNA hybridized with a single mRNA of 3.1 kilobases that was detectable in brains of neonatal and adult rats but not in kidney, liver, lung, and muscle of either. The sequence of the proteoglycan did not exhibit significant homology to any other known protein, indicating that the proteoglycan, designated neuroglycan C, is a novel integral membrane proteoglycan.

Proteoglycans are molecules in which glycosaminoglycan chains are covalently linked to core glycoproteins. They are major glycoconjugates, located at the cell surface and in extracellular spaces, and they are considered to have various regulatory effects on cells by binding to chemotrophic factors, cell adhesion molecules, and extracellular matrix glycoproteins (1,2). Early biochemical studies of the mammalian brain revealed the presence of a varied set of proteoglycans whose expression is regulated precisely during brain development (3,4). The developmentally regulated expression of such molecules has been implicated in the formation of several complicated structures in the brain, such as the mossy fiber (5), visual (6), and cortical barrel systems (7)(8)(9). Studies in vitro have also shown that neuronal proteoglycans are involved in the morphogenesis of neurons, for example, in neuritogenesis. They are capable of exerting both positive (10,11) and negative (12)(13)(14)(15)) effects on the outgrowth of neurites in vitro.
Our understanding of the functions of neuronal proteoglycans is expanding as the primary structures of core proteins are elucidated. Such analyses have revealed that a number of proteoglycan core proteins share similar sequence motifs and can be grouped into several families. Neurocan (16), brevican (17), versican (18), and BEHAB (19) are members of a soluble chondroitin sulfate proteoglycan family; their sequences include the tandem repeats of hyaluronan-binding domain. N-Syndecan (20) is a member of the syndecan family of heparan sulfate proteoglycans, which have extensively similar transmembrane and cytoplasmic domains. Moreover, it was recently revealed that a chondroitin sulfate proteoglycan, phosphacan/ 6B4 proteoglycan, is an extracellular variant of a receptor-like protein-tyrosine phosphatase, RPTP␤/ (21,22). Although primary structures of a number of core proteins have been reported, more proteoglycans in the brain remain to be characterized. In particular, the complete primary structures of only three membrane-bound proteoglycans have been reported, namely NG2 (23), glypican (24), and cerebroglycan (25), although there appear to be at least 16 core proteins in the membrane-bound fraction of the rat brain (4). Therefore, we attempted to raise monoclonal antibodies against membranebound proteoglycans and to isolate the cDNA clones that encoded for a proteoglycan-core protein.
Here, we report a novel membrane-spanning chondroitin sulfate proteoglycan that is recognized by a panel of monoclonal antibodies. The predicted protein is an integral membrane protein containing a signal peptide, a chondroitin sulfate-attachment domain, a cluster of basic amino acids, a cysteine-containing domain, a 24-amino acid transmembrane domain, and a cytoplasmic tail. The amino acid sequence of the core protein exhibits little similarity to other known proteins, indicating that the proteoglycan is a novel species of integral membrane proteoglycan. We propose the name neuroglycan C (NGC 1 from neuronal proteoglycan with chondroitin sulfate) for this proteoglycan.

EXPERIMENTAL PROCEDURES
Extraction of Membrane-bound Proteoglycans-Brains from 10-dayold Sprague-Dawley rats (usually 100 rats per experiment; Slc:SD strain; SLC Inc., Shizuoka, Japan) were homogenized with a Teflon-* 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EMBL Data Bank with accession number(s) U33553.
¶ To whom correspondence should be addressed. glass homogenizer in 2 ml per brain of ice-cold phosphate-buffered saline (PBS) that contained 20 mM EDTA, 10 mM N-ethylmaleimide, and 2 mM phenylmethylsulfonyl fluoride as protease inhibitors. Homogenization and subsequent steps were carried out at 4°C unless otherwise indicated. The homogenate was centrifuged at 27,000 ϫ g for 40 min. The pellet was subjected to rehomogenization in the same solution. After centrifugation, the PBS-insoluble material was homogenized in 2 ml per brain of PBS that contained the protease inhibitors plus 1% Nonidet P-40 at 0°C. The homogenate was stirred for 60 min on a magnetic stirrer and was then centrifuged at 27,000 ϫ g for 40 min. The supernatant was retained, and the pellet was treated again with the detergent-containing buffer. After centrifugation, both detergent-containing supernatants were combined and lyophilized.
Partial Purification of Membrane-bound Proteoglycans-The lyophilized residue obtained from the pooled supernatants was suspended in 1 ml per brain of 4 M urea, 50 mM Tris-HCl, pH 7.5, that contained 0.2% Nonidet P-40, 0.1 M NaCl, 2 mM EDTA, 1 mM N-ethylmaleimide, and 0.2 mM phenylmethylsulfonyl fluoride (urea buffer), and the suspension was dialyzed against the same buffer. After centrifugation at 27,000 ϫ g for 30 min, the supernatant was mixed with 1 ml per brain of DEAE-Sephacel (Pharmacia, Uppsala, Sweden), which had been equilibrated with the urea buffer, and the mixture was stirred on a magnetic stirrer for 2 h. The DEAE-Sephacel resin was recovered by centrifugation at 6,500 ϫ g for 10 min and was washed twice with the urea buffer. The resin was packed in a glass column, and proteoglycans were eluted from the resin with a linear gradient of NaCl from 0.1 to 0.7 M in the urea buffer.
The proteoglycan fraction (eluted by 0.45-0.68 M NaCl) was pooled and concentrated to 5 ml on a Diaflo YM-10 membrane (Amicon Corp., Danvers, MA). The concentrated solution was chromatographed on a column (1.6 cm, inner diameter, ϫ 100 cm) of Sepharose CL-4B (Pharmacia) in 4 M guanidine HCl, 50 mM Tris-HCl, pH 7.5, that contained 0.2% Nonidet P-40 and the protease inhibitors. The proteoglycan fraction (K av from 0.11 to 0.65) was pooled and concentrated to 3 ml on a Diaflo YM-10 membrane. Proteoglycans were precipitated from the solution by addition of 3 volumes of 95% ethanol that contained 1.3% (w/v) potassium acetate. The precipitate was washed twice with 75% ethanol that contained 1% (w/v) potassium acetate at 0°C, and then it was dried over P 2 O 5 in vacuo.
The dried material was dissolved in 10 ml of 4 M guanidine HCl, 50 mM Tris-HCl, pH 7.5 (guanidine HCl buffer), at room temperature. The solution was applied to a column (1.6 cm, inner diameter, ϫ 5.0 cm) of octyl-Sepharose (Pharmacia). Elution was performed with a linear gradient of Nonidet P-40 from 0 to 0.8% (v/v) in 100 ml of the guanidine HCl buffer. Fractions eluted at a detergent concentration from 0.3 to 0.6% were pooled, and proteoglycans in the pooled fraction were precipitated with ethanol and dried as described above.
The proteoglycan preparation was further purified by ultracentrifugation in a CsCl density gradient at an initial density of 1.38 g/ml in 4 M guanidine HCl, 50 mM Tris-HCl, pH 7.5, at 10°C in an RPS-65T rotor (Hitachi, Tokyo) at 150,000 ϫ g for 30 h. After centrifugation, the sample was collected as seven fractions. Fractions numbered 2 to 6 were pooled and dialyzed against PBS at 4°C. The amounts of hexuronate and protein in the proteoglycan solution were measured by the method of Bitter and Muir (26) and a protein assay kit (Bio-Rad), respectively. This purification procedure usually yielded about 3 mol of hexuronate (2.2 mg of protein) from 100 brains.
Soluble proteoglycans were extracted from 10-day-old rat brains with PBS that contained protease inhibitors and purified as described previously (3). In brief, proteoglycans were purified sequentially by column chromatography on DEAE-Sepharose and by gel filtration on Sepharose CL-4B. Fractions with K av that ranged from 0.45 to 0.60 were utilized for Western blot analysis.
Gel Electrophoresis and Western Blot Analysis-Membrane-bound proteoglycans were separated by SDS-polyacrylamide gel electrophoresis (PAGE) under reducing conditions with dithiothreitol on a 3% stacking gel with a 6% separating gel. They were transferred electrophoretically to a polyvinylidene difluoride (PVDF) membrane (Immobilon P, Millipore) at 60 V overnight in 25 mM Tris, 192 mM glycine (pH 8.3) that contained 20% (v/v) methanol at 4°C. To block nonspecific binding of antibodies, the membrane was incubated in 1% skim milk (DIFCO Laboratories, Detroit, MI) in PBS. The membrane was then treated overnight at 4°C with hybridoma-conditioned medium that contained mouse monoclonal antibodies and stained in PBS with the Vectastain ABC kit (Vector Labs, Burlingame, CA). In some experiments, the intensity of immunostaining was quantified with a computer (LC520; Apple) via an image scanner (GT-6500; Epson, Tokyo) and application of the National Institutes of Health image program (public domain).
Production of Monoclonal Antibodies-Monoclonal antibodies (mAbs) were produced by previously described methods (27). The preparation of membrane-bound proteoglycan was injected intraperitoneally into BALB/c mice with complete and then with incomplete Freund's adjuvant at 2-week intervals (50 g of protein per injection). After the fourth booster of injection, mice were sacrificed, and spleen cells were fused with PAI myeloma cells using polyethylene glycol. Hybridomas, selected in HAT medium, were cultured in RPMI 1640 medium (Sigma) that contained 10% fetal bovine serum. Culture supernatants were screened for the presence of antibodies that reacted with the preparation of proteoglycan by Western blotting. Positive hybridomas were cloned by the limiting dilution method. mAbs C1, C3, C5, and C15 reacted with a chondroitin sulfate proteoglycan with a 120-kDa core glycoprotein. All of the four antibodies were determined to be IgG 1 with a mouse monoclonal antibody isotyping kit (ImmunoType, Sigma). Preparation of tissue sections from the rat cerebral cortex and immunostaining of the tissue sections were performed by previously described standard methods (9).
Purification of the 120-kDa Core Glycoprotein-The 120-kDa core glycoprotein was partially purified by the method used for partial purification of proteoglycan-type protein-tyrosine phosphatases that was described by Maeda et al. (22). In brief, a postnuclear fraction prepared from whole brains of 8-day-old rat was solubilized with 1% CHAPS. The solubilized materials were applied to a column of DEAE-Toyopearl (Tosoh Corp., Tokyo), which was washed with 0.25 M NaCl, and then proteins were eluted with 0.6 M NaCl. The negatively charged material from the column was fractionated by CsCl density gradient ultracentrifugation to separate proteoglycans from standard proteins. After centrifugation, the gradient was collected in 10 fractions. In the three lowest fractions, proteins of 140 and 120 kDa were detected as the major species after digestion with chondroitinase ABC (see Fig. 2 in Ref. 22). Therefore, proteoglycans in these fractions were digested with chondroitinase ABC, and the core glycoproteins were separated by SDS-PAGE (6% polyacrylamide). The core glycoproteins were transferred to a PVDF membrane, and the band of the 120-kDa core glycoprotein was excised with a razor blade. The core glycoprotein of 120 kDa was recognized by the four monoclonal antibodies C1, C3, C5, and C15 on Western blots.
The purified 120-kDa core protein was subjected to amino acid sequence analysis. Fragments of the 120-kDa core protein were obtained by treatment with cyanogen bromide (CNBr) as described by Scott et al. (28). The intact core protein and the fragments were separated by SDS-PAGE, as described above, and transferred to a PVDF membrane at 0.3 A for 3 h at 4°C in 10 mM CAPS buffer (pH 11) that contained 10% methanol. After staining with 0.5% Ponceau Red in 1% acetic acid, bands of protein were excised from the membrane, and proteins were sequenced on an automatic protein sequencer (PPSQ-10; Shimadzu, Kyoto, Japan).
cDNA Libraries and Isolation of cDNA Clones-A gt11 cDNA library that was constructed with poly(A) ϩ RNA from the whole brain of an 18-day-old (P18) Sprague-Dawley rat as template and random primers (22) was used. Another gt11 cDNA library was constructed with poly(A) ϩ RNA from the whole brain of an 8-day-old (P8) Sprague-Dawley rat as template and oligo(dT) primers. These cDNA libraries were immunoscreened of 1 ϫ 10 6 plaques with a mixture of culture supernatants of mAbs C1, C3, C5, and C15. Immunoscreening was performed as described by Sambrook et al. (29) with the Vectastain ABC kit for immunodetection. After the first two clones, NGC1 (from the random primer-cDNA library) and NGC3 (from the oligo(dT) primer-cDNA library) had been isolated by immunoscreening, subsequent clones were obtained by plaque hybridization using 32 P-labeled cDNA probes. cDNA probes for plaque hybridization and Northern blotting were labeled with the Megaprime DNA labeling system (Amersham International, Amersham, UK). cDNA inserts from gt11 clones were purified and subcloned into the pBluescript II plasmid vector (Stratagene).
Sequencing of DNA-Subclones for sequencing were generated by the introduction of deletions by restriction enzymes. DNA sequencing was performed with Taq polymerase and dye-labeled primers using an automated DNA Sequencer (Alf II/Alfred combined system; Pharmacia). Editing and analysis of sequences were performed with GENETYX software (Software Development Co., Tokyo). The reading frame was verified by reference to our data for the N-terminal amino acid sequence of the 120-kDa core glycoprotein and for the CNBr-generated fragments. Comparisons with sequences in the GenBank (Release 87) and Swiss-Prot (Release 31) data bases were performed with a local alignment search tool (30).
Northern Blot Analysis-Total RNA was extracted from brains of rats on postnatal day 7 and adult rats, and also from kidney, liver, lung, and muscle of adult rats as described by Chomczynski and Sacchi (31). Poly(A) ϩ RNA was purified with oligotex-dT30 as described in the instruction manual from the manufacturer (TAKARA Biomedicals, Otsu, Japan). Poly(A) ϩ RNA (5 g) was subjected to electrophoresis on a 1% agarose gel that contained 2.2 M formaldehyde and transferred to a PVDF membrane (Immobilon N, Millipore). A 682-bp insert from NGC1 was labeled with [ 32 P]dCTP and used as a probe at a concentration of 3 ϫ 10 6 cpm/ml. The filter was washed finally in 0.2 ϫ SSC (SSC ϭ 0.15 M NaCl and 0.015 M sodium citrate) at 68°C and exposed to x-ray film (Fuji, Tokyo) for autoradiography.

FIG. 1. Western blot of a partially purified preparation of membrane-bound proteoglycans and a preparation of PBSsoluble proteoglycans from brains of 10-day-old rats.
The preparation of partially purified membrane-bound proteoglycans (A1) was digested with chondroitinase ABC (A2), chndroitinase ABC plus heparitinase I (A3 and B1), neuraminidase (B2), O-glycanase (B3), or N-glycanase (B4). In B, samples were digested sequentially with these glycosidases. Samples (2 nmol of uronic acid for the preparation of intact membrane-bound proteoglycans (A1, and 0.2 nmol for others) were subjected to electrophoresis on a 6% polyacrylamide gel, blotted onto a PVDF membrane, and stained with mAb C5. The positions of molecular mass markers are indicated in kDa.

FIG. 2. Spatiotemporal patterns of expression of NGC in the rat cerebral cortex.
A and B, a section of coronal tissue from the cerebral cortex of a 0-(A) and a 7-(B) day-old-rat was immunostained with mAb C5. Immunopositive material associated with neurons can be seen throughout the cortex. Scale ϭ 100 m. C, developmental changes in the relative amounts of the core glycoprotein of the novel proteoglycan. The intensities of the immunolabeled bands of protein on a Western blot were quantified by densitometry. The levels are shown as percentages of the level on P20.

RESULTS AND DISCUSSION
Characterization of Monoclonal Antibodies-We established four hybridoma cell lines (designated C1, C3, C5, and C15), each of which produced an antibody that reacted with a preparation of membrane-bound proteoglycans on a Western blot. All of the four mAbs recognized a very diffuse band that corresponded to an average molecular mass of 150 kDa when the preparation of membrane-bound proteoglycans that had been used as immunogen was examined (Fig. 1A, immunoblot with mAb C5). When the preparation of membrane-bound proteoglycans was digested with chondroitinase ABC, the band of a 120-kDa immunopositive protein became detectable. Treatment with heparitinase did not change the motility of this band. When a large amount of a preparation of PBS-soluble proteoglycans derived from brains of 10-day-old rats was used to prepare a Western blot, a 75-kDa core glycoprotein was weakly detected with mAb C5 (data not shown). mAb C5 did not stain any of the proteoglycan core glycoproteins prepared from extra-neuronal tissues, such as liver, lung, kidney, muscle, and cartilage (data not shown).
To examine the possibility that the proteoglycan might have oligosaccharide side chains in addition to chondroitin sulfate chains, the core glycoprotein was digested sequentially with three glycosidases, as shown in Fig. 1B, and the molecular mass of the deglycosylated core protein was estimated by SDS-PAGE. Each treatment with glycosidase resulted in an increase in the mobility of the band of the core protein on SDS-PAGE. The ability of mAb C5 to recognize the core protein was not affected by digestion with any of the glycosidases. Therefore, it appeared that the epitope recognized by mAb C5 was present on the polypeptide moiety.
The spatial expression of the proteoglycan in the rat cerebral cortex was examined by immunohistochemical staining with mAb C5. On postnatal day 0 (P0), immunolabelings with mAb C5 was associated with developing neuronal cells and was seen throughout the cerebral cortex ( Fig. 2A). A similar pattern of staining was seen in the cerebral cortex on P7 (Fig. 2B). As the brain matured, the intensity of immunostaining of the cortex decreased (data not shown). A similar spatial expression of the proteoglycan was also observed when sections were stained with other mAbs C1, C3, and C15. Fresh medium containing normal mouse IgG (5 g/ml, Sigma) was used in place of mAb solutions for negative control of immunostaining. The control sections were virtually free of immunolabelings.
To quantify the temporal expression of the proteoglycan in the cerebral cortex, PBS-insoluble extracts of the cerebral cortex at various developmental stages (from embryonic day 12 (E12) to adulthood) were digested with chondroitinase ABC and processed for immunoblotting, and the intensity of the immunolabeled band was quantified. A small amount of the proteoglycan was detected on E16 through E18, and the amount of the proteoglycan increased to reach a maximum level around P20 (Fig. 2C). After P20, the amount of the proteoglycan component decreased, and in the mature cerebral cortex this proteoglycan was expressed at approximately half of the peak level of expression. This biochemical data confirmed our immunohistochemical data.
Isolation of cDNA Clones-We purified the 120-kDa core glycoprotein, cleaved it into several peptides by treatment with CNBr, and determined the partial N-terminal amino acid sequences of the intact core glycoprotein and of two peptides. The N-terminal sequence of the intact core glycoprotein was VPAREAGSAIEAEEL. Entirely different N-terminal sequences were found in the two products of degradation by CNBr (15 and 24 kDa), (M)VPGGSISLRPRPGDPGKDLA and (M)GRFPGSP for the 15-and 24-kDa peptides, respectively.
One positive clone (NGC1) from the gt11 cDNA library (random primers) derived from a rat brain on P18 and another positive clone (NGC3) from the gt11 cDNA library (oligo(dT) primers) derived from a rat brain on P8 were initially isolated by immunoscreening with a mixture of the four mAbs, C1, C3, C5, and C15. NGC1 and NGC3 contained inserts of 682 bp (nucleotides 108 -790 in Fig. 3) and 1,398 bp (nucleotides 710 -2,107), respectively. The fusion protein with ␤-galactosidase from NGC1 reacted with only mAb C1 and not with mAbs C3, C5, and C15. By contrast, the fusion protein from NGC3 reacted with mAbs C3, C5, and C15 but not with mAb C1. The inserts of NGC1 and NGC3 were subcloned into the pBluescript II plasmid vector for further analyses (to yield pNGC1 and pNGC3, respectively). The authenticity of these clones was unequivocally established by identification of an amino acid sequence from the 120-kDa core glycoprotein and the 24-kDa CNBr-generated fragment within the amino acid sequence deduced from the insert in pNGC1 and by that from the 15-kDa CNBr fragment within the amino acid sequence deduced from the insert in pNGC3 (see Fig. 3, underlining).
With the same probe, mRNA for NGC was detected by Northern blot analysis (Fig. 4). A single transcript of 3.1 kb was detected in analyses of brains of 7-day-old and adult rats. This transcript was not detected in analyses of kidney, liver, lung, and muscle.
Predicted Structure of the Core Protein of NGC-The complete cDNA sequence encoding the core protein of NGC and the deduced amino acid sequence are shown in Fig. 3. The clones covered more than 2,107 bp, which included an open reading frame that encoded 544 amino acids. The calculated molecular mass was 58,612 Da. The nucleotide sequence around the ATG triplet at position 13 corresponded to the consensus sequence for an initiation site of translation (32). The 5Ј-region of 12 bp of the cDNA contributes to the 5Ј-untranslated region. There is a stretch of hydrophobic amino acids from the initial methionine to the first amino acid residue (valine) of the mature core protein, and it is probably a signal sequence. Hydropathy analysis of the predicted protein by the method of Kyte and Doolittle (33) revealed a second hydrophobic segment near the C terminus, composed of 24 amino acids (residues 426 -450), which was followed by two basic residues (Fig. 3, dashed line). This arrangement meets the criteria for a transmembrane domain proposed by Sabatini et al. (34). The 3Ј-untranslated region consists of 463 nucleotides and included a polyadenylation signal (AATAAA).
The size of the open reading frame was confirmed by reverse transcriptase-mediated PCR. All of the amplified DNA fragments using multiple pairs of primers (see "Experimental Procedures") had reasonable sizes, which are predicted from the cDNA sequence. Primers 1 and 2 used for reverse transcriptase-mediated PCR are positioned on the N-terminal sequence of the intact core glycoprotein determined by amino acid sequence analysis. In addition, primer 5 is positioned on the predicted C-terminal sequence. Therefore, the size of the open reading frame shown in Fig. 3 can be considered to be appropriate. After removal of oligosaccharide side chains, however, the estimated molecular mass of the core protein was 100 kDa (Fig. 1), which is still considerably larger than the calculated molecular mass (55.8 kDa) of the mature core protein with 514 amino acid residues encoded by the cloned cDNA. This discrepancy might be attributable to the anomalously slow electrophoretic migration of glycosylated proteins that results from decreased binding of SDS (35), since after digestion by the glycosidases some oligosaccharides might still remain on the core protein. Differences between actual and apparent molecular masses have been reported for the core proteins of other proteoglycans, such as syndecan (36), versican (18,37), and neurocan (16).
The N-terminal region, from amino acid residues 31-281, contained eight serine-glycine (SG) or glycine-serine (GS) dipeptide sequences. These dipeptides have been proposed to be core portions of the consensus sequence of attachment site for chondroitin sulfate (18,40). The amino acid sequences around the eight serine residues differed from the consensus sequences for attachment sites for glycosaminoglycans, which are SGXG (40) and (E/D)GSG(E/D) (18). However, these consensus sequences are not the only ones that allow attachment of glycosaminoglycans to core proteins. In neurocan, for example, the sequence around a serine residue to which a chondroitin sulfate chain is attached is EEVASGQED (16). The alignment of amino acids characteristic of attachment sites for chondroitin sulfate in reported proteoglycans, such as aggrecan, versican, syndecan, decorin, and collagen type IX, is SG/GS with preceding and following acidic amino acid residues. The importance of acidic amino acid residues in the attachment of chondroitin sulfate has also been noted elsewhere (18,40,41). The eight SG/GS dipeptides of the NGC core protein were arrayed in sequence with this consensus-type arrangement. NGC contained two additional dipeptide sequences (serine residues at positions 341 and 374) in its putative extracellular domain. However, these dipeptides were not associated with this consensus-type organization. Therefore, the N-terminal domain is putatively defined as the chondroitin sulfate attachment domain.
A short stretch of basic amino acids KRRKRRRRIR (residues 282-291) is found (double underlining in Fig. 3) adjacent to this putative chondroitin sulfate-attachment domain. The basic amino acid residues could contribute to the proteolytic processing of this proteoglycan since they correspond to sites of cleavage by serine proteases that might cooperate in the regulation of neurite outgrowth (42)(43)(44)(45)(46)(47). The soluble type of NGC might be such a proteolytic product since NGC mRNA was detected as a single 3.1-kb transcript on Northern blots of mRNAs from the rat brain (Fig. 4). Moreover, we reported previously the proteolytic processing of another chondroitin sulfate proteoglycan, neurocan (48).
The cytoplasmic domain of NGC contained threonine and serine residues that could be potential sites of phosphorylation by protein kinases. The amino acid sequences around the threonine residue at position 465 (KLRRTNK) and around the serine residue at position 521 (SPK) are similar to the consensus sequence for the sites of phosphorylation by protein kinase C (X(R/K)XX(T/S)X(R/K)) proposed by Graff et al. (49). Although it remains to be determined whether NGC is phosphorylated, it is possible that NGC might be involved in signal transduction. The putative cytoplasmic domain contains two tyrosine residues, but the sequences adjacent to these tyrosine residues differ from the reported consensus sequences for tyrosine phosphorylation (50). Fig. 5 is a schematic representation of the proposed domain structure of the NGC core protein.
A search of data base analyses at the amino acid level indicated that small portions of NGC are homologous to regions of the core protein of aggrecan, a large chondroitin sulfate proteoglycan of cartilage (51) In the cysteine-containing domain of NGC, we found three homologous clusters: 287-322 and 1827-1862 (31% identity/42% chemical similarity), 348 -386 and 1898 -1936 (23%/38%), and 401-411 and 1998 -2008 (55%/55%). All of the corresponding homologous clusters in aggrecan are also localized in one of the cysteine-containing domains of aggrecan. A schematic alignment of homologous sequences in rat aggrecan and NGC is shown in Fig. 6. Aggrecan consists of several structurally different domains, namely an immunoglobulin-like domain and tandem repeats of a hyaluronic acid-binding region in the Nterminal portion, a chondroitin sulfate-attachment domain in the central portion, and epidermal growth factor-like domains, a lectin-like domain, and a complement regulatory protein-like domain in the C-terminal portion (51). Several other chondroitin sulfate proteoglycans, such as versican (18), neurocan (16), brevican (17), and BEHAB (19), also include these domains. Therefore, they are all considered to be members of the aggrecan family. NGC is not a member of this family since the NGC core protein does not have any of these domains except for a chondroitin sulfate-attachment domain.