Mammalian Homologues of the Drosophila Slit Protein Are Ligands of the Heparan Sulfate Proteoglycan Glypican-1 in Brain*

Using an affinity matrix in which a recombinant glypican-Fc fusion protein expressed in 293 cells was coupled to protein A-Sepharose, we have isolated from rat brain at least two proteins that were detected by SDS-polyacrylamide gel electrophoresis as a single 200-kDa silver-stained band, from which 16 partial peptide sequences were obtained by nano-electrospray tandem mass spectrometry. Mouse expressed sequence tags containing two of these peptides were employed for oligonucleotide design and synthesis of probes by polymerase chain reaction and enabled us to isolate from a rat brain cDNA library a 4.1-kilobase clone that encoded two of our peptide sequences and represented the N-terminal portion of a protein containing a signal peptide and three leucine-rich repeats. Comparisons with recently published sequences also showed that our peptides were derived from proteins that are members of the Slit/MEGF protein family, which share a number of structural features such as N-terminal leucine-rich repeats and C-terminal epidermal growth factor-like motifs, and in Drosophila Slit is necessary for the development of midline glia and commissural axon pathways. All of the five known rat and human Slit proteins contain 1523–1534 amino acids, and our peptide sequences correspond best to those present in human Slit-1 and Slit-2. Binding of these ligands to the glypican-Fc fusion protein requires the presence of the heparan sulfate chains, but the interaction appears to be relatively specific for glypican-1 insofar as no other identified heparin-binding proteins were isolated using our affinity matrix. Northern analysis demonstrated the presence of two mRNA species of 8.6 and 7.5 kilobase pairs using probes based on both N- and C-terminal sequences, and in situ hybridization histochemistry showed that these glypican-1 ligands are synthesized by neurons, such as hippocampal pyramidal cells and cerebellar granule cells, where we have previously also demonstrated glypican-1 mRNA and immunoreactivity. Our results therefore indicate that Slit family proteins are functional ligands of glypican-1 in nervous tissue and suggest that their interactions may be critical for certain stages of central nervous system histogenesis.

Glypican-1, whose primary structure was first reported based on its cloning from human lung fibroblasts (1), was the initial member of a rapidly expanding family of glycosylphosphatidylinositol-anchored heparan sulfate proteoglycans that is currently composed of four other vertebrate proteins, cerebroglycan (glypican-2, Ref. 2), OCI-5 (glypican-3, Ref. 3), Kglypican (glypican-4, Ref. 4), and glypican-5 (5). We previously described a major heparan sulfate proteoglycan of nervous tissue (6, 7) that we later cloned and identified as the rat homologue of glypican-1 (8). Northern analysis demonstrated high levels of glypican-1 mRNA in brain and skeletal muscle, and in situ hybridization histochemistry showed that glypican-1 mRNA is especially prominent in cerebellar granule cells, large motor neurons in the brain stem, and CA3 pyramidal cells of the hippocampus (9). From this work and parallel immunocytochemical studies (9) we concluded that glypican-1 is predominantly a neuronal product in the late embryonic and postnatal rat nervous system. Glypican-1 was also found to be a dual modulator capable of enhancing the mitogenic response of fibroblast growth factor-1 but inhibiting the effects of fibroblast growth factor-7 in keratinocytes (10), and it can inhibit neurite outgrowth induced by amyloid precursor protein in vitro (11).
Genetic studies provide additional support for a role of glypicans in cell growth and development. Dally, the Drosophila homologue of glypican-1, is required for the control of cell division in the developing visual system and for morphogenesis of other tissues (12), and the human homologue of glypican-3/ OCI-5 (GPC3) was found to be mutated in patients with the Simpson-Golabi-Behmel overgrowth syndrome (13). We have also recently demonstrated a novel nuclear localization of glypican-1 in nervous tissue (14), suggesting that it may be involved in the regulation of cell division and survival by direct participation in nuclear processes.
Because the functional roles of glypican-1 in nervous tissue remain unknown, we have begun studies aimed at identifying ligands that may aid in understanding how it is involved in developmental and other neurobiological processes. By affinity chromatography of brain extracts on a matrix in which a recombinant glypican-Fc fusion protein was coupled to protein A-Sepharose, we isolated and partially cloned proteins whose sequences allowed us to identify mammalian homologues of the Drosophila Slit protein as ligands of glypican-1. Slit, which was initially identified by cross-hybridization using the sequence coding for tandem epidermal growth factor repeats of Notch, a gene involved in Drosophila neurogenesis (15), is necessary for development of midline glia and commissural axon pathways in Drosophila (16). Although information has only very recently become available concerning the functions of mammalian Slit proteins, our results suggest that interactions of these presum-ably extracellular proteins with cell surface glypican-1 may be important in axonal pathfinding and nervous tissue histogenesis.

EXPERIMENTAL PROCEDURES
Preparation of Glypican-Fc Fusion Protein Affinity Matrix-Human embryonic kidney 293 cells were transfected with a glypican-Fc fusion protein construct (14) using LipofectAMINE (Life Technologies, Inc.) and grown in serum-free Dulbecco's modified Eagle's medium containing 1% ITS ϩ (Collaborative Biomedical Products, Bedford, MA). The conditioned medium was continuously collected and frozen after centrifugation for 30 min at 27,000 ϫ g and addition of sodium azide to a concentration of 0.02%. The amount of the glypican-1 fusion protein in aliquots of the conditioned medium was estimated by Coomassie Blue staining following SDS-PAGE 1 in comparison with bovine serum albumin standards, after binding to an excess of protein A-Sepharose beads (Zymed Laboratories Inc.) using a ratio of 0.5 ml of medium/20 l of settled beads and elution twice by sample buffer.
To determine the proportion of the glypican-Fc fusion protein that was synthesized in a glycanated form, 293 cells in a six-well plate were transfected with the fusion protein construct, and after 24 h cells were washed twice with a short term labeling medium (methionine/cysteinefree Dulbecco's modified Eagle's medium supplemented with 1% ITS ϩ ) followed by incubation for 1 h with the same medium. The medium was subsequently changed for 4 h to labeling medium containing 125 Ci of [ 35 S]methionine/cysteine. The glypican-Fc fusion protein was purified from the conditioned medium by adsorption to protein A-Sepharose beads, and an aliquot of the beads was digested with heparitin-sulfate lyase (EC 4.2.2.8, Seikagaku America, Rockville, MD) at a concentration of 50 milliunits/ml in 0.1 M Tris buffer, pH 7.2, for 3 h at 37°C and compared with another aliquot incubated with buffer alone. (Enzyme assays using bovine kidney heparan sulfate as substrate and measurement of unsaturated disaccharide products by their absorption at 232 nm demonstrated that the heparitinase retains Ͼ85% of its initial activity after 6-h incubation under these conditions.) Both aliquots were boiled in sample buffer; proteins were separated by SDS-PAGE on a 5% gel; and after soaking for 30 min in 200 ml of 1 M sodium salicylate, the gel was dried and bands were quantitated by fluorography using the PhosphorImager (Molecular Dynamics, Sunnyvale, CA).
Pools of conditioned medium were later thawed and concentrated by pressure ultrafiltration on an Amicon PM-30 membrane in a stirred cell. After reclarification by centrifugation (30 min, 27,000 ϫ g), the glypican-Fc fusion protein was bound to protein A-Sepharose beads with gentle mixing at 4°C overnight, using a ratio of 1 mg of fusion protein (in 2-5 ml of concentrated medium)/ml of settled beads. The beads were thoroughly washed with cold PBS followed by 50 mM Trisbuffered saline, pH 8.0, containing 1% Nonidet P-40, 0.5% sodium deoxycholate, and 0.1% SDS. After re-equilibration in 0.1 M phosphate buffer, pH 8.0, the glypican-1 fusion protein was cross-linked to the protein A (17) at room temperature using a 30-fold molar excess of dimethyl pimelimidate (Pierce) added four times at 10-min intervals. The coupling reaction was terminated by washing the beads with 1 M Tris-glycine buffer, pH 7, followed by PBS, and the beads were stored in PBS containing 0.02% sodium azide. Before each use of the affinity matrix, any free fusion protein was removed by washing with 1 M NaCl in 50 mM PBS followed by 0.1 M glycine, and the beads were then again equilibrated in 50 mM PBS.
Affinity Chromatography of Rat Brain Extracts-In an initial experiment to determine which subcellular fractions might be enriched in glypican-1 ligands, brains of 30-to 130-day-old Sprague-Dawley rats were homogenized using a Teflon-glass tissue grinder in 4 volumes of Tris-HCl buffer, pH 8.25 at 4°C, containing 0.15 M NaCl and protease inhibitors (0.5 mM phenylmethylsulfonyl fluoride, 5 mM benzamidine, 50 mM EDTA, and 100 mM 6-aminohexanoic acid). The homogenate was centrifuged for 10 min at 1,000 ϫ g, and the supernatant (S1) was saved. The pellet was resuspended in the original volume of buffer, recentrifuged, and the pellet saved as P1. The S1 fraction was centrifuged for 30 min at 45,000 ϫ g, the pellet (P2) saved, and the supernatant centrifuged for 1 h at 200,000 ϫ g to yield a soluble fraction (S3) and pellet (P3). The P2 and P3 pellets were combined to give a mito-chondrial/microsomal fraction (P2/3). The P1 and P2/3 pellets were resuspended in the original Tris-HCl buffer (2.75 ml/g original brain) containing 1% CHAPS, 0.2 M NaCl, 1 mM EDTA, 10 M pepstatin A, and 10 M leupeptin and extracted with stirring overnight at 4°C. After centrifugation for 1 h at 200,000 ϫ g, the supernatants were saved as fractions S4N (a crude "nuclear" extract) and a membrane extract (S5M), respectively.
Later studies aimed at the isolation and amino acid sequencing of ligands used a simplified procedure. Brains of 30 -40-day-old rats were homogenized in 4 volumes of 25 mM PBS, pH 7.2, containing 5 mM EDTA, 100 M phenylmethylsulfonyl fluoride, 10 M leupeptin, and 10 M pepstatin. The homogenate was centrifuged for 10 min at 1,000 ϫ g and washed once. The washed P1 pellet was then extracted by stirring overnight at 4°C in PBS with protease inhibitors as described above but with the addition of 1% CHAPS, using 2.75 ml/g brain, and the extract was centrifuged for 2 h at 200,000 ϫ g. The resulting supernatant (designated S3N) was then used for affinity chromatography, after filtration through a pre-column of Sepharose CL-4B beads to remove any remaining particulate material or proteins that bind nonspecifically to the Sepharose.
Beads containing the glypican-Fc fusion protein were equilibrated in 50 mM PBS and mixed overnight with the brain extract at 4°C using gentle agitation, in a ratio of 10 ml of brain extract (3.6 g of brain)/ml of beads. Unbound proteins were removed by centrifugation followed by four washes (3 bead volumes each) with PBS. The beads were then transferred to a column, and bound proteins were eluted with 1 M NaCl. (Initial trials demonstrated that no additional protein was released by subsequent elution with 50 mM diethylamine, pH 11.5, in 0.15 M NaCl or by 0.1 M glycine, pH 3.) Proteins eluted with 1 M NaCl were concentrated by adsorption to StrataClean resin (Stratagene) using 10 l of beads/10 ml of eluate and released by boiling in SDS-containing sample buffer before use for SDS-PAGE (18).
Peptide Sequencing by Mass Spectrometry-Proteins in the 1 M NaCl eluate from the glypican-1 affinity column were electrophoresed on several lanes of an 8% 1-mm minigel, silver-stained using a protocol suitable for subsequent sequencing (19), and stored at 4°C. A major band with an apparent molecular size of ϳ200 kDa was excised from two to three lanes of the gel and digested in situ with trypsin (19). The complete tryptic peptide mixture was desalted and concentrated (20) on an Eppendorf Geloader pipette tip (Brinkman) packed with Poros RII resin (PE Biosystems, Framingham, MA) and eluted with 2 l of 50% methanol in 5% formic acid into a nano-electrospray sample needle. The unfractionated digest was analyzed by nano-electrospray (21) on a QTOF mass spectrometer (Micromass, Manchester, United Kingdom), and partial or complete sequences from 16 peptides were obtained by tandem mass spectrometry (22). Amino acid sequences of nine or more residues from 10 different peptides were used to search protein or EST data bases for matches.
cDNA Library Screening-The peptide sequences obtained by mass FIG. 1. Identification of specific glypican-1 ligands by affinity chromatography. A, silver staining of 10% SDS-PAGE gels showing two proteins, migrating at ϳ200 and ϳ22 kDa, that were isolated from rat brain using a glypican-Fc fusion protein affinity column, but do not appear in the 1 M NaCl eluate when the same brain extract was applied to protein A-Sepharose. B, digestion of the fusion protein affinity beads with heparitinase (ϩ) abolished binding of the 200-kDa protein, whereas there was no effect of incubating the fusion protein beads under the same conditions with buffer alone (Ϫ). spectrometry were compared with the nucleotide sequences in the nonredundant GenBank TM data base (EST Division), translated in all six reading frames using the NCBI BLAST server (http://www.ncbi.nlm.nih.gov/cgi-bin/BLAST/nph-newblast?Jformϭ1). Two translated EST sequences matched perfectly to our peptides 3 and 8 and were designated P3EST and P8EST, respectively. P3EST was amplified by PCR from a rat brain stem/spinal cord cDNA library (Stratagene) using primers designed from the EST sequence (sense, 5Ј-CCATCGATGCACCTCA-GAGTTCTTCAGCTC-3Ј; antisense, 5Ј-GCTCTAGACAAGTTGTTC-GAGTGGAGTCG-3Ј) and subcloned in pBluescript. P8EST could not be amplified from any of our three available rat brain cDNA libraries but was amplified from a neonatal mouse brain cDNA library (Stratagene) using primers designed from the EST sequence (sense, 5Ј-CCATCGAT-TCCTTGACGTGGCATCCCTG-3Ј; antisense, 5Ј-GCTCTAGACTCTAT-CACAGCTGTCCCCG-3Ј) and subcloned in pBluescript.
The P3EST sense and antisense primers were also used for PCR screening of amplified pools from a 6-week rat brain and an adult rat brain stem/spinal cord cDNA library (Stratagene). Plaques from PCRpositive pools were screened using a 400-base pair XbaI/ClaI restriction fragment of pBluescript/P3EST that was labeled in the presence of [␣-32 P]CTP (18 mCi/ml) using the Klenow fragment of DNA polymerase I, after purification by agarose gel electrophoresis and QIAEX extraction (Qiagen Inc., Chatsworth, CA).
Plaques were transferred to nitrocellulose filters, and DNA was immobilized by baking in vacuo. Filters were hybridized overnight at 42°C and washed three times at room temperature with 2 ϫ SSC, 0.1% SDS and twice at 65°C with 0.2 ϫ SSC, 0.1% SDS. Supernatants of positive plaques were first screened by PCR using the antisense primer of P3EST in combination with T3 or T7 vector primers, and clones showing the largest insert sizes were subjected to a second screening. Individual plaques were lifted and the plasmids isolated by in vivo excision.
Northern Analysis-pBluescript/P3EST and pBluescript/P8EST were linearized by ClaI digestion and transcribed into digoxigeninlabeled antisense RNA with T3 RNA polymerase (Promega) using the GENIUS 4 RNA labeling kit (Roche Molecular Biochemicals). The resulting probes were used for hybridization with Northern blots and for in situ hybridization histochemistry. mRNA from rat tissues and rat C6 glioma cells was prepared using the FastTrack mRNA isolation kit (Invitrogen Corp., San Diego, CA). Hybridization, washing, and detection with alkaline phosphataselabeled anti-digoxigenin antibodies were performed as described previously (23).
In Situ Hybridization Histochemistry-The digoxigenin-labeled sense RNA probes were prepared as described above except that plasmids were linearized by XbaI and transcribed with T7 RNA polymerase. In situ hybridization histochemistry was performed as described by Engel et al. (24).

Affinity Chromatographic Isolation of Glypican-1 Ligands-
Initial experiments surveyed extracts prepared from several subcellular fractions of brain (see "Experimental Procedures") for the presence of glypican-1 ligands, which were detected predominantly in the "crude nuclear fraction." This fraction was therefore used for all subsequent studies. Specific ligands were considered to be those proteins that bound only to the glypican-Fc fusion protein affinity beads, but not to protein A-Sepharose beads that did not contain the fusion protein. SDS-PAGE analysis and silver staining revealed a major specific ligand with an apparent molecular size of ϳ200 kDa (Fig.  1A), and a second band with a slightly slower mobility could frequently also be resolved (data not shown).
Methionine/cysteine labeling of the glypican-Fc fusion protein secreted by transfected 293 cells followed by SDS-PAGE before and after heparitinase treatment showed that approximately 40% of the fusion protein is glycanated (data not shown). These results indicate that in comparison with endogenous glypican-1 in rat brain and C6 glioma cells (14), the addition of heparan sulfate chains in transfected 293 cells lagged considerably behind the high level of expression of the glypican-1 core protein. Interaction of the 200-kDa ligand with the glypican-1 affinity matrix is at least partially mediated by the heparan sulfate chains, insofar as binding is abolished after treatment of the beads for 5 h with heparitinase (Fig. 1B). However, the 200-kDa band was not detected in eluates of brain proteins bound to heparin-agarose using identical conditions (data not shown).
Another specific ligand with an apparent molecular size of ϳ22 kDa was also detected, and at least five major bands were seen in the 40 -70 kDa range (Fig. 1A), but since these latter proteins bound equally well to protein A-Sepharose beads that did not contain the fusion protein they probably represent rat immunoglobulins or other nonspecific ligands. No sequence could be obtained from the 22-kDa band after transfer to a ProBlott membrane, indicating that it was N-terminally blocked.
Peptide Sequences Deduced from Mass Spectrometric Data-The 200-kDa band stained poorly with Coomassie Blue and sufficient amounts could not be obtained for transfer to a membrane for N-terminal Edman sequencing or for protease treatment and high performance liquid chromatography fractionation of peptides. We therefore used nano-electrospray collision-induced dissociation-tandem mass spectrometry (nano-electrospray CID-MS/MS), a more sensitive peptide sequencing technique that is capable of yielding useful sequence at the femtomole level from single silver-stained polyacrylamide gel bands (21). When applied to an unfractionated tryptic digest derived from several lanes of the 200-kDa protein, MS/MS spectra for 17 peptides were generated ( Fig. 2A). From these data, 16 complete or partial peptide sequences could be deduced (Table I). The MS/MS spectrum of the doubly charged [M ϩ 2H] 2ϩ peptide P8 ion is shown in Fig. 2B. The m/z 756.4 value for this ion corresponds to a molecular weight of 1510.6 for the peptide. The spectrum shows a nearly complete set of C-terminal sequence ions (labeled y n ) beginning with fragment y 1 that contains only Arg and ending with fragment y 10 that includes all of the residues up to and including Tyr. Although this amount of sequence was sufficient to identify the 200-kDa protein (see below), additional sequence information can be obtained by interrogating the mass difference between the highest observable y n ion and the molecular mass of the pep-tide. This difference corresponds to the sum of the in-chain masses of any N-terminal amino acids not yet accounted for by the MS sequencing (most often the first two residues). In the example shown (Fig. 2B), the difference of 227 Da can correspond to only three possible pairs of amino acids (Table I, considering Leu and Ile as a single possibility, and assuming no posttranslational modifications). In this case, the combination N,X (X ϭ L) matched the first two residues for the hit from the data base (Fig. 3). At low m/z, several N-terminal sequence ions (labeled b n ) are also evident, and the m/z value of the b 2 ion confirms the residual mass of the first two N-terminal residues (b 2 ϭ 227 ϩ H).
Using 3-4 residues of sequence data together with mass information (residual N-and C-terminal masses and peptide molecular weights), we generated short MS-derived peptide sequence tags (26) for the 16 peptides. These were searched against the current nonredundant protein data bases revealing that except for peptide 6, which is derived from keratin, there was no significant identity with previously identified proteins. Ten of the MS/MS spectra yielded sequences of nine or more amino acids (two of the peptide sequences differed by only a single amino acid), which we used to search EST data bases. This approach found ESTs from mouse embryo and myotube that matched peptides P3 and P8 (GenBank TM accession numbers AA396603 and AA645364, respectively). spectrometry For P1-10, the number in parentheses at the beginning of the sequence is the mass of the residual N-terminal amino acids (usually two residues) that could not be individually identified. Combinations of amino acids that will fit this mass are shown below the sequence, separated by commas to indicate that their order is unknown. In the sequences, X stands for either leucine or isoleucine, and glutamine residues (Q) are distinguished from lysine (K) based on the assumption that if this residue was lysine trypsin should have cut there. For P11-16, the sequences were assigned by fitting the peptide fragment ion data to the Slit protein sequences (Fig. 3). C* indicates acrylamidemodified cysteine.

P1
(  3. Alignments of peptide sequences obtained from the 200-kDa glypican-1 ligand with those of Slit proteins. The numbering refers to amino acids in rat Slit-1/MEGF4. M8EST represents the sequence that was amplified by PCR from a mouse brain cDNA library using primers designed from the mouse EST sequence. Peptides derived from the 200-kDa glypican-1 ligand are numbered as in Table I, and their sequences are underlined (except for peptide 9). Peptide 6 corresponds to the human keratin sequence (data not shown) and was presumably a result of contamination, whereas peptide 9 does not match perfectly with any reported protein and may represent either a new member of the Slit family or reflect amino acid polymorphism.
Cloning of the 200-kDa Ligand (Rat Slit-2)-Using the EST sequences for primer design and rat and mouse brain cDNA as templates, we amplified by PCR the corresponding rat and mouse sequences (designated P3EST and P8EST, respectively) and used the P3EST as a probe to screen a rat brain cDNA library. The longest clone obtained from screening (designated 1131, GenBank TM accession number AF141386) and the P8EST amplified from a mouse brain cDNA library show 91-93% amino acid identity to human Slit-2 (27), indicating that our partial sequences represent rat and mouse homologues of Slit-2 (Table II). Because some of our peptide sequences match better with MEGF4/Slit-1 than with Slit-2 (Fig. 3), it would appear that the peptide sequences in the 200-kDa gel band are derived from more than one Slit protein and that glypican-1 can bind to both Slit-1 and Slit-2 and possibly also to other related proteins.
Rothberg et al. (16) compared the LRRs and the conserved Nand C-terminal sequences surrounding the LRRs of the Drosophila Slit protein with other proteins containing LRRs. Similarity was found in several proteins involved in adhesive events such as the oligodendrocyte-myelin glycoprotein and the Toll protein of Drosophila, as well as with other proteins involved in extracelluar protein-protein interactions such as the platelet glycoproteins IX, Ib␤, and Ib␣ and small leucine-rich proteoglycans including decorin, biglycan, and fibromodulin. These conserved sequences flanking the LRRs can also be found in our partial sequence of rat Slit-2 (Fig. 4).
Tissue Distribution and Cellular Sites of Synthesis of Slit-2-P3EST and P8EST were used as templates to transcribe digoxigenin-labeled riboprobes for use in Northern analysis and in situ hybridization histochemistry. Northern analysis showed that both probes hybridized with 7.6-and 8.5-kb bands present in rat brain and C6 glioma cell mRNA (Fig. 5), confirming that both peptides (and ESTs) are derived from the same gene. Non-nervous tissues, including skeletal muscle, showed no message with the exception of lung, possibly due to the presence of bronchial smooth muscle.
In the hippocampus at 1 month postnatal, Slit-2 mRNA is present in CA1 and CA3 pyramidal neurons and in granule cells of the dentate gyrus (Fig. 6A), and consistent with the previously reported expression of glypican-1 in cerebellum (9), Slit-2 mRNA is seen in cerebellar granule cells (Fig. 6B). Al-though there is a weaker signal in white matter, which is probably in oligodendrocytes, glial cells do not appear to be a major source of Slit-2. In cerebrum at embryonic day 19 (E19), Slit-2 mRNA is seen primarily in the cortical plate and subplate (but not in the intermediate cortical layer), as well as in the thalamic nuclei, hippocampal formation, caudate putamen (Fig. 6C), and in the subcommissural organ (data not shown). In spinal cord at E13 and E16, Slit-2 mRNA is present predominantly in the ependymal and mantle layers, the floor plate, and dorsal root ganglia (Fig. 7A), and in E16 brain Slit-2 mRNA is detected in the trigeminal ganglion and the ventricular zone, including the entire ganglionic eminence (Fig. 7B). Slit-2 expression can be also seen in the embryonic retina (Fig. 7C) and optic stalk (data not shown).

TABLE II
Comparison of sequence similarities among members of the slit protein family Based on sequence comparisons, it is likely that both clone 1131 obtained by cDNA library screening using rat P3EST as a probe, and the P8EST amplified by PCR from a mouse brain cDNA library, represent the Slit-2 protein, whereas the two recently reported genes designated MEGF4 and MEGF5 (Ref. 28) appear to be rat homologues of Slit-1 and Slit-3, respectively. Numbers in parentheses indicate the number of amino acids or nucleotides in the sequences that were used for comparison. High degrees of identity are noted in bold, and the prefixes h and r in the column headings refer to the human and rat Slit genes. The GenBank™ accession numbers for the rat and human Slit/MEGF proteins are: rat Slit-1, AB017170; rat Slit-2 (clone 1131), AF141386; rat MEGF4, AB011530; rat MEGF5, AB011531; human Slit-1, AB017167; human Slit-2, AB0177168; human Slit-3, AB017169.

DISCUSSION
By affinity chromatography of rat brain extracts on a matrix containing a glypican-Fc fusion protein, we identified a 200-kDa ligand whose peptide sequences did not match any sequences then available in the data bases. During the course of our cloning of this ligand, other data became available indicating that the 200-kDa SDS-PAGE band contained peptide sequences derived from at least two proteins (rat Slit-1 and Slit-2) that are mammalian homologues of the Drosophila Slit protein. This overlapping of protein bands is not surprising insofar as all five of the human and rat Slit proteins that have been cloned up to now have very similar amino acid sequences and differ in size by only 11 or fewer (out of ϳ1,500) amino acids.
The Drosophila Slit protein is expressed by midline glia and is distributed along commissural axons. Reduction in slit expression results in a disruption of the developing midline cells and the commissural axon pathways. The presence of a putative signal peptide and the lack of a transmembrane domain together with other structural features indicate that the Slit proteins are secreted extracellular matrix proteins. Although Slit was not detected in a Tris-buffered saline extract of rat brain, like other extracellular matrix components it may be tightly bound to other cell surface or extracellular proteins and require detergents or dissociative conditions for extraction.
The amino acid sequence of the Slit proteins can be divided into four domains. These consist of four N-terminal LRRs, seven to nine epidermal growth factor-like repeats (seven in Drosophila and nine in vertebrates), a motif with a high degree of identity to agrin, laminin, and perlecan (designated the ALPS domain, Ref. 29), and a C-terminal cysteine-rich domain. LRRs are found in a number of intracellular and extracellular proteins and contribute to protein-protein interactions and cell adhesion (30,31). The epidermal growth factor-like motif has been identified as an extracellular binding domain involved in cell adhesion and receptor-ligand interactions (32), the ALPS domain is responsible for protein-protein interactions and selfaggregation of agrin, laminin, and perlecan (for a review, see Ref. 29), and the cysteine-rich domain is considered to be essential for dimerization of proteins such as von Willebrand factor (33). It is likely that the Slit proteins function to link multiple ligands and thereby mediate cell interactions. In view of the complex domain structure of the Slit proteins and their heparitinase-sensitive interactions with glypican-1, it will be important to identify which protein domain(s) may also be involved in this binding.
During our MS-based sequencing experiments, we attempted CID-MS/MS on a 2,011-Da peptide from the 200-kDa protein tryptic digest. Although we were unable to generate any sequence-specific fragment ions, the peptide readily lost neutral fragments characteristic of tyrosine phosphorylation (34). These preliminary data therefore suggest that Slit proteins may be phosphorylated.
We demonstrated that two major rat Slit-2 mRNA species (8.6 and 7.5 kb) are expressed in adult brain and lung and in rat C6 glioma cells and that the localization of this mRNA in the postnatal hippocampal formation and developing cerebellum is similar to that of glypican-1. In contrast to these results, Northern analysis of human Slit-2 showed a single 8.5-kb mRNA expressed predominantly in adult spinal cord but also FIG. 5. Expression of Slit-2 in rat tissues. Northern blots of mRNA from 7-day and adult brain, C6 glioma cells, and adult liver, heart, spleen, kidney, lung, and muscle were probed with digoxigenin-labeled P3EST and P8EST RNA transcripts.
FIG. 6. Localization of Slit-2 mRNA in rat postnatal hippocampus and cerebellum and embryonic cerebrum. In situ hybridization histochemistry of P28 rat hippocampal formation (A) shows mRNA in neurons of the dentate gyrus (DG) and the CA1-CA3 regions. In P8 cerebellum (B) the mRNA is primarily present in granule cells of both the external (arrow) and internal (arrowhead) granule cell layers. In E19 cerebrum (C) the mRNA is seen primarily in the cortical plate (CP) and subplate (arrows), as well as in the ventricular zone (VZ), hippocampal formation (H) and the caudate putamen (CA). Note that the intermediate cortical layer (ICL) does not show Slit-2 message. Bars, 300 m.
in fetal lung and kidney, and a 9-kb rat MEGF5/Slit-3 message was seen in brain, whereas a major 5.5-kb mRNA and a minor 9.5-kb species of human Slit-3 were found in adult endocrine tissues but were not detectable in brain (27). Although human Slit-1 has the same expression pattern as that of rat Slit-1, there may be species differences in the tissue distribution and alternative splicing of Slit-2 and Slit-3 in humans as compared with rodents. It is also noteworthy that whereas Drosophila Slit is expressed only by glia, rat MEGF4/Slit-1 and Slit-2 are predominantly neuronal products (Ref. 27 and Figs. 6 and 7).
In adult brain, rat MEGF4/Slit-1 is expressed in the hippocampus, cerebral cortex, and olfactory bulb but not in the cerebellum (27), where we detected Slit-2 mRNA. Because our various peptide sequences match both Slit-1 and Slit-2, it would appear that at least two Slit proteins were isolated by our affinity chromatographic procedure and that glypican-1 functions in nervous tissue may be mediated by its differential interactions with two or more members of the Slit protein family. Recent genetic and biochemical studies of Drosophila and mammals demonstrated that Slit proteins bind Robo, a repulsive guidance receptor on growth cones (35)(36)(37). At least one of the mammalian Slit proteins, Slit-2, is a repulsive molecule for olfactory bulb axons (37), embryonic spinal motor axons (35), and developing forebrain axons (38). Interestingly, a 140-kDa N-terminal fragment of Slit-2 was purified from a tissue extract that stimulates the elongation and branching of sensory axons and shown to be responsible for these effects, whereas full-length Slit-2 does not have this activity (39). Although it was also demonstrated that heparin can release Slit-2 from the cell surface (35), we found that Slit proteins could not be isolated from our brain extract when a heparin-agarose matrix was substituted for the glypican-Fc fusion protein affinity column. Because the 140-kDa N-terminal fragment of Slit-2 was not detected in our affinity column eluate, it is possible that glypican-1 binds full-length Slit proteins via both its core protein and the heparan sulfate chains and that this interaction regulates the proteolysis and thereby the generation of biologically active fragments of Slit proteins. The identification of interactions between glypican-1 and Slit proteins therefore not only provides the most direct evidence yet available for an involvement of glypican-1 in nervous tissue development, but also suggests a possible regulatory mechanism underlying the dual functionality of Slit proteins.