J Biol Chem, Vol. 274, Issue 30, 21217-21222, July 23, 1999

Differential Binding of Vascular Endothelial Growth Factor B Splice and Proteolytic Isoforms to Neuropilin-1^*

Taija Makinen^§, Birgitta Olofsson^§¶, Terhi Karpanen, Ulf Hellman, Shay Soker^**, Michael Klagsbrun^**, Ulf Eriksson^¶, and Kari Alitalo

From the  Molecular/Cancer Biology Laboratory, Haartman Institute, University of Helsinki, FIN-00014 Helsinki, Finland, the ^¶ Ludwig Institute for Cancer Research, Stockholm Branch, Box 240, SE-17177 Stockholm, Sweden, the  Ludwig Institute for Cancer Research, Uppsala Branch, Box 595, SE-751 24 Uppsala, Sweden, and the ^** Department of Pathology, Children's Hospital, Harvard Medical School, Boston, Massachusetts 02115

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Vascular endothelial growth factor B (VEGF-B) is expressed in various tissues, especially strongly in the heart, and binds selectively to one of the VEGF receptors, VEGFR-1. The two splice isoforms, VEGF-B₁₆₇ and VEGF-B₁₈₆, have identical NH₂-terminal cystine knot growth factor domains but differ in their COOH-terminal domains which give these forms their distinct biochemical properties. In this study, we show that both splice isoforms of VEGF-B bind specifically to Neuropilin-1 (NRP1), a receptor for collapsins/semaphorins and for the VEGF₁₆₅ isoform. The NRP1 binding of VEGF-B could be competed by an excess of VEGF₁₆₅. The binding of VEGF-B₁₆₇ was mediated by the heparin binding domain, whereas the binding of VEGF-B₁₈₆ to NRP1 was regulated by exposure of a short COOH-terminal proline-rich peptide upon its proteolytic processing. In immunohistochemistry, NRP1 distribution was found to be overlapping or adjacent to known sites of VEGF-B expression in several tissues, in particular in the developing heart, suggesting the involvement of VEGF-B in NRP1-mediated signaling.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Vascular endothelial growth factor (VEGF)¹ is a prime regulator of endothelial cell proliferation, angiogenesis, vasculogenesis, and vascular permeability (for reviews, see Refs. 1 and 2). The effects of VEGF are mediated by two high affinity receptor tyrosine kinases VEGFR-1 (Flt-1) and VEGFR-2 (KDR/Flk-1), mainly expressed on endothelial cells. Five different VEGF transcripts encoding polypeptides of 121, 145, 165, 189, and 206 amino acid residues are generated by alternative splicing from a single human VEGF gene (3-5). VEGF₁₂₁ is secreted as a soluble protein, whereas VEGF₁₆₅ has affinity for heparan sulfates and is partially sequestered into the pericellular matrix. Recently, a third receptor for the VEGF₁₆₅ isoform was identified from nonendothelial tumor cells, which bound a sequence encoded by exon 7 of the VEGF gene (6). This receptor was identical to Neuropilin-1 (NRP1), a type I transmembrane receptor for semaphorins/collapsins (7, 8). Coexpression of NRP1 with VEGFR-2 in transfected cells resulted in increased VEGF₁₆₅ binding and enhanced chemotaxis and mitogenicity in response to VEGF₁₆₅ (6). The NRP1 binding may explain why VEGF₁₆₅ is more potent than VEGF₁₂₁ as a mitogen for endothelial cells (9).

NRP1 is expressed in the tips of actively growing axons of particular classes of neurons, and studies of it have concentrated on its important role in axon growth and guidance in the developing embryo (10-12). NRP1 is, however, expressed in the developing embryo ubiquitously in endothelial cells of capillaries and blood vessels and in mesenchymal cells surrounding the blood vessels as well as in certain other nonneuronal tissues including the endocardial cells of the embryonic heart (6, 13). Furthermore, homozygous NRP1 knock-out embryos die of cardiovascular failure at embryonic day 10.5-12.5, and overexpression of NRP1 under the -actin promoter is lethal because of severe anomalies of the nervous and cardiovascular systems (13). The cardiovascular defects may result from modulation of VEGF bioactivity and VEGF-induced angiogenesis by abnormal NRP1 levels.

At present, the VEGF-family of growth factors contains four other known members, namely placenta growth factor (PlGF), VEGF-B/VRF, VEGF-C/VRP, and VEGF-D/FIGF (14). PlGF and VEGF-B bind selectively to VEGFR-1 and resemble most closely VEGF in sequence and in genomic structure (15, 16). Alternatively spliced PlGF and VEGF-B mRNAs each give rise to two protein isoforms, PlGF-1 and PlGF-2 and VEGF-B₁₆₇ and VEGF-B₁₈₆, respectively (16-18). The PlGF-2 and VEGF-B₁₆₇ isoforms have basic COOH-terminal domains responsible for binding heparan sulfate proteoglycans and a sequence homologous to the NRP1 binding exon 7-encoded peptide of VEGF.

PlGF-2 was recently shown to bind NRP1 (19), but it is not known whether NRP1 also binds other angiogenic growth factors of the VEGF family and whether it thus might have a more general role in the regulation of angiogenesis. The sequence homology between the COOH-terminal parts of VEGF₁₆₅ and VEGF-B₁₆₇ and the observation that VEGF-B₁₆₇ is sequestered into the pericellular matrix upon its secretion from cells suggested that, like VEGF₁₆₅ and PlGF-2, also VEGF-B might interact with NRP1. In this study we have carried out binding experiments using both soluble extracellular and transmembrane NRP1 proteins and show that NRP1 provides a receptor for VEGF₁₆₇ and for the proteolytically processed, but not for the full-length, form of VEGF₁₈₆.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Cell Culture and Materials-- Drosophila Schneider 2 (S2) cells were grown in DES^TM medium according to the supplier (Invitrogen). 293-T, 293-EBNA, and COS cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. PAE (a kind gift from Dr. Lena Claesson-Welsh) and PAE-NRP1 cells were grown in Ham's F-12 medium supplemented with 10% fetal calf serum. Anti-VEGF (mAb 293), rhVEGF₁₆₅, and rhVEGF₁₂₁ were purchased from R&D systems, anti-Myc antibodies (mAb 9E10) were from BABCO, and recombinant mVEGF₁₆₄ was kindly provided by Dr. Herbert Weich (Braunschweig, Germany). The affinity purified polyclonal rabbit anti-mouse NRP1 antibodies were a kind gift from Dr. Hajime Fujisawa (Nagoya, Japan). VEGF-B antibodies were affinity purified from rabbit antisera made against an NH₂-terminal peptide of mVEGF-B (20).

Construction of NRP1-Ig and NRP1-Myc Expression Vectors-- The extracellular domain of mouse NRP1 (248-2914 bp of mNRP1 cDNA, GenBank^TM accession number D50086, a kind gift from Dr. Fujisawa) was cloned into the pIgplus expression vector (Ingenius). A fragment encoding the COOH-terminal part (2738-2914 bp) of the extracellular domain was first ligated into the vector as an EcoRV-BamHI fragment and the 5' part (248-2738 bp) was then added as an EcoRV fragment (the 5'-EcoRV site derived from the pBluescript KS II vector). The sequence of the EC domain was thus in-frame with the sequences encoding the human IgG Fc part to allow the production of a dimeric NRP1-Ig fusion protein. Oligonucleotide 5'-TACAGGATCCCGAACAAAAACTCATCTCAGAAGAGGATCTGAATAGCTAAGAATTCACTT-3' and its complementary strand oligonucleotide encoding the Myc-peptide EQKLISEEDL followed by a stop-codon were annealed and ligated as a BamHI-EcoRI fragment into the pcDNA3.1(+)zeo vector (Invitrogen). A HindIII-BamHI fragment from NRP1/pIgplus (248-2738 bp of NRP1 cDNA) was then ligated into pcDNA3.1-Myc in frame with the sequence encoding the Myc-peptide.

Generation of VEGF-B Mutants-- The VEGF-B₁₈₆ mutants having a stop codon after every 15 nucleotides downstream of the exon 1-5 encoding sequence (nucleotides 1-502) and the single amino acid substitution mutant Arg-127  Ser were generated using single-stranded mutagenesis as described previously (20). A vector producing only the proteolytically processed cystine knot fragment of mVEGF-B₁₈₆ (1-127 amino acids) was amplified polymerase chain reaction by using primers 5'-TACAGGATCCGGCACCATGAGCCCC-3' and 5'-TACAGAATTCTTAGCGGGGCTGGGGAC-3', digested with BamHI and EcoRI, and ligated into pcDNA3.1(+)zeo vector (Invitrogen).

Transfections, Immunoprecipitation, and Soluble Receptor Binding-- 293-T cells were transfected with plasmids encoding the fusion proteins VEGFR-1-Ig, VEGFR-3-Ig or NRP1-Ig, or the NRP1-Myc protein, by using the calcium phosphate precipitation method, and conditioned medium was collected after a 24-32-h serum starvation. 293-T, COS, or 293-EBNA cells were similarly transfected with plasmids encoding the hVEGF₁₆₅, mVEGF-B₁₆₇, mVEGF-B₁₈₆, mVEGF-B_Ex1-5, or hVEGF-CNC proteins and metabolically labeled with 100 µCi/ml Pro-mix^TM L-³⁵S (Amersham Pharmacia Biotech) for 3-7 h beginning 24 h after transfection. Ten µg/ml heparin was added to the labeling medium to facilitate the release of heparan sulfate-binding growth factors (VEGF₁₆₅ and VEGF-B₁₆₇) from the cell surface and pericellular matrix. Metabolically labeled media (except from the VEGF transfection) were immunodepleted of endogenous VEGF and possible heterodimers by immunoprecipitating with monoclonal anti-VEGF antibodies. Equal aliquots of media containing the metabolically labeled growth factors were then used for immunoprecipitation or for receptor binding analysis. Growth factors were incubated for 3 h at room temperature with protein A-Sepharose bound receptor Igs in the binding buffer (0.5% BSA, 0.02% Tween 20). For competition studies, 2 µg of recombinant human VEGF₁₆₅ was added to the binding reaction. The Sepharose beads were then washed three times with ice-cold phosphate-buffered saline (PBS), 1% Triton X-100 and twice with PBS, and the bound proteins were analyzed in 15% SDS-polyacrylamide gel electrophoresis under reducing or nonreducing conditions.

Production and Purification of GST-Fusion Proteins-- Exon 6B (encoding amino acids 117-161) of mVEGF-B167 was amplified from the cDNA by polymerase chain reaction using the following primers: 5'-ACGTAGATCTAGCCCCAGGATCCTC-3' and 5'-ACGTGAATTCTCACCTACAGGTTGCTGG-3'. The amplified fragments were digested with the appropriate enzymes and ligated into the BamHI- and EcoRI-sites of pGEX-2T (Amersham Pharmacia Biotech). The construct was verified by sequencing and transformed into the Escherichia coli strain BL-21, and the GST fusion proteins were produced and purified according to the instructions of the manufacturer and dialyzed against PBS overnight. Glutathione-agarose-purified GST-exon 6B fusion protein was applied to a Hi-Trap^TM heparin-Sepharose affinity column (5 ml, Amersham Pharmacia Biotech), washed extensively, and eluted using a linear salt gradient of 0.15-1.5 M NaCl in 20 mM sodium-phosphate buffer, pH 7.2.

Production and Partial Purification of VEGF-B₁₆₇-- Full-length human VEGF-B₁₆₇ was amplified by polymerase chain reaction using the primers 5'-CGCCAGATCTCCTGTCTCCCAGCCTGAT-3' and 5'-GCGAATTCTCACCTTCGCAGCTTCCGGC-3' and subcloned into the BglII-EcoRI sites downstream of the signal sequence of the pMT/Bip/V5-HisC vector (Invitrogen). S2 cells were transfected with the resulting plasmid, and stable clones were selected according to the instructions of the manufacturer (Invitrogen). The expression of VEGF-B₁₆₇ was induced with 500 µM CuSO₄, and the medium was harvested 6 days thereafter. 100 ml of medium, centrifuged for 15 min at 10000 rpm and passed through a 0.45-µm Millipore filter, was chromatographed on a Resource S (Amersham Pharmacia Biotech) column equilibrated in 15 mM sodium phosphate buffer, pH 7.0, and eluted with a linear salt gradient (0-1 M NaCl) at a flow rate of 1 ml/min. The fractions containing VEGF-B were pooled and diluted to a final salt concentration of 0.3 M NaCl and then passed through a heparin-Sepharose column (1 ml of Hi-Trap heparin) and eluted using a linear salt gradient 0.6-1.5 M NaCl. The fractions were analyzed by Western blotting using specific antisera against human VEGF-B.

Competition of ¹²⁵I-hVEGF₁₆₅ Binding to NRP1 by GST-VEGF-B₁₆₇ Exon 6B-- Human recombinant VEGF₁₆₅ was radiolabeled to a specific activity of 2.5 × 10⁵ cpm ¹²⁵I/ng using the Iodogen reagent, as described earlier (20). Confluent PAE- and PAE/NRP1 cells seeded in 24-well plates were washed once with ice-cold binding buffer (F-12 medium, 0.5 mg/ml BSA, 20 mM Hepes, pH 7.4) and 0.5 ng/ml of labeled VEGF₁₆₅ with or without rhVEGF₁₆₅ (250 ng/ml), rhVEGF₁₂₁ (200 ng/ml), GST (25 mg/ml), or GST-exon6B (25 mg/ml) proteins were added in binding buffer. The cells were washed and lysed as described (20), and the radioactivity was measured in a -counter.

Purification of NRP1-Ig and Binding of Soluble NRP1-Ig to GST-exon6B-- NRP1-Ig was purified from the conditioned medium of transfected 293-T cells using protein A-Sepharose. The Ig fusion protein was eluted from the Sepharose using 0.1 M glycine, pH 3.0, neutralized with 1 M Tris, pH 8.0, and dialyzed against PBS overnight. 10 µg of purified GST or GST-exon6B was coated onto enzyme-linked immunosorbent assay plates (Nunc Maxisorp surface) for 90 min at room temperature. The protein solution was removed, and the coated wells were blocked with 5% BSA in PBS for 30 min. The plates were then washed three times with 0.5 mg/ml BSA in PBS and incubated with soluble NRP1-Ig (or VEGFR-3-Ig) at a concentration of 1 µg/ml. The binding was allowed for 2 h at room temperature. The plates were washed three times as above before addition of anti-human Ig conjugated with horseradish peroxidase (dilution 1:500). The anti-human Ig antibody was left for 40 min in the wells after which the plates were washed as above, including one additional wash with PBS. Then the substrate 1,2-phenylenediamine dihydrochloride was added according to the supplier (DAKO), and the reaction was stopped by addition of an equal volume of 0.5 M H₂SO_4, followed by the reading of the absorbance at 450 nm.

Mass Spectrometric Analysis of the Processed VEGF-B₁₈₆-- The VEGF-B₁₈₆ was isolated by SDS-PAGE and visualized by modified silver staining (21). After in-gel digestion using endoproteinase LysC from Achromobacter lyticus (Wako Chemicals GmbH, Neuss, Germany) (22), the peptide mixture was desalted and concentrated on a reversed phase support packed in a thin pipette tip. Analysis by MALDI-TOF mass spectrometry was done on a Bruker Biflex III instrument, using -cyano-4-hydroxycinnamic acid as matrix to prepare the sample as a "dried droplet." Calibration was done externally with a known peptide mixture.

Immunohistochemical Detection of NRP1-- Immunohistochemical detection of the NRP1 protein was performed on cryostat sections of day 12.5 embryos using the polyclonal affinity purified anti-mouse NRP1 antibodies. The staining was carried out as described previously (11, 13) except for the fixation of the sections, which was done in 2% paraformaldehyde for 15 min. Negative controls were done by omitting the primary antibodies.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

VEGF-B₁₆₇ and the Processed Form of VEGF-B₁₈₆ Bind to Soluble NRP1 Extracellular Domain-- To analyze VEGF-B binding to NRP1, we used expression vectors encoding the NRP1 extracellular domain fused to the IgG Fc domain or to the Myc epitope. Metabolically labeled media from transfected 293-T cells containing hVEGF₁₆₅, mVEGF-B₁₆₇, mVEGF-B₁₈₆, or mVEGF-B_Ex1-5 were incubated with protein A-Sepharose-bound NRP1-, VEGFR-1-, or VEGFR-3-Ig fusion proteins. VEGF-B₁₆₇ was precipitated by both NRP1-Ig and VEGFR-1-Ig (Fig. 1A). Interestingly, only the proteolytically cleaved form of VEGF-B₁₈₆ was bound to NRP1-Ig, whereas both the full-length and the processed forms were precipitated by VEGFR-1-Ig. VEGF-B_Ex1-5, consisting of the sequences encoded by exons 1-5 of the VEGF-B gene, which are present in both VEGF-B isoforms, was precipitated selectively by VEGFR-1-Ig but not by NRP1-Ig. This indicated that the NRP1-binding sites are different from the VEGFR-1 binding epitopes and that the former are located in the COOH termini of the two splice isoforms. Also, monomeric Myc-tagged NRP1 bound VEGF₁₆₅, VEGF-B₁₆₇, and the processed VEGF-B₁₈₆ fragment in a specific manner. When approximately equal amounts of NRP1-Ig and NRP1-Myc were compared for their ability to bind to and to precipitate labeled VEGFs, no significant differences could be detected (data not shown). This suggests that NRP1 does not need to be a preformed dimer for binding, or alternatively NRP1 extracellular domains can by themselves form dimers before ligand binding.

View larger version (43K): [in this window] [in a new window]   Fig. 1.   Precipitation of VEGF-B using soluble NRP1-Ig and VEGFR-Ig fusion proteins. A, schematic exon structures of the two VEGF-B cDNA variants encoding the 167- and 186-amino acid residue isoforms. Shown below are results of gel electrophoresis in reducing conditions of the metabolically labeled VEGF-B167, VEGF-B186, or VEGF-BEX1-5 proteins precipitated using the indicated fusion proteins or specific antibodies (IP). VEGF-B186 undergoes proteolytic processing, and only the processed NH2-terminal cystine knot growth factor fragment (arrow, see Ref. 20) binds to both NRP1 and VEGFR-1. In contrast, VEGF-BEx1-5, the fragment common for both isoforms, binds VEGFR-1 but not NRP1. B, gel electrophoresis in nonreducing conditions indicates weak NRP1 binding of the heterodimers between the processed and full-length forms of VEGF-B186 (FL+PF). The binding of both VEGF-B167 and VEGF-B186 is inhibited in the presence of 2 µg of recombinant human VEGF165. The presence of the two bands in VEGF-B167 (*) has been discussed elsewhere (20).

Analysis of the bound proteins in nonreducing conditions revealed that dimers of full-length VEGF-B₁₈₆ polypeptides did not bind to NRP1, in contrast to the dimers between full-length and processed forms of VEGF-B₁₈₆ (Fig. 1B). However, on the basis of the band intensities, heterodimer binding appeared much weaker than the binding of homodimers of the processed forms. The binding of both VEGF-B₁₆₇ and processed VEGF-B₁₈₆ to NRP1 was competed with an excess of recombinant human VEGF₁₆₅ (Fig. 1B), indicating that the interaction sites of VEGF and VEGF-B on NRP1 are at least overlapping, if not identical.

VEGF-B₁₆₇ Interacts with NRP1 through the Exon 6B-encoded Heparin Binding Domain-- The exon 6B encoded sequence of VEGF-B₁₆₇ resembles the heparin and NRP1-binding domain of VEGF₁₆₅, encoded by exon 7. To test the involvement of exon 6B in heparin and NRP1 binding, a fusion protein between glutathione S-transferase and mouse VEGF-B exon 6B-encoded sequence (GST-exon6B) was constructed. The fusion protein eluted from heparin-Sepharose at around 0.8 M NaCl, similarly to the wild type full-length human VEGF-B₁₆₇, indicating that the sequence encoded by this exon is probably responsible for the heparin binding ability in VEGF-B (Fig. 2, A and B). Furthermore, the GST-exon6B, but not GST or VEGF₁₂₁, competed for iodinated VEGF₁₆₅ binding on PAE-NRP1 cells (Fig. 2C). Also, immobilized GST-exon6B and VEGF₁₆₅ interacted with the soluble NRP1-Ig protein (Fig. 2D and data not shown).

View larger version (30K): [in this window] [in a new window]   Fig. 2.   VEGF-B167 interacts with NRP1 through the heparin binding domain. A, ion exchange chromatography of the GST-exon 6B fusion protein in a heparin-Sepharose column. An aliquot from the peak fraction of eluted GST-exon 6B was analyzed by silver staining (inset). B, elution at around 0.8 M NaCl was similar to the wild type full-length VEGF-B167. C, competition of iodinated VEGF165 binding to NRP1 expressing PAE-cells by GST, GST-exon 6B, mVEGF164, or hVEGF121 proteins. D, binding of NRP1-Ig by immobilized GST or GST-exon 6B.

Proteolytic Processing Regulates the Ability of VEGF-B₁₈₆ to Bind NRP1-- Plasmin cleavage of either VEGF or VEGF-B yields an NH₂-terminal cystine knot domain which contains the VEGFR binding epitopes, but plasmin-treated VEGF-B₁₈₆ expressed in COS-cells did not bind NRP1 (Fig. 3A). In addition, neither VEGF-B_Ex1-5 encoded by exons 1-5 nor the full-length VEGF-B₁₈₆ bound to NRP1 although the proteolytically processed VEGF-B₁₈₆ did. Because of the small size difference between VEGF-B_Ex1-5 and processed VEGF-B₁₈₆, the proteolytic cleavage site in VEGF-B₁₈₆ apparently occurred in the beginning of the exon 6A-encoded sequence (20). To map the cleavage site and the NRP1 interaction sites, a number of mutants were generated, introducing stop codons at every fifth amino acid residue after the common NH₂-terminal domain of 115 amino acid residues (Fig. 3B). Careful analysis of the metabolically labeled, immunoprecipitated VEGF-B mutants revealed that proteolytic processing first occurred in the Pro-130 mutant that contains an additional 15 residues after the exon 5-encoded sequence, and evidence of processing was even more visible in the longer mutants (Fig. 3B, left panel). All of the mutants were able to bind VEGFR-1 (data not shown), but NRP1 binding occurred only for the Pro-130 and longer mutants, and only the proteolytically cleaved forms of these mutants seemed to bind NRP1 (Fig. 3B, right panel). This suggested that the cleavage site is between residues 125-130 and that the binding to NRP1 requires 10-15 residues COOH terminal of the common, exon 1-5-encoded domain.

View larger version (31K): [in this window] [in a new window]   Fig. 3.   Proteolytic processing regulates VEGF-B186 binding to NRP1. A, binding of a plasmin digest of radiolabeled VEGF-B186 to soluble NRP1-Ig and VEGFR-1-Ig fusion proteins. B, schematic drawing of the VEGF-B186 mutants which were used to map the proteolytic cleavage site and the NRP1 interaction site. Metabolically labeled VEGF-B186 mutants expressed in 293-T cells were precipitated using specific antibodies (IP) or with NRP1-Ig fusion protein (NRP1) and analyzed under reducing (R) or nonreducing (NR) conditions. Proteolytic processing was seen in the mutants Pro-130 through Ile-145, and the proteolytically cleaved form of the same mutants showed also NRP1 binding.

Identification of the Proteolytic Cleavage Site in VEGF-B₁₈₆-- To more accurately define the site of proteolytic cleavage, VEGF-B₁₈₆ expressed in 293-T cells was purified from the conditioned medium using affinity chromatography employing NH₂-terminal VEGF-B antibodies cross-linked to protein A-Sepharose. The bound VEGF-B fragment was eluted by boiling in electrophoresis sample buffer. The proteolytically processed form was isolated from a silver-stained SDS-polyacrylamide gel and treated with LysC endoproteinase, and the digest was subjected to MALDI-TOF mass spectrometry analysis. Among other peptide masses found, one was 1676.02 Da (M+H⁺, monoisotopic), which corresponds to (K)PDRVAIPHHRPQPR (theoretical mass 1675.92 Da). This peptide would not appear in proteolytic digests with the highly specific LysC endopeptidase unless Arg-127 was the COOH-terminal amino acid in the processed VEGF-B₁₈₆. When the Arg-127 was mutated into a Ser residue and the mutant was expressed in 293-T cells, proteolytic processing as well as NRP1 binding were abolished, indicating that the NRP1 binding site is masked in the full-length VEGF-B₁₈₆ form but is accessible in the processed form (Fig. 4A). The reason for the slower mobility of the mutant polypeptide in comparison with the wild type VEGF-B₁₈₆ is not known, but it might be caused by additional glycosylation. Furthermore, when only the VEGF-B₁₂₇ polypeptide corresponding to the processed form was expressed as a recombinant protein, it was found to bind NRP1 (Fig. 4B). This data confirms the cleavage site at position Arg-127 and the requirement of a cleaved accessible COOH terminus for NRP1 interaction.

View larger version (24K): [in this window] [in a new window]   Fig. 4.   Proteolytic processing of VEGF-B186 occurs between Arg-127 and Ser-128. A, the top part shows a schematic drawing of the Arg-127 to Ser mutant (RSSS). Metabolically labeled wild-type, Ex1-5 and RSSS mutant proteins and the processed VEGF-B186 were precipitated with specific antibodies (IP) or with NRP1-Ig or VEGFR-1-Ig fusion proteins and analyzed under nonreducing conditions. B, comparison of the mobilities and NRP1 binding properties of the processed NH2-terminal fragment of VEGF-B186 and the corresponding recombinant VEGF-B127 form (1-127 amino acids). The unprocessed RSSS mutant does not bind NRP1 (panel A) whereas the VEGF-B127 form does (panel B). In nonreducing conditions, the mobility of the exon 1-5-encoded polypeptide dimer was slower than that of proteolytically processed VEGF-B186, probably because of differential folding. WT, wild type.

The masking of the NRP1 epitope in the full-length VEGF-B₁₈₆ could be because of protein folding or because of a sterical hindrance by the O-linked glycans of the COOH terminus. On the basis of the heterogeneous mobility of the mutant polypeptides of increasing length, the first indication of O-glycosylation was detected in the Ser-140 polypeptide (Fig. 3C). To test the effect of the O-linked glycans on NRP1 binding, full-length VEGF-B₁₈₆ was deglycosylated by neuraminidase or with a combination of neuraminidase and O-glycosidase. However, the resulting deglycosylated polypeptides did not acquire NRP1 binding ability (data not shown), which suggests that the glycans are not responsible for the masking of the NRP1 binding epitope in the full-length VEGF-B₁₈₆.

Paracrine and Autocrine Expression Patterns of NRP1 and VEGF-B during Development-- To study the relationship of NRP1 and VEGF-B expression in embryonic cells and tissues, we used specific polyclonal antibodies in immunoperoxidase staining of sections of day 12.5 mouse embryos. As shown in Fig. 5, A and B, the anti-NRP1 antibodies stained developing myocardial cells originating from the mesoderm in the developing heart, but gave no staining of the epicardium or of the endocardial cushion tissue, which is derived from the embryonic neural crest. Interestingly, the endothelia and the surrounding smooth muscle cell layer of several major and minor vessels such as the cerebral artery and its branches (Fig. 5D), intersomitic vessels, and the vasculature of the choroid plexus (Fig. 5E) gave very strong immunostaining with the anti-NRP1 antibodies. As is evident from previously published papers, NRP1 was also expressed in the neural tissues of the developing brain, for example in the cerebellum and in the hippocampus and along the developing spinal cord (data not shown; for review, see Ref. 11). Overlapping expression patterns thus exist with VEGF-B, which is especially abundant in cardiac myocytes of the developing heart, in the choroid plexus, and in the smooth muscle cells of large arteries as well as in developing muscle, bone, and cartilaginous structures (23).

View larger version (112K): [in this window] [in a new window]   Fig. 5.   Expression of NRP1 in day 12.5 embryos. NRP1 expression was detected by immunohistochemistry using polyclonal affinity purified anti-NRP1 antibodies. A, NRP1 is expressed in the developing heart. B, higher magnification of the heart shows that NRP1 is expressed both in the developing myocardium and in the endocardium, but not in the endocardial cushion tissue. C, no staining was detected in the negative control done by omitting the primary antibody. Strong NRP1 expression was also detected in endothelial cells of capillaries and blood vessels, for example in cerebral artery (D) and in vasculature of choroid plexus (E). h, heart; lu, lung; li, liver; a, atrium; v, ventricle; ec, endocardiac cushion; my, myocardium. Scale bar, 100 µm.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

In this study we demonstrate that VEGF-B binds to NRP1, a receptor for collapsins/semaphorins, which has been shown to function in guiding neuronal growth cones and as an isoform-specific receptor for two angiogenic factors of the VEGF-family, VEGF₁₆₅ and PlGF-2. Both VEGF-B splice variants, the heparin-binding VEGF-B₁₆₇ and soluble VEGF-B₁₈₆ forms, were able to bind to NRP1, but the binding of VEGF-B₁₈₆ was regulated by proteolytic processing. Only the processed form lacking the O-glycosylated COOH terminus was able to bind NRP1, whereas in VEGF-B₁₆₇ the heparin binding domain encoded by exon 6B was responsible for binding. The exon 6B-encoded domain of VEGF-B₁₆₇ is highly homologous to the NRP1 binding epitopes of the VEGF₁₆₅ and PlGF-2 isoforms, which also bind heparin. In contrast, in VEGF-B₁₈₆ an alternative splice acceptor site in exon 6 generates an insertion and a frameshift in the coding sequence; therefore the COOH terminus of VEGF-B₁₈₆ differs from the heparin binding domain of VEGF-B₁₆₇ (16). In the processed form of VEGF-B₁₈₆, the peptide encoded by the NH₂ terminus of exon 6A thus constitutes a novel NRP1 binding epitope. VEGF-B₁₈₆ does not seem to have affinity for heparin and the binding to NRP1 was not heparin-dependent, yet in the presence of heparin the binding appeared stronger.² Because the processing site was mapped to Arg-127, the specific NRP1 binding epitope constitutes minimally only 12 amino acid residues, with the COOH-terminal Arg residue probably being important for efficient binding.

Proteolytic processing is commonly used to regulate the bioavailability and activity of members of the platelet-derived growth factor (PDGF)/VEGF family. Both PDGF isoforms, PDGF-A and -B, are proteolytically processed after biosynthesis (24). Also VEGF-C and VEGF-D are sequentially processed at sites flanking the cystine knot growth factor domain. Furthermore, proteolytic processing of VEGF-C regulates its affinity to the VEGFR-3 and VEGFR-2 receptors (25). Also, proteolytic cleavage of VEGF-B₁₈₆ appeared to result in an increased VEGFR-1 affinity (20). Activation of NRP1 binding by proteolytic cleavage could further increase the VEGFR-1 binding affinity and thus modulate VEGF-B₁₈₆-induced biological effects, including regulation of protease activities in endothelial cells (20). The glycans of the COOH terminus were not responsible for the masking of the NRP1 binding epitopes in the full-length VEGF-B₁₈₆, as deglycosylation of the COOH terminus did not lead to acquisition of NRP1 binding ability. It should be noted that the NRP1 binding domains of VEGF₁₆₅, PlGF-2, and Sema III are highly basic (7). The exposure of a basic peptide sequence (RVAIPHHRPQPR) in VEGF-B₁₈₆ upon proteolytic cleavage might be enough to give the protein an ability to bind NRP1. The binding of the processed form of VEGF-B₁₈₆ by NRP1 was, however, weaker than that of VEGF-B₁₆₇ or VEGF₁₆₅. Approximately 20% of the immunoprecipitated proteolytically processed VEGF-B₁₈₆ protein was precipitated with NRP1 in the absence of heparin. In contrast, both specific antibodies and soluble NRP1-Ig fusion protein precipitated approximately equal amounts of VEGF-B₁₆₇ or VEGF₁₆₅.

NRP1 and VEGF-B were prominently co-expressed for instance in the myocardial cells of the developing heart and in the smooth muscle cell layers surrounding larger blood vessels (13, 23, and the this paper). In addition, NRP1 detected in endothelial cells of large vessels is located in juxtacrine relationship with the smooth muscle cell layer producing VEGF-B (23). It has been reported that NRP1 and the related NP-2 can form homo- and hetero-oligomers through the interaction between MAM domains, they bind different but overlapping subset of semaphorins, and they mediate chemorepulsion of axons of different classes of neurons (26-29). However, although NP-2 also binds VEGF₁₆₅, the expression pattern of NP-2 is different from that of NRP1 or VEGFR-1, e.g. being absent from endothelial cells of capillaries (26, 28).

The exact role of NRP1 binding in the biology of VEGFs is unknown, but NRP1 has been reported to enhance the mitogenic effects of VEGFR-2 upon VEGF₁₆₅ stimulation when the two receptors are coexpressed in transfected endothelial cells (6). In addition to VEGFR-2, NRP1 has been suggested to function as a ligand binding subunit of a transmembrane receptor(s) that mediate specific signals for different semaphorins (29, 30). Thus, NRP1 appears to regulate other signal-transducing receptors, and no intrinsic signaling may occur via NRP1 itself. Although the high conservation of the short cytoplasmic domain of NRP1 suggests importance of this part to NRP1 function, no binding partners or obvious protein homology domains have been reported within this portion. In fact, the cytoplasmic tail was shown to be unnecessary for the Sema III-induced biological responses via NRP1. The presence of Sema III binding a1/a2 (CUB) domain and the MAM domain responsible for NRP1 oligomerization and possible interaction with other transmembrane signal-transducing proteins were sufficient for Sema III-induced growth cone collapse (27, 29). It is not known whether VEGF binding is sufficient for signal transduction via a NRP1-associated hypothetical signal-transducing polypeptide chain although the enhanced mitogenic signaling of VEGFR-2 in NRP1-overexpressing cells suggests such a possibility.

According to the present results, VEGF-B, in analogy to the Sema III, could therefore induce biological responses in cells which express NRP1, but no VEGFRs. In addition, NRP1 might potentiate the effects of VEGF-B in cells where VEGFR-1 is expressed. Furthermore, as VEGF and Sema III have been shown to compete for NRP1 binding (28), and as we here show competition between VEGF and VEGF-B, there could be a more extensive cross-regulation of cellular signals between the two families of growth factors.

	ACKNOWLEDGEMENTS

We thank Marko Hyytiäinen for help with the fast protein liquid chromatography and Tapio Tainola and Paula Hyvärinen for expert technical assistance. Dr. Hajime Fujisawa is kindly acknowledged for providing the mouse NRP1 cDNA and anti-NRP1 antibodies.

	FOOTNOTES

^* This study was supported by grants from the University of Helsinki, the Finnish Cancer Organizations, the Academy of Finland, the Helsinki University Hospital (TYH 8105), the Sigrid Juselius Foundation, and the Swedish Medical Research Council (K99-03P-12070-03C) and by Karolinska Institutet.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^§ These authors contributed equally to the work.

To whom correspondence should be addressed: Molecular/Cancer Biology Laboratory, Haartman Institute, P. O. Box 21 (Haartmaninkatu 3), University of Helsinki, FIN-00014 Helsinki, Finland. Fax: 358-9-1912 6448; E-mail: Kari.Alitalo@Helsinki.FI.

² T. Makinen, B. Olofsson, U. Eriksson, and K. Alitalo, unpublished data.

	ABBREVIATIONS

The abbreviations used are: VEGF, vascular endothelial growth factor; NRP1, Neuropilin-1; VEGFR,VEGF receptor, PlGF, placenta growth factor; MALDI-TOF, matrix-assisted laser desorption ionization/time of flight; mAb, monoclonal antibody; bp, base pair(s); mVEGF, mouse VEGF; rhVEGF, recombinant human VEGF; BSA, bovine serum albumin; PBS, phosphate-buffered saline; GST, glutathione S-transferase; PDGF, platelet-derived growth factor; PAE, porcine aortic endothelial (cell).

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES