Functional regulation of semaphorin receptors by proprotein convertases.

PLEXIN genes encode receptors for secreted and membrane-bound semaphorins. It was proposed that the extracellular domain of plexins acts as an inhibitory moiety, preventing receptor activation. Here we show that plexin-B1 and plexin-B2 undergo proteolytic processing in their extracellular portion, thereby converting single-chain precursors into non-disulfide-linked, heterodimeric receptors. We demonstrate that plexin processing is mediated by subtilisin-like proprotein convertases, by inhibition with alpha1-antitrypsin Portland, and by mutagenesis of the substrate-cleavage sites. We provide evidence indicating that proprotein convertases cleave plexins in a post-Golgi compartment and, likely, at the cell surface. In addition, we find that both cell surface targeting and proteolytic processing of plexin-B1 depend on protein-protein interaction motifs in the cytoplasmic domain of the receptor. We then show that proteolytic conversion of plexin-B1 into a heterodimeric receptor greatly increases the binding and the functional response to its specific ligand semaphorin 4D/CD100. Thus, we conclude that cleavage by proprotein convertases is a novel regulatory step for semaphorin receptors localized at the cell surface.

Plexins function as cell surface receptors for all classes of semaphorins, either alone or in complex with neuropilins (Refs. [1][2][3][4]reviewed in Ref. 5). Semaphorins include secreted and transmembrane proteins that act as repulsive cues for axon guidance and are furthermore implicated in a variety of functions, spanning from immune response to angiogenesis and tumor progression (reviewed in Refs. 5 and 6). The human plexin family contains at least nine members, classified into four distinct subfamilies based on sequence similarity (1). In addition, although A-subfamily plexins are predominantly expressed in the developing nervous system, plexin-B1 (the Bsubfamily prototype and high-affinity receptor for semaphorin 4D) appears to be more ubiquitously distributed (7).
The extracellular domain of all plexins has features in com-mon with scatter factor receptors (MET gene family) and semaphorins (3). In addition to mediating ligand binding, it can associate with neuropilins (1,4) and scatter factor receptors (8).
It was reported (9) that deleting part of the extracellular domain of plexins results in a conformational change that activates receptor signaling. This implies that, in the absence of the ligand, the extracellular domain of plexins brings about a steric hindrance that inhibits receptor function. Interestingly, the extracellular domains of plexin-B1 and plexin-B2 contain a putative cleavage site for subtilisin-like proprotein convertases, located in the proximity of the transmembrane domain (1). Moreover, this site is phylogenetically conserved, because it is also found in fly plexin B (1). The family of subtilisin-like proprotein convertases (PCs) 1 includes furin and many other members (reviewed in Ref. 10). They are known to process a variety of transmembrane and secreted proteins, including scatter factor receptors and semaphorins, which are both phylogenetically related to plexins. For class 3 semaphorins, proteolytic cleavage was shown to regulate the axon-repelling activity (11). It has not yet been shown whether plexins B actually undergo proteolytic processing in cells. If this were true, they could either be converted into heterodimeric receptors or their extracellular domain could be released into the extracellular space. The intracellular domain of plexins is extremely well conserved within the family and across evolution and contains stretches that are distantly related to GTPase activating proteins (GAPs, Ref. 12). Furthermore, specific sequences in the cytoplasmic domains of plexins of the B subfamily are responsible for binding activated Rac1 (13)(14)(15) and PDZ domaincontaining proteins (16 -18).
In this study we have shown that plexin-B1 and plexin-B2 are found in cells and tissues in a heterodimeric form because of proteolytic cleavage by PCs. This event appears to require receptor localization at the cell surface, which in turn is regulated by sequences in the cytoplasmic domain. Finally, we have shown that the proteolytic processing of plexins by PCs significantly increases ligand binding and functional response.

EXPERIMENTAL PROCEDURES
cDNA Expression Constructs-Plexin-B1 expression construct (in pcDNA3 vector, Invitrogen) includes an in-frame VSV tag immediately after the signal peptidase site. Plexin-B2 expression construct (in pMT2 expression vector) includes a VSV tag at the COOH terminus of the protein. The secreted form of plexin-B2 (B2-EC) contains the whole extracellular domain of plexin-B2 (amino acids 1-1190) fused to glutathione S-transferase (GST) at the COOH terminus by cloning into the expression vector pMT2-GST (a kind gift from D. Shaap). Human PLEXIN-B1 and PLEXIN-B2 cDNAs were mutated in their subtilisinlike substrate-processing sites (R 1302 RRR 3 AAAA and R 1161 QKR 3 AQKA, respectively) to generate uncleavable forms by virtue of a recombinant PCR-based approach (19). The mutated sequences were verified by DNA sequencing. For gene transfer experiments, PLEXIN-B1 and PLEXIN-B1-uncleav cDNAs were subcloned in the lentiviral transfer vector pRRLsin.cPPT.hCMV.Wpre (kindly provided by L. Naldini, University of Torino).
Cells, Transfections, and Drugs-Human HepG2, 293T, HT-29, and SW-48 cells, canine kidney MDCK and murine NIH-3T3 cell lines were purchased from ATCC. The human GTL-16 gastric carcinoma cell line and the Suit2 pancreatic cell line were described in Refs. 20 and 21, respectively. Cells were maintained in Dulbecco's modified Eagle's medium, supplemented with 2 mM L-glutamine and 10% fetal bovine serum (Invitrogen). 293T cells were maintained in Iscove's modified Dulbecco's medium (Sigma) supplemented with 10% fetal bovine serum. NIH-3T3 cells were maintained in Dulbecco's modified Eagle's medium with 10% heat-inactivated bovine serum. cDNA transfections in 293T cells were carried out by calcium phosphate precipitation. MDCK cell clones stably expressing PLEXIN-B1 cDNA were obtained by calcium phosphate transfection and selection with G-418 (Invitrogen). Pools of HepG2 and GTL-16 cells expressing ␣1-antitrypsin Portland and NIH-3T3 cells expressing plexin-B1 or B1-uncleav were obtained by transduction with lentiviral vectors according to published protocols (22).
The inhibitor of PCs decanoyl-RVKR-chloromethylketone was purchased from BACHEM (CH). Monensin and Brefeldin were from Sigma.
Antibodies-IC2-specific polyclonal antibodies were raised in rabbits against a GST fusion protein containing almost the entire intracellular domain of human plexin-B1 (amino acids 1543-2135). EC6.9 monoclonal antibody, directed against the extracellular domain of plexin-B1, was obtained by DNA and recombinant protein immunization of mice. 2 Neither of these antibodies cross-reacts with plexin-B2. Monoclonal anti-VSV antibodies (clone P5D4) were purchased from Sigma.
Immunoprecipitation and Western Blotting-Cells were lysed on ice in EB buffer (20 mM Tris-HCl, pH 7.4, 5 mM EDTA, 150 mM NaCl, 10% glycerol, 1% Triton X-100) in the presence of a mixture of protease inhibitors (phenylmethylsulfonyl fluoride, leupeptin, aprotinin, pepstatin). Tissue samples were homogenized and lysed in EB buffer. Immunoprecipitations and Western blotting were performed according to standard protocols. Final detection was done with the ECL system (Amersham Biosciences). Silver staining of polyacrylamide gels was performed using the silver stain kit (Sigma).
Metabolic Labeling-Cells were starved for 1 h in serum-free medium (without cysteine and methionine) and pulsed from 20 min to 1 h in the presence of 200 Ci/ml [ 35 S]methionine/cysteine (Pro-mix, Amersham Biosciences), then washed, and chased for different times in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. Cells were lysed in EB buffer, and proteins were immunoprecipitated as described above and submitted to SDS-PAGE. After fixation, gels were incubated with 1 M Na-salicylate for 30 min at room temperature. After gel drying, detection was done by autofluorography or by a Phosphor-Imager system (Amersham Biosciences).
Labeling of Cell Surface Proteins by Biotinylation-Surface proteins were labeled for 30 min at 4°C using the ECL protein biotinylation module (Amersham Biosciences). After labeling, cells were rinsed twice with Dulbecco's modified Eagle's medium-1% bovine serum albumin to quench unreacted biotin and then lysed in EB buffer. Cell extracts were incubated with the appropriate antibodies or with immobilized streptavidin-agarose (Pierce). Protein samples were analyzed by SDS-PAGE and Western blotting; biotinylated proteins were revealed by peroxidase-linked streptavidin (avidin-HRP) and the ECL detection system.
Cell Surface Staining-NIH-3T3 cells expressing plexin-B1, plexin-B1-uncleav, or vector alone were seeded in 48-well dishes coated with poly(L)-lysine. The next day, cells were incubated with anti-VSV antibody diluted in culture medium for 1 h at 4°C. After washing, cells were fixed with 3.7% formaldehyde in phosphate-buffered saline for 20 min, and endogenous phosphatases were inactivated by incubation in a water bath at 65°C for 15 min. Cell-bound antibodies were then detected with alkaline phosphatase-conjugated anti-mouse antibodies (Molecular Probes) for 30 min at room temperature. Specifically bound alkaline phosphatase activity was eventually revealed by p-nitrophenylphosphate hydrolysis at 37°C in a reaction buffer containing 1 M Tris-HCl, pH 9.5, 1% bovine serum albumin, 1 mM MgCl 2 . Absorbance values were measured at a wavelength of 405 nm and corrected for background absorbance.
Semaphorin Binding and Collapse Assays-The Sema4D binding assay using NIH-3T3 cells expressing plexin-B1, B1-uncleav, or vector alone was performed as described (1,23). Briefly, cells seeded in 48-well dishes were incubated for 30 min with different concentrations of Sema4D fused to secreted alkaline phosphatase (SeAP). Cells were then washed three times, fixed with acetone-formaldehyde, and incubated at 65°C for 15 min to inactivate endogenous phosphatases. Cell-bound Sema4D was quantified by measuring SeAP activity on p-nitrophenylphosphate, as above. Scatchard plot analysis was performed using Equilibrate (by GertJan C. Veenstra).
The cell collapse assay was performed according to the protocol described in Ref. 4, incubating NIH-3T3 cells with Sema4D-SeAP for 30 min at 37°C, followed by incubation with NBT/BCIP (Promega). Stained cells were visualized with a Leica DMLB microscope, and images were captured using a Leica DC300F camera.

Plexin-B1 Is Predominantly Found in a Cleaved Form in
Cells and Tissues-PLEXIN-B1 encodes a large glycosylated protein, functioning as receptor for semaphorin 4D (Sema4D/ CD100) (1,7). Previous experiments have shown that plexin-B1 mRNA, unlike that of A subfamily members, is widely distributed in non-neuronal tissues (7). To further analyze the expression of plexin-B1 receptor in human cell lines and tissues, we raised specific polyclonal antibodies against its cytoplasmic sequence (IC2, described under "Experimental Procedures").
We first screened a number of cell lines derived from human tumors by Western blotting and detected plexin-B1 expression in GTL-16 (derived from gastric carcinoma), Suit2 (pancreatic adenocarcinoma), HepG2 (liver hepatoma), and HT-29 (colon adenocarcinoma) cell lines (Fig. 1A). In these cells, the IC2 antibody specifically recognized low amounts of a protein with a molecular mass of 300 kDa (most likely corresponding to the full-length receptor) together with a predominant smaller protein of molecular mass of 100 kDa that might correspond to a truncated form of plexin-B1 containing the cytoplasmic domain and a portion of the extracellular moiety. Control experiments with pre-immune serum did not reveal any bands (data not shown). We then analyzed normal human tissue samples (Fig. 1B) and confirmed that the predominant protein species recognized by the IC2 antibody is a 100-kDa polypeptide, whereas the p300 precursor protein is barely detectable. We ruled out that the 100-kDa fragment was generated because of post-lytic protein degradation, because it was also detected in protein extracts obtained by cell lysis in boiling 2.5% SDS (data not shown). We concluded that in vivo endogenous plexin-B1 is predominantly expressed as a truncated protein with a molecular mass of 100 kDa.
To confirm that the two proteins detected by IC2 were derived from plexin-B1, we analyzed the receptor in transfected cells. When lysates from 293T, MDCK, and NIH-3T3 cells expressing exogenous plexin-B1 were analyzed with IC2, we again detected two protein species (p300 and p100, Fig. 1C), identical to those found in cells and tissues. Moreover, results obtained with in vitro transcription/translation experiments from the same cDNA used in transfections ruled out the possibility that downstream ATG codons might be used for alternative translation initiation (data not shown). Therefore, we concluded that the truncated fragment of plexin-B1 likely derives from proteolytic processing.
Plexin-B1 and Plexin-B2 Are Synthesized as Single-chain Precursors and Proteolytically Processed into Non-disulfidelinked Heterodimers-We have previously reported (1) that the extracellular domain of plexin-B1 contains a tetrabasic amino acid sequence located in the proximity of the transmembrane domain (i.e. R 1302 RRR), which fits well with the consensus for cleavage by subtilisin-like PCs. Proteolytic processing in this site could therefore explain the consistent finding of cleaved plexin-B1 in cells and tissues. By analogy with scatter factor receptors, the proteolytic processing of plexin-B1 by PCs might lead to the conversion of a single-chain precursor into a heterodimeric receptor. This would result in an extracellular moiety (with a molecular mass of ϳ200 kDa, subunit ␣) and a transmembrane moiety (with a molecular mass of ϳ100 kDa, subunit ␤) that contains a small extracellular region and the plexin cytoplasmic domain ( Fig. 2A). In fact, we found that cells transfected with plexin-B1 express both the full-length receptor (p300) and two cleaved subunits (p200 and p100) at the cell surface (Fig. 2B). The identity of p100 as ␤ subunit was confirmed by detection with IC2 antibodies, whereas p200 ␣ subunit was identified by anti-VSV immunoblotting. In addition, we demonstrated the presence of plexin-B1 heterodimers on the surface of cells expressing endogenous receptors (Fig. 2C), using a monoclonal antibody directed against its extracellular domain (EC6.9). To confirm that the two subunits of plexin-B1 are associated in a complex, we show that the extracellular subunit is efficiently purified with an antibody (IC2) directed against the cytoplasmic domain of the receptor (Fig. 2D). In contrast to the plexin-B1 p300 precursor, the cleaved subunits were easily detectable at the cell surface (as demonstrated by surface biotinylation).
Plexin-B2, another member of the B subfamily of plexin receptors, also contains a polybasic sequence (R 1161 QKR) fitting with the consensus for PC-specific cleavage. Because an antibody directed against plexin-B2 was not available, we immunopurified an epitope-tagged form of plexin-B2 expressed in 293T cells. By Western blotting experiments using anti-VSV antibody (recognizing the COOH terminus of plexin-B2), we could detect two proteins with molecular mass of 240 and 80 kDa (Fig. 3A). Reminiscent of plexin-B1 precursor and ␤ subunit, p240 has the predicted size of a single-chain plexin-B2 precursor, and p80 may represent the plexin-B2 ␤ subunit. This would also imply the existence of an ␣ subunit of plexin-B2 at the cell surface. Indeed, by cell surface biotinylation, we could easily detect a protein with a molecular mass of ϳ170 kDa, likely corresponding to the cleaved ␣ subunit of plexin-B2, containing almost the entire extracellular domain (Fig. 3A). A large amount of p80 ␤ subunit is detected with anti-VSV antibodies among biotinylated cell membrane proteins, whereas only a minor fraction of uncleaved plexin-B2 is exposed at the cell surface (Fig. 3A, right panel). Similar to plexin-B1, we conclude that plexin-B2 is predominantly expressed at the cell surface as a heterodimer, whereas p240 probably represents the full-length precursor. Furthermore, to confirm that both p170 and p80 are derived from the proteolytic cleavage of p240 precursor, we carried out pulse-chase experiments. Fig. 3B shows that the plexin-B2 p240 precursor decreased during the chase, whereas p80 and p170 subunits accumulated with the same kinetics and could be co-purified. We concluded that p170 represents the ␣ subunit of plexin-B2, released by cleavage and yet associated with p80 ␤ subunit. In addition, we found that the receptor subunits are not covalently linked by disulfide bonds, as demonstrated by SDS-PAGE under non-reducing conditions of immunopurified plexin-B1 and plexin-B2 (Fig.  3C). The disulfide-linked heterodimeric receptor Met was included as a control (24).
Plexin Processing Is Caused by Site-specific Cleavage by Proprotein Convertases-Considering the presence of a specific consensus substrate-cleavage site for PCs in the extracellular domain of plexins B, we verified whether receptor processing could be inhibited by treating cells with convertase inhibitors. We first tested the PC-specific inhibitor decanoyl-RVKR-chloromethylketone (25). As shown in Fig. 4A, this inhibitor completely blocked the processing of both plexin-B1 and plexin-B2. Furthermore, we treated plexin-expressing cells with the PCinhibitor ␣1-antitrypsin Portland (␣1-PDX, Refs. 26 and 27). As shown in Fig. 4B, we observed a dramatic impairment of the proteolytic processing of both plexins upon co-expression with ␣1-PDX. We also observed increased amounts of the uncleaved precursors on the cell surface (data not shown). Analogously, ␣1-antitrypsin Portland inhibited receptor processing and formation of p100 in cells expressing endogenous plexin-B1 (Fig. 4C).
It is known that PCs require two basic charged residues in positions Ϫ1 and Ϫ4 with respect to the actual substratecleavage site, leading to the identification of the minimal consensus sequence RXXR (reviewed in Ref. 10). We have identified sequences in the extracellular domain of human plexin-B1 and plexin-B2 that fit this consensus and described a proteolytic processing consistent with the use of these sites. To confirm that plexin processing is caused by cleavage by PCs in the indicated positions, we mutated the specific consensus sequences in the extracellular domain of plexin-B1 and plexin-B2. As shown in Fig. 5, these mutated receptors, although normally targeted to the cell surface, cannot be proteolytically cleaved, indicating that PC-specific cleavage sites at the identified positions are absolutely required for processing. . Besides a barely detectable protein with a molecular mass of 300 kDa (p300) corresponding to full-length plexin-B1, we could detect a smaller protein with a molecular mass of 100 kDa (p100) corresponding to a truncated form of the receptor. C, human plexin-B1 was transfected into 293T, MDCK, and NIH-3T3 cells and analyzed by immunoblotting with IC2, as above.
Plexin-B2 Processing Depends on Protein Delivery at the Cell Surface-PC-dependent processing generally takes place during transit through the Golgi complex or within secretory vesicles en route to the cell membrane. However, proprotein convertases are also exposed on the plasma membrane and can be shed in the extracellular space or recycled into the cell (reviewed in Ref. 28). To determine whether the proteolytic cleavage of plexins-B takes place before or after transit through the Golgi apparatus, we treated plexin-B2-expressing cells with protein sorting inhibitors. Monensin A, which inhibits vesicular trafficking between Golgi apparatus and the plasma membrane, efficiently blocked the proteolytic processing of the receptor (Fig.  6A). Brefeldin A, which inhibits the trafficking between the endoplasmic reticulum and the Golgi apparatus, was also effective.
As a complementary approach, we expressed a secretable form of plexin-B2 extracellular domain, including its substratecleavage site, fused to glutathione S-transferase at the COOH terminus (B2-EC) (Fig. 6B). The molecule was efficiently processed and secreted, and two associated subunits accumulated in the conditioned medium, as demonstrated by silver staining and anti-GST Western blotting (Fig. 6B, left panels). Protein p170 (indicated by b) corresponds to the plexin-B2 ␣ subunit, and protein p30 (c) to the short extracellular sequence of the ␤ subunit fused to GST. The cleavage efficiency of B2-EC was comparable with that of the full-length plexin, inasmuch as some of it was found in the cell lysate as precursor (Fig. 6B,  right panels, a), probably because of protein overexpression. In contrast, we could not find any processed B2-EC protein intracellularly or associated with the cell surface, suggesting that the cleavage occurs after plexin delivery at the plasma membrane or immediately preceding secretion. Intriguingly, the presence of some uncleaved precursor bound to cell surface proteins suggests that proteolytic cleavage might induce a conformational change, leading to the secretion of the associated subunits into the medium. To demonstrate this assumption, we produced an uncleavable form of B2-EC by mutating the conserved substrate-cleavage site, analogously to what is shown in Fig. 5 for the full-size receptor. This uncleavable mutant remained attached to the cell surface and was absent from the medium (Fig. 6B, far right lanes).

Sequences in the Cytoplasmic Domain of Plexin-B1 Regulate Cell Surface Localization and Proteolytic Processing of the Receptor-It was reported recently that the COOH-terminal sequence of plexin-B1 associates with a PDZ-Rho-GEF (16 -18).
Here we show that a deleted receptor unable to interact with PDZ-Rho-GEF (B1⌬10, Ref. 16) is inefficiently targeted to the cell surface and is mostly found as a precursor protein (Fig. 6C). Notably, we observed that the deletion of the essential COOHterminal leucin residue in the plexin-B1 PDZ-binding consensus sequence, or the introduction of an unrelated COOH-terminal sequence, is not sufficient to block cell surface targeting of the receptor and consequently its proteolytic processing (data not shown).
Moreover, it has been shown that the cytoplasmic domain of plexin-B1 binds activated GTP-bound Rac1 (13)(14)(15). The interaction depends on the sequence located between the two conserved regions of the SP domain. Recently, it has been proposed that association with Rac-GTP regulates cell surface expression of plexin-B1 (29). Here we show that a mutated receptor unable to bind Rac-GTP (B1-GGA, Ref. 13) was virtually absent from the cell surface and accumulated intracellularly (Fig. 6C). Interestingly, the proteolytic cleavage of plexin-B1-GGA is totally abrogated, indicating that GTPase association is required for the physiological processing of plexin-B1 occurring on the plasma membrane.
Together these data indicate that the proteolytic cleavage of plexin-B1 by proprotein convertases is dependent on cell surface localization. We also show that specific sequences in the cytoplasmic domain of plexin-B1 are responsible for regulating cell surface targeting of the receptor, probably via interaction with activated Rac-1 and PDZ domain-containing proteins.
The Proteolytic Processing of Plexin-B1 Results in Increased Ligand Binding and Functional Response-To analyze the functional role of receptor processing, we expressed wild-type and uncleavable (B1-uncleav) forms of plexin-B1 in NIH-3T3 fibroblasts and tested their ability to interact with the specific ligand Sema4D. As shown in Fig. 7A, the wild-type receptor is largely present as a heterodimer, whereas the uncleavable plexin-B1 is found as single-chain precursor. The two receptor forms are expressed at comparable levels at the plasma mem- FIG. 2. Plexin-B1 cleaved subunits are associated in a complex. A, schematic representation of plexin-B1. The predicted processing site is indicated by an arrowhead. Epitopes recognized by antibodies EC6.9 and IC2, and the predicted molecular mass of the two subunits released by the cleavage (␣ and ␤), are also indicated. B, 293T cells transfected with plexin-B1 or control vector were surface-biotinylated. Proteins purified with anti-VSV antibodies were separated by SDS-PAGE, transferred to nitrocellulose, and decorated with avidin-HRP, followed by antibodies directed against the intracellular (IC2) or extracellular (anti-VSV) domains of the receptor. p300 precursor appears as a doublet; the upper band most likely corresponds to a post-translationally modified form of the receptor and is enriched at the cell surface. C, GTL-16, HepG2, and Suit2 cells expressing endogenous plexin-B1 and NIH-3T3 cells wild type or expressing the receptor were surface-biotinylated. Cell lysates were immunopurified using a monoclonal antibody raised against the extracellular domain of plexin-B1 (EC6.9) and further analyzed by avidin-HRP detection and IC2 immunoblotting. D, NIH-3T3 cells expressing plexin-B1 were labeled by surface biotinylation. Cell lysates were immunoprecipitated with IC2 antibodies recognizing the cytoplasmic domain of the receptor and analyzed as in panel C. brane, as quantified by surface staining with antibodies (Fig.  7B) and cell surface biotinylation (data not shown). However, when tested in a binding assay, the cells expressing the uncleavable receptor were significantly less efficient in binding Sema4D (Fig. 7C). By analyzing these data in a Scatchard plot, we calculated comparable affinity constants for wild-type and uncleavable receptors and an average 3-fold difference in B max values (at concentrations of ligand binding plateau). This indicates that the different binding ability of the uncleaved receptors should be attributed to a reduced number of functional binding sites at the cell surface.
Moreover, we observed a reduced ability of Sema4D to elicit a functional response in cells expressing the uncleavable receptor. This was determined by studying Sema-dependent cellular collapse of fibroblasts, by analogy to what has previously been described in other cell types (4,30). As shown in Fig. 7D, cells expressing processed plexin-B1 underwent cell contraction and rounding up upon treatment with Sema4D; in contrast, the single-chain receptor could not trigger an analogous response. In conclusion, we implicate the proteolytic processing of plexin-B1 in the formation of functional Sema4D binding sites at the cell surface. DISCUSSION We have previously reported that the human genome contains at least nine different plexin genes (1). Plexins belonging to the A subfamily are known to be predominantly expressed in the nervous system (7) and have been implicated in axon guidance. However we found that plexin-B1 is widely expressed in a variety of epithelial tissues and cell lines, suggesting that plexin-B1, and possibly other plexins, may have a general role in regulating cell migration and cell clustering/dissociation (reviewed in Ref. 31).
We found that endogenous plexin-B1 is mainly found as a heterodimer, because of proteolytic processing by PCs. Moreover, we showed that another member of the plexin-B subfamily, plexin-B2, is converted into a heterodimer by proprotein convertases. The cleavage is directed to specific sites in the extracellular domain of the receptors, fitting the consensus for PCs. In the extracellular domain of fly plexin B (i.e. R 1196 KKR) is also found a potential PC-specific processing site, suggesting phylogenetic conservation. The extracellular domains of other plexins do not contain bona fide cleavage sites for PCs. In fact, we did not observe proteolytic processing of other plexins in our   FIG. 3. Plexin-B2 is proteolytically cleaved and expressed at the cell surface as a heterodimeric complex. A, surface-biotinylated 293T cells expressing VSV-tagged plexin-B2 were analyzed by immunoblotting. Anti-tag antibodies detected a p240, corresponding to the full-length precursor of plexin-B2, and a smaller protein p80, corresponding to the receptor ␤ subunit. Avidin-HRP detected a minor fraction of the plexin-B2 precursor p240, together with p170 and p80 (the latter contains a very short extracellular sequence). In the third panel, cell surface proteins purified with avidin-agarose and immunoblotted with anti-VSV antibody are shown. B, pulse-chase analysis of plexin-B2 expressed in 293T cells. Cells were pulsed with [ 35 S]methionine/cysteine for 20 min and then chased for the times indicated; cell lysates were immunoprecipitated with anti-tag antibody, separated by SDS-PAGE, and submitted to autofluorography. C, 293T cells expressing plexin-B1 and plexin-B2 were surfacebiotinylated. Cell lysates were immunopurified with anti-VSV antibody and fractionated by SDS-PAGE in the presence or absence of reducing agents. Avidin-HRP detected plexin-B1 p300 precursor and p200/p100 cleaved subunits, as well as (barely detectable) plexin-B2 p240 precursor and p170 subunit, in both conditions. As a control, the disulfide-linked heterodimeric scatter factor receptor Met was immunoprecipitated from GTL-16 cells and analyzed under identical conditions. Met ␤ subunit, with a molecular mass of 145 kDa, was separated in the presence of reducing agents, whereas covalently linked ␣ and ␤ subunits run with a molecular mass of 190 kDa under non-reducing conditions. experimental conditions (data not shown). Members of the PC family identified to date include SPC-1/furin, PACE-4, PC2, PC1/3, PC4, PC5/6A-PC5/6B, and PC7. We demonstrate that plexin processing is inhibited by ␣1-antitrypsin Portland (␣1-PDX), known to specifically block a number of PCs, including SPC1/furin. LoVo colon adenocarcinoma cells lack expression of furin and have been used to discriminate proteins that are exclusively processed by this convertase from those that can be substrates of other family members (32,33). When plexin-B1 and plexin-B2 were expressed into LoVo cells, receptor processing was unaffected (data not shown), implicating the function of other PCs.
Several PCs are transmembrane proteins found in vesicles cycling between the Golgi apparatus and the plasma membrane (for a review, see Ref. 34). PCs localize transiently on the cell surface, from where they are recycled to endosomes and the trans-Golgi network. In addition, an active form of the enzyme is shed extracellularly (reviewed in Ref. 28). Here we report a new family of receptors that appear to be processed by PCs in a post-Golgi compartment and likely at the cell surface, suggesting that the activity of certain convertases may be regulated by their subcellular localization. This conclusion is based on the following evidence. First, an inhibitor that blocks vesicular trafficking from post-Golgi to the plasma membrane abrogates the processing. Second, the uncleaved precursor of the extracellular domain of plexin-B2 (B2-EC) can be found inside the cells and at the cell surface, whereas its processed form is only found in the conditioned medium. Third, mutated plexins that are impaired in subcellular trafficking and are not localized at the cell surface do not undergo proteolytic processing. Our results could also be explained by a processing step immediately preceding plexin localization at the cell surface. Notably, using uncleavable mutants we ruled out that the cleavage is required for surface localization. We demonstrate that specific sequences in the cytoplasmic domain of plexin-B1 are responsible for regulating cell surface targeting of the receptor and subsequent processing. The COOH-terminal sequence (likely through association with PDZ domain-containing proteins) greatly enhances cell surface localization of the receptor but is not directly required to regulate processing by PCs. The sequence mediating interaction with activated Rac1, instead, both regulates cell surface localization and is absolutely required for proteolytic processing of the receptor. This may be consistent with its role in localizing the receptor in specific membrane microdomains. GTPase-and PDZ domain-interacting sequences have been identified in plexin-B2 and in plexin-B3 as well; this suggests the existence of common regulatory mechanisms for this semaphorin receptor subfamily.
We showed that PC-mediated processing converts plexins B into heterodimers containing an ␣ and a ␤ subunit. The ␤ subunit is inserted in the plasma membrane and contains a . Cell lysates were immunoprecipitated with anti-VSV antibodies and analyzed by SDS-PAGE, followed by autofluorography. B, the extracellular domain of plexin-B2 (amino acids 1-1190) was fused to glutathione S-transferase (GST) to generate the secretable protein B2-EC. In the drawing, the predicted molecular mass of cleaved B2-EC fragments are indicated. The larger fragment (b, 170 kDa) corresponds to the ␣ subunit of plexin-B2, whereas the smaller protein (c, 30 kDa) includes GST and the short extracellular sequence downstream the substrate-cleavage site (indicated by an arrowhead). Plexin-B2-EC construct and its mutant in the PC site (uncleavable) were expressed in 293T cells, and the conditioned media were collected after 72 h from transfection. Secreted proteins were purified with glutathione-Sepharose and analyzed by anti-GST immunoblotting and silver staining. Lysates of surface-biotinylated cells were equally purified and analyzed with anti-GST antibodies and avidin-HRP. C, 293T cells were transfected with wild-type plexin-B1 or with the mutated receptors B1-⌬10 and B1-GGA and surface-biotinylated. Cell lysates were immunoprecipitated with anti-VSV and analyzed by immunoblotting with IC2 antibodies. Receptors exposed on the plasma membrane were revealed by avidin-HRP. Note that the ␤ subunit of B1-⌬10 runs slightly faster because of the deletion of its C-terminal tail.
FIG. 7. Uncleavable plexin-B1 displays reduced ligand binding ability and functional response. NIH-3T3 cells were transduced with plexin-B1, mutant plexin-B1 uncleav, or control vector and analyzed as follows. A, receptor expression was confirmed by Western blotting with IC2 antibody. B, cell surface localization of plexin-B1 and plexin-B1 uncleav was determined by immunostaining with anti-VSV, followed by alkaline phosphatase (AP)-coupled secondary antibody. AP activity was measured using p-nitrophenylphosphate as colorimetric substrate as described under "Experimental Procedures." Data are representative of three independent experiments. C, cells were incubated with different concentrations of Sema4D-AP for 30 min at 37°C, and specifically bound AP activity was quantified by colorimetric analysis, as above. Results shown are representative of three experiments performed in duplicate; error bars indicate the S.D. Scatchard plots are shown, and linear regression analysis was used to determine the line of best fit. Calculated K D values are indicated. D, cells incubated with Sema4D-AP as described in (C) were stained with AP substrate NBT/ BCIP. The cellular collapse triggered by plexin-B1 activation is significantly reduced in cells expressing plexin-B1 uncleav.
short extracellular sequence, plus the specific cytoplasmic domain of plexins. The ␣ subunit, including most of the extracellular domain, remains associated in the complex by force of weak bonds. The absence of disulfide bonds between the two subunits was confirmed by analysis under non-reducing conditions; in the case of plexin-B2, this is also consistent with the fact that no cysteine residues are found in the short extracellular sequence of the ␤ subunit. This leaves open the possibility that receptor heterodimers may dissociate spontaneously or under specific conditions. However, we did not find receptor fragments shed in the conditioned media of cells expressing plexin-B1 or plexin-B2, either basally or upon semaphorin stimulation (data not shown). Moreover, we did not observe changes in plexin-B1 processing upon ligand stimulation.
The specific functional role of plexins B processing by PCs is therefore a challenging question. PCs are implicated in the proteolytic processing of a number of proteins, including plexin ligands, the semaphorins (11), and plasma membrane receptors, such as scatter factor receptors, notably homologous to plexins (35). Importantly, PC-deficient animals display major developmental defects (36, 37; for a review see Ref. 10). However, in most cases, the mechanisms by which proteolytic cleavage regulates receptor function are unclear. The Notch receptor is a notable exception because its cleavage by PCs is required for efficient ligand binding (38). We addressed the issue of the specific function of plexin proteolytic processing in a number of ways. First, we demonstrated that this is a physiological event, consistently occurring in a variety of tissues and cell types. Second, we expressed wild-type and uncleavable plexin-B1 at comparable levels at the cell surface and found that higher amounts of Sema4D bind to cells expressing heterodimeric receptors, as compared with single-chain precursors. The affinity constant for the ligand does not seem to change with receptor cleavage. We thus speculate that the proteolytic processing of plexins B exposes additional ligand binding sites or promotes the formation of multimeric receptor complexes required for semaphorin binding. Rohm et al. (39) have previously reported another example of regulation of the number of semaphorin binding sites because of plexin-neuropilin interactions. Finally, we found that the functional response to the ligand is significantly increased in cells expressing plexin-B1 heterodimeric receptors, as compared with their uncleavable forms. Notably, cells expressing single-chain receptors did not collapse in response to low concentrations of Sema4D that were still effective in the presence of heterodimeric receptors.
In this study we started to address the issue of plexin diversity in mammalians. Whereas in the genome of invertebrates only two plexins are present, humans express at least nine different family members. We have previously shown that plexin-B1, the prototype of receptor subfamily B, is clearly distinguished by the ability to bind its semaphorin ligand in the absence of neuropilins, in contrast to plexins A. Now we show that plexins B are also distinguished by a heterodimeric structure due to proteolytic processing. It has been suggested that plexins of the A subfamily are activated by a conformational change because of neuropilin engagement by secreted semaphorins (9). We have not observed a similar behavior for plexin-B1. 3 This raises the possibility that different plexins follow distinct activation mechanisms. For instance, a conformational change induced by proteolytic processing of plexins B may allow for a more efficient ligand binding and functional response to transmembrane semaphorins, which directly interact with plexins without the need for neuropilins. The diversity of plexin receptors in vertebrates may thus reflect multiple functions and signaling modes deserving further investigation.