The Orthologous Human and Murine Semaphorin 6A-1 Proteins (SEMA6A-1/Sema6A-1) Bind to the Enabled/Vasodilator-stimulated Phosphoprotein-like Protein (EVL) via a Novel Carboxyl-terminal Zyxin-like Domain*

Neuronal development and apoptosis critically depend on the transformation of extracellular signals to intracellular actions resulting in cytoskeletal rearrangements. Ena/VASP (enabled/vasodilator-stimulatedphosphoprotein) proteins play an important role in actin and filament dynamics, whereas members of the semaphorin protein family are guidance signals in embryo- and organogenesis. Here, we report the identification of two novel transmembranous human and murine semaphorins, (HSA)SEMA6A-1 and (MMU)Sema6A-1. These semaphorin 6 variants directly link the Ena/VASP and the semaphorin protein family, since SEMA6A-1/Sema6A-1 is capable of a selective binding to the protein EVL (Ena/VASP-like protein). EVL is the third member of the Ena/VASP family of proteins that was identified sharing the same structural features as Mena (mammalian enabled) and VASP, although its functionality seems to be different from that of the other members. Here we demonstrate that SEMA6A-1/Sema6A-1 is colocalized with EVL via its zyxin-like carboxyl-terminal domain that contains a modified binding motif, which further stresses the existence of functional differences between EVL and Mena/VASP. In addition these findings suggest a completely new role for transmembranous semaphorins such as SEMA6A-1/Sema6A-1 in retrograde signaling.

During embryonic development, growth cones of outgrowing axons are guided with impressive accuracy to their appropriate target areas in the central and peripheral nervous system. Besides several other protein families, the semaphorins function in this respect as guidance cues to growth cones of navigating axons (1)(2)(3)(4). The Sema1a protein identified as the first member of the semaphorin family and formerly known as G-SemaI/FasIV was isolated some years ago in a screen for gly-coproteins active in axon guidance processes in the American grasshopper Schistocerca americana (5). It exhibited a repulsive and in later experiments also an attractive activity on certain populations of outgrowing pioneer axons in the grasshopper limb bud (6). In contrast to this, Sema3A/Sema3A was isolated only shortly thereafter, prepared from a soluble fraction of embryonic chicken brain. The purified Sema3A/ SEMA3A contained an activity capable of collapsing the growth cones of outgrowing dorsal root ganglia (7). A detailed analysis of these two proteins from such distantly related species as invertebrates and vertebrates identified the semaphorins as a new gene family active in neural and organ development (8).
Today the semaphorins represent a large family (Ͼ25 genes) of secreted and transmembranous glycoproteins defined by a common semaphorin domain of 500 amino acids (aa), 1 currently represented in eight different subclasses (subclasses 1-7 and V) (8,9). Only recently, members of the semaphorin gene family have also been implicated in neuronal apoptosis and cell death signaling (10,11).
Both processes, axon outgrowth during embryogenesis and cell degradation/elimination during programmed cell death, are characterized by a drastic remodeling of cytoskeletal elements. Unfortunately, proteins that link signals from the cell surface to the filament assembly-regulating machinery during semaphorin-mediated axon guidance processes are currently unknown.
A general role in filament dynamics control and actin-based motility is played by the proline-rich proteins of the Ena/VASP family (12). Enabled (ena) was identified as a genetic suppressor of the Abelson tyrosine kinase (abl) in Drosophila, whereas VASP (vasodilator-stimulated phosphoprotein) was identified as a target for cyclic AMP and GMP-dependent protein kinases and through its involvement in the motility of the intracellular bacteria Listeria monocytogenes (13,14). In contrast to these discoveries, based on functional approaches, the mammalian Ena homolog Mena (mammalian Ena) and the EVL (Ena/ VASP-like protein) protein were identified by their sequence similarity to Ena and VASP (12).
Proteins of the Ena/VASP family share a strictly conserved domain structure that consists of the amino-terminal EVH-1 (Ena-VASP homology), the carboxyl-terminal EVH-2 domain, and a proline-rich core domain. These domains are additionally discriminated by their broad functional diversity. The amino-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  terminal EVH-1 domain is involved in subcellular distribution and is capable of binding to a target domain displaying the motif (D/E)FPPPPX(D/E) (15). This target domain is found in the focal adhesion proteins vinculin and zyxin and also in the surface protein ActA of Listeria, where it is critically involved in the intracellular motility of these bacteria via the recruitment of the host cell VASP proteins to the bacterial surface (16). In contrast to this, the carboxyl-terminal EVH-2 domain can directly bind to F-actin and plays a crucial role in multimerization of Ena/VASP proteins (17,18). In addition, the proline-rich core of Ena/VASP directly binds the G-actin-associated protein profilin as well as the Src homology 3 domains of the Abelson and Src kinases (17)(18)(19). Therefore, Ena/VASP proteins function as a key element in the targeting of the filament synthesis machinery to areas of aggregation/concentration at cell sites associated with motility, growth, and adhesion such as focal adhesion points and lamellipodia.
Here we describe the new human and murine semaphorin 6 variants SEMA6A-1 and Sema6A-1 that bind to the Ena/VASPlike protein EVL. Interaction of these proteins results in the recruitment and localization of EVL to submembranous SEMA6A-1 structures of transfected cells. Our findings suggest that the newly identified semaphorin 6 variants play an important role in retrograde signaling to cytoskeletal elements controlling proteins like Ena/VASP.

MATERIALS AND METHODS
RNA and mRNA Isolation-Total RNA was prepared from tissues or cultured cells by the TRIzol procedure (Life Technologies, Inc.). Poly(A) ϩ RNA was isolated from total RNA with the poly(A)-Tract system (Promega).
Isolation of SEMA6A-1-cDNA was generated from total RNA of human SK-N-MC cells using the Superscript preamplification kit (Life Technologies, Inc.). PCR was performed in a total volume of 50 l using primers matching the Sema6A (GenBank TM accession number AF030430) sequence (base pairs 307-327 and 506 -526). Primers were as follows: 5Ј-primer, GACGTAGACACATGCAGGATG; 3Ј-primer, GA-GCGATGTTGGCATGTTTGG. PCR conditions were the following: denaturing at 94°C for 30 s, annealing at 52°C for 30 s, and elongation at 72°C for 45 s with a thermal cycler (MJ Research). PCR products were subcloned into pBSIIKS(Ϫ), sequenced by a commercial service, and verified. A fragment that exhibited a 90% identity to Sema6A was used to screen a SK-N-MC-ZAP Express library, which was generated from 2 g of mRNA (Stratagene). In total, four full-length SEMA6A-1 clones could be isolated.
3Ј-RACE Analysis and Cloning of (MMU)Sema6A-1-For 3Ј-RACE analysis, cDNA was generated from mouse total brain RNA using the SMART RACE cDNA amplification kit (CLONTECH). Subsequent PCRs were performed with the Advantage 2 polymerase (CLONTECH), and products were ligated into pBSIIKS(Ϫ) and sequenced.
Genomic Sequence Analysis-Genomic sequence encoding missing fragments of the sema6A-1 gene were amplified using Taq polymerase (Sigma) and human genomic DNA as template purified with the Genomic Tips System (Qiagen) from human SK-N-MC cells. Primers were designed based on the available genomic DNA sequence data (AC008524, AC010233.3, AC010296.2, and AC010497.3). Sequencing was performed by a commercial service. Assembling of contigs and sequence fragments was done using the DNAStar software (Lasergene) and the HUSAR software (DKFZ, Heidelberg).
In Situ Hybridization-For in situ hybridization, riboprobes corresponding to sequence positions aa 55-289 (0.7-kb more internal probe) or aa 869 -1030 (0.5-kb C-terminal probe) were used. Circular plasmids were linearized, and riboprobes were synthesized with T3 polymerase. Expression patterns were the same for both probes, and sense controls gave no detectable signals. In situ hybridization experiments were performed as described previously (20).
Northern Blot-Human fetal and adult tissue blots were purchased from CLONTECH. A DNA fragment corresponding to aa 869 -1030 of SEMA6A-1 was randomly labeled in a total volume of 20 l using Klenow enzyme (Roche Molecular Biochemicals) and [ 32 P]dCTP (3000 Ci/mmol) (Amersham Pharmacia Biotech) (21). Hybridizations were performed with the Easy Hyb solution (CLONTECH). For control hybridizations, a ␤-actin probe purchased from CLONTECH was used.
Subcloning of SEMA6A-1 and Generation of SEMA6A-1 Mutants-A XbaI/ScaI fragment corresponding to aa 99 -1030 was subcloned into the pFLAGCMV-1 vector (Sigma), which fuses an N-terminal FLAG tag to the protein that allows rapid detection of the fusion protein with the monoclonal ␣-FLAG antibody. Site I and site II mutations of SEMA6A-1 were generated with the PCR overlap procedure, and resulting products were subcloned into the pFLAG-SEMA6A-1 construct (22). Flanking primers were the following: 5Ј-primer, CTGGGTCCCCCGGGAGCCT-CCCTGTCTCAG; 3Ј-primer, CTGCCGCAGCGCTTCTTGGTCTGGTG-GGTA. Site I primers were as follows: 5Ј-primer, AGGGGAGCCAAC-GCGGCGGCCGCCCCGCAGAGG; 3Ј-primer, CGGGGCGGCCGCCGC-GTTGGCTCCCCTGCCAAA. Site II primers were as follows: 5Ј-primer, AAGCCGGCCGTAGCCGCCAAAGCATCCTTTGCTCCC; 3Ј-primer, AAAGGATGCTTTGGCGGCTACGGCCGGCTTTAGCGA. ⌬Cyt-2 was constructed by deleting the SmaI/XhoI fragment, which corresponds to the Cyt-2 terminal domain from the pFLAG-SEMA6A-1 construct, and religation of the remaining plasmid after a fill-in reaction for the overlapping DNA ends using Klenow enzyme (Roche Molecular Biochemicals). For verification, all mutant sites were sequenced.

SDS-Polyacrylamide Gel Electrophoresis and Western
Blotting-Proteins were separated by conventional vertical gel electrophoresis on SDS-polyacrylamide gels and transferred to nitrocellulose membranes (Schleicher & Schuell). For Western blot analysis, a monoclonal ␣-FLAG antibody (Sigma) was used at a concentration of 1:4000, ␣-EVL and ␣-Mena were used at 1:5000, and ␣-VASP (Alexis Corp.) was used at 1:4000. Detection of stained proteins was performed with ␣-mouse or ␣-rabbit secondary antibodies conjugated to horseradish peroxidase. Bands were visualized with the ECL detection system (Amersham Pharmacia Biotech).
Immunoprecipitation and Pull-down Assays-Cells transfected with pFLAG-SEMA6A-1, pFLAG-CMV1, or pFLAG-SEMA6A-1-mutants (site I, site II, ⌬Cyt-2) were solubilized in ice-cold extraction buffer (phosphate-buffered saline, 140 mM, 5 mM EDTA) containing 1% (v/v) Triton X-100 (23). Extracts were precleared with 15 l of protein A-Sepharose suspension (Sigma) for immunoprecipitation or nickel beads (Qiagen) for nickel bead pull-down assays. For immunoprecipitation, antibodies (␣-EVL, ␣-Mena, ␣-VASP) were added, and incubation was carried out overnight at 4°C. Protein A-Sepharose suspension (75 l) (Sigma) was added, followed by incubation for 2 h at 4°C. The beads were collected by centrifugation and washed three times in extraction buffer containing Triton X-100. Subsequently, the precipitated proteins were separated by SDS-polyacrylamide gel electrophoresis. In nickel bead pull-down assays, purified EVL(His) 6 (1 g/ml) was added to the lysate and incubated overnight at 4°C. Protein complexes were pelleted through the addition of nickel beads (50 l) and incubation for 1 h at 4°C. Beads were collected and washed as described above. Probes were subsequently separated by gel electrophoresis, blotted, and analyzed as described above.

RESULTS
SEMA6A-1 and Sema6A-1, Two New Semaphorin 6 Variants-In an attempt to identify and isolate repulsive guidance cues that are involved in degenerative processes such as neuronal apoptosis, we isolated cDNAs from a human cDNA library and a mouse RACE cDNA pool that encode the two new semaphorin 6 variants (HSA)SEMA6A-1 and (MMU)Sema-6A-1, respectively. 2 The open reading frames encode proteins of 1030 and 1005 aa with calculated molecular masses of 112.2 and 114.4 kDa. Both proteins are closely related to the murine semaphorin 6A (Sema6A), a protein known to repel embryonic sympathetic axons (Fig. 1A) (24,25). Interestingly, the predicted proteins SEMA6A-1 and Sema6A-1 were 142 and 127 aa longer when compared with the already known Sema6A. This is due to an additional peptide stretch within their cytoplasmic domain, which was to date unknown for members of the semaphorin protein family. Within the region of overlap, the proteins share 93% identity/95% similarity to Sema6A and 39 -62% homology to the other class 6 semaphorins (Fig. 1, A and B).
(HSA)SEMA6A-1 Is Localized in a Disease Region on Human Chromosome 5-To determine whether these newly identified variants result from alternative splicing, we performed BLAST data base searches on human genomic sequences. Here, we identified four working drafts (AC008524, AC010233.3, AC010296.2, and AC010497.2) of the human chromosome 5 and a mapped genomic survey sequence (AB002453) that localizes the human sema6A-1 gene to the disease locus 5q21-22, which is known to be deleted in certain forms of lung cancer (26). Gaps between genomic sequences containing segments of identity to SEMA6A-1 were amplified by PCR on human genomic DNA, sequenced, and analyzed. Thus, the assembled human gene sema6A-1 consists of 20 exons, including two untranslated exons, covering roughly 60 kb of genomic sequence on chromosome 5. The genomic structure gives no evidence for the existence of a shorter Sema6A-like human variant indicative of alternative splicing (Fig. 1A). In addition, we performed 3Ј-RACE analysis on human cDNA and were unable to amplify 3Ј-endings corresponding to the previously described Sema6A sequence. Finally, in situ hybridization with 3Ј-terminal Sema-6A-specific probes (base pairs 2721-2770, accession number AF030430) on murine adult and embryonic tissues did not exhibit any specific signal (Data not shown).
SEMA6A-1/Sema6A-1 Exhibit a New Cytoplasmic Zyxinlike Domain-An in depth analysis of the cytoplasmic tail uncovered that the SEMA6A-1/Sema6A-1 cytoplasmic domains can be subdivided into two peptide stretches named herein Cyt-1 and Cyt-2, respectively. Both domains contain no obvious signaling motifs. While Cyt-1 shares a degree of homology to the cytoplasmic domains of other semaphorin 6 proteins ranging from 25 to 35%, Cyt-2 displays no homology to any known semaphorin. Surprisingly, Cyt-2 exhibits a small region of homology (33% identity/49% similarity) to zyxin, a proline-rich protein that is present at focal adhesion points and that is directly involved in the regulation of filament dynamics via binding to Ena/VASP proteins (27)(28)(29) (Fig. 1, B and C).
Northern Blot Analysis of SEMA6A-1 Revealed Two Transcripts-To determine whether SEMA6A-1/Sema6A-1 expression is consistent with a role as a guidance signal or regulatory element during development and regeneration/degeneration in embryonic and adult tissues, we examined the sites of expression by RNA in situ hybridization (Sema6A-1) and Northern blotting (SEMA6A-1). Northern blot analysis with the carboxyl-terminal SEMA6A-1-specific probe (aa 869 -1030) indicated that two major SEMA6A-1 transcripts of equal intensity and of 5 and 7 kb in length are present in human embryonic and adult tissues (Fig. 2, A and B). The highest expression levels of SEMA6A-1 were detected in human embryonic brain and kidney, whereas only low to moderate expression was seen in developing lung and liver. Only small amounts of SEMA6A-1 transcripts were detected in human adult tissues with the exception of rather strong expression levels in the highly regenerative placental tissues (Fig. 2B). Employing oligonucleotides representing C-terminal sequences of the previously described shorter Sema6A, we were not able to detect any mRNA transcript in multiple tissue murine Northern blots. These experiments, RACE, in situ hybridizations, and Northern blot-ting, indicate that SEMA6A-1/Sema6A-1 is not the result of alternative splicing of the putative Sema6A gene.
SEMA6A-1/Sema6A-1 Is Predominantly Expressed during Development-In situ hybridizations revealed that Sema6A-1 displays a complex expression pattern in the mouse embryo. Neural embryonic tissues displayed high levels of Sema6A-1 mRNA expression in proliferating zones, in the diencephalon, retina, in dorsal root ganglia, and also in the trigeminal ganglion (Fig. 3, A and B). Remarkably, in particular areas of differentiation such as the diencephalon, expression persists into adulthood (Fig. 3, C and D). A more regionalized expression of Sema6A-1 was seen in coronal and sagittal sections from adult mouse brain (Fig. 3, C-I). Most prominently, high levels were observed in the olfactory system including piriform cortex, in basically all of the thalamic nuclei except the paraventricular nucleus and intermediodorsal nucleus, in hypothalamus, in amygdala, and in layers IV and VI of the cerebral cortex (Fig. 3, D-F). Sema6A-1 is also present in corpus callosum and in anterior commissure (not shown); thus, also glial cells express Sema6A-1 mRNA (Fig. 3G). In the cerebellum, all granule cells are positive, with an enhanced expression in the flocculus, and also in the brain stem, various motor nuclei highly express Sema6A-1, e.g. facial, trigeminal (not shown), and vagus nuclei (Fig. 3, H and I). We also screened various developmental stages of the mouse embryo with an oligonucleotide probe representing C-terminal sequences of the Sema6A cDNA but were not able to detect any signal above background (data not shown).

SEMA6A-1/Sema6A-1 Contains a Motif Similar to the Target Motif of the Ena/VASP EVH1 Domain-
The identification of a zyxin-like carboxyl-terminal domain raised the question of whether SEMA6A-1/Sema6A-1 could also bind to members of the Ena/VASP protein family. Given this, SEMA6A-1/ Sema6A-1 could, therefore, be directly involved in actin dynamics via the Ena/VASP cascade. Binding of zyxin and other proteins to the EVH1 domain of Ena/VASP proteins occurs via a peptide stretch that displays the conserved sequence motif DFPPPP (15,16,30). Protein sequence analysis revealed that SEMA6A-1/Sema6A-1 exhibits next to the region of homology to zyxin two modified sites of this sequence that may mediate binding to Ena/VASP (site I, DNPPPAP (aa 858 -964); site II, DVPPKP (aa 1010 -1015)).

SEMA6A-1 Colocalizes with EVL and Mena in HEK 293 and HT22
Cells-In order to address the question of whether SEMA6A-1/Sema6A-1 may be involved in mediating extracellular signals to members of the Ena/VASP protein family, we constructed an amino-terminally FLAG epitope-tagged SEMA6A-1 (FLAG-SEMA6A-1). Immunoblotting of FLAG-SEMA6A-1 expressed in human embryonic kidney cells (HEK293) detected a major protein band of 125 kDa in size, which closely corresponds to the predicted protein size (Fig. 4,  lane 2). Protein samples separated without ␤-mercaptoethanol showed that SEMA6A-1 forms dimeric complexes, a feature commonly observed for semaphorins (Fig. 4, lane 3) (31, 32). Higher molecular weight complexes that were also detected may be due to an artificial aggregation of these proteins, a phenomenon frequently observed for transmembranous proteins. Expression in human HEK293 followed by immunofluorescent analysis using ␣-FLAG and ␣-SEMA6A-1 revealed that the expression product of FLAG-SEMA6A-1 is localized at the cell surface (Fig. 5, A 1 and A 2 ). The analysis of the confocal images furthermore demonstrated the high specificity of the crude rabbit sera (Fig. 5A 3 ). Immunofluorescent analysis of transfected HEK293 and mouse hippocampal HT22 cells using ␣-FLAG/␣-Mena (in HEK293) or ␣-FLAG/␣-EVL (in HT22) double stains clearly revealed that SEMA6A-1 colocalizes with EVL (Fig. 5, B 2 and B 3 ) and Mena (Fig. 5, C 2 and C 3 ), members of the Ena/VASP family, further indicating a possible interaction between these proteins.
SEMA6A-1 Co-immunoprecipitates with EVL-Employing antibodies specific for Mena, VASP, and EVL, we tried to co-immunoprecipitate SEMA6A-1 from Triton X-100 extracts of transfected HEK293 and HT22 cells to prove a direct interaction of SEMA6A-1/Sema6A-1 and Ena/VASP proteins. Western blot analysis of the precipitated protein complexes revealed that the FLAG-SEMA6A-1 protein co-immunoprecipitates with EVL but neither with Mena nor with VASP (Fig. 6, A-C). In a second experiment, we supplemented Triton X-100 extracts of untransfected HT22 cells with purified native FLAG-SEMA6A-1 protein followed by ␣-EVL co-immunoprecipitation of the protein complexes. Subsequent Western blot analysis again revealed that SEMA6A-1 coprecipitates with the EVL protein (Fig. 6A).
The Ena/VASP-like Protein EVL Binds to Site II of SEMA6A-1-To confirm this molecular interaction and to identify the binding motif within the SEMA6A-1 sequence, we mutated all proline and aspartic acid residues within the two peptide stretches of site I and site II to uncharged alanine residues in both sites, resulting in the constructs FLAG-SEMA6A-1-mut-I (site-I; DNPPPAP mutated to ANAAAAA) and FLAG-SEMA6A-1-mut-II (site-II; DVPPKP mutated to AVAAKA). In addition, we generated a deletion construct lacking the complete Cyt-2 domain (FLAG-SEMA6A-1⌬Cyt-2). Extracts of SEMA6A-1-mutant-transfected cells were combined with purified His-tagged EVL protein, and protein complexes were precipitated with nickel beads. These pull-down assays revealed that the sequence DVPPKP is responsible and sufficient for binding of FLAG-SEMA6A-1 to EVL, since precipitation was not possible with the FLAG-SEMA6A-1-mut-II or the FLAG-SEMA6A-1⌬Cyt-2 constructs, whereas precipitation was not altered with the FLAG-SEMA6A-1-mut-I construct. These data derived from the co-immunoprecipitation experiments and pull-down assays confirm a direct interaction of SEMA6A-1 and EVL as already indicated by the colocalization of these proteins. DISCUSSION Guidance signals and filament dynamics controlling proteins like Ena/VASP have central functions during embryogenesis and other events such as degeneration and regeneration that require a remodeling of the cytoskeleton (33)(34)(35)(36). A number of proteins are able to bind to Ena/VASP including the integrinbinding protein zyxin, profilin, and the tyrosine phosphatase Dlar (37-39). The last was only recently shown to be involved in motor axon outgrowth (40). Semaphorin signaling to the cytoskeleton known to date requires binding of the extracellular semaphorin domain to cell surface receptors such as neuropilin and plexin (41,42). The resulting downstream cascade of intracellular actions potentially involves Ena/VASP proteins and leads to cytoskeletal reorganization, e.g. in the growth cone of outgrowing axons (12,33,34). So far, only little is known about the function of cytoplasmic domains of transmembranous semaphorins (subclasses 1 and 4 -6) and the involvement of these domains in retrograde signaling. Nevertheless, a first hint of a potential indirect linkage of tranmembranous semaphorins to cytoskeletal element-binding proteins like Ena/ VASP is given by the fact that the cytoplasmic domain of Sema6B binds c-Src, a protein kinase that phosphorylates Ena/ VASP (43).
The Binding of SEMA6A-1/Sema6A-1 to EVL Occurs via a Zyxin-like Peptide Stretch-Here, we identify the two new human and murine semaphorins SEMA6A-1 and Sema6A-1, and we demonstrate for the first time that these guidance signals FIG. 2. Northern blot analysis of SEMA6A-1 expression in human tissues. A and B, human multiple-tissue mRNA blots were hybridized with a SEMA6A-1-specific probe (aa 869 -1030) revealing two major transcripts of 5 and 7 kb. Hybridization with a 2-kb human ␤-actin probe was used to confirm that identical mRNA levels are given in all lanes (A and B, bottom). Positions of SEMA6A-1 mRNA are indicated by the arrowheads. LGP, lateral globus pallidus; ME, median eminence; Ml, mitral cell layer; Pir, piriform cortex; pn, preoptic nuclei; Re, retina; Sc, sclerotom, SFO, subfornical organ; teg, tegmentum of pons; Th, thalamus; VZ, ventricular zone; IV, fourth layer of cerebral cortex; V, trigeminal ganglion; VI, sixth layer of cerebral cortex; 7, facial nucleus; 10, dorsal motor nucleus vagus.
can directly and tightly bind to a member of the Ena/VASP family. The first evidence for a direct protein interaction between the cytoplasmic domain of SEMA6A-1 with intracellular protein was seen in immunofluorescent analysis that revealed that SEMA6A-1 colocalizes with EVL and Mena, two proteins of the Ena/VASP family at submembranous structures. Furthermore, we could confirm an interaction of SEMA6A-1 with EVL through several immunoprecipitation and pull-down assays using ␣-EVL antibodies. In addition, the mutation of two potential binding sites (sites I and II) clearly demonstrated that binding of SEMA6A-1 to EVL occurs via site II. Although the binding motif DVPPKP does not fully fit to the DFPPPP motif known from other binding partners, we propose that binding of SEMA6A-1 to EVL is due to the conserved glutamic acid and proline residues at positions 1, 3, and 6 that frame the hydrophobic residue (valine) at position 2 that is known to be essential for binding (15,17,30). It is already known that a DLPPPP motif is also sufficient for binding, and the valine residue at position 2 resembles the same hydrophobic group as leucin (15). The fact that Mena did not co-imunoprecipitate with SEMA6A-1 might indeed be due to the lack of a direct interaction of these proteins. It is known that EVL and Mena form heterodimers, which would explain the colocalization seen in the immunofluorescent analysis. Interaction between EVL and Mena might not be tight enough for protein co-precipitation.
Expression of SEMA6A-1/Sema6A-1 Is Consistent with a Role in Development and Regeneration/Degeneration-In addition to the biochemical characterization, we analyzed the regions of SEMA6A-1 and Sema6A-1 expression in human and mouse tissues (24). These are consistent with a more general role of SEMA6A-1 and Sema6A-1 in neuro-and organogenesis as well as in regenerative and degenerative processes; all of these areas are characterized by a highly dynamic rearrangement of cytoskeletal elements. The novel SEMA6A-1/Sema6A-1 variant described here is obviously not a product of alternative splicing of the previously published Sema6A sequence (24),  5-8). B, FLAG-SEMA6A-1 was not precipitated when ␣-Mena (1:1000) or ␣-VASP (1:400) antibodies were used (lanes 1 and 3). Mena and VASP protein controls are shown (lanes 2 and 4). C, combination of purified EVL(His) 6 protein (1 g/ml) with Triton X-100 extracts of FLAG-SEMA6A-1 (wt, lane 1) or FLAG-SEMA6A-1-mut-I (mut-I, lane 2) transfected cells demonstrated that precipitation is unchanged with the site I mutation and the wild type control. Precipitation was not possible for the site II mutation (FLAG-SEMA6A-1-mut-II) and the ⌬Cyt-2-mutation (FLAG-SEMA6A-1⌬Cyt-2) (lanes 3 and 4). Positions of SEMA6A-1, EVL, Mena, and VASP are indicated by arrowheads.
since we failed to detect Sema6A mRNA in Northern blots and in situ hybridizations. Moreover, our efforts in identifying the human orthologous C-terminal end of the murine Sema6A via RACE experiments using RNA from the human neuroblastoma cell line SK-N-MC failed. Furthermore, we were not able to detect the published Sema6A C-terminal end in RACE experiments using RNA from mouse brain.
SEMA6A-1/Sema6A-1 Is a Candidate for Specific Retrograde Signaling in the Semaphorin Protein Family-On the basis of our data, we suggest a model where SEMA6A-1/ Sema6A-1 plays a role in retrograde or receptor signaling to cytoskeletal elements besides its repulsive effect of its semaphorin domain on certain neuronal subtypes during development. Binding of extracellular factors to SEMA6A-1/Sema6A-1 may facilitate the recruitment of Ena/VASP-like proteins to the cell surface, resulting in altered filament dynamics. Further evidence for a repulsive role of SEMA6A-1/Sema6A-1 in receptor or retrograde signaling is provided by the fact that overexpression of Ena/VASP proteins in fibroblasts leads to a downregulated cell motility. 2 Similarly, binding of the well known guidance signal SEMA3A/Sema3A to the endothelial neuropilin-1 receptor, which mediates repulsive signals, also leads to a reduced cell motility (41). In addition, it was shown that a modulation of EVL function with respect to its promoting abilities of actin polymerization and depolymerization occurs via its ligand binding status and through phosphorylation by cyclic nucleotide-dependent protein kinases (49).
Furthermore, roundabout, the receptor for slit, a receptor/ ligand system that is active in midline axon guidance, exhibits a motif (DLPPPP) in its cytoplasmic domain that is also known to bind to Ena/VASP proteins (44,45). Besides this, supporting evidence for our model requiring SEMA6A-1/Sema6A-1 expression on axons, comes from immunostainings using SEMA6A-1/ Sema6A-1-specific antibodies that detect outgrowing fiber tracts of the dorsal root ganglia and olfactory bulb. 3 This kind of switched distribution of "receptor" and "ligand" is also known from the Eph-receptor family and its ephrin ligands (46 -48). Thus, SEMA6A-1/Sema6A-1 expressed on navigating axons may bind to extracellular signals while transducing them via the Ena/VASP cascade into cytoskeletal reorganizations within the growth cone.
Therefore, the identification of a novel zyxin-like protein domain as part of the transmembrane SEMA6A-1/Sema6A-1 that is capable of a selective binding to the Ena/VASP-like protein EVL suggests a potent function for these variants as receptor-like components or elements in retrograde signaling in cytoskeletal rearrangements during neurogenesis.