The WW Domain of Neural Protein FE65 Interacts with Proline-rich Motifs in Mena, the Mammalian Homolog of DrosophilaEnabled*

The neural protein FE65 contains two types of protein-protein interaction modules: one WW binding domain and two phosphotyrosine binding domains. The carboxyl-terminal phosphotyrosine binding domain of FE65 interacts in vivo with the β-amyloid precursor protein, which is implicated in Alzheimer disease. To understand the function of this adapter protein, we identified binding partners for the FE65 WW domain. Proline-rich sequences sharing a proline-proline-leucine-proline core motif were recovered by screening expression libraries for ligands of the FE65 WW domain. Five proteins of molecular masses 60, 75, 80, 140, and 200 kDa could be purified from mouse brain lysates by affinity to the FE65 WW domain. We identified two of these five proteins as the 80- and 140-kDa isoforms encoded by Mena, the mammalian homolog of theDrosophila Enabled gene. Using the SPOTs technique of peptide synthesis, we identified the sequences in Mena that interact with the FE65 WW domain and found that they contain the signature proline-proline-leucine-proline motif. Finally, we demonstrated that Mena binds to FE65 in vivo by coimmunoprecipitation assay from COS cell extracts. The specificity of the Mena-FE65 WW domain association was confirmed by competition assays. Further characterization of the FE65-Mena complex may identify a physiological role for these proteins in β-amyloid precursor protein biogenesis and may help in understanding the mechanism of molecular changes that underlie Alzheimer disease.

FE65 is a brain-enriched protein with the modular structure of a typical adapter containing two types of protein-protein interaction domains: the WW domain and two phosphotyrosine binding (PTB) 1 domains (Refs. 1 and 2; see Fig. 1). WW do-mains bind proline-rich proteins containing the PPXY or PPLP core motifs (where X signifies any amino acid) (3)(4)(5). PTB domains bind proteins containing the consensus sequence NPXY; in some instances, the tyrosine must be phosphorylated to mediate high-affinity binding (6 -8). Importantly, the carboxyl-terminal PTB domain of FE65 binds in vitro and in vivo to the cytoplasmic portion of the ␤-amyloid precursor protein (␤APP), a large transmembrane protein implicated in Alzheimer disease (1, 9 -11). ␤APP is a precursor protein of the ␤-amyloid peptide, the major constituent of the extracellular senile plaques in the Alzheimer brain. In pathological conditions, the level of ␤-amyloid peptide production and/or accumulation is increased dramatically compared with the normal physiological state. The formation of amyloid plaques correlates well with the onset of Alzheimer disease (12)(13)(14). The biogenesis of ␤APP is a complex process that involves specific proteolytic activities, as well as other steps, which include additional posttranslational modifications, trafficking, and secretion (15)(16)(17)(18)(19). Little is known about the partner molecules that interact with ␤APP and control its processing. In addition to FE65, three other proteins have been shown recently to interact with the cytoplasmic domain of ␤APP: the heterotrimeric G protein G o (20 -22), the neuron-specific X11 protein (23), and a 59-kDa ␤APP-binding protein 1 (24).
The modular, adapter-like structures of FE65 and X11 proteins suggest that they may link the cytoplasmic portion of ␤APP to cytosolic proteins in a manner similar to signal transduction cascades involving growth factor or hormone receptors and integrins (reviewed in Ref. 2). The molecular components and the significance of the putative ␤APP signaling remain to be defined, but it is possible that defects in this pathway could be involved in the pathogenesis of Alzheimer disease (2,(25)(26)(27).
Biochemical and genetic analyses of proto-oncogenes have elucidated many facets of signaling processes that underlie neuronal function and development (reviewed in Ref. 28). The study of the Drosophila homolog of the Abl proto-oncogene implicated this gene in the processes of axonal outgrowth and fasciculation (reviewed in Ref. 29). Using genetic modifier screens, mutations in several new genes have been identified that modify Abl-dependent phenotypes. These genes include Disabled, Prospero, Failed Axon Connection, and Enabled (Ena) (30,31). Ena was identified in a specific screen for dominant mutations that alleviate the Abl phenotype (32). The Ena protein is axonally localized and is a substrate for the Drosophila Abl kinase and a ligand for the SH3 domains of Abl and Src (33). Recently, two vertebrate proteins closely related to Ena were identified. These two proteins, mammalian Enabled (Mena) and Ena-VASP-like are closely related to Ena (34) as well as to vasodilator-stimulated phosphoprotein (VASP) (35). Together, these molecules make up the Ena/VASP family of proteins and share three distinct regions of similarity: the amino-terminal 115 amino acids (the Ena-VASP homology 1 (EVH1) domain), a proline-rich core, and the carboxyl-terminal 226 amino acids (the EVH2 domain). The EVH1 domain mediates subcellular targeting of Ena/VASP family proteins by engaging in protein-protein interactions with a distinct prolinerich motif (34,36). Ena/VASP family proteins are concentrated in focal adhesions and actin stress fibers and are found in areas of dynamic actin remodeling, such as lammellipodia and axonal growth cones. The subcellular distribution, interactions with profilin, and a small actin monomer-binding protein implicated in the regulation of actin dynamics suggest a role for Ena/VASP family proteins in the regulation of cell motility and morphology by modulating the actin-based cytoskeleton (Ref. 34; reviewed in Ref. 37). A specialized role for Mena in the nervous system is suggested by a 140-kDa neuron-specific isoform of Mena that is present in the developing and adult brain along with the broadly expressed 80-kDa form of Mena (34).
Here we report the identification of a cognate ligand for the WW domain of FE65. Using a combination of functional screens of mouse embryo cDNA expression libraries, the SPOTs technique of peptide synthesis, pull-down experiments from mouse brain lysates, and coimmunoprecipitation from mammalian cell extracts, we were able to characterize the potential ligands of the FE65 WW domain. We identified one of these ligands as Mena, and we demonstrate herein the interaction of the FE65 WW domain with Mena through specific proline-rich motifs that contain the PPLP core.
Mutagenesis of FE65 WW Domain-Mutant GST-FE65 WW fusion protein was generated in which tryptophan (Fig. 1, W) at position 69 was substituted with phenylalanine (F) and proline (P) at position 72 was substituted with alanine (A). These mutations of the FE65 WW domain were designed based on NMR structural data for the human YAP WW domain and on empirical data from mutants of human YAP WW domain that did not bind proline-rich ligands, although the structure of the domain was maintained (39,40). The pGEX-2TK-FE65WW construct was used as a template. The mutagenesis was performed using the Chameleon double-stranded, site-directed mutagenesis kit (Stratagene). PAGE-purified mutagenic primer used for the mutagenesis was as follows: 5Ј-pGGA GGC CCG GCC GGC GGG TTC GAA CTG GGT GGT CCC-3Ј.
Molecular Cloning of FE65 WW Domain Ligands-Mouse embryo cDNA expression libraries made in pEXlox vectors from 12, 14, and 16 day mRNA (Novagen) were screened with 32 P-labeled GST-FE65 WW mutant and wild-type fusion proteins by standard methods (41,42). Three to four rounds of enrichment were performed until pure and uniformly positive phages were isolated. Plasmids containing recovered cDNA inserts of positive clones were generated by cre-mediated excision. Both strands of the cDNA clones were sequenced by the method of Sanger et al. (38).
Analysis of Binding Motif by the SPOTs Method-The SPOTs technique of peptide synthesis on derivatized cellulose membrane was performed as described (43,44). All reagents and equipment, including amino acids, derivatized membranes, instruction manual, and software (SPOTs, Release 1.0Ј), were purchased from Cambridge Research Biochemicals and Genosys Biotechnologies, Inc. The Mena SPOTs membrane was a kind gift from Juergen Wehland and Ronald Frank (Gesellschaft fü r Biotechnologische Forschung mbH, Braunschweig, Germany). The membranes were blocked with the blocking buffer supplied by the manufacturer and thereafter probed with the 32 P-labeled GST, GST-FE65 WW, or GST-mYAP WW1 fusion protein.
Antibodies-The following antibodies were used for Western blot analysis: affinity-purified antisera raised against 19-mer peptide from carboxyl-terminal portion of Mena (34); antisera against mbh1 (a kind gift from Edward Ziff, New York University, New York, NY); monoclonal antibodies against gelsolin (Transduction Laboratories); monoclonal anti-MAP antibodies (Sigma; a kind gift from Gianni Piperno, Mount Sinai Medical Center, New York, NY); antisera against myosin binding protein H (a kind gift from Donald Fischman, Cornell University Medical School, New York, NY).
Pull-down and Western Blotting Experiments-Mouse brains were homogenized in RIPA buffer. Clarified lysates were diluted 10-fold with Tris/Tween buffer to the final protein concentration of 1 mg/ml and incubated with various GST fusion proteins bound to glutathione agarose beads (200 g of protein/precipitation reaction) for 12 h at 4°C. The beads were then washed twice in 30 volumes of PBS with 0.5% Triton X-100 and twice in PBS only. Protein complexes separated by 7.5% SDS-PAGE were subsequently transferred to nitrocellulose membrane (Bio-Rad). The blots were probed either with the 32 P-labeled GST fusion proteins or anti-Mena antibodies.
Coimmunoprecipitation-COS-7 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Hyclone) and 1% penicillin/streptomycin mixture in 5% CO 2 atmosphere at 37°C. 3 ϫ 10 6 cells were transfected by electroporation at 250 microfarads and 220 V with 20 g of CMV-HA-FE65 plasmid. 62 h after the transfection, the cells were harvested in ice-cold PBS and centrifuged at 2000 rpm at 4°C, and the pellet was dissolved in lysis buffer (10 mM Tris HCl, pH 7.5; 150 mM NaCl; 0.1 mM sodium vanadate; 50 mM NaF; 0.5% Nonidet P-40; 1 mM phenylmethylsulfonyl fluoride; 10 g/ml aprotinin, leupeptin, and pepstatin). The extracts were clarified by centrif-ugation at 16,000 ϫ g at 4°C, and 4 mg of supernatant were incubated for 1 h at 4°C with an anti-HA monoclonal antibody (a kind gift from Francesca Fiore) or with an unrelated monoclonal antibody (a kind gift from Caterina Turco, Università degli Studi di Napoli "Federico II"). Thereafter, 30 l of protein A-Sepharose resin (Pharmacia) were added to each sample of the extract-antibody mixture, and the immunocomplexes were eluted with 50 mM Tris HCl, pH 6.8, 2% SDS, 10% glycerol, 100 mM dithiothreitol, 0.01% bromphenol blue. The proteins were resolved by 7.5% SDS-PAGE and transferred to Immobilon-P membranes (Millipore). The filter was blocked in 5% nonfat dry milk in TBS-T buffer and incubated with anti-Mena antibodies at 1:1000 dilution for 1 h at room temperature. After being washed in TBS-T, the filter was exposed to horseradish peroxidase-conjugated protein A (Amersham Corp.) at a dilution of 1:5000 for 30 min at room temperature. The signals were detected by chemiluminescence using the ECL system (Amersham Corp.).
Competition Experiments-The precipitation from mouse brain lysates was performed as described above. During probing with 2.5 g/ml radioactively labeled GST-FE65 WW, the GST-13Pro fusion protein containing polyproline region of clone 13 was added at two arbitrarily chosen concentrations, 15 g/ml and 40 g/ml. Blots were then washed three times for 15 min and exposed.

FE65 WW Domain Interacts with Several Proteins Present in Mouse Brain
Lysates-To demonstrate that the putative WW domain of FE65 indeed interacts with the protein(s) present in the mammalian nervous system, we performed pull-down experiments by incubating wild-type GST-FE65 WW fusion protein with mouse brain lysates. GST and mutant GST-FE65 WW domain proteins were used in control experiments. Bound proteins from mouse brain lysates were analyzed by Western ligand blots probed with either radioactively labeled GST, mutant GST-FE65 WW, or GST-FE65 WW fusion proteins (Fig. 2). Precipitation with GST and mutant GST-FE65 WW did not show any specific interaction (Fig. 2, A and B), whereas precipitation with the wild-type GST-FE65 WW revealed at least five proteins, with molecular masses of 60, 75, 80, 140, and 200 kDa (Fig. 2C).
Identification of Specific Amino Acid Sequences Interacting with the WW Domain of FE65-As previously demonstrated (3)(4)(5), WW domains interact with proteins containing the PPXY or PPLP core motifs. The first step toward the identification of proteins interacting with the FE65 WW domain was the elucidation of the core motif required for this binding. Thus, we identified potential ligands by screening phage expression libraries and analyzed these proteins by the SPOTs technique of peptide synthesis. We screened 12-, 14-, and 16-day mouse embryo cDNA expression libraries with radioactively labeled GST-FE65 WW fusion protein. Twenty positive clones were isolated that did not bind radioactively labeled GST or mutant GST-FE65 WW. They were grouped into five independent clones (Table I). Two clones, 7 and 9-1, encoded an identical polyproline sequence with the PPLP motif, suggesting that the WW domain of FE65 may belong to the class of WW domains that interact with PPLP rather than PPXY core motifs. By sequence and Northern blot analyses, clones 7 and 9-1 appeared to contain binding sites within cryptic open reading frames present in untranslated regions of the cDNAs (data not shown). Clone 13 represents a novel cDNA with a long open reading frame that also contains the PPLP motif in the coding frame. Northern blot analysis revealed that it is testes-specific (data not shown), making it unlikely to be a ligand of the brain-enriched form of FE65. Clone 13 may represent a ligand of the WW domain of the related FE65L2 protein, which is present in both brain and testes. 2 Clone 48 corresponds to a ubiquitously expressed mRNA that does not encode a perfect PPLP motif but instead encodes several proline-rich regions that bind to WW domain of FE65 with relatively weaker affinity. Finally, one of the isolated clones encodes the carboxylterminal portion of mbh1, also known as macrophage capping protein 39 (46,47). mbh1 is a 45-kDa actin-binding protein phosphorylated at multiple serines and threonines that belongs to the gesolin/severin family (48 -50). The molecular function of this protein is unknown.
To confirm binding of cloned proteins, GST fusion proteins that represented full and deleted portions of the isolated clones were generated. Radioactively labeled GST-FE65 WW fusion protein, which showed specific binding to the GST-ligand fusion proteins, was used as a probe for Western ligand blot analysis of the deletion series. GST alone or mutant GST-FE65 WW were used as controls (data not shown). The results of the deletion analysis determined the minimal sequences within positive clones sufficient and necessary for binding (Table I).
Interestingly, but not surprisingly, the GST-mbh1 fusion protein of the full cloned mbh1 segment (450 base pairs) was required to bind GST-FE65 WW. This binding event might be explained by helix-loop-helix structure of mbh1 in this region (46). Distant amino acids, including prolines, might be juxtaposed by folding to form a binding motif sufficient to interact in vitro. The pull-down experiments reported in Fig. 2 do not show any band with an apparent molecular weight compatible with that of mbh1. Furthermore, we were not able to demonstrate the complex between mbh1 and FE65 in vivo by coimmunoprecipitation from PC12 cells or mouse brain lysates with anti-mbh1 antibodies. Therefore, it seems likely that the mbh1-FE65 WW domain interaction we observed in vitro is spurious.
To gain more insight into the polyproline sequence identified in the positive clones, we performed a "valine scan" of a prolinerich sequence of clone 13 by the SPOTs technique to determine which residues are required for binding to the WW domain of FE65. Valine was used for substitutions instead of the typical alanine replacement because the fragment of clone 13 contained alanine. As shown in Fig. 3, six consecutive prolines are necessary for significant binding in vitro between the WW domain and peptide ligands (spots 4 -7). Substitutions of leucine at position 8 of the target peptide with glutamic and aspartic acids (spots 19 and 20) result in a weaker binding to the WW domain of FE65. In contrast, replacements with basic amino acids, such as lysine and arginine (spots 25 and 31), lead to a relative increase in interaction between the FE65 WW domain and its ligand peptides. Spot 37 represents a binding motif found in clones 7 and 9-1 that binds strongly to GST-FE65 WW fusion protein. Peptides 38 -40 are proline-rich segments of sequence from clone 48. Only peptide 38 (VEPMLD-LLPPPPPPP) showed interaction on the filter assay. Spot 41 represents the proline-rich sequence found in formin and containing the PPLP motif. Leder and colleagues (4) have reported that this sequence interacts in vitro with certain SH3 as well as WW domains, and we observed relatively strong binding to WW domain of FE65 (Fig. 2, spot 41). Finally, a 15-mer peptide containing exclusively prolines (spot 42) interacts robustly with the FE65 WW domain, confirming our prediction that the presence of six or more consecutive prolines is sufficient for an oligopeptide to bind to the WW domain of FE65. Together, these experiments indicate that the consensus sequence for a ligand of FE65 WW domain may include a minimal sequence of six consecutive prolines or a longer sequence of polyprolines that could be interrupted by specific amino acids, such as leucine, lysine, or arginine.
Identification of 80-and 140-kDa Bands as Mena-To identify candidate FE65 WW ligands, we searched the GenBank TM and EMBL data bases for proteins that contain at least six consecutive prolines with probable core sequences of PPLP or PP(K/R)P, that are expressed in the brain, and of which the estimated molecular weights are similar to those identified by pull-down experiments with the FE65 WW domain (see Fig. 2). Several potential candidates were identified, including MAPs (MAP1, MAP2, MAP5, Tau, ␤-tubuline, and tyrosine-tubuline), gelsolin, myosin binding protein H, and Mena. Western blot analysis with specific antibodies against these molecules identified Mena and none of the other proteins among the FE65 WW pulled down proteins. FE65 WW-bound protein complexes were separated by SDS-PAGE and transferred onto nitrocellulose membrane, and the blot was probed with radioactively labeled GST-FE65 WW (Fig. 4A) and thereafter with anti-Mena antibodies (Fig. 4B). Two bands of 80 and 140 kDa observed in both cases appeared to be identical in terms of relative migration and intensity. To confirm that these signals actually represent Mena, we compared lysates made from normal mouse brains to lysates from Mena null mutant mice that were engineered by homologous recombination of the Mena gene (34). 3 The FE65 WW domain pull-downs from Mena knockout mouse brain lysates lacked the 80-and 140-kDa bands. The third protein, represented by lower band of 60 kDa, which is stained with antibodies against Mena, represents a nonspecific signal because other Mena antisera fail to detect this signal in total brain lysates. 3 FE65 and Mena Interact in Vivo-Currently available anti-FE65 antibodies are directed against the WW domain of the protein and thus may be inhibitory in coimmunoprecipitation 2 T. Russo, unpublished data. 3 F. Gertler, unpublished data. experiments due to their ability to compete with any protein for the binding to the WW domain. Therefore, we generated an expression vector in which the CMV promoter drives the expression of a HA-tagged FE65 protein. This construct was transfected transiently in COS-7 cells. Protein extracts from these cells were incubated with anti-HA antibodies, and immunocomplexes were analyzed by Western blotting using anti-Mena antibodies. As shown in Fig. 5, two bands of the 80-kDa isoform of Mena were detected in total cell lysate and in protein complexes immunoprecipitated by anti-HA antibodies but they were not detected by the unrelated control antibodies. It has been shown previously that the 80-kDa form of Mena migrates as a doublet (34). We noted an apparent difference in the pattern of the 80-kDa form of Mena precipitated with GST-FE65 WW domain from brain lysates (e.g. Fig. 4, lane 9: single band) and with HA-tagged FE65 from COS cells (Fig. 5, lane 4: doublet). This may be due to the presence of an additional form of Mena in COS-7 cells (Fig. 5, lanes 1 and 4).
Specificity of Mena-WW Domain Interaction-GST fusion proteins of the WW domains of FE65, mouse YAP (WW1), Nedd4 (WW2 and WW3), Ess1, Msb1, the unique domain of Yes (irrelevant sequence control), and GST alone were used in pull-down experiments with mouse brain lysates. A Western blot of the precipitated proteins was probed with anti-Mena antibodies (Fig. 6A). The WW domains of FE65 and Msb1 precipitated both the 80-and 140-kDa forms of Mena efficiently. The WW domains of YAP and Nedd4 (WW2) bound well only to the 140-kDa isoform, whereas GST alone, Nedd4 (WW3), Ess1, and the unique domain of Yes showed no interaction. The FE65 and Msb1 WW domains share a high degree of similarity and belong to one subclass of WW domains containing three consecutive aromatic residues in the middle of the WW domain sequence, for which the core of binding motif is PPLP (2). In contrast, YAP and Nedd4 WW domains bind to ligands containing the PPXY motif (2,5,39,51). Interestingly, the neural isoform of Mena contains the PPSY motif, which might be responsible for binding to YAP and Nedd4 WW domains. This suggestion is supported by observation that these WW domains bind much more strongly to the 140-kDa isoform of Mena than to the 80-kDa form (Fig. 6A), which lacks the PPSY sequence. Although the 80-kDa isoform was detected in pull-downs with mYAP WW1 and Nedd4 WW2, it was present at a very low level, which probably could be explained by the presence of heteromultimeric or homomultimeric Mena complexes.
FE65 WW Domain Interacts with Specific Polyproline Regions of Mena-To identify the binding motif through which Mena interacts with FE65, we probed a SPOTs filter containing overlapping 15-mer peptides, which represent the entire sequence of the neural isoform of Mena, with radioactively labeled GST-FE65 WW fusion protein (Fig. 7). As expected, spots 142 and 143, representing stretches of polyprolines interrupted with a single leucine, demonstrated the strongest binding. In addition, other proline-rich regions of Mena (spots 174 -176 and 180 -193) showed binding to FE65 WW domain, although with weaker affinity than spots 141-145 of Mena. Remarkably, all target peptides that bound to the WW domain of FE65 in this experiment contain the PPLP sequence (Fig. 7). Also, when a membrane supporting Mena peptides was probed with radioactively labeled GST-mYAP WW1 fusion protein, binding was detected only on spots 85-88, which contain the PPSY motif (data not shown). These results confirm that the binding of the 140-kDa isoform of Mena to the WW domains of mYAP and Nedd4, shown in a previous experiment, occurs through a sequence containing a PPSY core that is present in the alternatively included portion of the neuron-specific Mena isoform.
GST Fusion Protein Containing the Polyproline Region of Clone 13 (GST-13Pro) Competes with Mena for Binding to FE65 WW Domain-To confirm that binding between Mena and the FE65 WW domain does indeed occur through the proline-rich region of Mena with the PPLP core motif, we performed a competition assay in which GST-13Pro fusion protein was used (see Table I). Blots with proteins precipitated from mouse brain lysates with GST, mutant GST-FE65 WW, and wild-type GST-FE65 WW were incubated with radioactively labeled GST-FE65 WW alone (Fig. 8D) or with the addition of GST-13Pro fusion protein (Fig. 8, A and C). When increasing concentrations of competing protein were used, the interaction between Mena and the GST-FE65 WW fusion protein was proportionally decreased (Fig. 8, A and C). In addition FIG. 3. Mutational analysis of the P 1 P 2 P 3 P 4 P 5 P 6 P 7 L 8 P 9 A 10 P 11 P 12 P 13 Q 14 P 15 target peptide that binds to the WW domain of FE65. The SPOTs technique was used to synthesize a repertoire of mutated peptides. Spot 1 represents the wild-type sequence of the target peptide as isolated from clone 13 in screening of cDNA expression libraries. Spots 2-16 represent a valine scan of the target peptide; each amino acid of the target peptide sequence was sequentially substituted with valine. A sequential substitution of L 8 with all remaining amino acids corresponds to spots 17-36. Spots 37-42 represent spots with potential binding affinities to the WW domain of FE65 (see "Results"). The control spots, 43-48, indicate derivatized spots onto which no amino acids were applied. For blotting, the 32 P-labeled GST-WW domain of FE65 was used. A, autoradiogram of the membrane exposed for 5 min; B, autoradiogram of the membrane exposed for 15 min; C, orientation of the derivatized spots on which peptides were synthesized; D, individual sequences of the peptides corresponding to numbered spots.
to Mena, the 200-kDa protein was also competed (Fig. 8A). This result suggests that other ligands of FE65 WW domain may interact via a similar binding motif containing the PPLP sequence, although binding of the FE65 WW domain to 60-and 75-kDa proteins was not significantly affected by competition with GST-13Pro (Fig. 8, A and C). DISCUSSION We identified the Mena protein as one of the cognate ligands for the WW domain of FE65 adapter protein and mapped the sites of interaction on Mena to polyproline-rich regions containing the signature PPLP motif. More importantly, we documented the binding between Mena and the FE65 WW domain in vitro and in vivo by pull-down experiments from mouse brain lysates and by coimmunoprecipitations of the complex from COS cells overexpressing the FE65 gene product.
The following aspects of the work deserve brief comment: (a) specificity of the interaction between FE65 and Mena; (b) biological role of the FE65-Mena complex; and (c) functional implications of the FE65-Mena complex for the biogenesis of the ␤APP.
Given the numerous examples of specificity and degeneracy in the protein-protein interactions mediated by SH2, SH3, and known WW domains (3, 4, 52, 53), we consider the FE65-Mena complex a cognate pair. The major criterion for this conclusion is the expression of these two proteins in the same cellular compartment (34,54) and tissues (i.e. cytoplasm and neural tissues) and formation of the complex in vivo as revealed by its coimmunoprecipitation. Two other observations support the specificity of binding. First, the WW domain of FE65 belongs to the subset of WW domains that contain three aromatic positions in the middle of the linear sequence of the domain. This feature seems to correlate with the preference of the domain for polyproline ligands containing PPLP cores. In contrast, the WW domains with two consecutive aromatic amino acids in the middle bind ligands with PPXY cores (2-4). Our results are Mena knockout mouse brain lysate; lane 6, precipitation with wild-type GST-FE65 WW from Mena knockout mouse brain lysate; lane 7, precipitation with GST from control mouse brain lysate; lane 8, precipitation with mutant GST-FE65 WW from control mouse brain lysate; lane 9, precipitation with wild-type GST-FE65 WW from control mouse brain lysate. Nitrocellulose blot of precipitated proteins was probed with 32 P-labeled GST-FE65 WW fusion protein (A) and anti-Mena antibodies (B). The normalized amount of protein in each lane was confirmed by Coomassie stain (C). Two isoforms of Mena protein are indicated by arrows and numbers at the right. Numbers at left indicate molecular mass markers in kDa. consistent with this observation in that the strongest relative binding of the FE65 WW is to Mena peptides containing PPLP cores. Interestingly, the PPSY motif that is present in the neural isoform of Mena exhibited binding in vitro only to the WW domains of YAP and Nedd4. The unique stretches of homoprolines interrupted by leucines form binding sites to other modules, including the SH3 domains, EVH1 domains, and profilin. Although seemingly degenerate, these interactions may require specific core sequences within the polyproline regions as proposed in the "binary switch" hypothesis for overlapping core motifs for SH3 and WW domains (5, 51, 55). Because the proline-rich core of Mena binds to the profilin, WW, and SH3 and EVH2 domains are shown as patterned boxes. B, autoradiogram of the membrane exposed for 5 min; C, autoradiogram of the membrane exposed for 15 min; D, orientation of the derivatized spots on which peptides were synthesized; E, selected sequences of the peptides corresponding to numbered spots; the first five (1-5) and the last five (271-275) consecutive peptides are shown in addition to the sequences that scored positively in the binding assay.
domains, it will be important to determine which of these interactions are compatible and which are competitive. Second, the competition experiment, in which PPLP-containing polypeptides were used to dissociate interactions between Mena and FE65, supports our conclusion on the specificity of binding, although the control, "scrambled" peptide was not used due to an intrinsic difficulty in permuting homoprolines and the significant affinity of the FE65 WW domain for hexamer or longer polyprolines.
Three other proteins present in mouse brain lysate bind to the WW domain of FE65, one of which, the 75-kDa protein, either is a strong binder or is present in relatively higher quantity in the lysates. The identification of these proteins and insight into their biological properties will further our understanding of FE65 function. Perhaps different ligands of the FE65 WW domain compete with each other and thus modulate the transduction of the signal from ␤APP to the intracellular pathway (2).
The FE65 PTB2 domain interacts with carboxyl-terminal tail of ␤APP in vivo (1, 8 -10). Proteins that bind to PTB1 and WW domains of FE65 could be involved in biogenesis of ␤APP. Specifically, cellular trafficking of ␤APP and its derivatives is a complex process during which ␤APP, as well as products of ␤APP secretory cleavage, are transported to the endosomes/ lysosomes in the clathrin-coated vesicles (56). The sequence NPTY, which is present in the cytodomain of ␤APP and shown to be responsible for interaction with the FE65, was also found to be a specific targeting signal for internalization into the clathrin-coated vesicles of low-density lipoprotein receptor and other proteins (57). Therefore, FE65 may serve as an adapter protein that brings other partner molecules into the complex with ␤APP, which would effect ␤APP secretion, internalization, and/or trafficking. In light of this, Mena, being a cytoskeletal protein involved in microfilament assembly, is a good candidate to participate in cellular network of proteins interacting with ␤APP in the cytoplasm. Interestingly, expression of the neuronenriched isoform of Mena in fibroblasts induces the formation of actin-rich cellular outgrowths implicating Mena in the establishment of cytoskeletal microfilament connections (34). Similar to its relative, VASP, Mena has been shown to interact with the G-actin binding protein profilin. Furthermore, Mena, profilin, and vinculin (shown to bind to Mena in vitro) are concentrated in axonal growth cones (34). It is expected that the study of the Mena-FE65 complex may elucidate molecular events that affect or regulate ␤APP biogenesis and may provide clues to the molecular changes that underlie Alzheimer disease.