Multimerisation of receptor-type protein tyrosine phosphatases PTPBR7 and PTP-SL attenuates enzymatic activity

Dimerisation of receptor-type protein tyrosine phosphatases (RPTPs) represents an appealing mechanism to regulate their enzymatic activity. Studies thus far mostly concern the dimerisation behaviour of RPTPs possessing two tandemly oriented catalytic PTP domains. Mouse gene Ptprr encodes four different protein isoforms (i.e. PTPBR7, PTP-SL and PTPPBSgamma-42/37) that contain a single PTP domain. Using selective membrane permeabilisation we here demonstrate that PTP-SL, like PTPBR7, is a single membrane-spanning RPTP. Furthermore, these two receptor-type PTPs constitutively formed homo- and hetero-meric complexes as witnessed in chemical cross-linking and co-immunoprecipitation experiments, in sharp contrast to the cytosolic PTPPBSgamma-42 and PTPPBSgamma-37 PTPRR isoforms. This multimerisation occurs independently of the PTP domain and requires the transmembrane domain and/or the proximal hydrophobic region. Using overexpression of a PTPBR7 mutant that essentially lacks the intracellular PTP domain-containing segment, a monomer-mimicking state was forced upon full-length PTPBR7 immunoprecipitates. This resulted in a significant increase in the enzymatic activity of the PTPRR PTP domain, which strengthens the notion that multimerisation represents a general mechanism to tone down RPTP catalytic activity.     

PTPBR7 binding proteins in myelinating neurons of the mouse brain

Mouse protein tyrosine phosphatase PTPBR7 is a receptor-like, transmembrane protein that is localized on the surface of neuronal cells. Its protein phosphatase activity is reduced upon multimerization, and PTPBR7-deficient mice display motor coordination defects. Extracellular molecules that may influence PTPBR7 activity, however, remain to be determined. We here show that the PTPBR7 extracellular domain binds to highly myelinated regions in mouse brain, in particular the white matter tracks in cerebellum. PTPBR7 deficiency does not alter this binding pattern, as witnessed by RAP in situ staining of Ptprr^-/- mouse brain sections. Additional in situ and in vitro experiments also suggest that sugar moieties of heparan sulphate and chondroitin sulphate glycosaminoglycans are not critical for PTPBR7 binding. Candidate binding proteins were affinity-purified exploiting the PTPBR7 extracellular domain and identified by mass spectrometric means. Results support the suggested link between PTPRR isoforms and cerebellar calcium ion homeostasis, and suggest an additional role in the process of cell-cell adhesion.     

Altered MAP kinase phosphorylation and impaired motor coordination in PTPRR deficient mice

The neuronal protein tyrosine phosphatases encoded by mouse gene Ptprr (PTPBR7, PTP-SL, PTPPBSgamma-42 and PTPPBSgamma-37) have been implicated in mitogen-activated protein (MAP) kinase deactivation on the basis of transfection experiments. To determine their physiological role in vivo, we generated mice that lack all PTPRR isoforms. Ptprr-/- mice were viable and fertile, and not different from wildtype littermates regarding general physiology or explorative behaviour. Highest PTPRR protein levels are in cerebellum Purkinje cells, but no overt effects of PTPRR deficiency on brain morphology, Purkinje cell number or dendritic branching were detected. However, MAP kinase phosphorylation levels were significantly altered in the PTPRR-deficient cerebellum and cerebrum homogenates. Most notably, increased phospho-ERK1/2 immunostaining density was observed in the basal portion and axon hillock of Ptprr-/- Purkinje cells. Concomitantly, Ptprr-/- mice displayed ataxia characterized by defects in fine motor coordination and balance skills. Collectively, these results establish the PTPRR proteins as physiological regulators of MAP kinase signalling cascades in neuronal tissue and demonstrate their involvement in cerebellum motor function.     

Tyrosine-specific MAPK phosphatases and the control of ERK signaling in PC12 cells

Background

Spatio-temporal control of extracellular signal-regulated kinase (ERK) activity, a critical determinant of the cell's response to growth factors, requires timely dephosphorylation of its regulatory tyrosine and/or threonine residue by MAPK phosphatases. We studied the physiological role of kinase interaction motif (KIM)-containing protein tyrosine phosphatases (PTPs) in the control of EGF- and NGF-induced ERK activity in neuroendocrine PC12 cells.

Results

We found a single KIM-containing PTP to be endogenously expressed in rat PC12 cells: the transmembrane PTPRR isoform termed PCPTP1. Protein knock-down of PCPTP1, or fourfold overexpression of its mouse orthologue, PTPBR7, left EGF- and NGF-induced ERK1/2 activity in PC12 cells unaltered. Ectopic expression of cytosolic PTPRR isoforms, however, resulted in reduced EGF-induced ERK1/2 activity, an effect that was dependent on the phosphatase activity and the KIM-domain of these PTPs.

Conclusion

The finding that robust changes in tyrosine-specific MAPK phosphatase expression levels have minor effects on temporal ERK1/2 activity control in PC12 cells suggests that dual-specificity MAPK phosphatases may act as major regulators of growth factor-induced ERK1/2 signaling in these cells.     

Colocalisation of the protein tyrosine phosphatases PTP-SL and PTPBR7 with beta4-adaptin in neuronal cells

The mouse gene Ptprr encodes the neuronal protein tyrosine phosphatases PTP-SL and PTPBR7. These proteins differ in their N-terminal domains, with PTP-SL being a cytosolic, membrane-associated phosphatase and PTPBR7 a type I transmembrane protein. In this study, we further explored the nature of the PTP-SL-associated vesicles in neuronal cells using a panel of organelle markers and noted a comparable subcellular distribution for PTP-SL and the beta4-adaptin subunit of the AP4 complex. PTP-SL, PTPBR7 and beta4-adaptin are localised at the Golgi apparatus and at vesicles throughout the cytoplasm. Immunohistochemical analysis demonstrated that PTP-SL, PTPBR7 and beta4-adaptin are all endogenously expressed in brain. Interestingly, coexpression of PTP-SL and beta4-adaptin leads to an altered subcellular localisation for PTP-SL. Instead of the Golgi and vesicle-type staining pattern, still observable for beta4-adaptin, PTP-SL is now distributed throughout the cytoplasm. Although beta4-adaptin was found to interact with the phosphatase domain of PTP-SL and PTPBR7 in the yeast two-hybrid system, it failed to do so in transfected neuronal cells. Our data suggest that the tyrosine phosphatases PTP-SL and PTPBR7 may be involved in the formation and transport of AP4-coated vesicles or in the dephosphorylation of their transmembrane cargo molecules at or near the Golgi apparatus.     

Comparison of distinct protein isoforms of the receptor for advanced glycation end-products expressed in murine tissues and cell lines

The receptor for advanced glycation end-products (RAGE) is thought to be expressed ubiquitously as various protein isoforms. Our objective was to use Northern blotting, immunoblotting, and sensitivity to N-glycanase digestion to survey RAGE isoforms expressed in cell lines and mouse tissues in order to obtain a more comprehensive view of the RAGE expressome. Pulmonary RAGE mRNA (1.4 kb) was smaller than cell-line and tissue RAGE mRNA (6 kb-10 kb). Three anti-RAGE antibodies that recognized three distinct RAGE epitopes were used for protein studies (N-16, H-300, and αES). Lung expressed three predominant protein isoforms with apparent molecular masses of 45.1, 52.6, and 57.4 kDa (N-16/H-300) and four isoforms at 25.0, 46.9, 52.5, and 54.2 kDa (αES). These isoforms were expressed exclusively in lung. Heart, ileum, and kidney expressed a 44.0-kDa isoform (N-16), whereas aorta and pancreas expressed a 53.3-kDa isoform (αES). Each of these isoforms were absent in tissue extracts prepared from RAGE^−/− mice. Cell lines expressed a 70.0-kDa isoform, and a subset expressed a 30.0-kDa isoform (αES). Lung RAGE appeared to contain two N-linked glycans. Tissue and cell-line RAGE isoforms were completely insensitive to PNGase F digestion. Thus, numerous RAGE protein isoforms are detectable in tissues and cell lines. Canonical transmembrane and soluble RAGE appear to be expressed solely in lung (N-16/H-300). Non-pulmonary tissues and cell lines, regardless of the source tissue, both express distinct RAGE protein isoforms containing the N-terminal N-16 epitope or the αES RAGE epitope encoded by alternate exon 9, but lacking the H-300 epitope. This work was supported by NIH grants R01 GM37631 and GM68481.     

Distinct membrane localization and kinase association of the two isoforms of CD58

The adhesion molecule CD58 is involved in intercellular adhesion and in signal transduction. It is natively expressed in both a transmembrane form and a glycosylphosphatidylinositol (GPI)-anchored form, and hence provides a model for the study of two distinct membrane-anchored forms of the same protein in the same cell. We demonstrate here that the two isoforms of CD58 are localized in distinct membrane compartments. The GPI-anchored form localizes in lipid rafts, while the transmembrane form resides in nonraft domains. In addition to distinct membrane localization, the two isoforms of CD58 differ in their association with protein kinases. GPI-anchored CD58, residing in raft domains, is constitutively associated with protein kinases. However, cross-linking mediates a substantial increase in kinase activity which is predominantly associated with the transmembrane CD58 in nonraft membrane domains. The extensive inducible kinase activity, associated with transmembrane CD58, is demonstrated in wild-type cells as well as in GPI-deficient variant cells. Thus, although the transmembrane CD58 is excluded from rafts, it may trigger signaling independently of the GPI-linked isoform.     

Targeted mass spectrometric analysis of N-terminally truncated isoforms generated via alternative translation initiation

Alternative translation initiation is a mechanism whereby functionally altered proteins are produced from a single mRNA. Internal initiation of translation generates N-terminally truncated protein isoforms, but such isoforms observed in immunoblot analysis are often overlooked or dismissed as degradation products. We identified an N-terminally truncated isoform of human Dok-1 with N-terminal acetylation as seen in the wild-type. This Dok-1 isoform exhibited distinct perinuclear localization whereas the wild-type protein was distributed throughout the cytoplasm. Targeted analysis of blocked N-terminal peptides provides rapid identification of protein isoforms and could be widely applied for the general evaluation of perplexing immunoblot bands.     

Different functions of HOPS isoforms in the cell

Hepatocyte odd protein shuttling (HOPS) moves between nucleus and cytoplasm. HOPS overexpression leads to cell cycle arrest in G₀/G₁, and HOPS knockdown causes centrosome alterations, with subsequent abnormal cell division. Recently, we demonstrated that HOPS acts as a functional bridge in NPM-p19^Arf interactions. Here we show that HOPS is present in 3 different isoforms that play distinct intracellular functions. Although HOPS is a transmembrane ubiquitin, an isoform with intermediate molecular weight is cleaved from the membrane and released into the cytosol, to act as the shuttling protein. We identified a signal peptide peptidase structure in N-terminal membrane-bound HOPS that allows the regulated intramembrane proteolysis (RIP) system to control the relative amounts of the released, shuttling isoform capable of binding NPM. These results argue for distinct, isoform-specific functions of HOPS in the nucleolus, nucleus, and cytoplasm and provide insight into the dynamics of HOPS association with NPM, whose mutation and subsequent delocalization is found in 30% of acute myeloid leukemia patients.     

Cell-specific changes in expression of mRNAs encoding splice variants of aplysia cell adhesion molecule accompany long-term synaptic plasticity

Aplysia neurons express several splice variants of apCAM, a member of the Ig superfamily of cell adhesion molecules. The major transmembrane isoform is endocytosed in sensory neurons (SNs) during the early phases of long-term facilitation (LTF) of SN synapses evoked by serotonin (5-HT) or in the motor neuron L7 during the early phases of long-term depression (LTD) of SN synapses evoked by Phe-Met-Arg-Phe-amide (FMRFa). We used single cell RT-PCR to evaluate whether expression of mRNAs encoding for different apCAM isoforms in SNs and L7 is regulated during LTF produced by 5-HT, and LTD produced by FMRFa. Single SNs and L7s express mRNAs encoding for all major isoforms, but the proportion of each isoform expressed differs for the two cells. SN expresses more mRNA encoding for GPI-linked isoforms, while L7 expresses more mRNA encoding for the major transmembrane isoform. The neuromodulators produced significant changes in the proportional levels of mRNAs encoding for specific apCAM isoforms during the first 4 h after treatments without affecting overall levels of apCAM mRNA. 5-HT evoked changes that exaggerated cell-specific differences in isoform expression. FMRFa evoked changes that reduced cell-specific differences in isoform expression. The effects of the neuromodulators on apCAM mRNA expression were not detected when cells were cultured alone or when SNs were cocultured with another motor cell that failed to induce synapse formation (L11). The results suggest that rapid cell-specific regulation of splice variant expression may contribute to different forms of long-term synaptic plasticity.     

Differential effects of over-expressed neural cell adhesion molecule isoforms on myoblast fusion

We have used a transfection based approach to analyze the role of neural cell adhesion molecule (NCAM) in myogenesis at the stage of myoblast fusion to form multinucleate myotubes. Stable cell lines of myogenic C2 cells were isolated that express the transmembrane 140- or 180-kD NCAM isoforms or the glycosylphosphatidylinositol (GPI) linked isoforms of 120 or 125 kD. We found that expression of the 140-kD transmembrane isoform led to a potent enhancement of myoblast fusion. The 125-kD GPI-linked NCAM also enhanced the rate of fusion but less so when a direct comparison of cell surface levels of the 140-kD transmembrane form was carried out. While the 180-kD transmembrane NCAM isoform was effective in promoting C2 cell fusion similar to the 140-kD isoform, the 120-kD isoform did not have an effect on fusion parameters. It is possible that these alterations in cell fusion are associated with cis NCAM interactions in the plane of the membrane. While all of the transfected human NCAMs (the transmembrane 140- and 180-kD isoforms and the 125- and 120-kD GPI isoforms) could be clustered in the plane of the plasma membrane by species-specific antibodies there was a concomitant clustering of the endogenous mouse NCAM protein in all cases except with the 120-kD human isoform. These studies show that different isoforms of NCAM can undergo specific interactions in the plasma membrane which are likely to be important in fusion. While the transmembrane and the 125-kD GPI-anchored NCAMs are capable of enhancing fusion the 120-kD GPI NCAM is not. Thus it is likely that interactions associated with NCAM intracellular domains and also the muscle specific domain (MSD) region in the extracellular domain of the GPI-linked 125-kD NCAM are important. In particular this is the first role ascribed to the O-linked carbohydrate containing MSD region which is specifically expressed in skeletal muscle.     

Signal transduction by CD58: The transmembrane isoform transmits signals outside lipid rafts independently of the GPI-anchored isoform

The adhesion molecule CD58 is natively expressed in both a glycosylphosphatidylinositol (GPI)-anchored form and a transmembrane form. We previously demonstrated that the two isoforms of CD58 are differentially distributed in the cell membrane. The GPI-linked form resides in lipid rafts while the transmembrane form resides outside lipid rafts. Following cross-linking a fraction of transmembrane CD58 redistributes to lipid rafts. It has also been demonstrated that ligand binding to CD58 induces biological functions such as cytokine production and immunoglobulin isotype switching, indicating that cell–cell interactions result in CD58-mediated signal transduction. However, the signaling pathways involved in these activation processes are poorly defined. Here we show for the first time that cross-linking of CD58 induces protein tyrosine phosphorylation of BLNK, Syk and PLCγ, and activation of ERK and Akt/PKB. In addition, we studied how these signaling events relate to the distinct membrane localization of the two isoforms of CD58. We demonstrate that cross-linking of CD58 triggers signaling that is predominantly associated with transmembrane CD58 in nonraft microdomains. Moreover, signaling through transmembrane CD58 does not depend on coexpression of the GPI-linked isoform. Thus, despite the residence of its GPI-anchored isoform in lipid rafts and the translocation of a fraction of its transmembrane isoform to lipid rafts, CD58 signaling is triggered by the transmembrane isoform outside lipid rafts. These findings corroborate signaling outside lipid rafts, as opposed to the established notion that rafts function as essential platforms for signaling.     

Characterization of four mammalian numb protein isoforms. Identification of cytoplasmic and membrane-associated variants of the phosphotyrosine binding domain.

Numb is a membrane-associated, phosphotyrosine binding (PTB) domain-containing protein that functions as an intrinsic determinant of cell fate during Drosophila development. We have identified four isoforms of mammalian Numb with predicted molecular masses of 65, 66, 71, and 72 kDa that are generated by alternative splicing of the Numb mRNA. The different isoforms result from the presence of two sequence inserts within the PTB domain and the central region of the protein. The endogenous expression pattern of these isoforms, examined using specific antisera, varied in different tissues and cell lines. In addition, differentiation of P19 cells with retinoic acid leads to the specific loss of expression of the 71- and 72-kDa Numb proteins, suggesting that the expression of certain forms of Numb protein is regulated in a cell type-specific manner. Expression of Numb proteins fused to green fluorescent protein revealed that the form of the PTB domain with the alternatively spliced insert constitutively associated with the plasma membrane in polarized Madin-Darby canine kidney cells. In contrast, the isoform without the insert was cytoplasmic, suggesting that different PTB domain isoforms may regulate the subcellular localization of Numb proteins. The membrane localization may be due, in part, to differential affinity for acidic phospholipids. The distinct expression and localization patterns of the different mammalian Numb isoforms suggest that they have distinct functional properties.     

Cell‐specific changes in expression of mRNAs encoding splice variants of Aplysia cell adhesion molecule accompany long‐term synaptic plasticity

Aplysia neurons express several splice variants of apCAM, a member of the Ig superfamily of cell adhesion molecules. The major transmembrane isoform is endocytosed in sensory neurons (SNs) during the early phases of long‐term facilitation (LTF) of SN synapses evoked by serotonin (5‐HT) or in the motor neuron L7 during the early phases of long‐term depression (LTD) of SN synapses evoked by Phe‐Met‐Arg‐Phe‐amide (FMRFa). We used single cell RT‐PCR to evaluate whether expression of mRNAs encoding for different apCAM isoforms in SNs and L7 is regulated during LTF produced by 5‐HT, and LTD produced by FMRFa. Single SNs and L7s express mRNAs encoding for all major isoforms, but the proportion of each isoform expressed differs for the two cells. SN expresses more mRNA encoding for GPI‐linked isoforms, while L7 expresses more mRNA encoding for the major transmembrane isoform. The neuromodulators produced significant changes in the proportional levels of mRNAs encoding for specific apCAM isoforms during the first 4 h after treatments without affecting overall levels of apCAM mRNA. 5‐HT evoked changes that exaggerated cell‐specific differences in isoform expression. FMRFa evoked changes that reduced cell‐specific differences in isoform expression. The effects of the neuromodulators on apCAM mRNA expression were not detected when cells were cultured alone or when SNs were cocultured with another motor cell that failed to induce synapse formation (L11). The results suggest that rapid cell‐specific regulation of splice variant expression may contribute to different forms of long‐term synaptic plasticity. © 2000 John Wiley & Sons, Inc. J Neurobiol 45: 152–161, 2000     

Cytoplasmic Ubiquitin-Specific Protease 19 (USP19) Modulates Aggregation of Polyglutamine-Expanded Ataxin-3 and Huntingtin through the HSP90 Chaperone

Ubiquitin-specific protease 19 (USP19) is one of the deubiquitinating enzymes (DUBs) involved in regulating the ubiquitination status of substrate proteins. There are two major isoforms of USP19 with distinct C-termini; the USP19_a isoform has a transmembrane domain for anchoring to the endoplasmic reticulum, while USP19_b contains an EEVD motif. Here, we report that the cytoplasmic isoform USP19_b up-regulates the protein levels of the polyglutamine (polyQ)-containing proteins, ataxin-3 (Atx3) and huntingtin (Htt), and thus promotes aggregation of their polyQ-expanded species in cell models. Our data demonstrate that USP19_b may orchestrate the stability, aggregation and degradation of the polyQ-expanded proteins through the heat shock protein 90 (HSP90) chaperone system. USP19_b directly interacts with HSP90 through its N-terminal CS (CHORD and SGT1)/P23 domains. In conjunction with HSP90, the cytoplasmic USP19 may play a key role in triage decision for the disease-related polyQ-expanded substrates, suggesting a function of USP19 in quality control of misfolded proteins by regulating their protein levels.     

The N-terminal region of neuregulin isoforms determines the accumulation of cell surface and released neuregulin ectodomain

Two neuregulin-1 isoforms highly expressed in the nervous system are the type III neuregulin III-beta1a and the type I neuregulin I-beta1a. The sequence of these two isoforms differs only in the region that is N-terminal of the bioactive epidermal growth factor-like domain. While the biosynthetic processing of the I-beta1a isoform has been well characterized, the processing of III-beta1a has not been reported. In this study, we compared III-beta1a and I-beta1a processing. Both III-beta1a and I-beta1a were synthesized as transmembrane proproteins that were proteolytically cleaved to produce an N-terminal fragment containing the bioactive epidermal growth factor-like domain. For I-beta1a, this product was released into the medium. However, for III-beta1a, this product was a transmembrane protein. In cultures of cells expressing III-beta1a, the amount of neuregulin at the cell surface was much greater, and the amount in the medium was much less than in cultures expressing I-beta1a. Phorbol ester treatment and truncation of the cytoplasmic tail had markedly different effects on III-beta1a and I-beta1a processing. These results demonstrate an important role for the N-terminal region in determining neuregulin biosynthetic processing and show that a major product of III-beta1a processing is a tethered ligand that may act as a cell surface signaling molecule.     

Alternative splicing controls the mechanisms of FAK autophosphorylation

Focal adhesion kinase (FAK) is activated following integrin engagement or stimulation of transmembrane receptors. Autophosphorylation of FAK on Tyr-397 is a critical event, allowing binding of Src family kinases and activation of signal transduction pathways. Tissue-specific alternative splicing generates several isoforms of FAK with different autophosphorylation rates. Despite its importance, the mechanisms of FAK autophosphorylation and the basis for differences between isoforms are not known. We addressed these questions using isoforms of FAK expressed in brain. Autophosphorylation of FAK(+), which is identical to that of "standard" FAK, was intermolecular in transfected cells, although it did not involve the formation of stable multimeric complexes. Coumermycin-induced dimerization of gyrase B-FAK(+) chimeras triggered autophosphorylation of Tyr-397. This was independent of cell adhesion but required the C terminus of the protein. In contrast, the elevated autophosphorylation of FAK(+6,7), the major neuronal splice isoform, was not accounted for by transphosphorylation. Specifically designed immune precipitate kinase assays confirmed that autophosphorylation of FAK(+) was intermolecular, whereas autophosphorylation of FAK(+6,7) or FAK(+7) was predominantly intramolecular and insensitive to the inhibitory effects of the N-terminal domain. Our results clarify the mechanisms of FAK activation and show how alternative splicing can dramatically alter the mechanism of autophosphorylation of a protein kinase.     

Human corin isoforms with different cytoplasmic tails that alter cell surface targeting

Corin is a cardiac serine protease that activates natriuretic peptides. It consists of an N-terminal cytoplasmic tail, a transmembrane domain, and an extracellular region with a C-terminal trypsin-like protease domain. The transmembrane domain anchors corin on the surface of cardiomyocytes. To date, the function of the corin cytoplasmic tail remains unknown. By examining the difference between human and mouse corin cytoplasmic tails, analyzing their gene sequences, and verifying mRNA expression in hearts, we show that both human and mouse corin genes have alternative exons encoding different cytoplasmic tails. Human corin isoforms E1 and E1a have 45 and 15 amino acids, respectively, in their cytoplasmic tails. In transfected HEK 293 cells and HL-1 cardiomyocytes, corin isoforms E1 and E1a were expressed at similar levels. Compared with isoform E1a, however, isoform E1 was more active in processing natriuretic peptides. By cell surface labeling, glycosidase digestion, Western blotting, and flow cytometry, we found that corin isoform E1 was activated more readily as a result of more efficient cell surface targeting. By mutagenesis, we identified a DDNN motif in the cytoplasmic tail of isoform E1 (which is absent in isoform E1a) that promotes corin surface targeting in both HEK 293 and HL-1 cells. Our data indicate that the sequence in the cytoplasmic tail plays an important role in corin cell surface targeting and zymogen activation.     

Got diversity? Wiring the fly brain with Dscam

The Drosophila gene Dscam, encoding Down syndrome cell-adhesion molecule, is required for the development of neural circuits. Alternative splicing of Dscam mRNA potentially generates 38016 isoforms of a cell-surface recognition protein of the immunoglobulin superfamily. These isoforms include 19008 different ectodomains joined to one of two alternative transmembrane segments. Each ectodomain comprises a unique combination of three variable immunoglobulin domains. Biochemical studies support a model in which each isoform preferentially binds to the same isoform on opposing cell surfaces. This homophilic binding requires matching at all three variable immunoglobulin domains. These findings raise the intriguing possibility that specificity of binding by the Dscam isoforms mediates cell-surface recognition events required for wiring the fly brain.