Drosophila Ack targets its substrate, the sorting nexin DSH3PX1, to a protein complex involved in axonal guidance.

Dock, the Drosophila orthologue of Nck, is an adaptor protein that is known to function in axonal guidance paradigms in the fly including proper development of neuronal connections in photoreceptor cells and axonal tracking in Bolwig's organ. To develop a better understanding of axonal guidance at the molecular level, we purified proteins in a complex with the SH2 domain of Dock from fly Schneider 2 cells. A protein designated p145 was identified and shown to be a tyrosine kinase with sequence similarity to mammalian Cdc-42-associated tyrosine kinases. We demonstrate that Drosophila Ack (DAck) can be co-immunoprecipitated with Dock and DSH3PX1 from fly cell extracts. The domains responsible for the in vitro interaction between Drosophila Ack and Dock were identified, and direct protein-protein interactions between complex members were established. We conclude that DSH3PX1 is a substrate for DAck in vivo and in vitro and define one of the major in vitro sites of DSH3PX1 phosphorylation to be Tyr-56. Tyr-56 is located within the SH3 domain of DSH3PX1, placing it in an important position for regulating the binding of proline-rich targets. We demonstrate that Tyr-56 phosphorylation by DAck diminishes the DSH3PX1 SH3 domain interaction with the Wiskott-Aldrich Syndrome protein while enabling DSH3PX1 to associate with Dock. Furthermore, when Tyr-56 is mutated to aspartate or glutamate, the binding to Wiskott-Aldrich Syndrome protein is abrogated. These results suggest that the phosphorylation of DSH3PX1 by DAck targets this sorting nexin to a protein complex that includes Dock, an adaptor protein important for axonal guidance.

To form precise patterns of connections among cells, guidance receptors expressed on neuronal growth cones respond to extracellular signals provided by the developing nervous system (1). Interestingly, a particular guidance signal may act as either a repellent or an attractant dependent solely upon the activity of the intracellular signal transduction pathways of the neuron (2). Thus, understanding how individual neurons trans-late extracellular signals into the cytoskeletal rearrangements necessary for the directed movement of their growth cones requires the molecular identification of protein-protein interactions within the guidance receptor complex.
Previous reports have proposed that Dock, 1 the Drosophila orthologue of the mammalian SH3/SH2 adaptor protein Nck, links guidance signals to changes in the actin cytoskeleton in photoreceptor growth cones (3). Recently, using the epitopetagged SH2 domain of Dock, we biochemically purified five proteins (molecular masses of 270, 145, 74, 69, and 63 kDa), which are present in a putative complex with the SH2 domain of Dock. Further characterization revealed that p270 is a novel receptor-like protein that is a member of the immunoglobulin superfamily and shares extensive sequence similarity to the human Down's Syndrome cell adhesion molecule (Dscam). Additionally, we demonstrated that Dscam lies upstream of Dock signaling and is important for normal axonal pathfinding in the developing nervous system in the fly (4).
The second protein that was characterized from the collection of Dock SH2 domain-interacting proteins was the 63-kDa protein identified as the SH3 and PX domains containing protein, DSH3PX1 (5). DSH3PX1 is a member of the sorting nexin family and is able to interact with both Dock and Dscam (5). Importantly, we also demonstrated that the SH3 domain of DSH3PX1 interacts with Wiskott-Aldrich Syndrome protein (WASP), a protein involved in regulating the actin cytoskeleton (5). This finding suggests that DSH3PX1 may link the Dscam receptor complex to cytoskeletal modification machinery.
Here we describe the identification of the 145-kDa protein, which we have named Drosophila Ack. Mammalian ACKs are tyrosine kinases that are specifically activated by Cdc42 but not by Rac or Rho (6,7). ACK1 is activated by the stimulation of M3-muscarinic receptors via a mechanism that requires the Src-like kinase, Fyn (8). The association of ACK1 with Cdc42 may result in sustained activation of Cdc42 (6). In addition, ACK1 has recently been shown to interact with Nck via its proline-rich regions and with clathrin via a clathrinbinding motif (9). ACK2, a smaller splice variant of ACK1, is stimulated by bradykinin, epidermal growth factor, or integrin ␤ 1 -mediated cell adhesion (10). Similar to ACK1, ACK2 activity is regulated by Src, and it also interacts with clathrin (11,12). Recently, ACK-related tyrosine kinase-1, the Caenorhabditis elegans orthologue of ACK1, was shown to neg-atively regulate epidermal growth factor receptor signaling in a Grb-2-dependent manner (13).
This report describes the isolation, identification, and characterization of DAck. We demonstrate that DAck co-immunoprecipitates in a cellular complex with Dock and DSH3PX1 and confirm these observations with a systematic yeast two-hybrid analysis. The specific domains by which Dock directly interacts with DAck are identified in vitro by GST pull-down assays employing various GST-Dock fusion proteins. We also demonstrate that DSH3PX1 is a substrate for DAck in vivo and in vitro and determine that a major site of DSH3PX1 phosphorylation is a conserved tyrosine residue (Tyr-56) present in its SH3 domain. The phosphorylation of this tyrosine residue reduces the binding of DSH3PX1 with WASP, an interaction previously shown to require the DSH3PX1 SH3 domain (5). Instead, tyrosine-phosphorylated DSH3PX1 interacts with Dock via its SH2 domain. This finding suggests that phosphorylation of DSH3PX1 by DAck changes its preference for protein interaction partners.
S2 cells were transfected with the Drosophila expression vector pMT/ V5-His (Invitrogen) containing wild type or mutant DAck-coding sequences for the generation of stable transformants using FuGENE 6 (Roche Molecular Biochemicals) as described previously (5). To induce protein synthesis, CuSO 4 was added to the pMT/V5-His stable transformants at a final concentration of 500 M 48 h prior to harvest.
Two-hybrid Screen-The selected two-hybrid screen was run as described previously (5). The ␤-galactosidase assay was scored as blue color apparent by 15 min (ϩϩϩ), blue color apparent by 30 min (ϩϩ), blue color apparent by 45 min (ϩ), and either no blue color apparent or blue color apparent after 1 h. (Ϫ).
Antibody Affinity Purification, Co-immunoprecipitation, and Western Analysis-Polyclonal antisera were raised in rabbits against recombinant fusion proteins for Dock, DSH3PX1, and DAck (Cocalico, Reasmtown, PA). The p-Tyr antibody (4G10) and V5-epitope antibody were purchased from Upstate Biotechnology and Invitrogen, respectively. Affinity columns for purification of antisera were produced by coupling the appropriate recombinant fusion protein to an Affi-Gel support (Bio-Rad) according to the protocol supplied by the manufacturer. Eluted fractions containing antibody were combined and concentrated in 1ϫ phosphate-buffered saline using a Centricon plus-20 centrifugal filtration device (Millipore). Affinity-purified antibodies were stored in aliquots at Ϫ20°C.
GST-Dock/GST-SH3PX1 in Vitro Pull-down Assay-S2 cell extract was prepared by lysing 2 ϫ10 7 cells/ml RIPA buffer as described above. GST-tagged proteins for each Dock SH3 domain, its SH2 domain, and the SH3 domain of DSH3PX1 were expressed in bacteria and purified on glutathione-agarose (15). 1 ml of S2 cell extract (ϳ2 ϫ 10 7 cells/ml) prepared in lysis buffer (50 mM Tris, pH 8.0, 100 mM NaCl, 1% Nonidet P-40) was mixed with ϳ10 g of protein attached to beads, incubated for 1 h at 4°C with rocking, washed three times with lysis buffer (1 ml each), and resuspended in 50 l of Laemmli loading buffer. 5-10 l were analyzed on a 10% SDS-polyacrylamide gel and Western blotted using antibodies directed against DAck.
In Vitro Kinase Assay-S2 extracts used for DAck kinase reactions were made using RIPA buffer minus deoxycholate and SDS (RIPA kinase). Immunoprecipitations were as follows. Lysates were cleared by centrifugation at 14,000 ϫ g at 4°C for 20 min and incubated at a 4°C mixing for 1 h with 25 l of FLAG beads (Sigma). Beads were washed three times with 1 ml of RIPA kinase, four times with 1 ml of RIPA kinase containing 1 M NaCl, and 2 times with kinase buffer (25 mM Tris, pH 7.4, 3 mM MgCl 2 , 3 mM MnCl 2 , 0.5 mM EGTA, 0.1 mM NaVO 4 , 0.5 mM dithiothreitol). 10 l of FLAG beads were mixed with 10 g of bacterially expressed wild type or mutant DSH3PX1, incubated for 30 min at 25°C in the presence of 100 M ATP spiked with 10 Ci of [␥-32 P]ATP (6000 Ci/mmol). Mutants were made in GST-DSH3PX1 using the QuikChange TM site-directed mutagenesis kit (Stratagene). The reaction was terminated by adding Laemmli loading buffer and analyzed by SDS gel electrophoresis followed by autoradiography.
Pull-down Analysis Using Recombinant Dock and Immunoprecipitated WASP-The in vitro kinase reaction was performed as described previously with the exception that ϳ1-2 g of recombinant GST-DSH3PX1 was used in the kinase reaction along with DAck (released from the FLAG beads by the addition of 50 g of FLAG peptide) in a final volume of 100 l for 2-3 h at 25°C. Under these conditions the stoichiometry of phosphorylation of DSH3PX1 ranges between 30 and 50%. His-tagged Dock was purified from bacteria using Ni-agarose as described by the manufacturer (Qiagen). V5-tagged WASP stably expressed in S2 cells was immunoprecipitated from cells treated with dsRNA directed against DSH3PX1 (DSH3PX1 RNAi) as described previously (5). Approximately 0.5-1 g of His-tagged Dock and immunoprecipitated WASP were used for each binding reaction as analyzed by silver staining. Binding reactions (200 l) containing 20 l of the protein affinity resins and 20 l of the kinase reaction (Ϯ kinase) were rocked for 1 h at 4°C followed by three washes using RIPA kinase buffer. Beads were resuspended in 50 l of Laemmli sample buffer and analyzed by SDS-PAGE followed by Western analysis using anti-DSH3PX1 (1:5000).

RESULTS AND DISCUSSION
Dock was first described by Garrity et al. (3) as an adaptor protein playing a critical role in axonal pathfinding. Because tyrosine phosphorylation is known to play an important role in the development of the fly nervous system (16 -18), we were interested in the interactions directed by the SH2 domain of Dock. In an effort to identify tyrosine-phosphorylated proteins that interact with Dock, the SH2 domain of Dock was overexpressed as a His-tagged protein in Drosophila S2 cells (4,5). 2 Protein lysates were prepared from S2 cells and subjected to Ni-agarose affinity chromatography followed by purification on a p-Tyr-specific antibody (4G10) affinity column. Proteins eluted from the anti-p-Tyr affinity column were separated by SDS-PAGE electrophoresis and visualized by Coomassie Blue staining. A protein of 145 kDa was excised from the gel and digested with trypsin, and the peptides were purified by high pressure liquid chromatography for sequence analysis.
Sequence determination of the tryptic peptides obtained from p145 resulted in seven unambiguous amino acid sequences ( Fig. 1, underlined) that were used to search the Berkeley Drosophila genome data base for corresponding expressed sequence tags. Four overlapping expressed sequence tags, GH10884, GH07655, GH11561, and LD18821, in conjunction with 5Ј-RACE products were completely sequenced in order to obtain a full-length nucleotide sequence encoding p145. The cDNA contains an open reading frame of 3219 nucleotides encoding a tyrosine kinase (light gray box) with sequence similarity to mammalian kinases of the ACK family. Immediately following the kinase domain is an SH3 domain (dark gray box) followed by a COOH-terminal ubiquitin-associated (UBA) domain (black box) (Fig. 1). The UBA domain is a motif found in UV excision repair proteins, certain protein kinases, and proteins playing roles in the ubiquitination pathway (19). Although the role of this domain is not known, it has been suggested that UBA domains are involved in conferring target specificity to enzymes of the ubiquitination system (19). To date, there have been no reports of ACK ubiquitination, but we have observed that DAck is rapidly processed into smaller fragments in S2 cell extracts (data not shown).
Between the SH3 domain and the UBA domain, the COOHterminal sequence of DAck contains four proline-rich (PXXP) motifs (Fig. 1, clear boxes), which could be involved in interactions with SH3 domains. There are also six YXX motifs (boldface) that are thought to be important in vesicle trafficking particularly with regard to interactions with the adaptor proteins in clathrin-coated vesicles (Fig. 1) (20). Moreover, a putative clathrin-binding core motif LIDIS is also found in this region of DAck (Fig. 1, gray box with white lettering). Mammalian ACK1 and ACK2 have recently been shown to interact directly with clathrin using this conserved motif (9,11). The most distinctive differences between mammalian ACKs and DAck are the apparent absence of a Cdc42-binding domain and the presence of a UBA domain in the Drosophila enzyme. The apparent absence of a Cdc42-binding domain in DAck raises the intriguing possibility that this enzyme may be activated by a different mechanism from mammalian ACKs.
Because DAck and DSH3PX1 were purified from S2 cell extracts using the SH2 domain of Dock, we thought it was important to determine whether these molecules associate with one another in vivo. DSH3PX1 can be co-immunoprecipitated with both DAck and Dock from S2 cell extracts ( Fig. 2A) (5). Furthermore, DAck can also be co-immunoprecipitated with Dock (Fig. 2B). Because DAck contains several proline-rich motifs, we wanted to know whether the SH3 domains of Dock and/or its SH2 domain interact with DAck. GST fusions of SH3 domains and SH2 domain of Dock were produced in bacteria and were used in a pull-down assay in conjunction with S2 cell extracts as the source of interacting proteins. Under these conditions, Dock interacts with DAck via its SH2 domain (Fig.  2C). In the case of mammalian Nck and ACK1, it has been reported that the SH3 domains of Nck are able to bind to ACK1 in an overlay assay (9). The discrepancy between our results and Teo et al. (9) most probably results from the type of assay employed. It is possible that DAck isolated from S2 cell extract is conformationally unavailable for binding to the SH3 domains of Dock. However, we also tested the ability of the DSH3PX1 SH3 domain to interact with DAck. As can be seen in Fig. 2C, the SH3 domain of DSH3PX1 interacts strongly with DAck in S2 cell extracts, suggesting that at least one of the proline-rich (PXXP) motifs of DAck is available for protein-protein interactions.
Because DAck was isolated by virtue of its ability to interact with the Dock SH2 domain, we sought to dissect the potential protein-protein interactions between Dock complex proteins using a directed two-hybrid screen (5). Dock interacts strongly with full-length DAck but not with a COOHterminally truncated DAck construct consisting of amino acids 1-507 (Table I). It was not possible to use full-length DAck as a bait because of autoactivation. Because the DAck constructs encode the active kinase, the removal of the COOH-terminal sequence most probably removes an autophosphorylation site that is important for Dock SH2 domain binding (Fig. 2C). Full-length DSH3PX1 as well as the DSH3PX1 SH3 domain also interact strongly with full-length DAck but not with truncated DAck, presumably because of the removal of one or more of the COOH-terminal PXXP motifs of DAck. As reported previously (4, 5), Dock and the DSH3PX1 SH3 domain are capable of interacting with Dscam. However, DAck (amino acids 1-507) that contains the kinase domain plus the SH3 domain is not capable of interacting with any of the Dock complex proteins tested. Therefore, the binding partners for the DAck SH3 domain are most probably not present in this group. Interestingly, our twohybrid results suggest that Dock is able to interact with itself. This possibility would effectively increase the ability of Dock to bring together a variety of molecules simultaneously.
Because DAck co-immunoprecipitates ( Fig. 2A) and directly interacts (Table I) with DSH3PX1, we sought to determine whether DSH3PX1 is an in vivo substrate for this kinase. dsRNAs were used to ablate the expression of DAck and other intracellular Src-related tyrosine kinases and the resulting extracts separated by SDS-PAGE followed by Western analysis using antibodies directed against phosphotyrosine (Fig. 3A). Equal loading of DSH3PX1 was confirmed by stripping and reprobing the Western with antibody recognizing DSH3PX1 (Fig. 3A). In S2 cell extracts, the ablation of DAck, Src42A, or Src64 each reduce the level of tyrosinephosphorylated DSH3PX1. Simultaneous ablation of all three kinases completely obliterates Tyr phosphorylation of DSH3PX1. Alternatively, the removal of Shark and Abl did not effect the level of SH3PX1 tyrosine phosphorylation. These results suggest that Src42A, Src64, and DAck are upstream of DSH3PX1 phosphorylation. Our ability to coimmunoprecipitate DSH3PX1 and DAck leads us to speculate that DAck is an important cellular kinase for DSH3PX1 tyrosine phosphorylation. Furthermore, it has been reported that the Src-like kinase, Fyn, phosphorylates and activates mammalian ACK1 (8) and that ACK2 activity is similarly controlled by Src phosphorylation (12). Therefore, the reduction in DSH3PX1 phosphorylation observed in the Src RNAi examples may be the result of reduced DAck activity. To demonstrate that DAck can directly phosphorylate DSH3PX1, DAck engineered to contain an NH 2 -terminal FLAG tag was stably expressed in S2 cells. This kinase was immunoprecipitated from S2 cells using FLAG beads followed by extensive washing in high salt buffer to remove any co-precipitating proteins. Precipitated DAck was then added to an in vitro kinase assay containing bacterially expressed DSH3PX1 as the substrate (Fig. 3B). Under these experimental conditions, DAck is able to efficiently phosphorylate DSH3PX1. A kinase-dead version of DAck made by mutating Lys-156 to Ala, similarly expressed and immunoprecipitated from S2 cells, was not able to phosphorylate itself or DSH3PX1 (Fig. 3B). This finding indicated that DAck and not a co-precipitating kinase was responsible for DAck phosphorylation. In addition, immunoprecipitated DAck only phosphorylates itself in the absence of added substrate, demonstrating that no potential substrates have co-immunoprecipitated with the active kinase under the experimental conditions used. Taken together, DSH3PX1 is an in vitro substrate for DAck, and our RNAi data suggest that it is also an in vivo substrate.
To elucidate the role of DAck phosphorylation on DSH3PX1 function, the site of phosphorylation by DAck was determined. A comparison of the amino acid sequences among mammalian, C. elegans, and Drosophila SH3PX1 highlighted the conservation of several tyrosine residues including Tyr-9, Tyr-56, and Tyr-256 (Drosophila numbering) (5). These Tyr residues were mutated to Phe in the context of the bacterial expression vector, GST-DSH3PX1. Mutant and wild type proteins were expressed in bacteria and subjected to the in vitro kinase assay as described above. FLAG-tagged wild type DAck was immunoprecipitated from S2 extracts as described under "Materials and Methods" and used in the in vitro kinase assays along with the bacterially expressed wild type and mutant DSH3PX1 proteins (Fig. 3C). The mutation of Tyr-56 to Phe (Y56F) significantly reduced the ability of DAck to phosphorylate DSH3PX1, indicating that Tyr-56 is a major site of Tyr phosphorylation (Fig. 3C). Residual phosphorylation persists, suggesting that more than one Tyr in DSH3PX1 is phosphorylated by DAck. However, the mutation of Tyr-9 and Tyr-256 do not dramatically reduce the level of DSH3PX1 tyrosine phosphorylation, indicating that these residues are not candidates for DAck phosphorylation (Fig. 3C). It is interesting that Tyr-56 falls in the polyproline recognition site of the DSH3PX1 SH3 domain. Based on mutational studies on Src-family SH3 domains (21), we predict that the phosphorylation of Tyr-56 would interfere with the ability of the DSH3PX1 SH3 domain to bind prolinerich ligands. Because we previously demonstrated that DSH3PX1 interacts with WASP via its SH3 domain, we predict that the phosphorylation of Tyr-56 will interfere with WASP binding. Similar results have been reported for the SH3 do-  (22,23). In the case of PSTPIP, the phosphorylation of Tyr-367, which resides in its SH3 domain polyproline-binding pocket interferes with WASP binding (22). For Bruton's tyrosine kinase, autophosphorylation of a specific tyrosine residue in its SH3 domain abolishes its interaction with WASP while preserving its ability to interact with c-Cbl (23). These reports provide evidence that a common theme may exist for regulating SH3 domain-polyproline interactions involving WASP.
To test this theory, DSH3PX1 phosphorylated in vitro by DAck was allowed to interact with either Dock or WASP immobilized on affinity resins. After extensive washing, the amount of DSH3PX1 bound to each protein was visualized by Western analysis. In the case of Dock, His-tagged protein was produced in bacteria and attached to Ni-agarose resin. Because we could not produce recombinant WASP in bacteria, V5-tagged WASP was overexpressed and immunoprecipitated from S2 cells as described previously (5). As can be seen in Fig. 4A, phosphorylated DSH3PX1 binds efficiently to epitope-tagged Dock, whereas nonphosphorylated-DSH3PX1 does not. This demonstrates that the phosphorylation of DSH3PX1 by DAck generates a binding site for the Dock SH2 domain. On the other hand, nonphosphorylated DSH3PX1 binds efficiently to WASP, whereas P-DSH3PX1 displays reduced binding (Fig. 4A). The interaction between P-DSH3PX1 and WASP can be explained by the incomplete phosphorylation of DSH3PX1 in our kinase assay and from the ability of nonphosphorylated DSH3PX1 to dimerize with P-DSH3PX1. Therefore, the above data suggest that the binding of DSH3PX1 to WASP may be regulated by phosphorylation of Tyr-56. To further test this possibility, Tyr-56 was mutated to either aspartate or glutamate to mimic the incorporation of the negatively charged phosphate group at this position. Modeling of the DSH3PX1 SH3 domain indicates that Tyr-56 is solvent-exposed, suggesting that these mutations should not globally effect the structure of the SH3 domain. DSH3PX1 containing either the Y56D or the Y56E FIG. 3. DAck, Src42A, and Src64 are cellular tyrosine kinases for DSH3PX1. A, profile of tyrosine-phosphorylated DSH3PX1 in S2 cell extracts treated with dsRNAs for various tyrosine kinases. Protein extracts from untreated S2 cells or S2 cells treated with the indicated dsRNAs directed against cytoplasmic tyrosine kinases were analyzed by SDS-PAGE and Western blot using p-Tyr antibodies and DSH3PX1 antibodies. The top panel represents the p-Tyr signal obtained for DSH3PX1, and the bottom panel represents the signal obtained for anti-DSH3PX1. B, in vitro kinase assay using immunoprecipitated DAck. NH 2 -terminal FLAG-tagged kinase-dead or wild type DAck was stably expressed in S2 cells. Wild type and mutant DAck were immunoprecipitated from S2 cells using FLAG beads and were combined with ϳ5 g of recombinant GST-DSH3PX1 in an in vitro kinase reaction (see "Materials and Methods"). Phosphorylated proteins were separated by SDS-PAGE and stained with Coomassie Blue to verify equal loading. Phosphorylation was visualized by autoradiography overnight at Ϫ80°C with a screen. C, DAck in vitro kinase assay with DSH3PX1 mutants. Conserved Tyr residues in GST-DSH3PX1 were mutated to Phe, and the recombinant proteins were expressed and purified from bacteria. FLAG-tagged DAck was immunoprecipitated from S2 cells and mixed with recombinant DSH3PX1 in an in vitro kinase assay as described under "Materials and Methods." Phosphorylated proteins were separated by SDS-PAGE and stained with Coomassie Blue to verify equal loading. Phosphorylation was visualized by autoradiography overnight at Ϫ80°C with a screen.

FIG. 4. Phosphorylation of DSH3PX1 determines its binding partners.
A, equal amounts of nonphosphorylated (NP) GST-DSH3PX1 or phosphorylated (P) GST-DSH3PX1 were allowed to interact with bacterially expressed His-Dock or immunoprecipitated V5tagged WASP in a pull-down analysis. The stoichiometry of phosphorylation of DSH3PX1 for this experiment was 48%. The load of His-Dock and immunoprecipitated WASP is shown in the silver-stained panels. B, wild type (WT) GST-DSH3PX1 or the indicated Tyr-56 mutants were allowed to interact with WASP in a pull-down analysis. In each case, DSH3PX1 binding was analyzed by Western analysis. mutation is no longer able to bind to WASP (Fig. 4B). As expected, the conservative Y56F mutation does not affect the ability of DSH3PX1 to bind to WASP. Taken together, our results suggest that WASP most probably interacts with nonphosphorylated DSH3PX1, whereas phosphorylated DSH3PX1 is targeted for Dock interactions. It is intriguing that the dimers of DSH3PX1 could create bridges between PXXP motif containing proteins such as WASP and SH2 domain containing proteins such as Dock depending upon the phosphorylation state of their SH3 domains.
In summary, we have identified DAck as a member of a complex of proteins involved in axonal guidance via its association with Dock. It is important to note that experiments in fly embryos using dsRNAs directed against Dock and DAck result in similar axonal pathfinding defects as assayed by Bolwig's organ development. 2 Given the ability of DAck to interact with DSH3PX1, a potential sorting nexin that associates with the clathrin-coated adaptor protein 50 (5,24), and the ability of mammalian ACK1 to interact with clathrin (9,11), it is tempting to speculate that DAck is involved in regulating the extracellular presentation of Dscam and/or other Dock-associated receptors by endocytosis via clathrincoated pits. This speculation is further supported by the role of C. elegans ACK-related tyrosine kinase-1 in downregulating Let-23, the C. elegans epidermal growth factor receptor orthologue (13). Furthermore, we identify DSH3PX1 as a substrate for DAck and demonstrate that phosphorylation of DSH3PX1 probably increases its interaction with Dock while decreasing its interaction with WASP. The protein complex consisting of Dscam, Dock, DAck, and DSH3PX1 and their known protein-protein interactions are summarized in Fig. 5. In this scenario, DAck acts as a molecular switch to control DSH3PX1 protein-protein interactions. Nonphosphorylated DSH3PX1 interacts strongly with WASP, a known modulator of the actin cytoskeleton. When phosphorylated, DSH3PX1 interacts preferentially with Dock. In addition, the PX domain of DSH3PX1 interacts with phospholipids, thereby targeting DSH3PX1 to specific cellu-lar membranes (data not shown, reviewed in Ref. 25). We are interested in understanding how the Dock SH2 domain chooses among its binding partners, i.e. Dscam versus DAck versus DSH3PX1, and how the resulting protein complexes ultimately influence neurite outgrowth. For now, the complexity of the protein-protein interactions involving Dock preclude us from directly linking a specific Dock protein complex to specific changes in the actin cytoskeleton. Nevertheless, Dock and the proteins recruited by Dock are clearly instrumental in signaling changes in the actin cytoskeleton that are required for directed axonal growth. Green lines indicate protein-protein interactions specified by SH3 domains interacting with proline-rich sequences (horizontal hatched boxes). Red lines indicate protein-protein interactions involving p-Tyr residues interacting with SH2 domains. The loops in the Dscam schematic represent IgG 2 repeats, and the blackened ovals represent fibronectin repeats. In DSH3PX1, CC represents coiled-coil.