DSCAM functions as a netrin receptor in commissural axon pathfinding

Abstract

Down syndrome cell adhesion molecule (DSCAM) is required for axon guidance and dendrite arborization. How DSCAM functions in vertebrates is not well understood. Here we show that DSCAM is expressed on commissural axons and interacts with Netrin-1, a prototypical guidance cue for commissural axons. The knockdown of DSCAM by specific siRNA or blockage of DSCAM signaling by overexpression of a mutant lacking its intracellular domain inhibits netrin-induced axon outgrowth and commissural axon turning in vitro. SiRNA-mediated knockdown of DSCAM in ovo causes defects in commissural axon projection and pathfinding. In transfected cells, DSCAM by itself, in the absence of DCC, is capable of mediating netrin signaling in activating phosphorylation of Fyn and Pak1. These findings demonstrate an essential role of vertebrate DSCAM in axon guidance, indicating that DSCAM functions as a receptor of netrin-1. Our data suggest previously unexpected complexity in receptors that mediate vertebrate netrin signaling.

Netrins, a conserved family of secreted proteins, can promote axon outgrowth and guide growth cone navigation in species ranging from Caenorhabditis elegans to mammals (1–5, 37). Receptors for UNC-6/netrin have been identified in C. elegans as UNC-40 and UNC-5 (1, 6, 7). The mammalian homologs of UNC-40 are Deleted in Colorectal Cancer (DCC) and neogenin (8, 9). Netrins can act as either axon attractants or axon repellents. DCC/UNC40 can mediate both attractive and repulsive responses, whereas UNC-5 mediates repulsion (6, 8–13). In the embryonic spinal cord, DCC mediates the attractive effect of netrin in commissural axons (3, 9). DCC-deficient mice exhibit defects in commissural axon projections that are similar to those seen in netrin-1–deficient mice, with reduction in, shortening of, and misguidance of commissural axons (14, 15). Some commissural axons in DCC knockout mice still project to and cross the floor plate, however, suggesting that other guidance receptors may be involved in this process (14, 15).

The human DSCAM gene was originally identified as a gene associated with mental retardation (16). It encodes a protein of the Ig superfamily containing 10 Ig domains, 6 fibronectin type III (Fn III) domains, 1 transmembrane, and 1 intracellular domain (16). The Drosophila homolog of human DSCAM has been isolated in a screen for tyrosine-phosphorylated proteins interacting with Dock, an intracellular adaptor protein homologous to mammalian Nck (17). The fly Dscam gene has an amazing molecular diversity, with 38,016 potential alternative splicing isoforms, and is required for neuronal wiring (17–21; reviewed in ref. 22); however, the vertebrate DSCAM gene encodes only a few splicing isoforms (23). Mouse DSCAM is expressed widely in the developing nervous system (16, 24). Recent studies indicate that DSCAM plays an important role in neurite arborization, cell body spacing, and lamina-specific synaptic targeting in vertebrate retina (25, 26).

In a previous study, we found that human DSCAM can bind to p21-activated kinase 1 (Pak1) and stimulate Pak1 activity (27). Netrin-1 also activates Pak1 (28). These observations prompted us to test the role of DSCAM in netrin signaling. Here we report that DSCAM interacts with netrin-1. DSCAM is expressed on the spinal commissural axons as they extend to and across the floor plate. In vitro knockdown of DSCAM inhibits netrin-induced axon outgrowth and commissural axon attraction. Knockdown of DSCAM in ovo causes defects in commissural axon pathfinding in chick neural tube. These results indicate that DSCAM is a receptor required for netrin-dependent commissural axon outgrowth and pathfinding. While our work was under review, a study was published indicating that DSCAM functions as a netrin receptor in collaboration with DCC (29).

Results

Expression of DSCAM in Commissural Neurons of the Neural Tube.

In situ hybridization studies have shown that DSCAM is expressed throughout the developing mammalian nervous system (16, 30). To examine the expression of DSCAM protein, we carried out immunostaining of mouse spinal cords using a specific antibody against DSCAM (27) and compared the results with the staining pattern of TAG-1, an antibody specifically recognizing commissural axons (Fig. 1A, D, and G). At E11.5, DSCAM is expressed in the spinal cord, including the motor columns, motor axons, dorsal root ganglions, commissural axons, and ventral funiculus [Fig. 1B, C, E, F, and H; supporting information (SI) Fig. S1D]. In dissociated E11.5 commissural neurons, DSCAM was detected in the soma and on the cell membrane, axon, and growth cone (Fig. 1H, K and N; Fig. S1 A–C). A similar expression pattern was found in E15 cortical neurons (Fig. S1 J and M). Co-localization with TAG-1 indicated that DSCAM is expressed in commissural axons (Fig. 1C, F, and I). Confocal immunofluorescent microscopy showed that DCC and DSCAM are partially co-localized in commissural neurons (Fig. 1 J–O) and cortical neurons (Fig. S1 I–N). DSCAM expression in the embryonic spinal cord in chicken is similar to that in mice (Fig. S1 E–H).

Fig. 1.

DSCAM expression in the developing spinal cord. (A–F) Expression of TAG-1 (A, C, D, and F) and DSCAM (B, C, E, and F) in transverse sections of the E11.5 mouse spinal cord was detected by confocal microscopy after immunostaining with anti-TAG-1 and anti-DSCAM antibodies, respectively. Panels C, F, and I are overlay images of panels A with B, D with E, and G with H, respectively. Panels D, E, and F are higher-magnification images of the ventral spinal cord of A–C. (Scale bar: 100 μm.) (G–N) Expression of TAG-1 (G and I) and DSCAM (H and I) and of DCC (J and M) and DSCAM (K and N) in dissociated E11.5 spinal cord neurons. Panels L and O are the overlay images of panels J with K and M with N, respectively. (Scale bars: 10 μm.)

Biochemical Characterization of DSCAM Interaction With Netrin-1.

To examine a potential interaction between netrin-1 and DSCAM, we transfected HEK293 cells using plasmids expressing netrin-1 tagged with HA (netrin-HA) and Flag-tagged full-length human DSCAM (DSCAM-Flag), DSCAM mutants [lacking the extracellular domain (DSCAMΔN) or the intracellular domain (DSCAMΔC), as shown in Fig. 2A]), or DCC. As expected, netrin-HA was co-immunoprecipitated by DCC (Fig. 2B, lane 4). Either the full-length DSCAM or DSCAMΔC, but not DSCAMΔN, co-immunoprecipitated with netrin, demonstrating that the extracellular domain of DSCAM is required for its interaction with netrin [Fig. 2B (lane 3) and C].

Fig. 2.

Netrin-1 interacts with DSCAM and activates protein phophorylation in DSCAM-expressing cells. The control vector, or DSCAM (wild-type or mutant), together other plasmids, as indicated, were transfected into HEK293 cells, then treated with netrin or the control vehicle. Immunoprecipation (IP) and Western blot (WB) experiments were carried out using specific antibodies. (A) A diagram illustrating the full length and the mutant DSCAM proteins. (B) DSCAM co-immunoprecipitated with netrin-1, similar to DCC. (C) The full-length DSCAM or DSCAMΔC, but not DSCAMΔN, interacted with netrin-1 in co-immunoprecipitation. (D) Netrin-1 induced Pak1 phosphorylation in DSCAM-expressing cells. After transfection of HEK293 cells with DSCAM-Flag or/and HA-Pak1 plasmids, Pak1 protein was immunoprecipitated using HA antibody. Phospho-Pak1 and the total Pak1 protein were detected by Western blot analysis using anti–phospho-Pak1 (S423) and anti-HA antibodies, respectively. (E) Netrin-1 enhanced tyrosine phosphorylation of DSCAM. HEK293 cells transfected with DSCAM-Flag were treated with netrin for 5 minutes (lane 2) or 20 minutes (lane 3). DSCAM protein was immunoprecipitated using anti-Flag antibody, followed by Western blot analysis with phospho-tyrosine antibody (4G10). (F) DSCAM or DCC alone was sufficient to mediate netrin-induced phosphorylation of Fyn. DCC-myc or DSCAM-flag plasmids were transfected into HEK293 cells individually or in combination. After netrin stimulation, Fyn phosphorylation was detected by Western blot analysis using anti–phospho-tyrosine after immunoprecipitation using a specific anti-Fyn antibody.

DSCAM interacts with Pak1, a serine/threonine kinase (27). To investigate whether DSCAM is involved in netrin-induced Pak1 phosphorylation, we co-transfected DSCAM with Pak1 into HEK293 cells. DSCAM expression alone was sufficient to activate Pak1 phosphorylation (Fig. 2D; compare lanes 4 and 1), consistent with findings in our previous study (27). Addition of netrin-1 further increased phospho-Pak1 in the presence of DSCAM (Fig. 2D, lane 3), whereas netrin-1 alone in the absence of DCC or DSCAM did not stimulate Pak1 phosphorylation (Fig. 2D; compare lanes 2 and 1). Netrin treatment also stimulated tyrosine phosphorylation of DSCAM (Fig. 2E). In addition, phosphorylation of Fyn, a Src family kinase required for netrin signaling (31), was increased in the presence of either DCC or DSCAM after netrin stimulation (Fig. 2F; compare lanes 5 and 6 and lanes 3 and 4). In multiple experiments, we found no further increase in netrin-induced Fyn phosphorylation when both DSCAM and DCC were expressed in HEK293 cells (Fig. 2F; compare lanes 1 and 3). These findings indicate that DSCAM expression alone in HEK293 cells is sufficient to mediate netrin-induced phosphorylation of Pak1 and Fyn (Fig. 2).

Binding of Netrin-1 to DSCAM Expressed on the Cell Surface.

To examine whether netrin-1 binds to DSCAM protein expressed on the cell surface, we transfected the DSCAM cDNA plasmid (Fig. 3 B and I) into HEK293 cells, incubated these cells with netrin-1 protein, and then examined netrin-1 binding by immunostaining using a specific anti–netrin-1 antibody (Fig. 3 C and J). HEK293 cells expressing Robo1, another member of the Ig superfamily expressed on the commissural axons (32–34), did not bind to netrin (Fig. 3 F–H). Co-immunoprecipitation experiments also demonstrated that Robo1 did not bind to netrin (Fig. S2A). Similarly, no interaction was detected between DSCAM and Semaphorin3A, another neuronal guidance molecule (Fig. S2C). We further characterized the netrin–DSCAM interaction using an assay used previously to examine netrin–DCC and Slit–Robo interactions (9, 32). DSCAM-expressing cells were incubated with netrin-1 tagged with alkaline phosphatase (netrin-AP), and netrin binding to DSCAM was determined by measuring AP activity bound to the cells. An apparent dissociation constant (K_d) of approximately 12.6 nM was obtained for netrin–DSCAM interaction from the binding curve (Fig. 3 L and M), comparable to that in netrin–DCC and Slit–Robo interactions (9, 32). Together, our findings indicate that the interaction between DSCAM and netrin-1 is specific and that DSCAM does not interact promiscuously with other neuronal guidance molecules.

Fig. 3.

Binding of netrin to DSCAM-expressing cells. HEK293 cells were transfected with Robo-GFP or DSCAM plasmid and then incubated with netrin protein. Netrin bound to the cell surface was detected by the anti-netrin antibody and Cy3-conjugated secondary antibody. DSCAM expression was detected by the anti-DSCAM with Cy2 conjugated secondary antibody. Expression of Robo-GFP was visualized by GFP fluorescence. (A–D) Netrin binding to DSCAM-expressing cells: (A) bright field image; (B) DSCAM expression; (C) netrin-1 binding; (D) overlay of panels B and C. (E–H) The absence of netrin binding on Robo-GFP cells: (E) the bright field image; (F) Robo-GFP expression; (G) no binding of netrin; (H) overlay of panels F and G. (Scale bar, 100 μm.) (I–K) High-magnification confocal images (Z stacks) showing the binding of netrin to DSCAM-expressing cells: (I) expression of DSCAM; (J) netrin binding; (K) overlay of panels I and J. (L and M) Netrin-AP–containing medium was diluted to different concentrations and incubated with either DSCAM-expressing cells or control cells. Free and bound AP activities were measured as described in SI Materials. The apparent K_d was estimated as 12.6 nM.

Involvement of DSCAM in Netrin-Induced Axon Outgrowth.

To examine the function of DSCAM, we designed small interfering RNAs (siRNA) targeting the mouse and chicken DSCAM genes. After testing several siRNAs, we identified one siRNA that significantly reduced the expression of endogenous DSCAM in mouse and chicken neurons without affecting other proteins, such as DCC (Figs. S2D and S3). We used this DSCAM siRNA in subsequent experiments. Co-transfection of the wild-type human DSCAM with DSCAM siRNA restored DSCAM expression level (Fig. S3; data not shown), indicating that human DSCAM is resistant to the DSCAM siRNA.

To evaluate whether DSCAM is involved in axon outgrowth, we co-transfected a Venus YFP plasmid with either the control siRNA (Fig. S4 A, B, and E) or DSCAM siRNA (Fig. S4 C, D, and F) into E15 cortical neurons and then treated the neurons with netrin-1 or the control. Transfection of DSCAM siRNA, but not the control siRNA, inhibited netrin-induced neurite outgrowth without affecting the basal level of axon outgrowth (Fig. S4 A–D). In contrast, DSCAM siRNA did not inhibit BDNF-induced neurite outgrowth (Fig. S4 E and F), indicating a specific effect of DSCAM siRNA. These results demonstrate that DSCAM plays a role in netrin-induced neurite outgrowth in cortical neurons.

To examine the role of DSCAM in commissural axon outgrowth, we used chick dorsal spinal cord explant cultures. DSCAM siRNA or control siRNA together with YFP plasmid were introduced in ovo into chicken neural tubes at stage 12–15, and the YFP-labeled segments were dissected at stage 18–20. Axon outgrowth was quantified by measuring the numbers of axon bundles and the total axon length per explant. In explants transfected with YFP only or with YFP plus control siRNA, netrin-1 significantly induced axon outgrowth (Fig. 4A, B, C, I, and J). DSCAM siRNA, but not control siRNA, significantly inhibited netrin-induced axon outgrowth (Fig. 4 B–D, I, and J). Expression of wild-type human DSCAM, which is resistant to DSCAM siRNA, rescued the effect of DSCAM siRNA on netrin-induced axon outgrowth (Fig. 4 F, I, and J). These results indicate that DSCAM is required for netrin-induced commissural axon outgrowth in vitro.

Fig. 4.

Inhibition of netrin-induced axon outgrowth by DSCAM siRNA. (A–F) Axon outgrowth from YFP-positive dorsal spinal cord neurons transfected with YFP only (A and B), with YFP plus the control siRNA (C), with YFP plus DSCAM siRNA (D and E), or with YFP plus DSCAM-siRNA and wild-type human DSCAM (F) in the presence of netrin (B, C, E, and F) or the control (A and D). Netrin-induced axon outgrowth was inhibited by DSCAM-siRNA (E). The expression of wild-type human DSCAM plasmid reversed the inhibition of netrin-induced axon outgrowth by DSCAM siRNA (F). (G and H) Axon outgrowth from explants transfected with YFP plus control siRNA (G) or YFP plus DSCAM siRNA in the presence of anti-DCC antibody (H). (Scale bar: 100 μm.) (I and J) Quantification of the number of axon bundles and the total length of axons per explant for different groups: YFP plus control siRNA without netrin [5.66 ± 0.65 (0.60 ± 0.09 mm)], YFP only or YFP plus control siRNA with netrin [30.63 ± 1.70 or 34.2 ± 3.5 (3.97 ± 0.30 mm or 3.14 ± 0.54 mm)], DSCAM siRNA [5.76 ± 0.62 (0.59 ± 0.07 mm) without netrin and 16.99 ± 1.42 (1.73 ± 0.20 mm) with netrin], DSCAM siRNA plus wild-type human DSCAM with netrin treatment [26.89 ± 2.31 (3.30 ± 0.41 mm)], anti-DCC with netrin [13.40 ± 1.56 (1.33 ± 0.17 mm)], and DSCAM siRNA plus anti-DCC with netrin [8.88 ± 1.06 (0.75 ± 0.10 mm)]. Data are presented as mean ± SEM. Con, the control treated; Net, the netrin-1–treated group; WT, wild-type human DSCAM; Ab, anti-DCC antibody; siRNA, DSCAM siRNA; Ctl siRNA, control siRNA. The P values for the differences are < .001 between groups I and II (for both I and J), < .001 between groups I and III (for both I and J), < .001 between groups II and VI (for both I and J), < .001 between groups III and VI (for both I and J), .0025 (I) and .0048 (J) between groups VI and VII, and .02 (I) and .007 (J) between groups IV and VIII.

To explore the possibility that DSCAM may function together with DCC in netrin signaling, we used a DCC antibody capable of blocking DCC function (11). Anti-DCC antibody significantly reduced, but did not completely eliminate, netrin-induced axon outgrowth (Fig. 4 C and G; group IV in Fig. 4 I and J). Knockdown of DSCAM by siRNA in the presence of anti-DCC antibody further reduced the netrin-induced axon outgrowth (Fig. 4 E and H; group VIII in Fig. 4 I and J). These results demonstrate that both DCC and DSCAM are required for netrin signaling in stimulating axon outgrowth.

The Role of DSCAM in Axon Attraction by Netrin-1.

To examine the role of DSCAM in axon attraction by netrin-1, we performed a commissural axon turning assay as described previously (31, 35). As illustrated in Fig. 5A, the YFP plasmid was electroporated into the chick neural tube at stage 12–15. The YFP-labeled neural tube was isolated, and the spinal cord was laid out as an “open book” and co-cultured along with an aggregate of HEK cells. More than 90% of the axons projecting from the dorsal spinal cord electroporated at these stages were commissural axons, as demonstrated by immunostaining with the anti–axonin-1 antibody (35). With YFP expression alone, commissural axons turned toward the netrin-secreting cell aggregate (Fig. 5 A, C, and G), whereas most axons projected straight toward the floor plate when co-cultured with the control cell aggregate not secreting netrin (Fig. 5 B and G).

Fig. 5.

DSCAM plays a role in netrin-induced attractive turning of spinal cord axons. Electroporation of the YFP plasmid into chick neural tubes allows visualization of axons. (A) A schematic diagram of the open-book preparation of the spinal cord co-cultured with cell aggregates after electroporation. Netrin-induced axon attraction is illustrated by the turning of axons toward cell aggregates secreting netrin protein. (B) YFP-labeled axons projected straight toward the floor plate when co-cultured with an aggregate of the control HEK cells. (C) Netrin-secreting HEK cells attracted the commissural axons transfected with YFP. The red arrowheads indicate the turning axons. (D) Axons co-expressing YFP and DSCAMΔC projected straight toward the floor plate and were not attracted by the aggregate of netrin-secreting cells. (E) DSCAM siRNA inhibited nerin-mediated attraction of the commissural axons. The red arrows designate the representative axons projecting straight toward the floor plate. (F) Co-transfection of DSCAM siRNA with wild-type human DSCAM plasmid rescued netrin attractive response. (Scale bar: 300 μm.) (G) Quantification of axon turning as described in Materials and Methods. The numbers on the top of each bar (n) indicate the numbers of explants tested in the corresponding groups. Data are presented as mean ± SEM. The percentages of turning axons were 5.7% ± 0.9% for group I (B), 90.7% ± 1.4% for group II (C), 87.2% ± 2.1% for group III (not shown), 43.6% ± 4.7% for group IV (D), 34.7% ± 3.2% for group V (E), and 72.2% ± 3.2% for group VI (F). Con, control cell aggregates; Net, netrin-1–secreting cell aggregates. P values (Student t-test) are all < .0001 between groups I and II, groups III and IV, groups II and V, and groups V and VI.

Co-transfection of YFP with the wild-type DSCAM plasmid did not affect netrin-induced axon attraction (Fig. 5G; microphotograph not shown); however, DSCAMΔC (the mutant lacking the intracellular domain) significantly reduced netrin-induced axon attraction (Fig. 5 D and G). Electroporation of DSCAM siRNA, but not the control siRNA (data not shown), significantly inhibited axon turning toward netrin-secreting cells (Fig. 5 E and G). Furthermore, the expression of wild-type human DSCAM rescued the defects in axon turning caused by DSCAM knockdown (Fig. 5 F and G). In contrast, DSCAM siRNA had no impact on the repulsive effect of Sema3A on dorsal root ganglion axons (data not shown), suggesting that DSCAM is not a nonspecific receptor for different neuronal guidance cues. Taken together, these findings indicate that DSCAM is required for netrin-mediated commissural axon attraction.

The Requirement of DSCAM for Commissural Axon Projection.

We examined the effects of the DSCAM mutant and DSCAM siRNA on commissural axon projection in chick embryos in ovo. The YFP plasmid was electroporated into the neural tube along with wild- type DSCAM or mutant DSCAMΔC. Open-book preparations of spinal cords at stage 23 were immunostained with the anti–axonin-1 antibody (35, 36). More than 90% of the axons expressing YFP stained positive for axonin-1 (Fig. 6 A–O). By stage 23, most commissural axons expressing YFP had reached the floor plate (Fig. 6 G, H, I, P, and Q). The expression of wild type DSCAM did not affect commissural axon projection (Fig. 6 A, B, C, P, and Q); however, expression of DSCAMΔC significantly inhibited commissural axon projection (Fig. 6 D, E, F, P, and Q). Most of the commissural axons in the spinal cord transfected with YFP and control siRNA reached the floor plate (data not shown). In contrast, only a small fraction of commissural axons expressing DSCAM siRNA reached the floor plate (Fig. 6 J, K, L, P, and Q). The effect of DSCAM siRNA on commissural axon projection was reversed by co-transfecting wild-type human DSCAM plasmid (Fig. 6 M–Q).

Fig. 6.

DSCAM is required for commissural axon projection in vivo. Different combinations of plasmids and siRNAs were electroporated into the chick neural tube. (A–C) YFP with wild-type DSCAM plasmid. (D–F) YFP with DSCAMΔC. (G–I) YFP only. (J–L) DSCAM siRNA with YFP. (M–O) DSCAM-siRNA with human DSCAM and YFP plasmids. Panels A, D, G, J, and M show YFP images; panels B, E, H, K, and N show anti–axonin-1 antibody immunostaining; and panels C, F, I, L, and O are merged images. (Scale bar: 100 μm.) (P) Quantification of the percentage of axons reaching the floor plate for different groups: 81.8% ± 4.0% for wild-type DSCAM, 31.9% ± 3.7% for DSCAMΔC, 79.1% ± 2.3% for YFP only, 36.9% ± 4.7% for YFP and DSCAM siRNA, and 76.1% ± 3.4% for DSCAM siRNA and wild-type DSCAM. P values are < .001 between the groups compared: wild-type and DSCAMΔC, YFP and DSCAM siRNA, and DSCAM-siRNA and DSCAM-siRNA plus human DSCAM plasmid (Student t-test). (Q) Quantification of the average distance of axons from the floor plate in different groups: 7.39 ± 1.43 μm for wild-type DSCAM, 55.24 ± 7.72 μm for DSCAMΔC, 9.38 ± 1.26 μm for YFP, 49.25 ± 8.93 μm for DSCAM siRNA, and 16.31 ± 3.51 μm for DSCAM siRNA plus wild-type DSCAM. P values between groups compared are < .0001 for wild-type DSCAM and DSCAMΔC, < .01 for YFP and DSCAM-siRNA, and < .01 for DSCAM-siRNA and DSCAM siRNA plus wild-type DSCAM plasmid. The numbers on the top of each bar (n) indicate the numbers of embryos tested in the corresponding groups. The purple arrowheads indicate the commissural axons stalled before reaching the floor plate.

To further characterize the phenotype of DSCAM knockdown in vivo, we examined transverse sections of the chick spinal cords at stage 23 after electroporation (Fig. 7). When transfected with YFP alone or YFP with wild-type DSCAM, commissural axons projected normally toward the floor plate (Fig. 7 B and C); however, expression of DSCAMΔC or DSCAM siRNA led to misguidance and shortening of commissural axons (Fig. 7 D, F, and H). The mutant lacking the extracellular domain, DSCAMΔN, did not affect commissural axon projection (Fig. 7E). Expression of wild-type human DSCAM, but not DSCAMΔN, rescued the defects in commissural axon projection caused by the DSCAM siRNA (Fig. 7 G and H). These findings demonstrate that DSCAM is required for commissural axon projection and pathfinding.

Fig. 7.

DSCAM is essential for commissural axon pathfinding in vivo. (A) A diagram illustrating the commissural axon projection in a transverse section of the chick spinal cord. The chick neural tube was electroporated with YFP only (B); with YFP plus wild-type DSCAM (C); with YFP plus DSCAMΔC (D); with YFP plus DSCAMΔN (E); with YFP plus DSCAM siRNA (F); with YFP, DSCAM siRNA, and wild-type DSCAM (G); or with YFP with DSCAM-siRNA and DSCAMΔN (H). DSCAM siRNA not only inhibited the commissural axon extension, but also caused aberrant axon turning (F and H). The arrow in panel F and the arrowhead in panel H indicate the misguided commissural axons. Expression of wild-type human DSCAM rescued the defects caused by DSCAM siRNA in commissural axon projection and turning (G). DSCAMΔN expression did not reverse the effects of DSCAM siRNA (H). (Scale bar: 100 μm.)

Discussion

Our experiments provide several lines of evidence supporting the ligand–receptor relationship between netrin-1 and DSCAM. First, DSCAM is expressed on the developing commissural axons when they project toward and across the floor plate (Fig. 1). Second, netrin-1 interacts with the full-length or the extracellular domain of DSCAM (Fig. 2). Soluble netrin-1 binds to the surface of DSCAM-expressing cells with an apparent K_d similar to that of DCC–netrin interaction (Fig. 3). Third, DSCAM expression is sufficient to mediate netrin-induced phosphorylation of Pak1 and Fyn (Fig. 2), important signaling events for netrin function in neurons. Fourth, expression of the DSCAM mutant lacking its intracellular domain reduces the effect of netrin in inducing commissural axon turning and midline crossing (Figs. 5, 6, and 7). Fifth, knockdown of DSCAM by siRNA inhibits netrin-induced commissural axon outgrowth (Fig. 4), turning (Fig. 5), and midline crossing (Figs. 6 and 7).

In a cell surface binding assay, it is difficult to rule out the possibility that netrin-1 binds to a cell surface component induced by DSCAM expression rather than binding directly to DSCAM itself. Other adaptors or cofactors possibly may play a role in netrin signaling mediated by DSCAM. Nonetheless, our data support the need for a specific interaction between DSCAM and netrin for netrin signaling in commissural neurons. The overall phenotypic similarities in the DCC knockout and netrin-1–deficient mice suggest that DCC is important for netrin-1 signaling; however, in DCC^−/− mice, some commissural axons still cross the floor plate, suggesting a DCC-independent mechanism guiding commissural axons to the floor plate (14, 15). Our data suggest that DSCAM may contribute to such a DCC-independent mechanism.

Consistent with our findings, Ly et al. (29) recently reported that DSCAM is a receptor of netrin-1 and that DSCAM is required for netrin-induced commissural axon outgrowth (29). According to these authors, DSCAM siRNA did not affect axon turning, however. Our data demonstrate that DSCAM is required for both netrin-induced axon outgrowth and axon attraction in embryonic chick spinal cord (Figs. 4–7). Furthermore, DSCAM alone in the absence of DCC is sufficient to mediate netrin-induced phosphorylation of Fyn and Pak1 in transfected cells (Fig. 2). The results of our immunostaining experiments suggest that DCC and DSCAM do not have completely overlapping expression, and that different subsets of commissural neurons may exhibit differential DCC and DSCAM expression. Moreover, both DCC and DSCAM have different splicing isoforms that may exhibit differential expression among different subpopulations of neurons. It also is possible that the difference between these 2 studies reflects species differences between the rodents and chicks. Further investigation is needed to identify the factors contributing to the differences between our findings and those of Ly et al. (29).

Whether DSCAM functions alone or as a component of a multisubunit receptor complex remains unclear. Our data indicate that DSCAM is involved in netrin-induced phosphorylation of Pak1, Fyn, and DSCAM itself (Fig. 2). DCC also plays a role in Pak1 phosphorylation (28). The findings from our co-immunoprecipitation experiment suggests that DSCAM does not interact with DCC (Fig. S2B). Immunostaining revealed that DSCAM partially co-localized with DCC in dissociated cortical and spinal cord neurons (Fig. 1 J–O). Either DCC antibody or DSCAM siRNA alone was not sufficient to completely eliminate netrin responses. Blocking of DCC function by the specific DCC antibody, in conjunction with DSCAM siRNA, almost completely abolished netrin signaling in spinal cord axon outgrowth (Fig. 4). These findings suggest that both DCC and DSCAM play a role in netrin signaling. Further studies are needed to gain insight into the relationship between DSCAM and DCC in guiding commissural axon projection during neural development.

Signal transduction mechanisms downstream of DSCAM in the vertebrate nervous system remain poorly understood. In Drosophila, Dock adaptor interacts with the cytoplasmic domain of DSCAM and activates Pak1 (37), and the cytoplasmic domain of DSCAM is required to convert attachment to repulsion between sister axon branches (38). Unlike Drosophila Dscam, human DSCAM directly binds to and activates Pak1 (27). DSCAM also activates both JNK and p38MAP kinases (27). Our data indicate that DSCAMΔC (the mutant lacking the intracellular domain) inhibits commissural axon outgrowth, axon turning, and pathfinding in vitro and in vivo (Figs. 5–7), suggesting that the cytoplasmic domain of DSCAM is important for mediating netrin signaling. Tyrosine phosphorylation of FAK, Fyn, and p130^CAS is required for netrin-induced axon outgrowth and turning (31, 35, 39, 40). The netrin–DCC signaling complex includes Cdc42, Rac1, Pak1, and N-WASP (28). Our findings indicate that DSCAM is involved in netrin-induced phosphorylation of Pak1 and Fyn. More research is needed to investigate whether other signaling molecules, such as FAK, p130^CAS, Rac, Cdc42, or N-WASP, also act downstream of DSCAM, and also how the molecules downstream of different netrin receptors coordinate with one another to mediate netrin signaling in axon outgrowth and pathfinding.

Materials and Methods

Antibodies, Plasmids, and and siRNAs.

see SI Text.

Cell Culture, Transfection, Immunoprecipitation, and Immunostaining.

Cell culture, transfection, and immunoprecipitation were performed as described previously (32). Double staining of DSCAM with axonin-1 or TAG-1 or DCC was carried out using the same protocol as described in our previous study (35). Dissociated mouse cortical neurons were cultured and transfected as described previously (31). For details, see SI Materials and Methods.

Cell Surface Binding.

HEK293 cells grown in 12-well dishes were transfected with DSCAM, Robo-GFP, or DsRed plasmids using the calcium phosphate method as described previously (32). Approximately 40 h after transfection, cells were incubated with netrin-1– or Sema3A-containing media. Double staining of DSCAM and bound netrin-1 or Sema3A was carried out using the affinity-purified anti-DSCAM antibody and an anti–netrin-1 antibody or anti-myc antibody. The bound netrin-1 or Sema3A was visualized using a Cy3- or Cy2-labeled secondary antibody. Netrin-AP was used for equilibrium binding to cells expressing DSCAM or the vector control, and Scatchard analysis was carried out as described previously (9, 32). For details, see SI Materials and Methods.

Chick Spinal Cord Explant Culture, Axon Outgrowth Assay, Axon Turning Assay, and Commissural Axon Projection In Vivo.

White leghorn chicken embryos were collected and staged according to methods outlined by Hamburger and Hamilton (41). Chick spinal cord explant culture, electroporation, and analysis of axon outgrowth were performed as described previously (31, 35). The numbers of axons and total axon length were measured using the National Institutes of Health Image J software. Electroporation, open-book preparation, co-culture with control or netrin-secreting cell aggregates, immunostaining, and quantitative analyses of axon turning were performed as described previously (31, 35, 42). The percentage of turning axons was calculated from the numbers of YFP expressing axons turning toward the HEK cell aggregate divided by the total numbers of YFP expressing axons within 300 μm of the HEK cell aggregates. Transverse sections of spinal cords were immunostained to visualize commissural axons by confocal fluorescent microscopy. For details, see SI Materials and Methods.