Gradient-reading and mechano-effector machinery for netrin-1-induced axon guidance

Growth cones navigate axonal projection in response to guidance cues. However, it is unclear how they can decide the migratory direction by transducing the local spatial cues into protrusive forces. Here we show that knockout mice of Shootin1 display abnormal projection of the forebrain commissural axons, a phenotype similar to that of the axon guidance molecule netrin-1. Shallow gradients of netrin-1 elicited highly polarized Pak1-mediated phosphorylation of shootin1 within growth cones. We demonstrate that netrin-1–elicited shootin1 phosphorylation increases shootin1 interaction with the cell adhesion molecule L1-CAM; this, in turn, promotes F-actin–adhesion coupling and concomitant generation of forces for growth cone migration. Moreover, the spatially regulated shootin1 phosphorylation within growth cones is required for axon turning induced by netrin-1 gradients. Our study defines a mechano-effector for netrin-1 signaling and demonstrates that shootin1 phosphorylation is a critical readout for netrin-1 gradients that results in a directional mechanoresponse for axon guidance.


Introduction
Axon guidance is a critical step for the formation and regeneration of neuronal networks. More than a century ago Ramó n y Cajal identified the growth cone at the tip of extending axons, and proposed that it senses extracellular chemical cues and produces force for axon guidance (Cajal, 1890;Sotelo, 2002;Vitriol and Zheng, 2012). Accumulating evidence indicates that growth cones are indeed guided by extracellular molecules (Huber et al., 2003;Lowery and Van Vactor, 2009;Kolodkin and Tessier-Lavigne, 2011) and generate traction forces (Chan and Odde, 2008;Koch et al., 2012). Furthermore, analyses with microfluidic devices have shown that growth cones can navigate in response to extremely shallow gradients of diffusible and substrate-bound chemical cues in the microenvironment (Baier and Bonhoeffer, 1992;Rosoff et al., 2004;Xiao et al., 2014).
Shootin1, recently renamed shootin1a (Higashiguchi et al., 2016), is a brain-specific protein involved in axon outgrowth (Toriyama et al., 2006;Sapir et al., 2013). At the leading edge of growth cones, actin filaments (F-actins) polymerize and disassemble proximally, thereby undergoing retrograde flow (Forscher and Smith, 1988;Katoh et al., 1999). Shootin1a interacts with F-actin retrograde flow via cortactin (Weed and Parsons, 2001) and couples the F-actin flow with extracellular adhesive substrates (Shimada et al., 2008;Kubo et al., 2015) through the cell adhesion molecule L1-CAM (Rathjen and Schachner, 1984;Kamiguchi et al., 1998). We previously reported that Pak1 (Manser et al., 1994), upon activation by netrin-1, phosphorylates shootin1a (Toriyama et al., 2013); this in turn enhances shootin1a interaction with F-actin flow, thereby producing traction force on the substrate (Toriyama et al., 2013;Kubo et al., 2015). However, whether shootin1a mediates eLife digest Neurons communicate with each other by forming intricate webs that link cells together according to a precise pattern. A neuron can connect to another by growing a branch-like structure known as the axon. To contact the correct neuron, the axon must develop and thread its way to exactly the right place in the brain. Scientists know that the tip of the axon is extraordinarily sensitive to gradients of certain molecules in its surroundings, which guide the budding structure towards its final destination.
In particular, two molecules seem to play an important part in this process: netrin-1, which is a protein found outside cells that attracts a growing axon, and shootin1a, which is present inside neurons. Previous studies have shown that netrin-1 can trigger a cascade of reactions that activates shootin1a. In turn, activated shootin1a molecules join the internal skeleton of the cell with L1-CAM, a molecule that attaches the neuron to its surroundings. If the internal skeleton is the engine of the axon, L1-CAMs are the wheels, and shootin1a the clutch. However, it is not clear whether shootin1a is involved in guiding growing axons, and how it could help neurons 'understand' and react to gradients of netrin-1.
Here, Baba et al. discover that when shootin1a is absent in mice, the axons do not develop properly. Further experiments in rat neurons show that if there is a little more netrin-1 on one side of the tip of an axon, this switches on the shootin1a molecules on that edge. Activated shootin1a promote interactions between the internal skeleton and L1-CAM, helping the axon curve towards the area that has more netrin-1. In fact, if the activated shootin1a is present everywhere on the axon, and not just on one side, the structure can develop, but not turn. Taken together, the results suggest that shootin1a can read the gradients of netrin-1 and then coordinate the turning of a growing axon in response.
Wound healing, immune responses or formation of organs are just a few examples of processes that rely on cells moving in an orderly manner through the body. Dissecting how axons are guided through their development may shed light on the migration of cells in general. Ultimately, this could help scientists to understand disorders such as birth abnormalities or neurological disabilities, which arise when this process goes awry. axon guidance in vitro and in vivo remains unknown. In addition, how shootin1a associates with the substrates through L1-CAM is uncharacterized (Kubo et al., 2015).
Here, we combined gene knockout, protein interaction assays, force microscopy, speckle imaging and microfluidics to define a gradient-reading and mechano-effector machinery for netrin-1-induced axon guidance. We show that shootin1a is expressed at high levels in developing forebrain commissural axons and that Shootin1 knockout mice display abnormal guidance of these axons, a phenotype similar to that of Netrin-1 knockout mice. Notably, very small spatial differences in netrin-1 concentration elicited highly polarized directional phosphorylation of shootin1a within growth cones. Netrin-1-elicited shootin1a phosphorylation promoted direct interaction between shootin1a and L1-CAM, thereby generating traction force for growth cone motility. Furthermore, disturbance of the spatially regulated shootin1a phosphorylation within growth cones inhibited axon turning, but not axon outgrowth, induced by netrin-1 gradients. Our data demonstrate that shootin1a, through its spatially regulated phosphorylation within growth cones, mediates the gradient reading and mechanoresponse for netrin-1-induced axon guidance.

Results
Shootin1a is localized at high levels in axonal tracts of developing mouse brain To assess a role for shootin1a in axon guidance, we first analyzed its localization in the developing mouse brain. Consistent with a previous report (Toriyama et al., 2006), immunoblot analyses detected a low level of shootin1a in embryonic day (E) 13.5 mouse brains (Figure 1-figure supplement 1A). The expression increased remarkably between E13.5 and E16.5, remained high through postnatal days (P) 0-12, and then decreased to a low level in the adult. Immunohistochemical analyses with shootin1a-specific antibody detected shootin1a localization widely in P0 brain, with high levels of immunolabeling in the axonal tracts, such as the corpus callosum, anterior commissure, Figure 1 continued horizontal sections of the ventral forebrain of wild-type and Shootin1 knockout mice at P0 stained for Nissl substance. (E) Coronal sections of wild-type and Shootin1 knockout mouse brains at P0 immunolabeled with anti-L1-CAM antibody (green). Ectopic axonal projections were observed in the neocortex (arrowheads). In the knockout mice, the prominent axonal tracts observed in the intermediate zone of the neocortex of wild-type mice (arrows) were undetectable and ectopic axonal projections were observed (arrowheads). Lower panels show enlarged views of the rectangles. (F) Coronal sections of wild-type and Shootin1 knockout mouse brains at P0. DiI crystals (magenta) were placed into the neocortex (asterisks) to label callosal axons. An arrowhead indicates incomplete contralateral projections of callosal axons. (G) Horizontal sections of wild-type and Shootin1 knockout mouse brains at P0 immunolabeled with anti-L1-CAM antibody (green). In Shootin1 knockout mice, the bundling of the commissural axons was disrupted (arrowheads). (H) Horizontal sections of wild-type and Shootin1 knockout mouse brains at P0. DiI crystals (magenta) were placed in the anterior piriform cortex (asterisks) to label the anterior limb of the anterior commissure. Defasciculation and misprojection of the commissural axons are indicated by the arrow and arrowhead, respectively. Dashed lines indicate the anterior limb of the anterior commissure. Abbreviations: AC, anterior commissure; ACa, anterior limb of the anterior commissure; ACp, posterior limb of the anterior commissure; CC, corpus callosum; F, fimbria; Fx, fornix; KO, Shootin1 knockout mouse; MCP, middle cerebellar peduncle; Spt, Septum; VMH, ventromedial hypothalamic nucleus; WT, wild-type mouse. Scale bars: 500 mm. DOI: https://doi.org/10.7554/eLife.34593.003 The following source data and figure supplements are available for figure 1: hippocampal commissure, fornix, fimbria and middle cerebellar peduncle ( Figure 1A and Figure 1figure supplement 1B). We also detected high levels of shootin1a immunoreactivity that colocalized extensively with that of the axonal marker L1-CAM (Chung et al., 1991;Klingler et al., 2015) in the corpus callosum and anterior commissure at E16.5 ( Figure 1B). However, we could not detect shoo-tin1a immunoreactivity in the ventral commissure of the spinal cord at E12 (arrow, Figure 1-figure supplement 2A) when high levels of both netrin-1 and DCC are expressed (Keino-Masu et al., 1996;Kennedy et al., 2006;Bin et al., 2015;Dominici et al., 2017). Shootin1 knockout mice display abnormal projection of forebrain commissural axons To analyze further the roles of shootin1a in the developing brain, we generated Shootin1 knockout mice (Figure 1-figure supplement 3). Southern blot analysis confirmed that the first exon of the Shootin1 gene had been replaced with the b-galactosidase (LacZ) and neomycin resistance (Neo r ) genes ( Figure 1-figure supplement 3A and B). Immunoblot analysis demonstrated a complete loss of shootin1a protein (Figure 1-figure supplement 3C); shootin1a immunoreactivity was undetectable in Shootin1 knockout mouse brain sections (Figure 1-figure supplement 3D). Shootin1 knockout mice were born but 13.3% of them died during P0-P20 (n = 98). Coronal and horizontal sections of Shootin1 knockout brains revealed dysgenesis of the corpus callosum, anterior commissure and hippocampal commissure with a penetrance of 69.2% (n = 26) ( Figure 1C and D and Figure 1-figure supplement 1C); their thicknesses were significantly reduced by Shootin1 knockout ( Figure 1-figure supplement 1D). We also observed multiple defects in the brain of Shootin1 knockout mice, including agenesis of the septum ( Figure 1C); the detailed analyses of these phenotypes will be reported separately.
To analyze the commissure structures in detail, we visualized axon bundles using anti-L1-CAM antibody. In addition to decreased thickness of the axonal tracts at the midline of the corpus callosum (yellow line, Figure 1E), the prominent axonal tracts observed in the intermediate zone of the neocortex of wild-type mice (arrows, Figure 1E) were undetectable in Shootin1 knockout mice. In contrast, ectopic axonal projections toward the cortical plate were observed in Shootin1 knockout mice (arrowheads, Figure 1E). Consistent with the L1-CAM immunolabeling, DiI tracing also showed incomplete contralateral projections of callosal axons (arrowhead, Figure 1F). In the anterior commissure, L1-CAM immunolabeling showed a decrease in the thickness of the axon bundle that crosses the midline (yellow line, Figure 1G) as well as disruption of the anterior and posterior limbs of the commissure (arrowheads, Figure 1G). DiI tracing of the anterior limbs of the anterior commissure demonstrated defasciculation (arrow, Figure 1H) and misprojection (arrowhead, Figure 1H Shallow gradients of netrin-1 elicit highly polarized shootin1a phosphorylation within growth cones Extracellular gradients of netrin-1 induce growth cone attraction in vitro Hong et al., 1999;Bhattacharjee et al., 2010). To analyze the growth cone response under netrin-1 gradients, we prepared a device with microjet arrays that can generate gradients of diffusible molecules in the culture medium (Bhattacharjee et al., 2010) (Figure 2A). The substrates for neuronal culture, glass coverslips, were coated sequentially with polylysine and L1-CAM-Fc as reported (Shimada et al., 2008;Toriyama et al., 2013;Kubo et al., 2015). To estimate the soluble netrin-1 gradients, we used bovine serum albumin (BSA) labeled with the fluorescent tracer Alexa Fluor 594 or Alexa Fluor 488 as a proxy for netrin-1. As the molecular weight of BSA (66.0 kDa) is similar to that of mouse netrin-1 (68.2 kDa), we expect that the gradient of BSA mimics that of netrin-1 in the device. As reported (Bhattacharjee et al., 2010), our device generated stable gradients of BSA in the medium ( Figure 2B and Video 1). The difference in the BSA concentrations at the source side end and the other end of the area that expands 400 mm at the center of the linear gradient (red rectangle, Figure 2A), estimated by the fluorescence intensity, was 15% ( Figure 2B). As the regular growth cone width of cultured hippocampal neurons is about 10 mm (Katsuno et al., 2015), we (C) Neurons cultured in the device were labeled with CMAC (blue) and exposed to gradients of netrin-1 and Alexa Fluor 488-BSA (green) for 30 min. They were then fixed and immunolabeled with an antibody that recognizes shootin1a phosphorylation at Ser249 (red). The right panels show the fluorescent signals of CMAC and phospho-shootin1a in the growth cone located in the corresponding dashed rectangle. Yellow lines and dotted lines indicate the boundary and center line of the growth cone, respectively. A higher level of phospho-shootin1a immunolabeling was observed on the netrin-1 source side. Bar: 10 mm. (D) Quantification of relative phospho-shootin1a immunolabeling levels (phosopho-shootin1a immunoreactivity/CMAC staining) in the netrin-1 source side (high side) and control side (low side) of single growth cones. n = 12 growth cones. Data represent means ± SEM; ***p<0.01 (unpaired Student's t-test). DOI: https://doi.org/10.7554/eLife.34593.009 The following source data and figure supplements are available for figure 2: Source data 1. Quantification of relative phospho-shootin1a immunolabeling levels related to Figure 2D. estimate that gradient steepness (the percentage change in concentration) (Rosoff et al., 2004) of netrin-1 that covers growth cones located in the red rectangle area (Figure 2A) is about 0.4%. A previous study reported that netrin-1 attaches to polylysine-coated substrates, thereby mediating chemotropic axon guidance (Moore et al., 2012). To examine whether the present assay system produces netrin-1 gradients on the substrate, we next analyzed the attachment of netrin-1 to substrates coated with L1-CAM. As reported (Moore et al., 2012), incubation of glass coverslips with netrin-1 (200 or 300 ng/ml) led to netrin-1 attachment on the polylysine-coated substrate within 15 min, and the attachment further increased after incubation for 7 hr ( . We also confirmed that our device produces a netrin-1 gradient attached to the substrate in a manner dependent on the incubation time and that the difference in concentration across the growth cone is about 0.6% at 10 min and 0.8% at 430 min (Figure 2-figure supplement 1C), which is similar to that of BSA ( Figure 2B). However, the amount of attached netrin-1 was 39% of that on polylysine after 420 min incubation ( Figure 2 -figure supplement 1A and B), suggesting that at least 61% of the applied netrin-1 is not attached to the substrate under our conditions. Together, these results indicate that gradients of both soluble and substrate-bound netrin-1 are produced in our culture system.
We previously reported that netrin-1 induces Pak1-mediated phosphorylation of shootin1a at Ser101 and Ser249 in axonal growth cones (Toriyama et al., 2013). Using the microjet device, we examined the localization of netrin1-elicited shootin1a phosphorylation in growth cones. Hippocampal neurons cultured in the device for 1.5-2 days were labeled with the fluorescent volume marker 7amino-4-chloromethylcoumarin (CMAC) and exposed to a netrin-1 gradient for 30 min. The neurons were then fixed and immunolabeled with an antibody that recognizes shootin1a phosphorylation at Ser249. We selected axons located near the center of the gradients (red rectangle, Figure 2A) and that were oriented approximately perpendicular to the netrin-1 gradient. To our surprise, quantification of the phospho-shootin1a immunoreactivity and CMAC staining revealed a highly polarized localization of the phosphorylated shootin1a within growth cones ( Figure 2C). The relative level of the phosphorylated shootin1a (phosopho-shootin1a immunoreactivity/CMAC staining) was 71% higher on the netrin-1 source side than on the control side (p=0.002, n = 12) ( Figure 2D), and contrasted markedly with the shallow gradients of extracellular netrin-1 estimated by the fluorescent tracer ( Figure
To determine whether the shootin1a phosphorylation modulates the shootin1a-L1-CAM interaction, we analyzed the interaction between the phosphorylated shootin1a and L1-CAM-ICD. An in vitro binding assay with purified proteins showed that the interaction of L1-CAM-ICD with phosphomimic shootin1a (shootin1a-DD), in which Ser101 and Ser249 were replaced by aspartate, was stronger than that with wild-type shootin1a (shootin1a-WT) ( Figure 3B): the apparent dissociation constant for shootin1a-DD (K d = 46.4 ± 4.4 nM) was 2.9-fold lower (p<0.02, n = 3 independent experiments) than that of shootin1a-WT (K d = 133.5 ± 13.6 nM) ( Figure 3C). In vitro phosphorylation and binding assays also demonstrated that phosphorylation of shootin1a by Pak1 promotes its interaction with L1-CAM-ICD ( Figure 3D and E). Ectopic expression of a constitutively active Pak1 in HEK293T cells increased phosphorylation of myc-shootin1a at Ser101 and Ser249 in these cells ( Figure 4A and B); this in turn promoted the interaction between shootin1a and L1-CAM-ICD ( Figure 4A and B). Conversely, expression of a dominant-negative Pak1 decreased shootin1a phosphorylation and inhibited the interaction ( Figure 4A and B). Furthermore, stimulation of neurons by netrin-1 increased the phosphorylation of shootin1a at Ser101 and Ser249 ( Figure 4C and D); this led to a concomitant increase in the interaction between endogenous shootin1a and L1-CAM ( Figure 4C and D). In axonal growth cones, phosphorylated shootin1a was highly colocalized with L1-CAM ( Figure 4E). Altogether, our data demonstrate that netrin-1-induced shootin1a phosphorylation promotes direct interaction between shootin1a and L1-CAM.

Shootin1a-L1-CAM interaction mediates netrin-1-induced F-actinadhesion coupling and mechanoresponse
To address the role of the shootin1a-L1-CAM interaction, we next analyzed the shootin1a region that interacts with L1-CAM. An in vitro binding assay with purified proteins showed that residues 1-125 of shootin1a (shootin1a (1-125)) were essential and sufficient to bind to L1-CAM ( Figure 5A). On the other hand, we previously reported that another region, shootin1a (261-377), is responsible for its interaction with cortactin, which links shootin1a to F-actin flow ( Figure 5A) (Kubo et al., 2015). Consistent with these data, when expressed in hippocampal neurons, myc-shootin1a (1-125) was highly colocalized with L1-CAM in axonal growth cones ( Figure 5B). As shootin1a (1-125) interacts with L1-CAM but not with cortactin, we expected that it can be used as a dominant negative mutant that disrupts the shootin1a-L1-CAM interaction. Indeed, shootin1a (1-125) overexpressed in HEK293T cells bound to L1-CAM-ICD, and inhibited the interaction between shootin1a and L1-CAM-ICD ( Figure 5C).
Coupling between F-actins and substrate reduces the speed of F-actin flow in growth In vitro binding assay using purified shootin1a-WT (100 nM) and purified GST-L1-CAM-ICD (100 nM). Proteins were incubated with Glutathione Sepharose 4B and GST-L1-CAM-ICD was eluted. The eluate was then analyzed by SDS-PAGE and CBB staining; 0.2% of the input proteins were also analyzed. (B and C) In vitro binding assay using purified shootin1a-WT or purified shootin1a-DD and purified GST-L1-CAM-ICD. Shootin1a-DD or shootin1a-WT at increasing concentrations was incubated with GST-L1-CAM-ICD and Glutathione Sepharose 4B. GST-L1-CAM-ICD was eluted. After SDS-PAGE, the eluate was immunoblotted with anti-shootin1 antibody or stained with CBB (B), and the bound shootin1a-DD and shootin1a-WT were then quantified (C). Data represent means ± SEM (n = 3 independent experiments). (D and E) In vitro binding assay using Pak1-phosphorylated purified shootin1a and purified GST-L1-CAM-ICD. Shootin1a-WT (100 nM) or Pak1-phosphorylated shootin1a-WT (100 nM) was incubated with GST-L1-CAM-ICD and Glutathione Sepharose 4B. GST-L1-CAM-ICD was eluted. After SDS-PAGE, the eluate was immunoblotted with anti-shootin1 antibody or stained with CBB (D). Input proteins (1%) were Figure 3 continued on next page cones (Suter et al., 1998;Toriyama et al., 2013). Using shootin1a (1-125) as a dominant negative mutant, we examined whether the shootin1a-L1-CAM interaction is involved in netrin-1-induced mechanical coupling between F-actin flow and the substrate. Hippocampal neurons expressing mRFP-actin were cultured on coverslips coated with L1-CAM, and F-actin flow in the growth cone was monitored by live-cell fluorescence microscopy ( Figure 6A, Video 2). In neurons overexpressing a control protein myc-GST, the fluorescent features of mRFP-actin moved retrogradely at 4.5 ± 0.1 mm/min (mean ± SE, n = 30 fluorescent features), as previously reported (Shimada et al., 2008). Overexpression of myc-shootin1a (1-125) in hippocampal neurons increased the velocity of F-actin flow ( Figure 6A and B), suggesting that shootin1a (1-125) inhibits the F-actin-adhesion coupling. Netrin-1 stimulation significantly decreased the velocity of F-actin flow in control growth cones, reflecting promotion of the F-actin-adhesion coupling. On the other hand, netrin-1 accelerated the flow in the presence of overexpressed myc-shootin1a (1-125) ( Figure 6A and B); this can be explained by the inhibition of the F-actin adhesion coupling as well as a simultaneous increase in actin polymerization by netrin-1-induced activation of Cdc42 and Rac1 (Shekarabi et al., 2005;Brianç on-Marjollet et al., 2008). These results suggest that shootin1a-L1-CAM interaction mediates netrin-1-induced F-actin-adhesion coupling.
We further monitored mechanoresponses of growth cones using traction force microscopy. Hippocampal neurons were cultured on L1-CAM-coated polyacrylamide gels with embedded 200 nm fluorescent beads. Traction forces under the growth cones were monitored by visualizing forceinduced deformation of the elastic substrate, which is reflected by displacement of the beads from their original positions. As reported (Toriyama et al., 2013), the reporter beads under the growth cones moved dynamically, reflecting the traction force on the substrate ( Figure 6C, Video 3); and the force was increased significantly by netrin-1 stimulation ( Figure 6C and D and Figure 6-figure supplement 1A). Importantly, inhibition of the shootin1a-L1-CAM interaction by overexpressing myc-shootin1a (1-125) significantly decreased traction forces and abolished the netrin-1-induced increase in traction forces ( Figure 6D). Overexpression of myc-shootin1a (1-125) also decreased axon length and abolished the netrin-1-induced axon outgrowth (Figure 6-figure supplement 1B and Figure 6E), suggesting that shootin1a-L1-CAM interaction is involved in the netrin-1-induced generation of traction force for growth cone migration.

Netrin-1-induced axon attraction requires shootin1a
Next, we examined whether shootin1a is involved in netrin-1-induced axon guidance. Hippocampal neurons were stimulated with a netrin-1 gradient for 7 hr; the gradient was applied approximately perpendicularly to the extending direction of axons (Figure 7-figure supplement 1A). The right panel (Figure 7-figure supplement 1A) depicts the migration of individual axonal growth cones. Consistent with previous reports Serafini et al., 1994;Bhattacharjee et al., 2010), the majority of the axonal growth cones migrated toward the netrin-1 source (Figure 7  also analyzed with anti-shootin1, anti-pSer101-shootin1 or anti-pSer249-shootin1 antibody. Quantitative data for bound shootin1a are shown in (E) (n = 3 independent experiments). Data represent means ± SEM; ***p<0.01 (unpaired Student's t-test). DOI: https://doi.org/10.7554/eLife.34593.016 The following source data is available for figure 3: Source data 1. Quantitative data for Kd value related to Figure 3C. DOI: https://doi.org/10.7554/eLife.34593.017 Source data 2. Quantitative data for bound shootin1a related to Figure 3E. DOI: https://doi.org/10.7554/eLife.34593.018 . These data are consistent with a previous report (Mai et al., 2009) that netrin-1 attached to the substrate induces axon turning of cultured hippocampal neurons. Thus, we conclude that gradients of both soluble and substrate-bound netrin-1 contribute to axon turning of hippocampal neurons in our assay system. We next stimulated hippocampal neurons expressing control miRNA or shootin1a miRNA, which inhibits shootin1a expression, with a netrin-1 gradient. The majority of the axonal growth cones of control neurons migrated toward the netrin-1 source ( Figure 7A, Video 6). The mean axon outgrowth velocity was 26.1 ± 3.0 mm/h ( Figure 7C), and the net change in the angle of the growth cone toward the netrin-1 source was 32.7 ± 2.2˚(n = 11) ( Figure 7D). Repression of shootin1a by RNAi not only reduced the axon outgrowth velocity (5.6 ± 0.9 mm/h, p<0.01) ( Figure 7B and C, Video 7) but also inhibited the growth cone turning toward the netrin-1 source (2.1 ± 1.2˚, p<0.01, n = 13) ( Figure 7B and D). Similar results were obtained using hippocampal neurons prepared from Shootin1 knockout mice (Figure 7-figure supplement 2). Together, these data indicate that netrin-1-induced axon guidance of hippocampal neurons on an L1-CAM substrate requires shootin1a.

Netrin-1-induced axon attraction requires shootin1a-L1-CAM interaction
We further examined a role of shootin1a-L1-CAM interaction in netrin-1-induced axon guidance. Hippocampal neurons overexpressing EGFP (control) or EGFP-shootin1a (1-125), which inhibits the shootin1a-L1-CAM interaction ( Figure 5C), were stimulated with a netrin-1 gradient for 7 hr ( Figure 7E and F, left panels). The majority of the axonal growth cones of control neurons migrated toward the netrin-1 source ( Figure 7E, Video 8). The mean axon outgrowth velocity was 32.4 ± 6.6 mm/h (Figure 7-figure supplement 3A), and the net change in the angle of the growth cone toward the netrin-1 source was 28.8 ± 3.8˚(n = 17) (Figure 7-figure supplement 3B). On the other hand, inhibition of the shootin1a-L1-CAM interaction by overexpressing EGFP-shootin1a (1-125) not only reduced the axon outgrowth velocity (13.8 ± 3.4 mm/h, p<0.05) but also inhibited the growth cone turning toward the netrin-1 source (0.6 ± 1.1˚, p<0.01, n = 16) ( Figure 7F, Figure 7-figure supplement 3 and Video 9). These data suggest that the interaction between shootin1a and L1-CAM mediates netrin-1-induced axon guidance.  (1%) were also analyzed with anti-pSer101-shootin1, anti-pSer249-shootin1, or anti-myc antibody. Quantitative data for phosphorylated and bound shootin1a are shown in (B) (n = 3 independent experiments). Data represent means ± SEM; ***p<0.01; **p<0.02; *p<0.05 (One-way ANOVA with Tukey's post hoc test). (C and D) Co-immunoprecipitation of shootin1a and L1-CAM in cultured cortical neurons. After incubation of neurons with 4.4 nM netrin-1 or BSA (control) for 1 hr, cell lysates were prepared and incubated with anti-shootin1 antibody (right panel). The immunoprecipitates were immunoblotted with anti-shootin1 or anti-L1-CAM antibody. The cell lysates (5%) were also analyzed with anti-pSer101-shootin1, anti-pSer249-shootin1, or anti-shootin1a antibody (left panel). Quantitative data for phosphorylated shootin1a and bound L1-CAM are shown in (D) (n = 3 independent experiments). Data represent means ± SEM; **p<0.02; *p<0.05 (Unpaired Student's t-test). (E) Fluorescence images of an axonal growth cone labeled with anti-pSer249-shootin1a (magenta) and anti-L1-CAM (green) antibodies. The cells were observed using a TIRF microscope. An enlarged view of the filopodium in the rectangle is shown in the lower panel. Arrowheads indicate phosphorylated shootin1a colocalized with L1-CAM. Bar: 5 mm (in the inset, 2 mm). DOI: https://doi.org/10.7554/eLife.34593.019 The following source data is available for figure 4: Source data 1. Quantitative data for phosphorylated and bound shootin1a related to Figure 4B. DOI: https://doi.org/10.7554/eLife.34593.020 Source data 2. Quantitative data for phosphorylated shootin1a and bound L1-CAM related to Figure 4D. DOI: https://doi.org/10.7554/eLife.34593.021 Shootin1a-L1-CAM interaction mediates netrin-1-induced axon guidance on laminin Laminins are widely used substrates for axon guidance assays (Turney and Bridgman, 2005;Nichol et al., 2016); L1-CAM on growth cones interacts directly with laminin presented on the substrate (Hall et al., 1997;Abe et al., 2018). To examine whether the shootin1a-L1-CAM interaction mediates netrin-1-induced axon guidance generally, we performed an axon guidance assay on an alternative substrate, laminin. Growth cones of cultured hippocampal neurons on laminin turned in response to netrin-1 gradients (Figure 7-figure supplement 4A-C), as they did on L1-CAM (Figure 7-figure supplement 1A). To assess shootin1a-mediated clutch coupling on laminin, we measured F-actin retrograde flow in growth cones on laminin. Consistent with previously reported (1-125) inhibits the interaction between shootin1a and L1-CAM-ICD. HEK293T cells were transfected with vectors to express myc-shootin1a and FLAG-L1-CAM-ICD; some of them were also co-transfected with a vector to overexpress myc-shootin1a (1-125) as indicated. Cell lysates were prepared and incubated with anti-FLAG antibody. The immunoprecipitates were immunoblotted with anti-myc or anti-FLAG antibody. The cell lysates (1%) were also analyzed with anti-myc antibody. The graph (right) shows quantitative data for bound shootin1a-DD (n = 3 independent experiments). Data represent means ± SEM; *p<0.05 (unpaired Student's t-test). DOI: https://doi.org/10.7554/eLife.34593.022 The following source data is available for figure 5: Source data 1. Quantitative data for bound shootin1a-DD related to Figure 5C. Together, these data indicate that netrin-1-induced axon guidance, which is mediated by the shoo-tin1a-L1-CAM interaction, is not limited to growth cones on L1-CAM. (1-125) before (control) or after netrin-1 stimulation (see Figure 6-figure supplement 1A for the direction of the traction forces, n = 14 growth cones). One-way ANOVA with Tukey's post hoc test was performed. (E) Three hours after plating, hippocampal neurons overexpressing myc-GST (control) or myc-shootin1a (1-125) were incubated with BSA (control) or 4.4 nM netrin-1 for 40 hr, and then immunolabeled by anti-myc antibody (see Figure 6-figure supplement 1B). Axon length was then analyzed (n = 909 neurons). One-way ANOVA with Schaffer's post hoc test was performed in the left graph, while an unpaired Student's t-test was used in the right graph. Data represent means ± SEM; ***p<0.01; *p<0.05; ns, not significant. Bars: 5 mm (in the kymographs of A, 2 mm). DOI: https://doi.org/10.7554/eLife.34593.024 The following source data and figure supplements are available for figure 6: Source data 1. Quantitative data for F-actin retrograde flow speed related to Figure 6B. DOI: https://doi.org/10.7554/eLife.34593.027 Source data 2. Quantitative data for the magnitude of the traction forces related to Figure 6D. DOI: https://doi.org/10.7554/eLife.34593.028 Source data 3. Quantitative data for axon length and axon outgrowth by netrin-1 related to Figure 6E. Netrin-1-induced axon attraction requires polarized shootin1a phosphorylation within growth cones Finally, we analyzed the role of polarized shootin1 phosphorylation within growth cones elicited by netrin-1 gradients. As shown above ( Figure 7A-D), repression of shootin1a by expression of shoo-tin1a miRNA inhibited axon outgrowth and growth cone turning toward the netrin-1 source. Expression of RNAi-refractory shootin1a-WT in neurons expressing shootin1a miRNA rescued the reduction of axon outgrowth as well as growth cone turning ( Figure 8A, C and D, Video 10), indicating that shootin1a regulates both axon outgrowth and growth cone turning. Our previous work has shown that axon outgrowth is regulated by shootin1a-mediated clutch coupling (Shimada et al., 2008;Kubo et al., 2015). As shootin1a-DD, the constitutively active shootin1a, mediates clutch coupling and force generation in the absence of PAK1 activity (Toriyama et al., 2013) but cannot be regulated by phosphorylation, displacement of wild-type shootin1a with shootin1a-DD would disturb netrin-1-induced polarized shootin1a regulation without disturbing the clutch coupling. As shown by Figure 8B-D and Video 11, disturbance of polarized shootin1a phosphorylation within growth cones by this displacement inhibited growth cone turning toward the netrin-1 source without reducing axon outgrowth velocity. These results demonstrate that the disturbance of axon turning caused by depletion of shootin1a ( Figure 7D and Figure 7-figure supplement 2D) or by dominant negative shootin1a (Figure 7-figure supplement 3B) is not attributed only to the inhibited axon outgrowth, and suggest that the polarized phosphorylation of shootin1a within growth cones is required for the directional axon guidance induced by netrin-1 gradients ( Figure 8E).

Discussion
Since the seminal proposal by Ramó n y Cajal (Cajal, 1890) that the growth cone senses extracellular chemical cues and produces force for axon guidance, considerable progress has been made in understanding the signaling events at the axon guidance machinery located within the growth cone. However, the gradient-reading as well as mechano-effector machinery that converts the environmental spatial chemical signals into the directional force for axon guidance has remained unclear. Here we have shown that shallow gradients of netrin-1 elicited highly polarized shootin1a phosphorylation within growth cones. Netrin-1-elicited shootin1a phosphorylation promoted direct interaction between shootin1a and L1-CAM, thereby generating traction force for growth cone motility. Source data 1. Quantitative data for axon outgrowth velocity related to Figure 7C. DOI: https://doi.org/10.7554/eLife.34593.048 Source data 2. Quantitative data for turning angle of axon toward the netrin-1 source related to Figure 7D. Furthermore, the spatially regulated phosphorylation of shootin1a within growth cones was required for axon turning induced by netrin-1 gradients. These results suggest that shootin1a constitutes a gradient-reading and mechano-effector machinery involved in netrin-1-induced axon guidance.

Mechano-effector machinery for netrin-1-regulated axon guidance
The present study defines a mechano-effector for netrin-1-induced axon guidance. Previous reports proposed that an increase in the mechanical coupling between F-actin retrograde flow in the growth cone and cell adhesions transmits the force of F-actin flow onto the adhesive substrates for growth cone migration (Mitchison and Kirschner, 1988;Suter and Forscher, 2000). In relation to this notion, a recent study reported that axon guidance cues affect F-actin-adhesion coupling locally within the growth cone to influence axon outgrowth and guidance (Nichol et al., 2016). In addition, we previously reported that shootin1a associates with the F-actin flow at the growth cone through its direct interaction with cortactin, and showed that netrin-1-induced phosphorylation of shootin1a by Pak1 promotes shootin1a-F-actin interaction through cortactin (Toriyama et al., 2013;Kubo et al., 2015).
Here, we demonstrated that netrin-1-induced shootin1a phosphorylation by Pak1 also promotes direct interaction between shootin1a and L1-CAM. Previous studies reported that netrin-1 induces dimerization of its receptor, DCC, thereby activating molecules including NCK1, FAK and FYN (Stein et al., 2001;Ren et al., 2004;Lai Wing Sun et al., 2011). This in turn induces activation of Video 6. A time-lapse movie of a hippocampal neuron expressing control miRNA, under a gradient of netrin-1. See the legend for Figure 7A. The gradient of Alexa Fluor 594-BSA in Figure 7A is not shown. DOI: https://doi.org/10.7554/eLife.34593.050 Video 7. A time-lapse movie of a hippocampal neuron expressing shootin1a miRNA, under a gradient of netrin-1. See the legend for Figure 7B. The gradient of Alexa Fluor 594-BSA in Figure 7B is not shown. DOI: https://doi.org/10.7554/eLife.34593.051 Cdc42 and Rac1, and their downstream kinase Pak1 (Li et al., 2002;Shekarabi and Kennedy, 2002;Shekarabi et al., 2005;Brianç on-Marjollet et al., 2008;Demarco et al., 2012) ( Figure 8E). Thus, under the activation of Pak1 via these signaling pathways, the shootin1a phosphorylation enhances both the shootin1a-adhesion and shootin1a-F-actin interactions that lead to increased F-actin-adhesion coupling ( Figure 8E). This double regulation would enable efficient regulation of forces for axon guidance in response to netrin-1. At present, no information is available on the three-dimensional structure of shootin1a. In addition, it is unknown how shootin1a-L1-CAM and shootin1a-cortactin interactions are promoted by the phosphorylation of shootin1a. Future investigations of the molecular structure of shootin1a, and of how the structures of the domains mediating interaction with L1-CAM and cortactin are affected by phosphorylation, will lead to a better understanding of this mechano-effector machinery.

Gradient-reading machinery for netrin-1-regulated axon guidance
The ability of cells to sense small spatial differences in environmental cues is essential for proper axon guidance as well as directional cell migration, but the molecular mechanism underlying it remains a major question (Quinn and Wadsworth, 2008;Hegemann and Peter, 2017). Initial studies reported that growth cones can respond to 1% gradients of repulsive tectal membranes presented on the substrate (Baier and Bonhoeffer, 1992) and 5-10% gradients of diffusible axon guidance molecules including netrin-1 (Ming et al., 1999). More recently, analyses employing microfluidic devices demonstrated that growth cones have even higher sensitivities to chemical gradients: Video 8. A time-lapse movie of a hippocampal neuron expressing EGFP, under a gradient of netrin-1. See the legend for Figure 7E. The gradient of Alexa Fluor 594-BSA in Figure 7E is not shown. DOI: https://doi.org/10.7554/eLife.34593.052 Video 9. A time-lapse movie of a hippocampal neuron expressing EGFP-shootin1a (1-125), under a gradient of netrin-1. See the legend for Figure 7F. The gradient of Alexa Fluor 594-BSA in Figure 7F is not shown. DOI: https://doi.org/10.7554/eLife.34593.053  Figure 7A and B, Figure 8A and B (n = 47 growth cones). (D) Turning angle of axon toward the netrin-1 source was obtained from the analyses in Figure 7A and B, Figure 8A  concentration can induce highly polarized phosphorylation of shootin1a within growth cones (pink), as a readout of highly sensitive gradient-reading processes. A netrin-1 gradient on the substrate would also contribute to polarized shootin1 phosphorylation (yellow). This process is achieved through a signaling pathway including DCC, Rac1/CDC42, Pak1 and shootin1a. The polarized phosphorylation of shootin1a within a growth cone locally promotes shootin1a-L1-CAM and shootin1a-cortactin interactions. These interactions in turn enhance asymmetrically the coupling between F-actin retrograde flow and the adhesive substrate and increase traction force (blue arrows) on the side of the netrin-1 source, thereby leading to a decision for the migratory direction (white arrow). Data represent means ± SEM; ***p<0.01; ns, not significant (one-way ANOVA with Schaffer's post hoc test). Bars: 50 mm.
DOI: https://doi.org/10.7554/eLife.34593.054 The following source data is available for figure 8: Source data 1. Quantitative data for axon outgrowth velocity related to Figure 8C. DOI: https://doi.org/10.7554/eLife.34593.055 Source data 2. Quantitative data for turning angle of axon toward the netrin-1 source related to Figure 8D. DOI: https://doi.org/10.7554/eLife.34593.056 Video 10. A time-lapse movie of a hippocampal neuron expressing shootin1a miRNA + RNAi refractory shootin1a-WT, under a gradient of netrin-1. See the legend for Figure 8A. The gradient of Alexa Fluor 594-BSA in Figure 8A is not shown. DOI: https://doi.org/10.7554/eLife.34593.057 Video 11. A time-lapse movie of a hippocampal neuron expressing shootin1a miRNA + RNAi refractory shootin1a-DD, under a gradient of netrin-1. See the legend for Figure 8B. The gradient of Alexa Fluor 594-BSA in Figure 8B is not shown. DOI: https://doi.org/10.7554/eLife.34593.058 for example, they can turn in response to 0.1-0.4% gradients of NGF (Rosoff et al., 2004), and 0.5 and 0.1% gradients of substrate-bound laminin and ephrin-A5, respectively (Xiao et al., 2014). This study presents the framework of a highly sensitive gradient-reading machinery for axon guidance ( Figure 8E). Our data demonstrate that a 0.4% (250:251) difference in netrin-1 concentration induces a 71% difference in shootin1a phosphorylation within growth cones, as a key readout of the spatial signal. This polarized phosphorylation locally promotes shootin1a-L1-CAM and shootin1a-cortactin interactions within growth cones and asymmetrically promote traction force (blue arrows, F Figure 8E) on the side of the netrin-1 source, leading to a decision for the migratory direction (white arrow, Figure 8E). Our data suggest that the netrin-1 gradient-reading process in growth cones is achieved through a signaling pathway that includes DCC, Rac1/Cdc42, Pak1 and shootin1a ( Figure 8E); however, how these molecules amplify very small spatial differences in netrin-1 concentration remain to be determined. It has been proposed that amplification of local signals through combined positive and negative feedback loops contribute to sense shallow gradients of extracellular chemicals (Yang et al., 2016;Hegemann and Peter, 2017) and that polarized assembly of signaling molecules may play a key role in it (Quinn and Wadsworth, 2008;Hegemann and Peter, 2017). Previous studies in C. elegans reported that local netrin-1 signals induce polarized distribution of the DCC orthologs UNC-40 within cell bodies (Adler et al., 2006;Ziel et al., 2009;Wang et al., 2014). However, growth cones of mouse cortical neurons did not show polarized localization of DCC under netrin-1 gradients (Taylor et al., 2015). In addition, FRET visualization of Cdc42 and Rac1 signals has not, so far, revealed a distinct polarized activation of these molecules in growth cones (Picard et al., 2009;Rappaz et al., 2016). Tracing the spatial signals from phosphorylated shootin1a back upstream to DCC will lead to a better molecular understanding of the gradient-reading mechanism involved in netrin-1-regulated axon guidance.

Shootin1a-mediated axon guidance in the brain
The present study has shown that shootin1a is expressed at high levels in developing forebrain commissural axons and that Shootin1 knockout mice display dysgenesis and misprojection of these axons. We analyzed their projections with the axonal marker L1-CAM (Chung et al., 1991;Klingler et al., 2015) and DiI tracing (Klingler et al., 2015). As L1-CAM interacts with shootin1a, ablation of shootin1a expression could lead to a change in L1-CAM localization in these axons; however, both the L1-CAM labelling and DiI tracing analyses revealed misprojection of these axons. Similar defects of callosal and anterior commissural axons were reported in knockout mice for Netrin-1 (Serafini et al., 1996) as well as DCC (Fazeli et al., 1997), Rac1 (Chen et al., 2007;Kassai et al., 2008) and L1-CAM (Demyanenko et al., 1999), thereby providing evidence for the notion that shootin1a cooperates with these molecules in axon guidance. Thus, the defects in the forebrain commissural axons in Shootin1 knockout mice are consistent with the in vitro observations that shootin1a is required for netrin-1-induced axon outgrowth and guidance.
However, although netrin-1 mRNA is distributed along the paths of forebrain commissural axons (Serafini et al., 1996), netrin-1 gradients have not yet been reported in these brain regions. In addition, as Shootin1 knockout mice exhibit multiple defects in the brain, we cannot conclude that the dysgenesis of the forebrain commissures is only due to the axon outgrowth and guidance deficits observed in in vitro assays. Generation of conditional knockout mice as well as detailed mapping of netrin-1 will facilitate future analyses of the shootin1a-mediated axon guidance in the brain. In contrast to the forebrain commissural axons, we could not detect distinct shootin1a localization in the ventral spinal commissural axons and could not observe their abnormality in Shootin1 knockout mice, suggesting that shootin1a is not required for guidance of the spinal commissural axons.

Netrin-1-induced axon guidance through chemotaxis and haptotaxis
Axon guidance and cell migration are directed by spatial gradients of soluble chemicals (called chemotaxis) (Gundersen and Barrett, 1979;Mortimer et al., 2008) and by chemicals presented on adhesive substrate or neighboring cells (termed haptotaxis) (Carter, 1967;Baier and Bonhoeffer, 1992). The present study indicates that gradients of both soluble and immobilized netrin-1 are produced on L1-CAM-coated substrate, thereby contributing to axon turning of hippocampal neurons. As soluble and immobilized chemical cues act as ligands to activate intracellular signaling pathways in growth cones (Huber et al., 2003;Lowery and Van Vactor, 2009;Kolodkin and Tessier-Lavigne, 2011), the gradient-reading machinery involving shootin1a may explain both axonal chemotaxis and haptotaxis elicited by shallow gradients of netrin-1 (red and yellow arrows, Figure 8E).
On the other hand, recent studies reported that netrin-1 locally presented by neural progenitors is required for guidance of the spinal commissural axons (Dominici et al., 2017;Varadarajan et al., 2017) and that these axons respond to substrate-bound netrin-1 (Moore et al., 2012) and steep gradients of netrin-1 (Sloan et al., 2015) in vitro. These reports underscore the importance of haptotaxis in netrin-1-induced axon guidance of spinal commissural neurons. Concerning the axon guidance mechanism of these neurons, Moore et al. (2012) reported that immobilization of netrin-1 is required; the immobilized netrin-1 is proposed to play a key role in mechanical activation of FAK, thereby leading to activation of signaling pathways including Crk-associated substrate (CAS) for axonal haptotaxis. The details of how axon guidance is regulated by immobilized netrin-1 remain important issues for future analyses.

DiI tracing
1,1'-dioctadecyl-3,3,3',3'-tetramethyl-indocarbocyanine dye (DiI; Invitrogen) crystals were placed in the dorsomedial cortex in a rostrocaudal series for tracing of the corpus callosum, or in the anterior olfactory nucleus/anterior piriform cortex for tracing of the anterior limb of the anterior commissure (Klingler et al., 2015). After the placement, brains were incubated in 4% FA at 37˚C for 3 months, and then 100 mm sections were prepared by vibratome (Leica). The sections were mounted on glass slides, and observed under a fluorescence microscope.

Cultures and transfection
Hippocampal neurons prepared from E18 rats were cultured on glass coverslips coated sequentially with polylysine and L1-CAM-Fc as described (Shimada et al., 2008;Toriyama et al., 2013;Kubo et al., 2015). For the experiments in Figure 7-figure supplement 4, we cultured neurons on glass coverslips coated sequentially with polylysine and laminin as described (Toriyama et al., 2006;Abe et al., 2018). All experiments except for the measurement of forces were carried out on glass surfaces. For the immunoprecipitation and immunoblot analyses in Figure 4C, we used cortical neurons, which also respond to netrin-1 (Li et al., 2008), as the experiments required large numbers of neurons. They were prepared from E18 rat embryos using the same protocol as above. The neurons were transfected with vectors using Nucleofector (Lonza) before plating. HEK293T cells (ATCC, RRID:CVCL_0063, cell identities were authenticated by STR profiling and cells were tested negative for mycoplasma using the TAKARA PCR mycoplasma detection set Cat# 6601) were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and transfected with plasmid DNA by the calcium phosphate method.

Axon guidance assay
A microfluidic device that generates netrin-1 gradients in culture medium was produced according to a previous report (Bhattacharjee et al., 2010), with modification. Briefly, it was fabricated with polydimethylsiloxane (PDMS; Silpot 184, Dow Corning Toray, Japan) and a glass coverslip; the device consists of an open rectangular cell culture area and two microchannels on the long sides of the culture area ( Figure 2A). The micro-molds of the channel pattern were lithographically fabricated on a photoresist (SU-8 3025, MicroChem, USA) spin-coated on a 70 mm thick silicon wafer. PDMS sheets were obtained from this mold, which had been treated with silicone oil (Barrier coat No. 6, ShinEtsu, Japan) to facilitate their removal. A PDMS sheet, coated with 1 mm thick PDMS glue (KE103, ShinEtsu, Japan), was then bonded to a glass coverslip coated sequentially with polylysine and L1-CAM-Fc. To generate netrin-1 gradients in the cell culture area, flows of culture medium (7.5 mm/min) with or without 4.4 nM (300 ng/ml) netrin-1 and 2 mM fluorescent tracer (Alexa Fluor 594-BSA or Alexa Fluor 488-BSA) were applied to the microchannels on either side of the open cell culture area (black arrows, Figure 2A). As reported (Bhattacharjee et al., 2010), the microfluidic device generated stable gradients of the tracer in the culture medium ( Figure 2B). For live imaging of neurons expressing miRNA, EGFP fluorescence was used as an indicator of miRNA expression. The turning angle of an axon toward the netrin-1 source was obtained by calculating the difference between the angles of the axonal tip at the first and last time-points of the observations (fl) (Bhattacharjee et al., 2010) (Figure 7D).

RNAi
For RNAi experiments, we used a Block-iT Pol II miR RNAi expression kit (Invitrogen). The targeting sequence of shootin1a miRNA and its effectiveness were reported previously (Toriyama et al., 2006). As described previously (Shimada et al., 2008), to ensure high-level expression of miRNA before neurite elongation, hippocampal neurons prepared from E18 rat embryos and transfected with the miRNA expression vector were plated on uncoated polystyrene plates. After a 20 hr incubation to induce miRNA expression, the cells were collected and then cultured on coverslips.

Protein preparation and in vitro kinase assay
Recombinant proteins were expressed in Escherichia coli as GST fusion proteins and purified on Glutathione sepharose columns, after which GST was removed by PreScission protease. L1-CAM-Fc was prepared as described (Shimada et al., 2008). Kinase reactions were carried out in 20 ml kinase buffer (50 mM HEPES pH 7.5, 10 mM MgCl 2 , 2 mM MnCl 2 , 1 mM DTT, 125 mM ATP, in the presence or absence of 10 mCi [g-32 P]ATP) containing 250 ng active Pak1 and 2.1 mg purified shootin1a as described (Kubo et al., 2015).

In vitro binding assay
Purified GST-L1-CAM-ICD and shootin1a were incubated overnight at 4˚C in reaction buffer (20 mM Tris-HCl pH 8.0, 100 mM NaCl, 1 mM EDTA, 1 mM DTT). After centrifugation for 15 min at 17,400 g at 4˚C, the supernatants were incubated with Glutathione Sepharose 4B beads for 2 hr at 4˚C. The beads were washed three times with wash buffer (20 mM Tris-HCl pH 8.0, 300 mM NaCl, 1 mM EDTA, 1 mM DTT) and once with TED buffer (20 mM Tris-HCl pH 8.0, 1 mM EDTA, 1 mM DTT). For elution, the Sepharose beads were incubated with 25 ml of elution buffer (15 mM reduced glutathione pH 8.0, 20 mM Tris-HCl pH 8.0, 100 mM NaCl, 1 mM EDTA, 1 mM DTT) for 2 hr at 4˚C. For the binding assay in Figure 3A, we incubated 100 nM shootin1a and 100 nM GST-L1-CAM-ICD in 10 ml reaction buffer. After further incubation with Glutathione Sepharose 4B (bed volume 500 ml), GST-L1-CAM-ICD was eluted by 20 mM glutathione buffer (pH 8.0). After concentrating the 10 ml eluate with a centrifugal filter (Millipore), using half of the eluate, we could detect the interaction between shootin1a and L1-CAM-ICD by CBB staining. For the binding assay in Figure 3B, the supernatants were analyzed by immunoblot. Apparent dissociation constants were calculated by non-linear regression using GraphPad Prism 6 (GraphPad Prism Software).

Fluorescent speckle imaging and traction force microscopy
The speckle imaging data in were obtained using neurons cultured on coverslips coated with L1-CAM-Fc or laminin as described (Shimada et al., 2008). Traction force microscopy was performed as described (Toriyama et al., 2013;Abe et al., 2018). Briefly, neurons were cultured on polyacrylamide gels with embedded fluorescent microspheres (200 nm diameter; Invitrogen). Time-lapse imaging of fluorescent beads and growth cones was performed at 37˚C using a confocal microscope (LSM710; Carl Zeiss) equipped with a C-Apochromat 63x/1.2 W Corr objective. The growth cone area was determined by EGFP fluorescence or from DIC images. Traction forces under the growth cones were monitored by visualizing force-induced deformation of the elastic substrate, which is reflected by displacement of the beads from their original positions, and expressed as vectors. The force vectors detected by the beads under individual growth cones were then averaged, and were expressed as vectors composed of magnitude and angle () ( Figure 6-figure supplement 1A, left panel) (Toriyama et al., 2013). To compare the forces under different conditions, the magnitude and angle () of the force vectors of the individual growth cones were statistically analyzed and expressed as means ± SEM, separately (Abe et al., 2018). They were also analyzed by one-way ANOVA with Tukey's post hoc test.

Analyses of netrin-1 attached on the substrate
Netrin-1 attached on the substrate was analyzed as described (Moore et al., 2012) with modifications. Glass coverslips coated with polylysine or coated sequentially with polylysine and L1-CAM-Fc were incubated with culture medium and netrin-1 containing a 10-His tag at the C-terminus, in the absence or presence of 2 mg/ml heparin. The glasses were washed with PBS, and blocked for 1 hr at room temperature with 0.1% BSA in PBS. They were then labeled with anti-His antibody and secondary anti-mouse antibody conjugated with Alexa Fluor 488. Fluorescence images were acquired using a fluorescence microscope (IX81; Olympus) equipped with an EM-CCD camera (Ixon DU888; Andor), using a plan-Fluar 20 Â 0.45 NA objective (Olympus), and quantified using ImageJ (Fiji version).

Statistical analysis
All statistical analysis were performed using Microsoft Excel, Statistical software R (RRID:SCR_ 001905) and Graphpad prism 7 (RRID:SCR_002798). Significance was determined by the two-tailed unpaired Student's t-test in most cases. For multiple comparisons, we used one-way ANOVA with Schaffer's post hoc test or Tukey's post hoc test.