Phosphatidylinositol 3-Kinase Facilitates Microtubule-dependent Membrane Transport for Neuronal Growth Cone Guidance

The activity of PI3K is necessary for polarized cell motility. To guide extending axons, environmental cues polarize the growth cone via asymmetric generation of Ca2+ signals and subsequent intracellular mechanical events, including membrane trafficking and cytoskeletal reorganization. However, it remains unclear how PI3K is involved in such events for axon guidance. Here, we demonstrate that PI3K plays a permissive role in growth cone turning by facilitating microtubule (MT)-dependent membrane transport. Using embryonic chick dorsal root ganglion neurons in culture, attractive axon turning was induced by Ca2+ elevations on one side of the growth cone by photolyzing caged Ca2+ or caged inositol 1,4,5-trisphosphate. We show that PI3K activity was required downstream of Ca2+ signals for growth cone turning. Attractive Ca2+ signals, generated with caged Ca2+ or caged inositol 1,4,5-trisphosphate, triggered asymmetric transport of membrane vesicles from the center to the periphery of growth cones in a MT-dependent manner. This centrifugal vesicle transport was abolished by PI3K inhibitors, suggesting that PI3K is involved in growth cone attraction at the level of membrane trafficking. Consistent with this observation, immunocytochemistry showed that PI3K inhibitors reduced MTs in the growth cone peripheral domain. Time-lapse imaging of EB1 on the plus-end of MTs revealed that MT advance into the growth cone peripheral domain was dependent on PI3K activity: inhibition of the PI3K signaling pathway attenuated MT advance, whereas exogenous phosphatidylinositol 3,4,5-trisphosphate, the product of PI3K-catalyzed reactions, promoted MT advance. This study demonstrates the importance of PI3K-dependent membrane trafficking in chemotactic cell migration.

The correct wiring of the nervous system relies critically on the navigation of developing axons to their destinations. The growth cone, the tip of elongating axons, recognizes extracellular guidance cues to form proper neuronal connections (1,2): graded distribution of guidance cues causes growth cone turning toward the higher concentration (attraction) or lower concentration (repulsion) of the cues. The graded distribution also elicits asymmetric increases in cytosolic Ca 2ϩ concentrations across the growth cone, with higher Ca 2ϩ concentrations on the side of the growth cone facing the higher concen-trations of the guidance cues (3)(4)(5)(6). Such asymmetric Ca 2ϩ signals mediate attractive or repulsive turning depending on the source of Ca 2ϩ . Two types of Ca 2ϩ release from the endoplasmic reticulum, Ca 2ϩ -induced Ca 2ϩ release (CICR) 2 through ryanodine receptors and inositol 1,4,5-trisphosphate (IP 3 )-induced Ca 2ϩ release (IICR) through IP 3 receptors, are sufficient to trigger growth cone attraction (6,7). For example, CICR mediates netrin-1-induced attraction, and IICR mediates NGF-or BDNF-induced attraction (6,8,9). On the other hand, Ca 2ϩ influx from the extracellular space through transient receptor potential channels or cyclic nucleotidegated channels has been implicated in growth cone repulsion (10 -14). In this way, Ca 2ϩ serves as a critical messenger that polarizes the growth cone for guided migration.
PI3K, an enzyme that catalyzes the production of phosphatidylinositol 3,4,5-trisphosphate (PIP 3 ) from phosphatidylinositol 4,5-bisphosphate (PIP 2 ), is involved in various cellular functions during embryonic development, including proliferation, cell migration, and axon guidance (15)(16)(17)(18). It has been reported that PI3K activity is required for growth cone attractive turning responses to guidance cues such as NGF, BDNF, and netrin-1 (6,9,18). The involvement of PI3K in repulsive turning depends on the type of guidance cues: PI3K is required for repulsion by myelin-associated glycoprotein but not by semaphorin 3A (18,19). Despite the importance of PI3K, how it controls growth cone responses to guidance signals remains elusive. NGF and BDNF are able to elicit Ca 2ϩ signals in growth cones in the absence of PI3K activity (6,9), suggesting that PI3K acts downstream of Ca 2ϩ signals in attractive axon guidance. Our previous study identified a mechanism by which Ca 2ϩ signals cause attractive turning: growth cones turn by transporting membrane vesicles toward the new direction (20). Here, we examined whether PI3K is involved in such membrane trafficking and show that PI3K controls microtubule (MT) dynamics, thereby facilitating vesicle transport in growth cones during attractive turning. dish coated with L1-Fc chimeric proteins that consisted of the whole extracellular domain of L1 and the Fc region of human immunoglobulin G (21). Cultures were maintained in Leibovitz L-15 medium (Invitrogen) supplemented with N2 (Invitrogen), 750 g/ml bovine serum albumin (Invitrogen), and 20 ng/ml NGF (Promega) in a humidified atmosphere of 100% air at 37°C.
Growth Cone Turning Assay-Growth cone turning induced by focal laser-induced photolysis (FLIP) of caged compounds was assessed as described previously (6,7). In some experiments, the following reagents were applied to the culture medium at least 20 min before time-lapse imaging: 10 M LY294002, 60 nM wortmannin, 10 nM nocodazole (Merck KGaA, Darmstadt, Germany), and 5 nM tetanus toxin (List Biological Laboratories, Campbell, CA).
FM1-43 Imaging-DRG neurons were exposed to high potassium (60 mM KCl) medium containing 5 M FM1-43 (Invitrogen) for 5 min at 37°C, followed by washes with Leibovitz L-15 medium. Images of FM1-43-labeled intracellular vesicles in a growth cone were collected every 3 s using a ϫ100 objective (UPlanSApo, numerical aperture of 1.40, Olympus, Tokyo, Japan) on an inverted microscope (IX81, Olympus) equipped with an electron-multiplying CCD camera (binning set at 1 ϫ 1, Hamamatsu Photonics, Hamamatsu, Japan). Time series stacks of fluorescent images were acquired with MetaMorph Version 7.1 (MDS Analytical Technologies, Downingtown, PA) using the "stream acquisition" function under the control of an electronic stimulator (Nihon Koden, Tokyo). FM1-43 puncta showing directional translocation from the central domain (C-domain) to the peripheral domain (P-domain) of growth cones were defined as centrifugally transported vesicles.
Caged Ca 2ϩ or caged IP 3 was photolyzed by exposing a growth cone to UV light (330 -385 nm). The size of the UV light-irradiated region (ϳ10 m in diameter) was controlled by a pinhole placed in the light path, and one side of a growth cone was exposed to the UV light (500-ms duration). In some experiments, 2 M xestospongin C (Merck KGaA), 10 M LY294002, 60 nM wortmannin, or 10 nM nocodazole was applied to the culture medium at least 20 min before time-lapse imaging.
Immunocytochemistry-After a 20-min drug treatment, neurons were fixed at 37°C for 30 min in a solution containing 80 mM Na-PIPES (pH 6.9), 1 mM MgCl 2 , 1 mM EGTA, 1 mM GTP, 3% sucrose, 0.1% glutaraldehyde, and 4% formaldehyde (20). The cells were permeabilized with 0.1% Triton X-100 at room temperature for 60 min and blocked with 10% goat serum in cytoskeleton buffer (10 mM MES (pH 6.1), 138 mM KCl, 3 mM MgCl 2 , 2 mM EGTA) (22) at room temperature for 30 min. The cells were incubated with anti-␤-tubulin monoclonal antibody (1:1000 dilution in cytoskeleton buffer; Chemicon, Billerica, MA) overnight at 4°C. Primary antibody binding was visualized with Alexa Fluor 594-conjugated secondary antibody (10 g/ml; Invitrogen). Differential interference contrast (DIC) and fluorescent images were acquired with a ϫ100 objective (UPlanSApo) and a CCD camera (ORCA-ERG with binning set at 1 ϫ 1, Hamamatsu Photonics) on an inverted microscope (IX81). Distribution of MTs in the growth cone P-domain was quantified by measuring the number and length of MTs (supplemental Fig. 2). The C-domain/P-domain boundary and the area of the P-domain were determined on DIC images. Individual MTs were traced from the C-domain/P-domain boundary to their tips, and the length of all individual MTs was averaged in each growth cone. The total length was obtained by summing up the length of all individual MTs in a growth cone. The number and total length of MTs in each growth cone were divided by the area of the P-domain. These three parameters of MTs were normalized to the mean of control growth cones to even out the interexperiment variance. The combined data of three independent experiments (n ϭ 15 growth cones for each experiment) are shown in Fig. 4
EB1 Imaging-Fluorescent images of growth cones expressing EGFP-or mCherry-tagged EB1 were acquired every 3 s with a ϫ100 objective (UPlanSApo) and a CCD camera (ORCA-ERG with binning set at 2 ϫ 2) on an inverted microscope (IX81). To quantify MT dynamics, we counted the number of EB1 comets that passed across the C-domain/Pdomain boundary and migrated into the P-domain. In each growth cone, the number of EB1 comets advancing into the P-domain was divided by the length of the boundary. We also measured the life time and speed of EB1 comets that had crossed the boundary. The life time was defined as the period that an EB1 comet spent from its crossing the boundary until its disappearance, and the values of all such comets were averaged in each growth cone. The speed of EB1 comets in each growth cone was defined as the mean speed of five randomly selected comets whose life time exceeded 6 s. The effects of the following drugs on MT dynamics were assessed by comparing these parameters in growth cones before and after a 10-min drug treatment: 10 M LY294002, 60 nM wortmannin, 10 M PIP 3 (Cayman Chemical, Ann Arbor, MI), and 10 M PIP 2 (Cayman Chemical).
PIP 3 Imaging-Growth cones expressing both Akt-PH-EGFP and DsRed-Mem were visualized by total internal reflection fluorescence microscopy as described previously (23). Fluorescent images were acquired with a ϫ100 objective (PlanApo TIRFM, numerical aperture of 1.45, Olympus) and a CCD camera (ORCA-AG with binning set at 2 ϫ 2, Hamamatsu Photonics) on an inverted microscope (IX81). Using DsRed-Mem as a membrane marker (23,24), PIP 3 density in the cytoplasmic leaflet of the plasma membrane was assessed by calculating the Akt-PH-EGFP/DsRed-Mem emission ratio (defined as R). Drug-induced changes in PIP 3 density were monitored by dividing R by that before drug treatment (R pre ).
Statistics-Data are expressed as the mean Ϯ S.E. Statistical analyses were performed using GraphPad Prism Version 4.01. p Ͻ 0.05 was judged statistically significant.

PI3K Is Required for Growth Cone Turning Induced by CICR
and IICR-We have previously reported that, on an L1 substrate, photolysis of caged Ca 2ϩ or caged IP 3 on one side of the growth cone elicits localized CICR or IICR, respectively, and is sufficient to trigger growth cone attraction (6,7,25). Localized photolysis of caged compounds has been used extensively to replicate growth cone turning responses to physiological cues and to analyze molecular mechanisms underlying axon guidance (20,23,26,27). The involvement of PI3K in axon turning was examined using embryonic chick DRG neurons that had been loaded with a caged Ca 2ϩ compound, onitrophenyl-EGTA, or with a caged IP 3 compound, D-myoinositol 1,4,5-triphosphate P 4 ,P 5 -(1-(2-nitrophenyl)ethyl) ester (Fig. 1). Repeated FLIP (3-s intervals) of either of these caged compounds on one side of the growth cone resulted in attractive turning. The CICR-induced attraction was abolished in the presence of LY294002 or wortmannin. Also, the IICR-induced attraction was blocked by LY294002. The inhibitory effect of wortmannin on IICRinduced attraction was reported in our previous study (6). We confirmed that the two PI3K inhibitors significantly decrease PIP 3 density in the growth cone plasma membrane (supplemental Fig. 1). The involvement of PI3K in axon guidance was further examined using the PH domain of Akt, which blocks the PI3K signaling pathway (28,29): growth cones expressing the Akt PH domain did not exhibit attractive turning responses to CICR (Fig. 1C). These results indicate the necessity of PI3K activity for Ca 2ϩ -induced growth cone attraction.
PI3K Is Required for CICR-elicited Centrifugal Vesicle Transport-We next addressed the issue of how PI3K is involved in growth cone attraction downstream of Ca 2ϩ signals. We have previously reported that CICR induces attractive turning via asymmetric centrifugal vesicle transport and subsequent exocytosis (20). Thus, the effect of PI3K inhibitors on CICR-elicited vesicle transport was examined (Fig. 2). Intracellular vesicles were visualized with FM1-43, a lipophilic fluorescent dye, while CICR was generated by photolyzing caged Ca 2ϩ on one side of the growth cone. Consistent with our previous observation (20), CICR increased the number of FM1-43-labeled vesicles migrating centrifugally from the Cto P-domain only on the side with CICR (Fig. 2, A and B, Control). UV irradiation in the absence of caged Ca 2ϩ loading had no effect on vesicle transport (Fig. 2B, Blank). Pretreatment with PI3K inhibitors abolished CICR-elicited vesicle transport (Fig. 2B, ϩLY294002 and ϩWortmannin), indicating that PI3K is required for centrifugal vesicle transport during growth cone attraction.
IICR Elicits Centrifugal Vesicle Transport in a PI3K-dependent Manner-So-called "Ca 2ϩ nanodomains or microdomains," spatially restricted Ca 2ϩ elevations, exert distinct intracellular responses depending on the distance between the open Ca 2ϩ channels and Ca 2ϩ -sensitive effectors (30). There- fore, although both CICR and IICR elicit growth cone attraction, their downstream mechanisms may be different. We tested whether IICR-elicited attraction depended on centrifugal vesicle transport and subsequent exocytosis in growth cones. Similar to the effect of CICR, IICR generated by photolyzing caged IP 3 facilitated centrifugal transport of FM1-43labeled vesicles only on the side with IICR (Fig. 3, A and B,  Control). This facilitation was completely blocked by treatment with an IICR inhibitor (Fig. 3B, ϩXestospongin C). Also, perturbation of MT dynamics with nocodazole negated IICRelicited vesicle transport (Fig. 3B, ϩNocodazole). Consistent with this observation, nocodazole-treated growth cones did not respond to FLIP-induced IICR and showed practically straight migration (0.4 Ϯ 3.4°, n ϭ 16 growth cones). Because CICR-elicited growth cone attraction requires tetanus toxin-sensitive exocytosis, we also examined the effect of tetanus toxin on IICR-elicited attraction. In the presence of tetanus toxin, IICR did not induce attractive turning (1.1 Ϯ 3.6°, n ϭ 16 growth cones). The effect of these two drugs on IICR-elicited attraction was analyzed statistically by comparing the turning angles of drug-treated growth cones with those of control growth cones (17.1 Ϯ 3.8°, n ϭ 16) (Fig. 1B, Control) using Dunnett's multiple comparison test: p Ͻ 0.01 for nocodazole and p Ͻ 0.01 for tetanus toxin. Collectively, these results strongly suggest that CICR and IICR elicit growth cone attraction via a common mechanism: MT-dependent centrifugal vesicle transport and subsequent exocytosis. Ca 2ϩ signals were generated every 3 s by repetitive UV photolysis of caged Ca 2ϩ at the area denoted by yellow circles. Arrowheads indicate FM1-43labeled vesicles that migrated centrifugally on the side with Ca 2ϩ signals (near side). Blue lines depict the growth cone outline. Digits represent seconds after the start of UV photolysis. Scale bars ϭ 5 m. B, frequency of centrifugal vesicle migration on the near and far sides of the growth cone before (Pre) and after (UV) the start of repetitive UV irradiation. Caged Ca 2ϩ loading was omitted in Blank. Numbers in parentheses indicate the number of growth cones examined. The Bonferroni multiple comparison test was used to compare 1) the frequency between both sides of the growth cone and 2) the frequency before and after photolysis on each side. #, p Ͻ 0.05 (near side versus far side during UV photolysis); *, p Ͻ 0.05 (Pre versus UV on the near side).
We then examined the effect of PI3K inhibitors on IICRelicited movement of FM1-43-labeled vesicles in growth cones and showed the requirement of PI3K for IICR-induced membrane trafficking (Fig. 3B, ϩLY294002 and ϩWortmannin). Therefore, we concluded that PI3K is required for CICRand IICR-elicited centrifugal vesicle transport, the critical mechanical event for growth cone attraction.
PI3K Controls MT Dynamics in the Growth Cone P-domain-Because Ca 2ϩ -triggered vesicle transport depends on MT dynamics, we hypothesized that PI3K facilitates vesicle transport by controlling MT dynamics. MTs are concentrated in the axon shaft and C-domain and are sprayed out into the P-domain (31). In our study, the C-domain/P-domain boundary (shown in figures as dashed lines) was determined based on DIC images, and MTs distributed in the P-domain were analyzed (supplemental Fig. 2). In a control growth cone treated with 0.06% Me 2 SO, many MTs advanced into the P-domain (Fig. 4A). On the contrary, fewer MTs were observed in the P-domain of growth cones treated with PI3K inhibitors. Quantitative analyses showed that the total length of MTs in the P-domain was significantly reduced by inhibition of PI3K (Fig. 4B). This reduction was attributable to the reduced number of MTs located in the P-domain because the length of individual MTs was unaffected (Fig. 4, C and D).
The role of PI3K in MT dynamics was further investigated in DRG neurons expressing EB1 tagged with fluorescent proteins. EB1 is a member of the plus-end tracking protein family that associates with the actively polymerizing plus-ends of MTs (32,33). Therefore, EB1-EGFP has been used extensively to study MT dynamics in various cell types, including neurons (34 -36). Importantly, EB1 binding to the plus-end of MTs is unaffected after inhibition of PI3K (22), indicating that EB1 imaging is a reliable method to assess MT dynamics even in the absence of PI3K activity. We examined the behavior of EB1 comets in growth cones before and after treatment with PI3K inhibitors or phosphoinositides. Although many EB1 comets showed centrifugal migration in an untreated growth cone, such comets tended to be less frequently observed in the same growth cone after treatment with PI3K inhibitors (Fig. 5 and supplemental Movies 1 and 2). The number of EB1 comets crossing the C-domain/P-domain boundary toward the P-domain was normalized by the length of boundary, and this normalized number was used as an index of the frequency of MT advance into the P-domain. The frequency of boundary crossing was decreased by the PI3K inhibitors LY294002 and wortmannin (Fig. 6A) but not by control treatment with vehicle (Fig. 6A, DMSO). These results are consistent with the immunofluorescence data showing that the number of MTs in the P-domain was reduced after PI3K inhi-

PI3K-dependent Membrane Trafficking for Growth Cone Guidance
bition. Because PI3K catalyzes the conversion of PIP 2 to PIP 3 , we examined the effect of these phospholipids. The addition of PIP 3 to the culture medium caused a significant increase in Akt-PH-EGFP recruitment to the cytoplasmic surface of the growth cone plasma membrane (supplemental Fig. 1), indicating that exogenous PIP 3 can mimic activation of intra-  cellular PI3K signaling cascade. Conversely, PIP 2 application attenuated Akt-PH-EGFP recruitment presumably because PIP 2 decreased the amount of PIP 3 through activation of PTEN (phosphatase and tensin homolog deleted on chromosome 10) (37,38). PIP 3 treatment of growth cones increased the frequency of MT advance, whereas vehicle (H 2 O) treatment had no detectable effect (Fig. 6A and supplemental Movie 3). In contrast, PIP 2 slightly decreased the frequency of MT advance ( Fig. 6A and supplemental Movie 4). We also quantified the life time and speed of EB1 comets that had crossed the C-domain/P-domain boundary (see "Experimental Procedures" for details). The PI3K inhibitors LY294002 and wortmannin shortened the life time and reduced the speed of EB1 comets (Fig. 6, B and C). Furthermore, PIP 3 prolonged the life time and increased the speed of EB1 comets, although PIP 2 had no detectable effect (Fig. 6, B and C). Consistent with the effects of pharmacological agents, inhibition of the PI3K signaling pathway by the Akt PH domain also decreased the frequency of boundary crossing, life time, and speed of EB1 comets in the growth cones ( Fig. 7 and supplemental Movies 5 and 6). Taken collectively, these results indicate that PI3K facilitates MT ad-vance into the growth cone P-domain and thereby contributes to centrifugal membrane trafficking for axon turning.

DISCUSSION
Ca 2ϩ mediates growth cone turning responses to various guidance cues, e.g. NGF and BDNF attract growth cones via FIGURE 6. PI3K facilitates MT advance into the growth cone P-domain. Shown is the effect of PI3K inhibitors, PIP 3 , and PIP 2 on EB1 dynamics in growth cones. As controls, the growth cones were treated with vehicles: dimethyl sulfoxide (DMSO) for PI3K inhibitors and H 2 O for PIP 3 and PIP 2 . A, the number of EB1 comets that crossed the unit length of the C-domain/P-domain boundary was compared in each growth cone before (Pre) and after (Post) drug treatment. Each line represents a drug-induced change in a single growth cone. B and C, shown are the drug-induced changes in the life time (B) and speed (C) of EB1 comets that had crossed the boundary. *, p Ͻ 0.05 (paired t test); **, p Ͻ 0.01; ***, p Ͻ 0.001. generation of IICR (6,9), and netrin-1 attracts growth cones via CICR (8). Here, we have demonstrated how PI3K is involved in attractive axon guidance downstream of CICR and IICR. We also showed that CICR and IICR share common downstream mechanisms for turning, in which membrane vesicles are transported centrifugally along MTs on the side with the Ca 2ϩ signals. This type of membrane trafficking drives attractive turning toward guidance cues via asymmetric addition of membrane components and functional molecules to the growth cone surface (20). In this study, we have identified the critical role of PI3K in Ca 2ϩ -triggered MT-dependent vesicle transport during growth cone attraction. This mechanism is likely to exist in axon guidance mediated by various chemoattractants. In contrast, the requirement of PI3K for repulsive turning depends on the type of guidance cues (18,19). Recent studies have implicated PI3K in endocytosis that drives growth cone repulsion downstream of Ca 2ϩ signals (23,39). Although defining the precise role of PI3K in repulsive guidance requires future studies, including those that address whether the generation of repulsive Ca 2ϩ signals depends on PI3K activity, this kinase may control membrane trafficking for bidirectional axon guidance.
Our study shows that the PI3K product PIP 3 promotes MT advance into the growth cone P-domain. PIP 3 activates Akt, which inactivates glycogen synthase kinase-3␤ (35). Control of MT dynamics by PI3K may be mediated by glycogen synthase kinase-3␤, which negatively regulates MT-binding proteins, including adenomatous polyposis coli and CRMP2 (collapsin-response mediator protein-2) (40,41). It was reported that the PI3K/glycogen synthase kinase-3␤ pathway plays a role in neuronal polarization and axon elongation via regulating MT-binding proteins (22,41). Also, functional perturbation of adenomatous polyposis coli on one side of the growth cone causes asymmetric growth cone expansion and its turning to the side with the increased area (42), suggesting the linkage between MT-dependent membrane trafficking and polarized migration. Besides the effect of PI3K on MT-binding proteins, PI3K could influence MT advance by altering the dynamics of F-actin. The position of MT plus-ends in the growth cone P-domain is determined by the balance between MT polymerization and retrograde F-actin flow because MTs are associated with F-actin and thereby pushed back continuously toward the C-domain (43). The retrograde F-actin flow is powered by myosin motors, in particular myosin II (44). PI3K inhibition causes myosin II activation (45) and would presumably accelerate the speed of retrograde F-actin flow. This might be an alternative mechanism that underlies the reduced frequency of MT advance into the P-domain after PI3K inhibition.
Mechanisms of directed migration in neuronal growth cones and non-neuronal cells share several common features. For example, localized Ca 2ϩ signals at the leading edge of fibroblasts point to the direction of migration (46). Furthermore, MT dynamics and membrane trafficking have been implicated in directed cell migration (47)(48)(49). PI3K activity is required for efficient cell migration toward increasing concentrations of chemoattractants (50,51), although more recent studies have suggested that PI3K is dispensable when chemoattractant gradients are steep (52,53). During chemotaxis, PI3K activation is restricted to the region of the cell facing the source of chemoattractant, where it promotes F-actin assembly and plays an instructive role in leading edge formation and cell migration (51,54,55). Besides this mode of action of PI3K, our findings in neuronal growth cones imply that PI3K may also contribute to chemotactic migration of non-neuronal cells through facilitating MT-dependent membrane transport toward the cell front.
It remains unclear whether PI3K plays an instructive role in axon guidance because there is no direct evidence of spatially restricted activation of PI3K during growth cone turning. Our previous study showed that an extracellular gradient of NGF causes asymmetric IP 3 production across the growth cone most likely via asymmetric NGF binding to its high affinity receptor TrkA (6). Considering that PI3K can also be activated upon NGF binding to TrkA (56), it is possible that PI3K activity becomes asymmetric during NGF-induced growth cone attraction. Similarly, receptors for other attractive cues such as BDNF and Wnt-4 can activate the PI3K pathway (56 -58). Therefore, localized PI3K activation on the side facing the source of attractive cues would polarize the growth cone by facilitating MT advance and centrifugal vesicle transport.
In summary, we have reported the precise role of PI3K in growth cone attractive turning: PI3K facilitates MT advance into the P-domain and therefore allows attractive Ca 2ϩ signals to promote centrifugal vesicle transport. Our study contributes to a better understanding of the molecular mechanisms underlying axon guidance and suggests the mode of PI3K action in cell chemotaxis in general.