c-Abl Tyrosine Kinase Is Required for BDNF-Induced Dendritic Branching and Growth

Brain-derived neurotrophic factor (BDNF) induces activation of the TrkB receptor and several downstream pathways (MAPK, PI3K, PLC-γ), leading to neuronal survival, growth, and plasticity. It has been well established that TrkB signaling regulation is required for neurite formation and dendritic arborization, but the specific mechanism is not fully understood. The non-receptor tyrosine kinase c-Abl is a possible candidate regulator of this process, as it has been implicated in tyrosine kinase receptors’ signaling and trafficking, as well as regulation of neuronal morphogenesis. To assess the role of c-Abl in BDNF-induced dendritic arborization, wild-type and c-Abl-KO neurons were stimulated with BDNF, and diverse strategies were employed to probe the function of c-Abl, including the use of pharmacological inhibitors, an allosteric c-Abl activator, and shRNA to downregulates c-Abl expression. Surprisingly, BDNF promoted c-Abl activation and interaction with TrkB receptors. Furthermore, pharmacological c-Abl inhibition and genetic ablation abolished BDNF-induced dendritic arborization and increased the availability of TrkB in the cell membrane. Interestingly, inhibition or genetic ablation of c-Abl had no effect on the classic TrkB downstream pathways. Together, our results suggest that BDNF/TrkB-dependent c-Abl activation is a novel and essential mechanism in TrkB signaling.


Introduction
Regulation of dendritic arborization is essential for brain function since it determines the receptive field of neurons [1]. Extrinsic neuronal signals that promote dendritic development include soluble ligands such as neurotrophins that bind to specific membrane receptors. Brain-derived neurotrophic factor (BDNF) binds to the tyrosine kinase receptor TrkB [2] promoting neurite extension, arborization and growth of dendrites, the formation and establishment of synaptic connections, survival, and differentiation of neuronal populations [3][4][5][6][7]. The binding of BDNF to TrkB stimulates its dimerization and transphosphorylation at Tyr residues, which in turn recruits signaling adaptors to transduce the signal inside the cell. The classical downstream signaling pathways triggered after TrkB-activation are the mitogen protein-activated protein kinases (MAPK), phosphoinositide-3-kinase (PI3K), and phospholipase-C gamma (PLC-γ) pathways. After TrkB activation by its ligand, the receptor undergoes endocytosis and transport to the neuronal soma. The internalized neurotrophin bound to its Trk receptor continues signaling in endosomes by activating components of the Ras-MAP kinase, PLC-γ, and PI3K pathways, thus participating in long-range signaling and transcriptional events [8]. Subsequently, the signaling endosome follows its post endocytic transport, and the receptor can be recycled into the plasma membrane or degraded by lysosomes [9][10][11][12]. between 10 and 30 µm away from the soma ( Figure 1B). Furthermore, we found that the branching points in the BDNF-treated neurons were twofold higher than those in control neurons ( Figure 1D). This increase promoted by BDNF is dependent on the TrkB receptor, since when we used K252a, an inhibitor of Trk receptors, the BDNF-induced increase in intersections and the number of branches was prevented (Supplementary Figure S1B,C). Therefore, BDNF treatment stimulates dendrite growth and arborization as we have previously shown [8,37,38]. Interestingly, when neurons were treated with BDNF in the presence of the c-Abl inhibitors, Imatinib or GNF2, BDNF-induced dendritic branching was reduced to similar levels of unstimulated neurons ( Figure 1A). Neurons treated only with c-Abl inhibitors showed a slight decrease in dendritic arbor complexity as compared to the DMSO condition ( Figure 1C). Furthermore, Imatinib and GNF2 prevented the increase in dendritic growth and branching induced by BDNF ( Figure 1B,D) as well as the increase in the number of primary dendrites ( Figure 1E). Therefore, c-Abl inhibition abolished BDNF-promoted enhancement of dendritic growth and complexity.   To confirm the requirement of c-Abl expression in the dendritic arborization increase induced by BDNF, we used hippocampal neurons derived from c-Abl KO embryos. Seven DIV c-Abl KO and WT hippocampal neurons were treated with BDNF for 48 h, and primary neurites and branching points were evaluated. Consistent with the results obtained with c-Abl inhibitors, the absence of c-Abl expression in neurons reduced BDNF-induced dendritic growth to basal levels ( Figure 1F). The Sholl analysis as well as the number of branching points quantifications confirmed this observation ( Figure 1G,H).
It has been reported that abl/arg double-null mice exhibit defective neural tube morphology and gross alterations in neuroepithelial actin cytoskeleton structures, suggesting that c-Abl knockout neurons could have alterations in dendritic tree development [25]. To discard any confounding effects attributable to developmental alterations in the dendritic arbor of c-Abl KO mouse neurons, we knocked down c-Abl expression by transfecting a shRNA targeting this protein into WT neurons, and then we evaluated the neuronal morphology after BDNF treatment. Consistently, hippocampal neurons transfected with the c-Abl-targeting shRNA did not increase their dendritic branching in response to BDNF treatment (Supplementary Figure S2).
Interestingly, neurons treated with a c-Abl activator (DPH) showed more branched dendrites compared to control neurons ( Figure 1J). After 48 h of DPH treatment, neurons showed a significant increase in dendrite intersections between 10 and 75 µm of the soma ( Figure 1K). In addition, DPH increased the number of branching points ( Figure 1L) and primary dendrites ( Figure 1L). Consistent with the reported role of c-Abl activation on apoptosis induction, we observed that some neurons died after 48 h of treatment with DPH. However, we verified that DPH only produced a 10% reduction of viability in MTT assays (Supplementary Figure S3A). Thus, the use of two different pharmacological inhibitors of c-Abl and two different approximations to reduce c-Abl expression demonstrated that c-Abl activity and expression are required for BDNF/TrkB-induced dendritic arborization in vitro and suggests that c-Abl activation is required for BDNF/TrkB-induced dendritic branching.

BDNF Increases c-Abl Activation and the Interaction of TrkB with c-Abl
It has been described that c-Abl interacts with the TrkA receptor in the juxtamembrane region of TrkA [39]. To explore a potential interaction between the TrkB receptor and c-Abl, we performed immunoprecipitation experiments to evaluate whether c-Abl associates with TrkB after BDNF stimulation. Interestingly, BDNF promoted an interaction between the TrkB receptor and c-Abl (Figure 2A). No association was observed when the immunoprecipitation assay was performed with a control IgG. To determine whether c-Abl is activated in response to BDNF, we examined c-Abl phosphorylation at tyrosine residue 412 [40,41] by immunostaining. Immunofluorescence against phosphorylated c-Abl showed that BDNF (50 ng/mL BDNF for 30 min) increased c-Abl phosphorylation in the soma and the dendrites of hippocampal neurons ( Figure 2B). In addition, we confirmed the effect of BDNF on c-Abl activation by western blot analysis. Neurons stimulated with BDNF showed a significant increase in the levels of c-Abl phosphorylation at tyrosine 412 after 30 min ( Figure 2C,D). Concomitant with c-Abl activation by BDNF, we observed a significant increment in the phosphorylation of one of the best-characterized substrates of c-Abl, the Crk adaptor protein (phospho-CrkII Tyr221) after 15 min of BDNF stimulation ( Figure 2E). We used 5 µM DPH, a c-Abl activator, for 60 min as a positive control to induce c-Abl activity. Even though DPH induced dendritic branching ( Figure 1J-L), it did not increase TrkB phosphorylation ( Figure 2C). These results collectively suggest that BDNF activates c-Abl directly downstream of TrkB activation.

BDNF-Induced c-Abl Activation Requires TrkB Activity but Does Not Involve the TrkB Downstream Signaling Pathways Regulated by PI3K, ERK1/2, and PLC-γ Signaling
Since BDNF increases c-Abl activation and interaction with TrkB, we investigated whether TrkB signaling induces c-Abl activation. To this end, neurons were exposed to BDNF and co-treated or not with the pan-Trk tyrosine kinase inhibitor K252a ( Figure 3A), and c-Abl phosphorylation was evaluated by western blotting. The increase in phosphorylation at Tyr 412 of c-Abl induced by BDNF was abolished by the K252a inhibitor, suggesting that Trk kinase activity is required for c-Abl phosphorylation. Furthermore, we evaluated BDNF-induced c-Abl phosphorylation in hippocampal neurons derived from TrkB F616A mouse embryos, which express a modified TrkB receptor whose activity is sensitive to the kinase inhibitor 1NM-PP1. Neuronal cultures were pre-treated for 1 h with 1NM-PP1 (1 µM) and next, incubated with 50 ng/mL BDNF in the presence or absence of 1NM-PP1. In agreement with the results obtained with K252a, we observed that 1NM-PP1-mediated TrkB inhibition prevented the activation of c-Abl induced by BDNF, which was previously observed by immunofluorescence and western blotting analysis ( Figure 3B,C). These results confirm that BDNF-induced c-Abl activation is TrkB-dependent.   Phospho-c-Abl (Tyr-412), c-Abl, Phospho-CrkII (Tyr221), CrkII, Phospho-Akt (Ser473), and GAPDH levels were analyzed by western blot. n = 3 (F) Seven DIV rat hippocampal neurons were pretreated with 5 µM PLC-γ inhibitor, U73122, for 60 min, and then stimulated with 50 ng/mL of BDNF for 30 min. Phospho-c-Abl (Tyr-412), c-Abl, Phospho-CrkII (Tyr221), Phospho-PKC (Thr-538), PKC, and GAPDH levels were analyzed by western blot. n = 2.
The phosphorylation sites on the TrkB cytosolic tail serve as specific anchoring sites for different intracellular adapter proteins, which contain Src 2 (SH2) and phosphotyrosinebinding domains (PTB). These adapter proteins mediate the activation of TrkB downstream signaling pathways, including RAS/MAPK (ERK1/2), PI3K/Akt, and PLC-γ/PKC [41]. To investigate whether c-Abl phosphorylation is dependent on one of these pathways, we evaluated the activation of c-Abl after BDNF stimulation in the presence or absence of inhibitors for these TrkB downstream signaling pathways. Surprisingly, BDNF-induced phosphorylation of c-Abl was observed in the presence of the neuronal treatments with LY294002 a PI3K inhibitor ( Figure 3D), U73122, an ERK1/2 inhibitor ( Figure 3E) or U0126, a PLC-γ inhibitor ( Figure 3F). These inhibitors effectively reduced the target signaling pathway since we observed a dose-dependent reduction of Akt, PKC, and ERK1/2, phosphorylation. Therefore, the activation of the ERK1/2, PI3K, and PLC-γ pathways is not involved in the BDNF/TrkB-induced c-Abl activation, suggesting a direct activation of c-Abl by TrkB.

c-Abl Restricts Basal Levels of TrkB and Downregulates Surface TrkB Levels but Does Not Affect BDNF-Induced Endocytosis
To obtain further insight into the mechanism by which c-Abl allows BDNF/TrkBinduced dendritic branching, we examined the requirement of c-Abl activity for TrkB endocytosis and endocytic trafficking. We started by evaluating whether c-Abl activity affects the levels of TrkB receptors in neurons after BDNF stimulation. We evaluated TrkB receptor levels in neurons treated with the c-Abl inhibitor Imatinib (5 µM) and performed activation time curves with BDNF (50 ng/mL). We observed an increased basal level of TrkB when c-Abl activity was reduced. Neurons maintained this trend at 1 h of stimulation with BDNF; however, at longer times no differences were observed in the levels of the TrkB receptors ( Figure 4A). Along the same lines, we observed that the basal levels of TrkB in c-Abl KO neurons were increased compared to WT neurons. When we stimulated with 50 ng/mL of BDNF, the decrease in TrkB levels over time did not show differences between neurons with or without c-Abl expression ( Figure 4B).
Then, we explored if the increased basal levels of TrkB in c-Abl null neurons or during c-Abl inhibition were due to alterations in TrkB receptor endocytic trafficking. We pretreated neurons with c-Abl inhibitors (Imatinib or GNF-2) for 1 h and then stimulated with BDNF for 30 min and measured levels of TrkB receptor on the plasma membrane using a surface biotinylation assay. The basal levels of TrkB increased 2 and 1.5 times by the treatments with the c-Abl inhibitors Imatinib and GNF2, respectively, compared to control ( Figure 4C,D). Interestingly, BDNF treatment induced a reduction of TrkB receptors in the plasma membrane in control neurons and in the neurons treated with the c-Abl inhibitors, and both conditions reached a similar amount of surface TrkB receptor by the end of the BDNF stimulus. This result shows that TrkB internalization in response to BDNF increases significatively (3-fold) when c-Abl is inhibited.
We then investigated whether TrkB receptors located in the plasma membrane were sorted to the lysosomes in the presence or absence of c-Abl after BDNF treatment using an antibody feeding assay [37]. To evaluate this, we transfected c-Abl-KO neurons with a Flag-TrkB coding plasmid, and after 24 h of expression we treated the transfected neurons with a Flag-specific antibody at 4 • C (20 min). The excess antibody was then washed off, and the neurons were stimulated with BDNF for 3 h at 37 • C. After washing and fixation, we evaluated the co-localization of Flag immunostaining with an antibody against Lamp1. The colocalization of TrkB and Lamp1 signals indicates TrkB presence in lysosomes.
Although a slight, non-statistically significant decrease in TrkB and Lamp1 colocalization was observed in c-Abl-KO neurons compared to WT neurons in basal conditions, no significant differences were observed after BDNF treatment ( Figure 4E). Our results show that inhibition of c-Abl or reduction of c-Abl expression increases the levels of TrkB receptor on the cell surface. Consistent with these results, when hippocampal neurons were treated with DPH for 30 min, the c-Abl activator reduced the levels of TrkB in the plasma membrane to the same level as the cells treated with BDNF ( Figure 4F). Furthermore, we performed treatments with DPH and BDNF in non-permeabilized neurons to evaluate the amount of endogenous TrkB in the cell surface. This was achieved by using an antibody directed against the extracellular domain of the receptor. As expected, the immunofluorescence quantification showed a decrease in the signal for TrkB (red) in both BDNF-treated neurons (Supplementary Figure S3B,C) and DPH-treated neurons, suggesting that both treatments induce TrkB internalization. Finally, neurons transfected with TrkB-Flag and labeled with anti-Flag antibodies at 4 • C showed that after 30 min of DPH treatment at 37 • C, TrkB accumulated in intracellular compartments independently of BDNF treatment ( Figure 4G). Altogether, these results suggest that in the absence of BDNF, the basal levels of c-Abl activity downregulate the levels of TrkB receptors at the cell surface.

BDNF Downstream Signaling Is Not Affected by c-Abl Activity
Although c-Abl expression and activity regulates the trafficking of TrkB in the absence of BDNF, it did not seem to regulate BDNF-dependent internalization and intracellular trafficking and does not explain the requirement of c-Abl for BDNF-TrkB-dependent enhancement of dendritic arborization. Since c-Abl phosphorylation was not regulated by TrkB downstream signaling pathways, we evaluated whether c-Abl is upstream of ERK1/2, PI3K, or PLC-γ. It is well known that the ERK1/2 and PI3K signaling pathways are required for BDNF-induced dendritic growth after TrkB activation [42,43]. To pursue this aim, we first analyzed if c-Abl activity modulates BDNF-induced TrkB activation. The increase of TrkB phosphorylation at Tyrosine 515 induced by BDNF was observed both in control neurons and in the presence of the c-Abl inhibitors, Imatinib or GNF2 ( Figure 5A,B). This suggests that c-Abl activity is not required for TrkB activation. Next, we focused on PI3K/Akt and ERK1/2 phosphorylation induced by BDNF. Interestingly, the presence of Imatinib and GNF2 did not affect the activation of these signaling pathways ( Figure 5C,D).  To confirm these results with a loss of function approach we evaluated the phosphorylation levels of TrkB, ERK1/2, and Akt in response to BDNF treatment in c-Abl KO neurons. Consistently, with our previous results using c-Abl inhibitors, hippocampal c-Abl-KO neurons showed similar levels of activation of TrkB in the presence of BDNF compared to wild-type neurons ( Figure 5E,F). Furthermore, the magnitude and time course of TrkB, Akt, and ERK1/2 activation induced by BDNF (0, 1, 3, or 6 h) were similar in WT and c-Abl null neurons ( Figure 5G,H). Therefore, c-Abl function is not required for the activation of classical BDNF-TrkB downstream signaling cascades.

c-Abl Increases the Retrograde Movement of TrkB Vesicle in Neurites
Studies have established that the post-endocytic trafficking of TrkB vesicles is required for dendritic arborization, by increasing local signaling in dendrites and synapses [37,38,[44][45][46].
To study the effect of c-Abl on TrkB post-endocytic vesicular trafficking, hippocampal neurons were co-transfected with a c-Abl-GFP construct to overexpress c-Abl and TrkB-mCherry. Then, live cell imaging was performed to assess the anterograde and retrograde speed of TrkB-mCherry moving vesicles. The quantification of the vesicular transport speeds showed that c-Abl overexpression increased the speed of TrkB vesicles moving in the retrograde direction but had no effect on anterograde speed ( Figure 6A). To confirm that c-Abl affects TrkB vesicular velocity, c-Abl-KO neurons were transfected with TrkB-mCherry fusion protein, and the same analysis of live cell images was performed. Interestingly, the retrograde speed of TrkB-mCherry fusion protein was significantly reduced in c-Abl knockout neurons as compared to wild-type neurons ( Figure 6B). In addition, when analyzing the TrkB-m-Cherry vesicles transfected in c-Abl-KO neurons, we observed TrkB vesicles with elongated and tubular shapes, different from the circular morphology of the vesicles in WT neurons. (Figure 6C). Taken together, analysis of TrkB vesicle trafficking suggests that c-Abl activity is required for an efficient retrograde TrkB transport rate in neurites.

Discussion
Development and maintenance of the dendritic arbor are crucial for proper brain function [47]. It is known that the c-Abl tyrosine kinase participates in dendrogenesis [30,31]; however, the signaling pathways that regulate its activation are not fully understood. In the present study, we identified c-Abl as a new TrkB receptor downstream signal component required for the inductive effect of BDNF on dendritic growth and arborization. Consistent with the literature, the c-Abl activator DPH induced an increase in the arborization and growth of dendritic trees. Interestingly, we observed that two c-Abl inhibitors that have different inhibition mechanisms, Imatinib and GNF2, have the same effect on preventing BDNF-induced dendritic arborization. Furthermore, we observed similar results using neurons transfected with a shRNA targeting c-Abl, and c-Abl-KO neurons, confirming the need for c-Abl activity for BDNF-induced TrkB activation to effectively promote dendritic arborization. Using western blot and immunofluorescence in TrkBF616A hippocampal neurons, we observed that BDNF promotes c-Abl phosphorylation in a TrkB-dependent manner. Interestingly, ERK1/2, PI3K, and PLC-γ signaling pathways were not required for c-Abl activation, suggesting that c-Abl acts through an independent pathway downstream of TrkB. Moreover, by surface membrane biotinylation, we observed that inhibition of c-Abl activity increases the availability of TrkB in the membrane in the absence of its ligand. This process did not affect BDNF-dependent endocytosis of the receptor or BDNF-induced sorting to lysosomes. Together, our results suggest that BDNF/TrkB-dependent c-Abl activation is a novel mechanism that is essential for effective TrkB signaling leading to dendrite arborization. Moreover, our results indicate that c-Abl function downstream of BDNF/TrkB is independent of the more studied TrkB downstream pathways: MAPK, PI3K, and PLC-γ. It is possible that other c-Abl-regulated processes are contributing to BDNFdependent dendritic branching including regulation of cytoskeleton and postendocytic trafficking of the receptor. Consistent with this last possibility, we found that the absence of c-Abl decreases the retrograde vesicular speed of the TrkB receptor, and, oppositely, the overexpression of an active c-Abl-GFP increases the retrograde vesicular speed of the receptor. Retrograde vesicular TrkB receptor transport has been described as a key event for BDNF-TrkB-induced dendritic arborization.
Reciprocal interactions between c-Abl and tyrosine kinase receptors, including EGFR, PDGFRB, and EPHB2, have been demonstrated in previous studies [20][21][22]. Interestingly, studies in HEK293 cells overexpressing c-Abl and TrkA have shown that these two proteins interact in the juxtamembrane region of the TrkA receptor [39]. This interaction has also been demonstrated in yeast in two hybrid experiments [48]. This suggests that the interaction between c-Abl and members of the Trk receptors family is a conserved mechanism among these two kinase proteins. Since we only observed TrkB-c-Abl association under BDNF stimulation, we could speculate that c-Abl is a signal transducer for BDNF-TrkB signaling. Consistently, using a pan-Trk inhibitor or neurons derived from TrkBF616A knock-in mice, we showed that BDNF promotes c-Abl activation and CrkII phosphory-lation, a downstream target of c-Abl. Similarly, studies in PC12 cells have shown that NGF induces CrkII phosphorylation in a c-Abl-dependent manner, impacting processes of cellular morphogenesis [23]. However, Crk is not required for BDNF-induced dendritic arborization [49], suggesting that TrkB-c-Abl signals through other pathways to regulate dendritic arborization. To elucidate the pathway involved in c-Abl activation, we pharmacologically inhibited each classic downstream pathway of TrkB including MAPK, AKT, and PLC-γ. To our surprise, the inhibition of these pathways did not prevent c-Abl phosphorylation induced by BDNF, and vice versa, c-Abl inhibition did not affect the activation of the TrkB signaling pathways. Altogether, our findings suggest that c-Abl plays a role in the regulation of neuronal morphology as an independent downstream pathway triggered by BDNF activation of TrkB. Further experiments are required to elucidate the complete TrkB-c-Abl signaling axis and the precise molecular mechanisms it activates to mediate dendritic growth.
One fundamental question in the neurotrophic factor field is how neurotrophins coordinate all the complex cellular processes that are involved in neurite growth, such as receptor endocytosis, endosomal trafficking, and cytoskeleton dynamics regulation. c-Abl has been implicated in all three of these processes, making it a very interesting candidate for downstream mediation of TrkB activation.
Endocytosis is a highly dynamic processes that requires tight regulation of the cytoskeleton. The c-Abl tyrosine kinase has been implicated in receptor tyrosine kinase signaling and endocytosis in non-neuronal cell types. For example, c-Abl stabilizes the EGF receptor (EGFR) in the membrane, negatively affecting its endocytosis and traffic to the lysosomes in a ligand-dependent manner [20]. c-Abl has also been shown to stabilize the transferrin receptor in the cell membrane of MEF cells [50]. On the other hand, c-Abl positively regulates BCR internalization through CrkII phosphorylation and Rac1 activation [51], demonstrating that c-Abl is capable of regulating growth factor receptor endocytosis via the regulation of actin dynamics. When we biotinylated surface membrane proteins in neuronal cultures, both Imatinib and GNF2 considerably increased the available levels of TrkB in the membrane. However, the inhibition of c-Abl did not affect BDNF-promoted receptor internalization, suggesting that c-Abl restricts the availability of TrkB in the plasma membrane independently of BDNF. Consistently with the increase of receptor availability, when c-Abl was inhibited, we observed an over-activation of the canonical TrkB downstream signaling pathways. An enticing possibility is that c-Abl could be regulating TrkB endocytosis via activation of cyclin-dependent kinase 5 (Cdk5). We and others have shown c-Abl can activate Cdk5 [27] and mediate physiological and pathophysiological processes [52], but their joint contribution to BDNF-induced dendritic branching has not been evaluated to date. Our results show that BDNF induces an increase in active Cdk5 (phosphorylated on Tyr15) which can be partially prevented by inhibiting c-Abl with GNF-2, an allosteric inhibitor (Supplementary Figure S4). Cdk5 plays key roles in tyrosine kinase receptor endocytosis by regulating endophilin-associated machinery [53]. BDNF-induced TrkB endocytosis is regulated through a tripartite interaction between Endophilin A1, retrolinkin, and the WAVE1 complex [54], suggesting a possible role for Cdk5 in specifically regulating TrkB endocytosis. Normal Trkb signaling can be disrupted by prolonged phosphorylation or lack of internalization of the receptor [10,55,56], suggesting that the increase of membrane-associated TrkB availability could be detrimental for proper BDNF signaling. The deregulated activation of c-Abl in disease states could sequester TrkB in the cell by inhibiting BDNF/TrkB signaling.
Post-endocytic trafficking of Trk receptors is required for proper signaling and neuronal function [11,57]. Indeed, several lines of evidence have shown that the effects of BDNF/TrkB on the activation of key downstream targets are dependent on sustained signaling from endosomes after internalization of BDNF/TrkB [8,58,59]. For example, it has been shown that Rab5 and Rab11 activity is required for BDNF-mediated dendritic branching as it maintains persistent activation of TrkB downstream signaling pathways [8,38]. Dendritic Rab5-positive early endosomes co-localize with TrkB and increase retrograde movement in response to BDNF treatment, most likely to convey the trophic signal to the nucleus and promote the transcription of neuronal growth and plasticity-related genes. Rab11 was also shown to be necessary to maintain TrkB phosphorylation, ERK1/2 activation, and CREB phosphorylation, all of which are required for BDNF to promote dendritic arborization. Interestingly, the expression of a Rab11 dominant-negative mutant significantly changes the transcriptional response to BDNF, decreasing the mRNA levels of proteins required for dendritic development, such as Arc [60]. Therefore, a possible explanation for the changes observed in dendritic arborization when manipulating c-Abl activity and expression could be related to changes in post-endocytic trafficking dynamics. When we evaluated total TrkB levels in the absence of c-Abl activity, we observed an upregulation of the receptor's levels, which could be explained by decreased endosomal trafficking into lysosomes. This observation prompted us to evaluate post-endocytic trafficking dynamics of TrkB in c-Abl neurons. Interestingly, we found that c-Abl-null neurons present a decreased retrograde speed in dendrites and a tendency towards a decrease in the location of TrkB in Lamp1 positive endosomes, and conversely, c-Abl overexpression increases retrograde trafficking in dendrites. Therefore, active c-Abl appears to promote retrograde trafficking of TrkB, which is a critical step in the process of BDNF-promoted dendritic arborization. A possible explanation for the increase in retrograde trafficking could be c-Abl-mediated regulation of molecular motors. While there are no reports of molecular motor-c-Abl interactions in neurons, the downstream target of c-Abl, Cdk5, has been shown to regulate both kinesin [61][62][63] and dynein activity in neurons [64,65]. Given the mixed microtubule polarity present in dendrites [66], a fine balance between the activation of both molecular motors is required for effective retrograde transport to take place. A hypothetical TrkB-c-Abl-Cdk5 signaling axis could regulate post-endosomal trafficking of the receptor to ensure effective BDNF-TrkB signaling.
It has been shown that BDNF induces dendrite branching and filopodial formation in developing neurons through the activation of actin binding proteins to promote actin polymerization [67,68] and Rho GTPase activity [69][70][71]. In this work, we showed that BDNF activates c-Abl to promote dendritic growth. In that context, it has been shown that c-Abl induces neurite growth modulating the dynamics of the actin cytoskeleton, either directly through its domain of interaction with F-actin [24] or indirectly through the inactivation of actin cytoskeletons regulatory proteins such as RhoA [29,31,72,73], Dok1 [74], WAVE [41,75], and Cdk5 [27]. Interestingly, Cdk5 can phosphorylate TrkB directly in response to BDNF, and the effects of this phosphorylation are relayed to the actin cytoskeleton through the attenuation of Cdc42 activity [16] and activation of Rac1 [76]. The activation of this signaling pathway is required for dendritic growth. It is entirely possible that c-Abl mediates actin cytoskeleton destabilization through RhoA inactivation while simultaneously regulating Rac1 and Cdc42 activity via Cdk5 phosphorylation. c-Abl could therefore be mediating tight temporal and spatial control of actin cytoskeleton dynamics to promote growth of the dendritic arbor.

Ethics Statement
All procedures were reviewed and approved by the Bioethics and Care of Laboratory Animals Committee of the Pontificia Universidad Católica de Chile (Protocol #150721002), which follows the local guidance documents generated by the National Research and Development Agency (ANID) and the Guide for the Care and Use of Laboratory Animals published by NIH of US Public Health Service.

Animals
Primary hippocampal and cortical neurons were prepared from rat or mouse embryos obtained from the institutional (CIBEM-UC) vivarium. TrkB F616A knock-in mutant mice were obtained from The Jackson Laboratory and bred in-house as homozygous mating pairs. This strain was generated as previously described [77]. The mice have C57BL/6 genetic background and did not display any gross physical or behavioral abnormalities. Homozygous c-Abl loxp /c-Abl loxp mice were kindly donated by Dr. Anthony J Koleske (Yale School of Medicine, US) and bred within our animal facility. c-Abl loxp /c-Abl loxp were bred with Nestin-Cre + mice, which were obtained from The Jackson Laboratory. This strain was originated and maintained on a mixed B6.129S4, C57BL/6 background and did not display any gross physical or behavioral abnormalities. The c-Abl null embryos (c-Abl loxp /c-Abl loxp , Nestin-Cre + , called from here on c-Abl-KO), their siblings c-Abl loxp /c-Abl loxp that express c-Abl (called from here on c-Abl-WT), and the TrkB F616A mice were housed at a 12/12 h light/dark cycle at 24 • C with ad libitum access to food and water. Genotyping was performed using PCR-based screening [42]. Primers:

Neuronal Treatments
To activate TrkB signaling, 7 DIV hippocampal neurons were deprived of B27 for one hour and then treated with 50 ng/mL BDNF for 5, 15, 30, and 60 min. In addition, DPH, an allosteric c-Abl activator, was used at a final concentration of 5 µM (Sigma Aldrich). To inhibit the TrkB receptor, hippocampal neurons were deprived of B27 and pretreated for 1 h with K252a (Tocris, Bristol, UK) at 0.2 µM, and then stimulated for 30 min with BDNF 50 ng/mL. In addition, 7 DIV hippocampal neurons from TrkB F616A knock-in embryos were used. TrkB F616A knock-in mice have a point mutation introduced into TrkB to convert phenylalanine to alanine at position 616 (F616A), which allows pharmacological and temporal inhibition of TrkB signaling via the highly membrane-permeable small molecule 1NM-PP1 [77]. The 1NM-PP1 inhibitor (0.5 µM, Cayman, Ann Arbor, MI, USA) was applied for 1h in neurobasal media without B27, and then neurons were stimulated for 30 min with BDNF (50 ng/mL). BDNF was maintained throughout all the treatments.
Dendritic arborization was induced in 7 DIV neurons by stimulation with 50 ng/mL of BDNF (Alomone, Jerusalem, Israel) for 48 h. Inhibitors for c-Abl, Imatinib, and GNF2 were used at a final concentration of 5 µM (Novartis, Basel, Switzerland) and for Trk receptors, K252a 0.2 µM (Tocris) was applied 1 h before applying BDNF and maintained throughout all the treatment. The c-Abl activator DPH was applied at a final concentration of 5 µM for 48 h.
For loss of function experiments, we used c-Abl null neurons or reduced c-Abl expression by transfecting neurons with short hairpin RNA plasmids targeting c-Abl (sh-c-Abl) or a control sh-scramble (c-Abl, sc-270357-SH, Santa Cruz Biotechnology, Santa Cruz, CA, USA). For transient transfections, hippocampal neurons were seeded at a density of 30 × 10 3 cells per well/in 12 mm coverslips and at 5 DIV neurons were transfected for 2 h in serum-free Opti-MEM using the Lipofectamine 2000 reagent (Life Technologies, Inc., Delhi, India) and treated at 7 DIV with BDNF (50 ng/mL).
To assess downstream TrkB signaling, hippocampal or cortical neurons were deprived of B27 and pretreated (or treated with a vehicle) with the following inhibitors at final concentrations of 1.5 or 10 µM: (a) a potent general PI3K inhibitor, (LY294002, Calbiochem), (b) a highly selective inhibitor of both MEK1 and MEK2 (U0126), and (c) a potent PLC-γ inhibitor (U73122). Then, neurons were stimulated with BDNF (50 ng/mL) for 30 min in the presence of the above-mentioned inhibitors.

Immunofluorescence
After treatment, neurons were rinsed twice with ice-cold PBS and fixed with 4% paraformaldehyde containing 4% sucrose in PBS for 20 min at room temperature. Later, cells were permeabilized for 10 min with 0.2% Triton X-100 in PBS and incubated in 3% bovine serum albumin in PBS (blocking solution) for 30 min at room temperature. Immunostaining was performed using primary specific antibodies by overnight incubation at 4 • C. After being washed 3 times in PBS, neurons were incubated with secondary antibodies (1:1000) for 1 h at room temperature. The cell coverslips were mounted in Dako mounting medium (CS703, Agilent Technologies, Santa Clara, CA, USA), visualized using a Nikon Eclipse Ti2 confocal microscope, and processed and quantified using ImageJ software (National Institutes of Health, Bethesda, MD, USA).

Dendritic Arborization Analysis
After treatments and staining of neurons using an anti-Map2 antibody (1:500) and secondary antibody Alexa-555 (1:1000), dendritic arborization was analyzed by Sholl analysis as previously described [36]. The visualization was performed by confocal microscopy using a Nikon Eclipse C2 confocal microscope connected to a computer with NIS-Elements C software. Images were acquired using a 60× objective at 1024 × 1024 pixel resolution, and 5-7 optical slices were captured along the z-axis every 0.5 µm. Z-stacks were integrated, and the images were segmented to obtain binary images. Ten concentric circles with increasing diameters (5 µm each step) were traced around the cell body, and the number of intersections between dendrites and circles was counted and plotted for each diameter. Analysis was performed using the ImageJ plugin developed by the Anirvan Gosh Laboratory (http://biology.ucsd.edu/labs/ghosh/software (accessed on 3 March 2014)). The number of total primary dendrites and branching points of all dendrites were manually counted from the segmented images.

Western Blotting
Seven DIV neurons, plated at a density of 10 6 cells/cm 2 , were incubated in B27-free neurobasal medium for 1 h and then treated with 50 ng/mL of BDNF (Alomone). After washing, neurons were lysed in RIPA buffer (50 mM Tris, 150 mM NaCl, 1 mM EGTA, 1 mM EDTA, 0.5% sodium deoxycholate, 1% NP-40, and 0.1% SDS) and supplemented with protease inhibitors cocktail (Roche) and phosphatase inhibitors cocktail (Roche, Basel, Switzerland). The homogenates were cleared by centrifugation at 14.000 rpm for 10 min. Protein extracts (50-70 µg) were loaded onto 10% SDS-PAGE and transferred to nitrocellulose membranes (Fisher Thermo Scientific). The membranes were blocked with 3% BSA in PBS for 1 h at room temperature, followed by overnight incubation at 4 • C with primary antibodies. After incubation with the appropriate HRP-conjugated secondary antibody (1:3000, Thermo Scientific), membranes were incubated with ECL substrate for detection of HRP enzyme activity (Thermo Fisher Scientific) and visualized in a Syngene gel documentation system. Images were quantified by ImageJ analysis (National Institutes of Health). 2-4 biological replicates were performed for every experiment. The number of repetitions for every experiment is stated in the corresponding figure description. The Student's t-test was used for statistical significance assessment.

Live-Cell Imaging of mCherry-TrkB
Six DIV c-Abl-KO and c-Abl-WT mouse neurons were transfected with 0.8 µg of mCherry-TrkB plasmid (the cDNA encoding mCherry-TrkB was kindly provided by Dr. Chengbiao Wu of the University of California, San Diego [UCSD]) for 2 h using Lipofectamine 2000 in B27-free Opti-MEM medium. During imaging, cultures were kept in Tyrode media (124 mM NaCl, 5 mM KCl, 2 mM CaCl 2 , 1 mM MgCl 2 , 30 mM D-glucose and 25 mM HEPES, pH 7.4) at 37 • C. Live-cell imaging was performed on a Nikon Eclipse C2 confocal microscope equipped with a live-cell temperature controller (LCI cu-501) and digital camera connected to a computer NIS-Elements C software. Images of a single neuron transfected with mCherry-TrkB were acquired using a 60× objective at intervals of 7 s for 3 min to establish the basal level of distribution and dynamics. TrkB-m-Cherry signal mobility was examined using the Manual Tracking ImageJ plug-in (National Institutes of Health, Bethesda, MD, USA).

Immunoendocytosis of Flag-TrkB
Hippocampal neurons (5 DIV) were transfected with 0.5 µg of Flag-TrkB plasmid (gift from Prof. Francis Lee, NYU, USA). After 48 h, hippocampal neurons were incubated at 4 • C for 10 min and then treated with an anti-Flag mouse antibody (1:750) for 20 min. Neurons were washed briefly with warm neurobasal medium and incubated with BDNF (50 ng/mL) for 30 min. Then, neurons were fixed with 4% paraformaldehyde containing 4% sucrose in PBS at room temperature and incubated with a donkey anti-mouse IgG conjugated to Alexa555 (1:1000) without permeabilization. Finally, samples were blocked and permeabilized as described above and immunostained with a donkey anti-mouse IgG-Alexa488 (1:500).

Surface Biotinylation Assay
Hippocampal neurons were pre-treated with GNF2 or Imatinib 5 µM for 1h and then treated with BDNF (50 ng/mL) for 30 min. Neurons were then transferred to 4 • C, rinsed with chilled PBS, and incubated with 0.8 mg/mL biotin (EZ-link ® Sulfo-NHS-LC Biotin, Thermo Scientific, Waltham, MA, USA) for 30 min. Subsequently, biotin was quenched with 50 mM NH 4 Cl for 10 min, cells were rinsed two times with PBS, lysed in RIPA buffer (150 mM NaCl, 1% Nonidet p-40, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris, pH 8) and protease (Complete Mini Protease Inhibitor Cocktail; Roche) and phosphatase inhibitors (PhosSTOP Phosphatase Inhibitor Cocktail; Roche), and centrifuged for 1 min, 14,000 rpm at 4 • C. Each lysate was incubated overnight with 30 µL of NeutrAvidin-coupled agarose beads (NeutrAvidin Agarose Resins; Thermo Scientific). Beads were washed with ice-cold lysis buffer, and then biotinylated proteins were eluted with 2 × SDS sample buffer. Cell-surface and total protein lysates were subjected to SDS-PAGE and western blot analysis.

Statistical Analyses
All data are presented as the mean ± Standard Error of the Mean (SEM). Sholl's analysis curves were compared with two-way repeated measures ANOVA, followed by Bonferroni's multiple comparisons. Statistical analyses were performed using a Student's t-test or one-way ANOVA followed by appropriate multiple comparisons test depending on the number of groups used in each experiment. Details of the specific tests used, level of significance, and number of replicates are indicated in each figure legend. Statistical analyses were performed using GraphPad Prism 8 (Scientific Software). The significance level was p < 0.05 for all treatments.