TIAM-1 differentially regulates dendritic and axonal microtubule organization in patterning neuronal development through its multiple domains

Axon and dendrite development require the cooperation of actin and microtubule cytoskeletons. Microtubules form a well-organized network to direct polarized trafficking and support neuronal processes formation with distinct actin structures. However, it is largely unknown how cytoskeleton regulators differentially regulate microtubule organization in axon and dendrite development. Here, we characterize the role of actin regulators in axon and dendrite development and show that the RacGEF TIAM-1 regulates dendritic patterns through its N-terminal domains and suppresses axon growth through its C-terminal domains. TIAM-1 maintains plus-end-out microtubule orientation in posterior dendrites and prevents the accumulation of microtubules in the axon. In somatodendritic regions, TIAM-1 interacts with UNC-119 and stabilizes the organization between actin filaments and microtubules. UNC-119 is required for TIAM-1 to control axon growth, and its expression levels determine axon length. Taken together, TIAM-1 regulates neuronal microtubule organization and patterns axon and dendrite development respectively through its different domains.


Author summary
In neurons, microtubules form an acentrosomal organization to support polarized trafficking and development of neuronal processes. While dendritic branches and the axon grow differentially, it is largely unknown how microtubules organize with actin architectures to promote distinct axon and dendrite development. In this work, we show that a multi-domain RacGEF protein TIAM-1 controls dendrite patterning by its N-terminal domains and suppresses axon development by its C-terminal GEF domains. TIAM-1 is required for maintaining microtubule polarity in dendritic regions and suppressing accumulation of axonal cargos and microtubules. In somatodendritic regions, TIAM-1 binds UNC-119 and maintains the organization between actin and microtubule. TIAM-1 prevents axonal distribution of UNC-119 and thus controls axon development. This work

Introduction
In neurons, actin filaments and microtubules (MTs) are fundamental frameworks for movement, subcellular compartmentalization, and outgrowth of neurites such as dendrites and axons [1]. Axonal outgrowth is influenced by many layers of regulation, including the production of membrane and cytoplasmic components such as vesicles and cytoskeletons, the transport of these membrane components and cytoskeletons, and the dynamics of the growth cone responding to local chemical cues and adhesion choices [2,3]. In the growth cone, actin filaments form filopodia and lamellipodia in the peripheral region surrounding the central microtubule bundles [4][5][6]. Actin and microtubule dynamics as well as the interplay between these two cytoskeletons are essential for growth cone behaviors [1,[7][8][9]. The regulation of MT dynamics and stability drive early neurite formation and influence neuronal polarity [8,10]. Polarized MT configuration is the basis for motor-mediated trafficking of axon and dendrite specific cargos [11]. As main tracks for motors, microtubules orient uniformly in plus-end-out direction in axons and often in minus-end-out direction in dendrites of invertebrates but exhibit mixed polarity in vertebrate dendrites [12]. Other than orientation, MTs are diversified by different compositions of tubulin isotypes and posttranslational modifications such as tyrosination, polyglutamylation, or acetylation which affect the interaction between motors and MTs [13]. However, how these different MTs assemble into a well-organized network in neurons remains largely unknown.
MT-interacting proteins and motors maintain MT organization in neurons [14,15]. The minus-end-directed motor dynein keeps uniform MT polarity in axons by removing minusend-out MTs in Drosophila [16,17]. The plus-end-directed motor Kinesin-1 mediates the transport of dendritic growth cone microtubule-organizing center (MTOC) and regulates dendritic MT orientation in C. elegans [18,19], while other kinesin motors like mitotic Kinesin-6 and -12 as well as Kinesin-2 also regulate MT orientation in dendrites [20][21][22]. A MT-interacting protein TRIM46 is required to construct the uniform axonal MT bundles in mammalian neurons [23]. Recently, we find that MT bundles are organized by UNC-33/CRMP to form a high-order spindle-like network in the cell body with extensions to axons and dendrites in worm neurons for efficient mitochondrial transport [24]. UNC-33 interacts with UNC-119 and UNC-44/ankyrin to form a membrane-associated complex that anchors MTs to the cortex [25].
In this study, we characterize the roles of actin regulators in the polarized development of axon and dendrite, including ACT-3 (Actin), PFN-1 (Profilin), UNC-60 (Cofilin/ADF), FHOD-1 (Formin), WASP, WAVE, MIG-2 (RhoG), and TIAM-1 (RacGEF). We find that UNC-60 (Cofilin/ADF) dedicatedly suppresses axon growth, and WAVE specifically promotes dendritic arbor growth but not axon growth. Among these regulators, interestingly, TIAM-1 supports dendrite growth while suppresses axon growth. In previous studies, TIAM-1 was reported to regulate neuronal protrusion and dendrite arborization [44,45]. Although TIAM-1 is a GEF factor, its GEF activity is not required for normal dendritic development [46]. Instead, a recent study showed the GEF activity of TIAM-1 is required for dendritic regeneration after laser ablation [47]. Here, we further show that TIAM-1 patterns neuronal development through its different domains: N-terminal EVH1 domain, middle PDZ domain, C-terminal GEF domain. Its N-terminal EVH1 domain regulates high-order dendritic branches; its middle PDZ domain is required for primary dendrite development; its C-terminal GEF domain controls axon outgrowth. TIAM-1 interacts with UNC-119 and regulates neuronal actin and microtubule organization. TIAM-1 restricts axonal distribution of UNC-119, which is necessary for controlling axonal outgrowth.
However, mutants of wsp-1 (WASP) and pfn-1 (Profilin) showed no significant change in axonal and dendritic growth (S1 Fig). The fhod-1 (Formin) mutant had a mild defect (10% decrease) in dendritic growth (S1 Fig). In contrast, the unc-60 (Cofilin/ADF) mutant (su158) displayed axonal overgrowth (with 29% increase) but no increase in total dendrite length ( Fig  1E and 1H-1I). Interestingly, wve-1 (WAVE) mutant (ok3308) showed 32% decrease of total dendritic length but no obvious change in axonal growth (Fig 1F and 1H-1I). Therefore, while actin supports both axonal and dendritic processes, specific actin regulators could preferentially promote axonal growth or dendritic arbor development. Different from dedicated axonal or dendritic regulators, we found that the RacGEF TIAM-1 regulates dendritic development and axonal growth in an opposite manner; the tiam-1 (ok772) mutant showed significant axonal outgrowth with about a two-fold increase in length but severely defective dendritic development (Fig 1G-1I). This interesting bimodal phenotype suggests that TIAM-1 likely regulates axonal and dendritic development through different mechanisms.

Different regions of TIAM-1 regulate the development of distinct neuronal subcompartments
Based on sequence analysis of cDNA clones, tiam-1 has three transcripts, tiam-1a, tiam-1b, and tiam-1c [44] (Fig 2A). We edited C7 codon of tiam-1a into a stop codon to generate the ntu22 mutant allele by CRISPR-Cas9. Furthermore, we generated the ntu14 mutant allele by editing both the C7 codon of tiam-1a and the W7 codon of tiam-1b into stop codons (Fig 2A). ntu22 showed a reduction of high-order dendritic branches but no defects in either axonal growth or primary dendrite development (S2A and S2G-S2I Fig), indicating the role of tiam-1a in regulating high-order dendrite development. In the ntu14 mutant, not only high-order dendrite formation was severely disrupted, but also the length of primary dendrite reduced, suggesting that tiam-1b is further required for primary dendrite development (S2B and S2G-S2I Fig). Since the mutant phenotype of ntu14 has no significant difference to the putative null allele ok772 that has a deletion that disrupts all three transcripts [44], tiam-1a and tiam-1b could be the main transcripts responsible for PVD development.
To study the functions of different domains, we generated extrachromosomal arrays of three transgenes that express TIAM-1A, TIAM-1B, or TIAM-1C, respectively and assayed their activities in ok772 mutant (Fig 2A). The full-length form TIAM-1A contains an N-terminal EVH1 domain, a middle region containing a PDZ domain, and a C-terminal DHPH GEF domain. Compared to TIAM-1A, the TIAM-1B lacks the N-terminal EVH1 domain, and TIAM-1C further loses a part of the middle region, containing a truncated PDZ domain and a DHPH GEF domain at the C-terminal region (Fig 2A). While TIAM-1A expression effectively restored both axonal and dendritic growth defects, TIAM-1B expression did not rescue distal high-order dendrite morphogenesis defects, indicating the function of the EVH1 domain in high-order dendrite development (Fig 2B-2C and 2F-2H). In contrast, TIAM-1C expression did not rescue dendritic growth defects but significantly suppressed the axonal overgrowth defect (Fig 2D and 2F-2H). These results suggested that each region of TIAM-1 may be responsible for the development of different neuronal subcompartments; the EVH1 domain is required for the development of high-order dendrites, the PDZ-domain-containing middle region is necessary for primary dendrite development (comparing TIAM-1B with TIAM-1C), and the C-terminal region with the DHPH domain is required for axonal growth (Fig 2I). To test whether axonal growth and dendritic development are respectively controlled by different regions of TIAM-1, we expressed the C-terminal truncated form TIAM-1N539 in ok772. Indeed, the expression of TIAM-1N539 restored dendritic development but failed to rescue the axonal outgrowth defect (Fig 2E-2H).
We further introduced different isoforms into tiam-1 mutant by the single-copy insertion technique that integrates transgene into the same locus of the chromosome to ensure consistent expression of isoforms [53]. These single-copy transgenic strains showed a stronger but similar rescuing effect to extrachromosomal arrays (S2C- S2I Fig). Notably, while TIAM-1N539 and TIAM-1C restored dendrite and axon development respectively, TIAM-1A was significantly more effective, suggesting that although different domains of TIAM-1 are required for regulating different sub-neuronal compartments, they perform better when they work together.

TIAM-1 regulates polarized microtubule organization
Since polarized cytoskeleton organization and transport system are required for axonal and dendritic development [11], we examined whether TIAM-1 regulates the specific development of cytoskeleton structures to affect the polarized deployment of vesicular organelles such as synaptic vesicles and mitochondria in axonal and dendritic differentiation. We labeled Given that synaptic vesicles and mitochondria are cargos of kinesin motors, we examined the distribution of the plus-end motor UNC-116 (Kinesin-1) and UNC-104 (Kinesin-3), which are required for the transport of mitochondria and synaptic vesicles, respectively [24,[56][57][58]. In PVD neurons, the anterior dendrite possesses MTs in minus-end-out orientation, while MTs are plus-end-out in the posterior dendrite and the axon [59]. These plus-end motors UNC-116 and UNC-104 localized to the axonal tips and the terminals of posterior dendritic branches of wild type (arrowheads in Figs 3A and S3J). In tiam-1(ok772), their signals were severely diminished in the posterior dendrite but accumulated in the axon (Figs 3B-3C and S3J-S3L). These results indicated that the tiam-1(ok772) mutation disrupts the plus-end motor-mediated transport system in the dendrite but enhances it in the axon.
We next examined the microtubule polarity by tracing the movement of the microtubule plus-tip marker EBP-2::GFP. We found that the loss of TIAM-1 has no effect on the direction of EBP-2 comets in the anterior primary dendrite and axon, but the EBP-2 puncta running into the axon significantly increased, suggesting that there are more growing microtubules in the axon in tiam-1(ok772) mutant (Fig 3E-3H). Interestingly, in tiam-1(ok772) mutant, most of the EBP-2 comets ran in a reversed direction compared to wild type in the posterior dendrite ( Fig 3D-3E, 3I and S1 and S2 Videos), suggesting that microtubule orientation is reversed in the posterior dendrite of tiam-1(ok772) mutant. Taken together, these results support the idea that TIAM-1 is required for organizing polarized microtubules to direct plus-end motors and their cargos.

TIAM-1 colocalizes with MT and actin bundles
Proper MT organization requires interacting proteins to tether and stabilize MTs to the cell membrane or cortical actin filaments [60,61]. In order to understand how TIAM-1 regulate MT organization, we constructed and expressed a GFP-tagged full-length TIAM-1A to examine its subcellular localization. We found that TIAM-1A::GFP specifically localizes to the somatodendritic compartment with punctate patterns in the primary and high-order dendrites but does not localize to the axon ( Fig 4A). We also labeled TIAM-1B and TIAM-1C as well as TIAM-1N539 with GFP to examine their localization in tiam-1 mutant PVD neuron. Different from the punctate TIAM-1A::GFP, TIAM-1B::GFP was diffuse and mainly localized in primary dendrites, the cell body, and the axon (S4A and S4B Fig different transgenes. One-way ANOVA. Error bar represents SEM. n = 10. n.s., not significant, � P<0.05, �� P<0.01, ��� P<0.001, compared to wild type or tiam-1(ok772). (I) Proposed model for the function of different TIAM-1 domains in PVD neuron development: While the C-terminal DHPH domain is required for the axonal growth (the pink region), the middle PDZ domain is necessary for the growth of primary and proximal high-order dendrites (the green region). The N-terminal EVH1 domain is essential for the growth of distal high-order dendrites (the blue region). EVH1 and PDZ domains potentially enhance the C-terminal domain activity in suppressing axonal growth, and similarly the C-terminal domain potentially enhances the activity of EVH1 and PDZ domains in promoting dendritic development (grey arrows).
A recent study has shown that UNC-119 is required for anchoring MT to the cell cortex [25]. We further examined the relative localization pattern of TIAM-1 and UNC-119. Interestingly, we found UNC-119 was highly colocalized with TIAM-1A::GFP in dendritic branches (Fig 4F and 4G). According to these observations, it is likely that TIAM-1 is associated with UNC-119.

TIAM-1 is essential for the organization of actin filaments and MTs
In wild-type PVD neurons, MTs labeled by RFP::TBA-1 were enriched in primary dendrites and the axon with a distribution pattern not overlapping with actin filaments labeled by GFP:: UtrCH, which mainly distributed in high-order dendrites (Fig 5A). In tiam-1 mutant, the distribution of MTs were reduced in primary dendrites and strongly enhanced in the axon ( Fig  5B and 5C). Strikingly, actin filaments highly accumulated in the primary dendrites and the axon in tiam-1 mutant (Figs 5B, 5D and S6A). These ectopic actin filaments abnormally intermingled with MTs in primary dendrites (Fig 5B'). In contrast, wild-type cortical actin filaments surround MTs in primary dendrites (Fig 5A'). In wild-type cell body, MTs form a regular organization between the nucleus and cortical actin filaments (Fig 5A"), but, in tiam-1 mutant, MTs detached from the cortex with irregular turns and circles (Figs 5B" and S6B). Taken together, these results suggested that TIAM-1 is required for the proper distribution of actin filaments and the organization of the cytoskeletons in PVD neuron.

TIAM-1 regulates axon development through UNC-119
Given that TIAM-1 colocalized and interacted with UNC-119, we hypothesized that TIAM-1 regulates axonal/dendritic polarized development by localizing UNC-119 in dendrites and preventing it from axon mislocalization. We tested if the elevation of the UNC-119 level is sufficient to promote axonal growth by expressing unc-119 transgene and the results were positive (Fig 7A, 7B and 7G). We further genetically ablated unc-119 in tiam-1 mutant to test whether UNC-119 is required for the dysregulated axonal outgrowth by using the putative null mutant allele ed3 [64]. In the tiam-1; unc-119 double mutant, the ablation of unc-119 effectively suppressed axonal length, compared to tiam-1 single mutant (Fig 7C-7E and 7G). Regarding dendrite development, tiam-1; unc-119 double mutation did not enhance the growth defects caused by tiam-1 single mutation, suggesting they act in the same pathway in dendrite development (Fig 7C-7E and 7F). To sum up, these results showed that UNC-119 levels determine axon length and are required for TIAM-1 to regulate axonal growth.

Discussion
Neuronal cytoskeletons are the basis for polarity establishment and the development of distinct axonal and dendritic structures to form functional synaptic connections. How cytoskeleton regulators promote polarized actin and MT organization in neuronal cells remains to be elucidated. Here, we examined the role of several actin regulators in axonal/dendritic development and identified UNC-60/Cofilin as a specific axonal regulator and WVE-1/WAVE as a specific dendritic regulator. Among these regulators, we find that the RacGEF TIAM-1 regulates both axonal and dendritic development but in opposite ways; it suppresses axonal growth but promotes dendritic formation. Our functional analysis shows that TIAM-1 directs distinct developmental patterns of neuronal subcompartments by its protein domains. While the Nterminal EVH1 domain of TIAM-1 promotes distal dendritic development, its middle PDZ domain is essential for primary dendrite formation. And, surprisingly, its C-terminal region is responsible for regulating axon development but not dendrite development. We further find TIAM-1 interacts with the cortical anchor complex component UNC-119 and colocalizes with actin filaments and MTs. Based on our findings, we propose a working model that TIAM-1 regulates MT and actin organization to direct axonal and dendritic development (Fig 7H). UNC-119 is essential for TIAM-1 to regulate axonal outgrowth.  -1(ok772) (B). In wild-type PVD, F-actin was mainly distributed in high-order dendritic branches, while MT was mainly distributed in the primary dendrites (white brackets) and the axon (yellow brackets). In tiam-1(ok772), F-actin and MT were abnormally distributed in the primary

Different TIAM-1 protein domains direct distinct axonal and dendritic development
In previous studies, TIAM-1 has been shown to regulate the development of dendrite and dendritic spine by its GEF activity in mice hippocampal neurons and worm PVD neurons [45,65]. But, Bülow group recently shows TIAM-1 GEF activity is not required in dendritic morphogenesis since a TIAM-1 mutation (T548F) that abolishes its GEF activity has no effect on dendrite development although the role of GEF domain in neuronal development remains unclear [46]. In this work, we surprisingly find the GEF domain-containing C-terminal fragment TIAM-1C regulates axon growth but not dendrite development. We further find the N-terminal fragment TIAM-1N539 without the GEF domain can restore a portion of dendrite development but not able to regulate axon development. These results suggest that the C-terminal GEF domain controls axon development, while EVH1 and PDZ domains regulate dendrite development. Interestingly, our results further indicate that EVH1 domain directs the development of distal highorder dendritic branches, and PDZ domain supports primary dendrite growth, respectively. Therefore, here we report that TIAM-1 is a multifunctional protein that directs distinct developmental patterns of neuronal subcompartments through its diverse functional domains.

tiam-1 isoforms play different roles in neuronal morphogenesis
In this study, we find that tiam-1 isoforms have different physiological roles in neuronal morphogenesis. Similarly, there are two isoforms of tiam-1 mammalian homolog Tiam2, Tiam2l (full-length form) and Tiam2s (short form), in mice and humans, functioning differently [66,67]. While Tiam2l suppresses t-haploid transmission, Tiam2s expression strongly enhances the transmission ratio of t-allele, suggesting that tiam-1 isoforms act differently in spermatogenesis [66]. Moreover, the mammalian Tiam1 has four isoforms that are detected in the cerebral cortex and hippocampus and their expressions are increased in Down syndrome mouse model, while the function of individual isoforms in the nervous system is unknown [68]. In neuronal development, we show that the full-length tiam-1a is responsible for highorder dendrite formation, while tiam-1b is required in primary dendrite formation. Since TIAM-1B lacks the EVH1 domain, which is often present in actin cytoskeleton regulators like Ena/VASP and WASP [62], it is possible that TIAM-1A utilizes this region to promote actin organization in distal dendrites. Actin cytoskeletons appear to interact with MT bundles differently in primary dendrites, where actin filaments surround MT bundles under the cell membrane, in comparison with actin-rich distal dendritic processes emanated from MTs (Fig 5A). We propose that the loss of the EVH1 domain makes TIAM-1B relocate to primary dendritic region and change its way to interact with cytoskeletons and promote primary dendrite formation (S4B Fig).

TIAM-1 organizes actin and MT cytoskeletons for axonal/dendritic polarized development
Although the MT cytoskeleton supports neuronal morphogenesis and serves as tracks for directional trafficking of cargos to promote dendritic and axonal differentiation, the molecular mechanisms that regulate distinct axonal/dendritic MT organizations remain largely unknown [10,15]. In neuronal cells, different actin structures interact and coordinate with MTs through crosslinking proteins and complexes [1]. It is likely that some crosslinking mechanisms regulate MT organization and coordinate dynamics of actin structures for axon or dendrite development and positioning organelles. Indeed, UNC-119 interacts with UNC-33 and UNC-44 to form a complex that anchors MTs to cell cortex [25]. Also, we recently showed that UNC-33 assembles a high-order MT organization for axonal transport [24].
In this work, we show that TIAM-1 is essential to maintain plus-end-out MTs in posterior dendrites and regulates axon specific cargos including synaptic vesicles and mitochondria. Notably, we find that TIAM-1 puncta localize with MTs and actin bundles and also interact with UNC-119. Based on these findings, it is possible that TIAM-1 stabilizes dendritic plusend-out MTs through the cortical anchor complex in the posterior dendrite. Consistent to this model, dendritic microtubule orientation is indeed disrupted in cortical anchor complex mutants, including unc-33, unc-44, and unc-119 [25,59]. In addition, since the minus-end binding protein PTRN-1/CAMSAP proteins stabilize MTs and regulate neuronal polarity [69][70][71][72][73], it is interesting to investigate the potential regulation between TIAM-1 and PTRN-1 in future studies.

TIAM-1 regulates axon development via UNC-119
We find the distribution of cortical anchor complex subunits UNC-119 and UNC-33 shifts to the axon by ablating tiam-1 (Fig 6A-6F). UNC-119 has been reported to be enriched in nervous system in invertebrates and vertebrates [64,74] and participate in axonal development [74,75]. Here we show that the loss of UNC-119 leads to shorter axons while overexpressing UNC-119 induces the extension of the axon, suggesting that UNC-119 is necessary and sufficient to promote axonal growth. It is possible that the increase of axonal UNC-119 recruits cortical anchor complex and stabilizes the tight interaction between cortical actin rings and MT arrays in axons. Consequently, the more MTs are stabilized as tracks, the more motors and cargos are recruited into the axons, and thus it facilitates axonal growth. Since TIAM-1 specifically distributes in dendrites and strongly interacts with UNC-119, the interaction between TIAM-1 and UNC-119 may stabilize the cortical anchor complex in dendrite and maintain the balance between dendritic and axonal development.

Strains
Worms were raised on NGM plates seeded with Escherichia coli OP50 at 20˚C. The Bristol N2 strain was used as wild type and all transgenic strains created for this study are shown in S1 Table. PVD neuron in wild type (D) and tiam-1(ok772) (E). Brackets indicate axon and asterisks indicate cell body. Scale bar, 10 μm.

CRISPR/Cas9 mediated genome editing
For introducing nonsense mutations into endogenous C7 loci of tiam-1a(ntu22), the CRISPR/ Cas9 mediated genome editing method will be utilized [78]. Briefly, Cas9 protein (10 μg/μl), tracrRNA (4 μg/μl), crRNA_rol-6 (4 μg/μl), crRNA_ tiam-1a (4 μg/μl) were mixed and incubated for 10 mins at 37˚C. Then the mixture was added with the repair DNA templates (1 μg/ μl) and injected into ntuIs1 (OCY295 strain). Rollers from the F1 generation were selected. F2 worms were lysed and the putative mutation sites in tiam-1 locus were sequenced. To simultaneously edit C7 codon of tiam-1a and W7 codon of tiam-1b into stop codons to generate tiam-1(ntu14), crRNA_ tiam-1b and the repair DNA templates for tiam-1b were used. All reagents were ordered from Integrated DNA Technologies (IDT). The crRNA sequences and repair oligo sequences are shown in S1 Table. Immunoprecipitation and immunoblotting To express TIAM-1 N-terminal domain and UNC-119 for Co-IP assays, Human Embryonic Kidney Cells 293 (HEK293T) were cultured with a standard protocol in DMEM supplemented with 10% FBS. 2 x 10 5 cells/well (in a 6-well dish) were transfected with the expression plasmid containing 1.5 μg TIAM-1N539 and 1 μg 3xFLAG-UNC-119-MYC using HyFectin transfection reagent kit. Two days after transfection, cells were lysed in 250 μl of lysis buffer (50mM Tris-Base, 150mM NaCl, 1% IGEPAL CA-630, pH 7.4, protease inhibitor cocktail (MedChem-Express)), and the lysates were then sonicated for 10 minutes for fully lysis. The resulting lysates were centrifuged at 16,200 ×g for 10 minutes at 4˚C, and the supernatants were incubated with 20 μl of anti-FLAG M2 Affinity Gel (Sigma-Aldrich) at 4˚C with agitation overnight. After washing three times with the lysis buffer, the samples were eluted with SDS loading buffer and were boiled for 3 min. SDS-PAGE and western blot were then performed using standard protocols. Anti-HA (Proteintech) and anti-FLAG M2 (Sigma-Aldrich) mouse antibodies were used at 1:2000 dilutions and 1:1000 dilutions, respectively, and HRP-conjugated goat antibodies to mouse were used at 1:10,000 dilutions (Proteintech). Details for plasmids used in co-IP assays can be found in S1 Table. Image acquisition For quantification, fluorescence images of live C. elegans were acquired using Zeiss Axio Imager microscope (Carl Zeiss, Germany) with X-cite light source and 63x/1.4NA oil objective and Evolve 512 EMCCD camera (photometrics) at 23˚C. The z stacks were collected and maximum intensity projections were used for further analysis. For each worm, 4-5 images were stitched together to cover the whole body (Adobe Photoshop). Worm images were straightened and measured by ImageJ. Worms at mid-L4 stage were immobilized with 2-3 μl of 5 mM levamisole (Sigma-Aldrich) on 3% agarose pads.
For confocal images in Figs 1-3 and 7 and S1-S3, worms were immobilized with 25mM levamisole on 8% agarose pads and imaged by a Zeiss LSM710 microscope (Carl Zeiss, Germany) with 63x/1.4NA oil objective with 488 nm and 561 nm laser lines.
For confocal images of Airyscan mode in Figs 4-6 and S4 and S5, worms were immobilized with 50mM levamisole on 8% agarose pads and imaged by Zeiss LSM880 confocal microscope (Airyscan SR mode).
Confocal images were acquired as Z-stack serials, and several images were taken and stitched together to cover the PVD neuron. Images were processed by Zen software to project maximum intensity projections for figures and further quantification.

Quantification and statistical analysis
Neuronal morphology analysis. For relative axonal or dendritic length, images of 10 worms for each genotype at mid-L4 stage were scored for their total dendritic length and axonal length as well as body length (from mouth to anus) by ImageJ. The dendritic or axonal length was divided by body length to get the relative dendritic or axonal length (Figs 1, 2, 7, S1 and S2).
Quantification of mitochondria and synaptic vesicles. For the distribution of mitochondria and the number of synaptic vesicles, mitochondria and synaptic vesicle clusters in PVDs of 10 worms for each genotype at mid-L4 stage were scored. Numbers of synaptic vesicle clusters, total mitochondrial numbers in all neurites, percentages of axonal mitochondrial (axonal mitochondrial number/ total neurite mitochondrial number), the density of synaptic vesicle (synaptic vesicle number/ axonal length), and density of mitochondria in axon (mitochondrial number/ axonal length) were quantified (S3 Fig). Quantification of motor fluorescence signals. For quantification of UNC-116::GFP and UNC-104::GFP, the GFP signals in the posterior dendrites and axons of 10 worms were circled as regions of interest (ROIs) by ImageJ to acquire total fluorenscence. Additional ROIs with the same size were measured as background autofluorescence inside the worm. Axonal and dendritic GFP intensity values were acquired by subtracting the background fluorescence from total fluorescence. The percentages of axonal UNC-116::GFP or UNC-104::GFP were calculated by the formula: axonal GFP intensity/ (dendritic + axonal GFP intensity) ×100% (Figs 3 and S3).
Quantification of colocalization. For the colocalization analysis between TIAM-1A::GFP (or GFP) and RFP::TBA-1 (or RFP::UtrCH), 16-18 z-stack Airyscan confocal images of the cell body region (~6 μM in z axis) were analyzed by Imaris 9.9.0. The intensity threshold that best represents the signal level for both GFP and RFP channels were set in the "Coloc" tool, and the percentages of GFP intensity of voxels colocalized with RFP were calculated (Fig 4B-4E).
Quantification of cytoskeletons. For quantification of RFP::TBA-1 and GFP::UtrCH, the intensity of both fluorescence reporters was quantified by using ImageJ. The segmented line tool was used to trace RFP::TBA-1 and GFP::UtrCH signals in the whole neuron including primary dendrites, high-ordered branches, and axons to create ROIs and the line width is 10 pixels (= 1.6 μm). To measure background from worm body, additional ROIs of the equal size areas were measured as background autofluorescence in the neighboring region inside the worm avoiding neurites. The RFP::TBA-1 and GFP::UtrCH signals in three neuronal compartments (primary dendrite, high-ordered branches, and axon) were acquired by subtracting background intensity. Ratios of RFP::TBA-1 and GFP::UtrCH in three compartments were calculated (signal in a sub-neuronal compartment was divided by the sum of three neuronal compartments) and analyzed (Fig 5C and 5D).
Quantification of UNC-119 and UNC-33. For the quantification of UNC-119::mCherry and UNC-33::GFP, the intensity of both fluorescence reporters was quantified using ZEN3.4 (blue edition). The Draw Spline Contour tool was used to draw outlines of neurites to create ROIs. The signals in the axon and the 60 μm of the anterior primary dendrite (start from the cell body) were measured. To measure background levels, additional ROIs of equal areas were defined in a neighboring region avoiding neurites inside the worm. The UNC-119::mCherry and UNC-33::GFP intensity were subtracted by background signals and divided by the area of ROI to calculate average intensity. The average intensity of axonal UNC-119::mCherry and UNC-33::GFP were divided by dendritic signals to generate the axon/dendrite ratio of UNC-119::mCherry and UNC-33::GFP (Fig 6).
Quantification of EBP-2 comet direction. EBP-2::GFP was recorded at the rate (1 frame per second) for 2 minutes to further generate Kymographs by ImageJ with the "Multi Kymograph" plugin. To quantify EBP-2::GFP comet direction, the proximal primary dendrite region within 50 μm from cell body was recorded. For scoring axonal EBP-2::GFP comet events, the comets that ran into the axon within 15 μm from cell body were counted (Fig 3).
Analysis of the distance from the TIAM-1A::GFP to the cell surface. To analyze the distance between TIAM-1A::GFP puncta and the cell surface, the 3D surface reconstruction of TIAM-1A::GFP and the cell surface were built by Imaris 9.9.0. The 'Surfaces' function was used to build the TIAM-1 puncta and the cell surface according to the TIAM-1A::GFP and RFP::UtrCH signals, respectively. After building the 3D surface reconstruction, the 'short distance' in Statistics tools was used and the distance of each TIAM-1 puncta to cell surface was quantified to generate S5 Fig. One-way ANOVA, Student's t-test, or Fisher's exact test were performed for statistics. Diagrams were made by GraphPad Prism 7.0 software. Resources: Shih-Peng Chan.