Bimodal Control of Dendritic and Axonal Growth by the Dual Leucine Zipper Kinase Pathway

The dual leucine zipper kinase pathway acts on two separate transcriptional programs to dichotomously control dendritic and axonal growth.


Introduction
The separation of the dendritic and axonal compartments in neurons is prerequisite to the function of neural circuits. Although the difference between dendrites and axons is a cornerstone of modern neuroscience, as theorized in the ''neuron doctrine'' by Ramon y Cajal [1], our molecular understanding of how neuronal compartmentalization is achieved remains limited. This knowledge, however, is crucial for understanding the assembly of neural circuits. Moreover, it is needed to develop strategies that will correct defective dendrites or axons with subcellular precision, and to alter the wiring of neural circuits in animal models in order to interrogate the functions of the nervous system.
Previous studies have demonstrated the existence of regulators dedicated to dendrite or axon growth in the same neuron, referred to as ''dedicated mechanisms'' herein. For instance, the transcription complex, p300-SnoN, specifically promotes axon growth in the cerebellar granule neurons [2]. In contrast, transcription factor NeuroD is dedicated to dendritic growth in mammalian cerebellar granule neurons [3]. Likewise, bone morphogenetic protein 7 (BMP7) specifically promotes dendritic growth in several types of neurons in culture [4,5]. In Drosophila, the transcription factor Dendritic arbor reduction 1 (Dar1) promotes dendritic, but not axonal, growth [6]. In addition, dendritic and axonal growth exhibit differences in their dependence on the secretory pathway [7].
Besides the dedicated mechanisms, another way to differentiate dendritic and axonal growth is through bimodal regulators that control dendritic and axonal growth in opposite directions [8][9][10][11]. Different from the dedicated mechanisms, the bimodal mechanisms may coordinate growth of the two neuronal compartments. However, how the function of a molecule or signaling pathway diverges into distinct dendritic and axonal regulations is poorly known.
In this study we report that the dual leucine zipper kinase (DLK) signaling pathway is a novel bimodal regulator for dendritic and axonal growth in vivo. The core players in the DLK signaling pathway are the DLK and the Pam/Highwire/RPM-1 (PHR) family of E3 ubiquitin ligases that suppress DLK expression. The PHR-DLK signaling module plays an important role in axon development, as demonstrated by studies in C. elegans [12][13][14], Drosophila [15][16][17], zebrafish [18,19], and mammals [20][21][22]. Loss of the Drosophila homologue of DLK-1, Wallenda (Wnd), suppresses the axonal overgrowth caused by loss of the PHR protein Highwire (Hiw) [15,17]. Consistently, overexpression of Wnd promotes axonal growth of motoneurons in Drosophila larvae [17]. In Drosophila adult mushroom body neurons, Hiw-Wnd pathway is required for the segregation of axon branches in response to guidance cues [23]. In addition to the roles in axon development, recent studies have discovered a conserved function of the DLK pathway in axon regeneration [24][25][26][27][28] and degeneration in several species [29][30][31][32]. Although these exciting findings have established critical roles for the DLK pathway in axon development, regeneration, and degeneration, whether the DLK pathway regulates dendrites remains unknown.
Here we show that the DLK pathway directs the growth of axons and dendrites in opposite directions in the class IV dendritic arborization (C4da) neurons in Drosophila. By inhibiting Wnd functions, Hiw restricts axonal growth but promotes dendritic growth. The opposite effects of the Hiw-Wnd pathway on axons and dendrites are achieved through two distinct transcription factors: Fos, which mediates the regulation of axonal growth, and Knot (Kn), which mediates the regulation of dendritic growth. Collectively, these results demonstrate that a single signaling pathway can differentiate dendritic and axonal growth through two independent transcriptional programs.

hiw Plays a Dichotomous Role in Differentiating Dendrite and Axon Growth
All functional studies of the PHR-DLK pathway in neurons have so far focused on axons. We first set out to determine whether the PHR gene hiw is involved in dendrite development using Drosophila as a model system.
The C4da neurons in Drosophila larva are a well-established in vivo system for studying the molecular mechanisms of dendrite and axon development. The dendrites and axons of these neurons are distinguishable from each other at both molecular and organelle levels in a way that resembles mammalian neurons [33]. Moreover, these neurons are amenable to single-cell genetic manipulations [33,34], which is important for comparing dendritic and axonal development in vivo. In each hemi-segment of a larva, there are three C4da neurons (ddaC, v'ada, and vdaB), whose cell bodies are located respectively in the dorsal, lateral, and ventral parts of the body wall. The axons of the three C4da neurons extend to the ventral nerve cord (VNC) where the terminals form a ladder structure ( Figure S1A). At single-cell resolution, the axon terminal of each C4da neuron consists of an anterior projection that extends within one segment length. ddaC and vdaB neurons also extend a contra-lateral branch and sometimes a posterior branch ( Figure S1A9) [35]. Collectively, the axon terminals of the three C4da neurons form a fascicle that connects two adjacent neuropils ( Figure S1A9).
To examine the role of hiw in dendritic development, we labeled the C4da neurons in hiw mutant larvae using a C4daspecific marker, ppk-CD4::tdTomato [34,36]. We found that dendritic growth was dramatically reduced in the null allele hiw DN and, to a lesser extent, in the hypomorphic hiw ND8 mutants ( Figure 1A and B). Both total length and number of termini of dendrites were significantly reduced in hiw DN and hiw ND8 mutants ( Figure 1B).
Consistent with the known function of hiw in suppressing axonal growth [15,16], hiw mutations led to exuberant growth of axon terminals in C4da neurons. In hiw mutant larvae, thickened connective fascicles were observed in the C4da neuropil ladder ( Figure S1B). In wild-type larvae, there was no hemi-segment that contained more than three longitudinal connectives between the axon entry point of abdominal segment 5 (A5) and that of A6 ( Figure S1A9,B, and D). In contrast, 100% of hiw mutant C4da neuropils exhibited more than three connectives ( Figure S1B and D), which could either arise from an increased number of axon branches from neurons in the same segment or from overextended axons that normally remain in other segments.
Our further analysis showed that the effects of hiw mutations on dendritic and axonal growth are not a result of defective dendrite and axon identities. The axon-specific marker, Kinesin-b-galactosidase [37,38], remained exclusively localized to the axons of C4da neurons that were mutant for hiw ( Figure S2A). Furthermore, the initial growth and pathfinding of axons to the VNC or the extension of minor dendritic processes remained unaltered in embryos devoid of both maternal and zygotic hiw ( Figure S2B). Thus, hiw appears to be dispensable for early development, including the initial specification of axon and dendrite. Taken together, these results suggest that hiw plays a dichotomous role in differentiating dendrite and axon growth after their identities have been specified.

Hiw Regulates Dendritic and Axonal Growth in a Cell-Autonomous Manner
Previous studies of axon development have discovered both cell-autonomous [16,17] and non-cell-autonomous roles of hiw [23]. To determine whether hiw functions cell-autonomously in C4da neurons and to examine the axon and dendrite defects at single-neuron resolution, we generated hiw mutant neurons with the Mosaic Analysis with a Repressible Cell Marker (MARCM) technique [39]. Consistent with the reduced dendritic growth in hiw mutant larvae, we observed a reduction of high-order dendritic branches in hiw loss-of-function mutant neurons ( Figure 1C). Moreover, fewer dendritic branches arrived at the segment border as compared to wild-type. hiw mutations caused a 43% reduction in total dendrite length and 40% reduction of the number of dendrite termini ( Figure 1E). In contrast to their dendritic defects, hiw mutations resulted in a 2.4-fold increase of axon terminal length ( Figure 1D and F) as compared to wild-type. The axon terminals of hiw mutant neurons typically spanned multiple segments, whereas the vast majority of wild-type C4da neurons extended axonal branches between their own segments and the anterior neighboring segments ( Figure 1D). Noticeably, although the axon terminals of hiw mutant neurons grew exuberantly, they preserved normal guidance within the C4da neuropil tracts.

Author Summary
Dendrites and axons are the input and output compartments of a neuron, respectively. Understanding how dendrites and axons are separated during neuronal development may help in developing strategies to correct defective neurons in neurological disorders and injuries. We show here that an evolutionarily conserved molecular pathway dichotomously controls dendritic and axonal growth. A key molecule in this pathway, dual leucine zipper kinase (DLK), suppresses dendritic growth but promotes axonal growth. While DLK is known to be a key regulator of axon growth and regeneration, this study reveals its roles in dendritic growth for the first time. In addition, we find that the DLK pathway diverges through two separate downstream programs that control the expression of other genes. These insights can help target this pathway to specifically promote axon regeneration without affecting dendritic growth. Overall, these results help provide a new framework for understanding neuronal compartmentalization and the diversity of neuronal morphology.
In agreement with the MARCM results, overexpressing Hiw in C4da neurons rescued both dendritic and axonal defects in hiw mutant larvae to a level comparable to wild-type ( Figure S3), further confirming that the loss of hiw in C4da neurons is responsible for the dendritic and axonal defects. Overexpression of Hiw alone did not significantly alter axonal or dendritic growth ( Figure S3), suggesting that hiw is necessary but insufficient to instruct dendritic growth and restrict axon growth. Taken together, these results demonstrate that Hiw functions as a cell-intrinsic bimodal regulator of dendritic and axonal growth in C4da neurons. Wnd Mediates the Functions of Hiw on Both Axonal and Dendritic Growth Two parallel downstream pathways are known to mediate axon overgrowth induced by loss of PHR proteins. First, the PHR orthologs in C. elegans (rpm-1) and Drosophila (hiw) suppresses the worm dlk-1 and the fly DLK wallenda (wnd), respectively, to restrain axonal growth in motoneurons [14,17]. Second, the worm rpm-1 regulates a trafficking pathway that consists of the Rab guanine nucleotide exchange factor (GEF) GLO-4 and the Rab GTPase GLO-1, which restrict axon extension in mechanosensory neurons and synaptogenesis in motoneurons [40]. In order to delineate the mechanism underlying the bimodal control of dendritic and axonal growth by hiw, we tested the involvement of these two pathways in axon and dendrite growth in C4da neurons. While wnd loss-of-function mutations on their own did not alter axonal ( Figure Figure 2). Hence, increased expression of the Wnd kinase is sufficient to inhibit dendritic growth and promote axonal growth.
We also examined the potential involvement of the Rab trafficking pathway by testing Drosophila homologs of glo-4 and glo-1 in axon and dendrite development in C4da neurons. In C. elegans, glo-4 mutants exhibited axon overextension similar to that in rpm-1 mutants [40]. Overexpressing the Rab GTPase Glo-1, which is activated by Glo-4, partially rescued axon termination defects in rpm-1 mutants [40]. The Drosophila homologs of glo-4 and glo-1 are claret (ca) and lightoid (ltd), respectively [41]. The ca mutant MARCM clones devoid of maternal contribution exhibited axons and dendrites that were indistinguishable from wild-type clones ( Figure S4A-D). In addition, overexpressing Ltd failed to rescue either axon or dendrite defects in hiw mutants ( Figure S4E-H). These results suggest that Drosophila C4da neurons use the DLK (Wnd) pathway, rather than the Ca-Ltd vesicle trafficking pathway, to mediate hiw function in axonal and dendritic growth.

The Fos Transcription Factor Mediates the Hiw-Wnd Control of Axonal Growth
How might the Hiw-Wnd pathway control axonal and dendritic growth differently in the same neurons? In Drosophila motoneurons, the Hiw-Wnd pathway requires the transcription factor Fos [17]. Fos is phosphorylated by Bsk (JNK) [42], which positions it as the downstream kinase of the Wnd-Hep7-JNK kinase cascade [17]. Overexpressing a dominant negative form of Fos partially suppresses axonal overgrowth at the NMJ of hiw mutants [17]. Because of this, we decided to examine whether Fos is required by Wnd to promote axonal growth in C4da neurons.
To test the role of Fos with loss-of-function mutants, and to bypass lethality caused by fos null mutations kay 1 [43,44], we generated kay 1 MARCM clones in the presence or absence of a UAS-Wnd transgene that overexpresses Wnd (OE Wnd). kay 1 alone did not alter axonal growth ( Figure 3A), but completely suppressed the axon overextension caused by Wnd overexpression ( Figure 3A and C), which suggests that fos is required for Wnd-induced axonal overgrowth. In contrast to the axonal role of Fos, kay 1 did not block the dendritic reduction caused by Wnd overexpression. The total dendritic length of MARCM clones that overexpressed Wnd in the kay 1 background (OE Wnd+kay 1 ) was indistinguishable from that of Wnd-overexpressing clones ( Figure 3B, B9, and C), and the number of dendrite termini was further reduced from that of Wndoverexpressing clones. Interestingly, the kay 1 mutation alone caused a mild reduction in dendritic length and branch number ( Figure 3B, B9, and C). This result suggests that, although Fos does not mediate the dendritic functions of the DLK pathway, it plays a minor role in supporting dendritic growth. Taken together, these results suggest that Wnd acts through Fos to specifically promote axonal growth.

Wnd Suppresses the Expression of the Transcription Factor Knot
In order to understand how the function of DLK pathway diverges into dendritic and axonal regulations, we hypothesized that the divergence occurred at the transcriptional level, and therefore tested the transcription factors that are known to regulate dendritic growth in da neurons. Among them, the Krüppel-like factor Dar1, the homeodomain transcription factor Cut (Ct), and zinc-finger transcription factor Knot (Kn, as known as Collier) have been shown to be essential for dendritic growth in C4da neurons. Loss-of-function mutations in each of these transcription factors severely reduce dendritic growth in C4da neurons [6,[45][46][47][48]. We first tested whether expression levels of these transcription factors in C4da neuron nucleus were altered in hiw loss-of-function mutants. No significant difference in the levels of Dar1 [6] or Cut [45] was observed between wild-type and hiw mutant C4da neurons ( Figure S5A-C). In contrast, the nuclear levels of Kn, which belongs to the evolutionarily conserved Collier/Olf1/EBF (COE) family, were significantly reduced in both hiw mutant neurons and Wnd-overexpressing neurons ( Figure 4A and B) Kn is required for the expression of the ENaC ion channel Pickpocket (Ppk) in C4da neurons [46][47][48]. Kn loss-of-function mutations reduce ppk transcription [46] and suppress ppk promoter activity as assayed with a ppk-eGFP transgene ( Figure 4D) [47,48]. Furthermore, misexpression of Kn induces ectopic ppk-eGFP expression in neuron types that do not normally express ppk-eGFP [46][47][48]. Therefore, the ppk-eGFP transgenes may be used as readout for Kn transcriptional activity. Consistent with the reduced Kn expression by hiw mutations or Wnd overexpression, we found a 37% reduction in ppk-eGFP fluorescence intensity in the soma of hiw mutant C4da neurons and a 68% reduction in those of Wnd-overexpressing neurons ( Figure 4C and D). Furthermore, overexpressing Kn rescued the reduced expression of ppk-eGFP in hiw mutant or Wndoverexpressing neurons ( Figure 4C and D). The correlation between ppk-eGFP fluorescence intensity and Kn levels suggests that the Hiw-Wnd pathway controls Kn transcriptional activity by regulating its protein levels. Nevertheless, it does not rule out the possibility of posttranslational regulation of Kn activity by Hiw-Wnd. Taken together, Hiw suppresses Wnd function, thus maintaining high levels of Kn protein in C4da neurons, which is required for dendritic growth.

Knot Mediates the Hiw-Wnd Control of Dendritic Growth
It has been demonstrated that loss-of-function mutations of kn cause reduction in dendritic length and branch numbers [46][47][48]. We tested potential genetic interactions between hiw and kn in controlling dendritic growth. C4da dendrites developed normally in both hiw DN/+ heterozygous and kn KN4/+ heterozygous larvae ( Figure 5A and B), in which Kn expression and ppk-eGFP levels remained comparable to wild-type ( Figure S5D-F). In contrast, the hiw DN/+ ; kn KN4/+ transheterozygous larvae exhibited dramatically reduced dendritic growth ( Figure 5A-B), revealing a strong genetic interaction between hiw and kn.
We investigated the nature of the genetic interaction by epistasis analysis. Kn overexpression resulted in a mild 16% reduction of C4da dendritic length ( Figure 5C-D), possibly due to destabilized microtubules as a result of increased expression of the microtubule severing protein Spastin [6,48]. Nevertheless, overexpressing Kn in hiw DN MARCM clones (hiw DN +OE Kn) rescued dendritic defects from 45% of reduction to 25% in dendritic length, and from 44% of reduction to 29% in dendrite termini number, as compared to wild-type ( Figures 5C-D), suggesting that Kn acts downstream of Hiw to control dendrite growth.
In contrast, Kn overexpression had no effect on axonal growth in either wild-type or hiw mutant MARCM clones ( Figure 5E-F). Taken together, our results suggest that the Hiw-Wnd pathway acts through Kn to regulate dendritic, but not axonal, growth.  There are four classes of dendritic arborization (da) neurons in Drosophila larva, which are categorized based on the complexity of dendritic branching [49]. Hiw mutations elevated the expression of puc-lacZ [50], a reporter for Wnd activity [26], in all four classes ( Figure S6A and B), suggesting that the Hiw-Wnd pathway is functional in all these neurons. However, Kn is only expressed in the class IV, and undetectable in other classes of da neurons [46][47][48]. If hiw acted via Kn to control dendritic growth, hiw mutations would not alter the dendritic morphology in class I (C1), class II (C2), and class III (C3) da neurons. Indeed, we observed that hiw mutant MARCM clones of C1-C3 da neurons all exhibited normal dendritic growth (Figures S7C and D, S8C and D, S9C and D), even though Hiw still restricts axonal growth in these neurons ( Figures S7A and B, S8A and B, S9A and B). These observations further suggest that the Hiw-Wnd pathway regulates dendritic growth in Kn-expressing neurons.
We next determined whether Kn expression endows neurons with the ability to respond to dendritic growth control by Wnd. Consistent with previous reports that ectopic expression of Kn in class I da (C1da) neurons leads to excessive dendritic branching and extension [47,48], the total dendrite length was increased by 55% and the number of dendritic branches was doubled in the C1da neurons overexpressing Kn (OE Kn) compared to wild-type. Such dendritic overgrowth was considerably reduced when Wnd was overexpressed in the same neurons ( Figure 6A-B), with the increase in total dendrite length inhibited from 55% to 10%. As a control, a kinase-dead form of Wnd failed to suppress Kn-induced dendritic overgrowth.
Similar to the effects in C4da neurons ( Figure 4A-B), we detected a reduction of the nuclear Kn levels in C1da neurons expressing both Kn and Wnd ( Figure 6C-D). It is noteworthy that, in these C1da neurons, Kn was expressed by the Gal4/UAS system, which bypasses endogenous transcriptional control. Thus, up-regulated Wnd kinase is likely to suppress Kn expression via posttranscriptional mechanism. Collectively, these results suggest that Hiw-Wnd pathway regulates dendritic growth in Knexpressing neurons by controlling the expression of Kn.

Discussion
In this study, we found that a single signaling pathway, consisting of the PHR E3 ubiquitin ligase Hiw and its downstream dual leucine kinase Wnd, serves not only as a negative regulator in axon growth but also as a positive regulator in dendrite growth in vivo. This is the first report, to our knowledge, to show a role for the DLK pathway in dendrite development. We further discovered that the functional divergence of this pathway is achieved through two transcription factors, Kn and Fos, which mediate the dendritic and axonal regulation, respectively.

Three Distinct Modes of Regulations of Axonal and Dendritic Growth
Taking into account the current study with previous studies, three distinct modes of axonal and dendritic growth regulation have been identified: shared, dedicated, and bimodal ( Figure 7A).
Dedicated mechanisms provide the basis for specifically regulating the morphogenesis of only axons or only dendrites. Molecular controls at work in dedicated mechanisms can be divided into (1) axon-dedicated mechanisms, including p300 and SnoN transcription complex [2]; and (2) dendrite-dedicated mechanisms, including transcriptional factors NeuroD [3] and Dar1 [6], growth factor BMP7 [4,5], and small GTPase Rab17 [55]. Manipulation of dedicated mechanisms leads to specific changes in the growth of either axons or dendrites, but not both. Thus, axonal growth per se does not regulate dendritic growth, and vice versa.
In contrast to dedicated mechanisms, bimodal mechanisms oppositely regulate axons and dendrites, and may serve to coordinate the growth of these separate compartments. Previous studies of different types of neuronal cultures have discovered three bimodal regulators: Sema3A [8,9], CLASP2 [10], and Rit [11]. In this study we have identified an in vivo bimodal regulatory mechanism that involves DLK kinase. The bimodal action of the DLK signaling pathway is achieved through two ''dedicated'' transcriptional programs. These two programs are likely to be independent because manipulating their activities rescues either dendritic or axonal defects, but not both, in hiw mutants. We also observed that transgenic Hiw and Wnd were present in the axon terminals in addition to the cell body but not in dendrites ( Figure  S6C and D), raising the intriguing possibility that elevated Wnd function in the axon terminals might impact transcriptional activities in the cell body, and consequently influence denritic growth.
It is likely that various bimodal controls exist in different neuron types. Moreover, it is possible that these bimodal controls intersect with each other. For instance, since the actions of Sema3A are mediated through cGMP/cAMP levels [9], another bimodal regulator might also influence cGMP/cAMP levels. It will be interesting to determine whether cGMP/cAMP are involved in PHR-DLK pathway for bimodal control of dendritic and axonal growth.

The DLK Pathway May Coordinate Dendritic and Axonal Growth After Axon Injury
Despite the requirement of DLK functions in axonal growth after axon injury [24][25][26][27][28], DLK is dispensable for axonal growth during development in the neuron types examined so far [14,17]. Consistently, we find that loss of dlk/wnd does not alter either dendritic or axonal growth in Drosophila C4da neurons. Rather, the overabundance of DLK/Wnd caused by defective PHR/Hiw functions leads to axonal overgrowth as well as dendritic reduction. Since axon injury leads to an overabundance of DLK/Wnd function [26,28], it is conceivable that the elevated activity of DLK/Wnd induced by axon injury not only promotes axon regeneration [24][25][26][27][28] but also restrains dendritic growth or prunes exiting dendritic branches to compensate for the increased demand of membrane or cytoskeleton supplies for axonal growth. This notion is consistent with previous studies that show dendrite retraction following axotomy in Drosophila da neurons [56] and mammalian cultured neurons [57,58].

Two Transcription Programs Directed by Kn and Fos Endow Bimodal Regulation of PHR-DLK Pathway
Although it is known that the zinc finger transcription factor Kn is essential for dendritic growth, the signaling mechanism that regulates Kn in neurons is unknown. In this study, we show that Kn specifically mediates dendritic regulation by the PHR-DLK pathway, which is supported by three lines of evidence. First, kn   . Regulatory mechanisms underlying dendritic and axonal growth. (A) Three distinct mechanisms regulating dendritic and axonal growth. Shared mechanisms control dendrite and axon co-growth. Dedicated mechanisms direct compartment-specific growth. Bimodal mechanisms differentially regulate dendritic and axonal growth. (B) A model that postulates the differential control of dendritic and axonal growth by the DLK pathway, which is based on the present study. In this model, DLK plays a dual role in neuron morphogenesis. Up-regulated DLK, caused either by PHR mutations or DLK overactivation, promotes the growth of axon terminals but restricts that of high-order dendritic branches. Such a dichotomous function is the result of signaling divergence into two transcriptional programs that are each dedicated to either dendritic or axonal growth. Fos serves a permissive role in the axonal regulation by DLK, whereas Kn specifically mediates the dendritic regulation by DLK. doi:10.1371/journal.pbio.1001572.g007 genetically interacts with hiw and functions downstream of hiw and wnd to regulate dendritic growth. Second, the Hiw-Wnd pathway regulates Kn expression in C4da neurons. Third, the Kn expression pattern is consistent with the presence of the Hiw-Wnd regulation of dendrite growth. Kn is selectively expressed in a subset of neurons [46][47][48]59]. Consistent with Kn expression pattern, hiw mutations caused dendrite defects only in the Knexpressing class IV neurons, and not in the other classes of da neurons that lack Kn. Interestingly, ectopic expression of Kn in class I neurons, which do not normally express Kn, is sufficient to endow the Hiw-Wnd regulation. These results strongly suggest that the PHR-DLK pathway regulates Kn to control dendrite development.
In contrast to Kn, the transcription factor Fos specifically mediates axonal regulation through Hiw-Wnd pathway. We found a two-fold role for fos in neuronal development. On the one hand, eliminating fos specifically causes dendritic reduction without affecting axon terminal length in C4da neurons. This indicates that endogenous Fos is specifically required for dendritic growth during normal development. On the other hand, the requirement of fos could switch to be axonal when augmented Wnd activity leads to exuberant axonal growth.
In summary, the Hiw-Wnd pathway can exert bimodal or dedicated control over dendritic and axonal growth, depending on the presence of the transcription factors that mediate its subcellular compartment-specific functions. If transcription factors for both dendritic and axonal growth are present, Hiw-Wnd signaling functions as a bimodal modulator ( Figure 7B). This model provides guidance for further investigation of the molecular basis of the diversity of neuronal morphology and the differential development of dendrites and axons.
Confocal imaging was performed with a Leica SP5 confocal system. Only da neurons from abdominal segment 4 to 6 were imaged for quantification of dendrites and axons to ensure consistency.
To compare protein expression levels in C4da neurons, larvae of different genotypes in the same experimental group were processed simultaneously. The same setting for image acquisition was applied to the same experimental group and signal saturation was minimized. Fluorescence intensities of different genotypes were normalized to wild-type (Figures 4 and S5) or the OE Wnd KD control group ( Figure 6).

Quantifications and Statistical Analysis
To quantify protein levels, mean fluorescence intensity of the region of interest in each channel was measured with NIH ImageJ software. For axon terminal and dendritic morphology, manual tracing was conducted with Neurolucida software. Branches shorter than 5 mm were excluded. For consistency, da neurons located between segment A4 and A6 from size-matched third instar larvae were imaged and analyzed in all experiments.
In all of the bar charts of quantification, the numbers in the bars indicate the sample numbers. Values and error bars indicate mean 6 SEM. Two-tailed unpaired student t-test was used. p values were indicated as: not significant (NS) p.0.05, * p,0.05, ** p,0.01, *** p,0.001.