Abelson, enabled, and p120catenin exert distinct effects on dendritic morphogenesis in Drosophila

Authors

  • Wenjun Li,

    1. Gladstone Institute of Neurological Disease, San Francisco, California
    2. Department of Neurology, University of California, San Francisco, California
    Current affiliation:
    1. Institute for Nutritional Sciences, Chinese Academy of Sciences, Shanghai 200031, China
    Search for more papers by this author
  • Yan Li,

    1. Gladstone Institute of Neurological Disease, San Francisco, California
    2. Department of Neurology, University of California, San Francisco, California
    Search for more papers by this author
  • Fen-Biao Gao

    Corresponding author
    1. Gladstone Institute of Neurological Disease, San Francisco, California
    2. Department of Neurology, University of California, San Francisco, California
    • Gladstone Institute of Neurological Disease, 1650 Owens St., San Francisco, CA 94158
    Search for more papers by this author

Abstract

Neurons exhibit diverse dendritic branching patterns that are important for their function. However, the signaling pathways that control the formation of different dendritic structures remain largely unknown. To address this issue in vivo, we use the peripheral nervous system (PNS) of Drosophila as a model system. Through both loss-of-function and gain-of-function analyses in vivo, we show here that the nonreceptor tyrosine kinase Abelson (Abl), an important regulator of cytoskeleton dynamics, inhibits dendritic branching of dendritic arborization (DA) sensory neurons in Drosophila. Enabled (Ena), a substrate for Abl, promotes the formation of both dendritic branches and actin-rich spine-like protrusions of DA neurons, an effect opposite to that of Abl. In contrast, p120catenin (p120ctn) primarily enhances the development of spine-like protrusions. These results suggest that Ena is a key regulator of dendritic branching and that different regulators of the actin cytoskeleton exert distinct effects on dendritic morphogenesis. Developmental Dynamics 234:512–522, 2005. © 2005 Wiley-Liss, Inc.

INTRODUCTION

The dendritic branching pattern is an important indicator of neuronal identity and is critical for the proper function of the nervous system (Ramón y Cajal, 1995; Masland, 2004). How are the shapes of different neurons specified during development and what are the major signaling pathways involved in this important process remain largely unknown (reviewed in McAllister, 2000; Cline, 2001; Scott and Luo, 2001; Wong and Ghosh, 2002; Gao and Bogert, 2003; Jan and Jan, 2003; Miller and Kaplan, 2003).

The nonreceptor tyrosine kinase Abelson (Abl) regulates cell adhesion and actin cytoskeleton dynamics, both of which are important in neuronal morphogenesis (Lanier and Gertler, 2000; Hernández et al., 2004). In Drosophila, Abl is highly localized in the axons of the central nervous system (CNS) (Gertler et al., 1989) and is expressed in many other cell types (Bennett and Hoffmann, 1992). Abl interacts genetically with Disabled (Gertler et al., 1989), Profilin (Wills et al., 1999a), Capulet (Capt), a homolog of the adenylyl cyclase-associated protein (Wills et al., 2002), and the microtubule plus-end tracking protein Orbit/MAST/CLASP (Lee et al., 2004) in controlling axonal growth in the embryonic CNS of Drosophila. Abl modulates a number of membrane receptors. Abl genetically interacts with Notch to produce defects in axonal extension, and Disabled directly binds to Notch in vitro (Giniger, 1998). Abl associates with the cytoplasmic domains of Dlar, a neuronal receptor protein tyrosine phosphatase (Wills et al., 1999b) and of Roundabout (Robo), a key molecule in repulsive axon guidance (Bashaw et al., 2000). Abl also plays an important role in regulating epithelial morphogenesis in Drosophila (Grevengoed et al., 2001).

A key molecule in the Abl pathway is Ena, which was identified as a dosage-sensitive suppressor of abl mutations (Gertler et al., 1990). Ena binds directly to Abl and functions as a substrate for the Abl kinase (Gertler et al., 1995); its in vivo function is regulated by phosphorylation through Abl (Comer et al., 1998). Ena directly affects axonal growth (Gertler et al., 1995) and its function is opposite to that of Abl (Wills et al., 1999b; Bashaw et al., 2000). Ena regulates the formation of filopodia on axonal growth cones (Lebrand et al., 2004) and directly modulates actin polymerization (Grevengoed et al., 2003; Krause et al., 2003). How Ena and Abl affect dendritic branching is much less studied.

To study the role of the actin cytoskeleton regulators in dendritic development in vivo, we use the peripheral nervous system (PNS) of Drosophila embryos/larvae as a model system. Each dorsal cluster consists of 12 sensory neurons including four external sensory (ES) neurons and eight multidendritic (MD) neurons. A subset of MD neurons, dendritic arborization (DA) neurons, elaborate highly branched dendritic trees just underneath the epidermis (Bodmer and Jan, 1987) that can be labeled with green fluorescent protein (GFP) using the UAS-Gal4 system (Brand and Perrimon, 1993), allowing visualization in living Drosophila embryos or larvae (Gao et al., 1999, 2000). Each MD neuron, including those in the dorsal cluster, exhibits a characteristic pattern of dendritic arborization (Bodmer and Jan, 1987; Sweeney et al., 2002; Grueber et al., 2002). This model system has been used to identify novel mutations that affect dendritic morphogenesis (Gao et al., 1999) and for dissecting genetic pathways involved in this important process (reviewed in Gao and Bogert, 2003).

Here, we show that spine-like protrusions on some DA neurons are actin-rich but contain no or little microtubules, in contrast to dendritic branches that contain both actin filaments and microtubules. We also report the identification and characterization of mutations in ena that lead to alterations in dendritic branching in Drosophila. We demonstrate that Ena and Abl regulate dendritic branching of DA sensory neurons in opposite ways. Unexpectedly, ena interacted genetically with p120catenin (p120ctn), a regulator of cell adhesion that binds to the cytoplasmic tail of all classic cadherins (Peifer and Yap, 2003). We also demonstrate that the primary function of p120ctn is to enhance the formation of spine-like protrusions on a subset of DA neurons. These results suggest differential control of various dendritic structures by a network of actin cytoskeleton regulators.

RESULTS

Morphological and Molecular Differences Between DA Neuron Dendritic Branches and Spine-Like Protrusions in Drosophila

In the dorsal cluster of the Drosophila PNS, different DA sensory neurons cover different dendritic fields with their unique branching patterns (Bodmer and Jan, 1987; Sweeney et al., 2002). For instance, ddaA neurons in the dorsal cluster (Fig. 1A) and v'pda neurons in the ventral cluster (not shown) exhibit numerous spine-like protrusions (Sweeney et al., 2002), while ddaE (Fig. 1B) and ddaF (not shown) neurons in the dorsal cluster and vpda neurons in the ventral cluster (Fig. 1C) extend many long, unbranched smooth dendrites. The spine-like protrusions are morphologically unique fine dendritic structures: they are short, thin processes extending from dendritic shafts with a more or less right angle (Fig. 1D). The spine-like protrusions on DA neurons do not contain detectable microtubules as shown by the absence of Tau-GFP labeling (Fig. 1E) or by the absence of microtubule-associated protein Futsch (data not shown). Interestingly, these protrusions are highly enriched with actin (Fig. 1F), just like mammalian spines. On the contrary, labeling by actin-GFP decreases in higher-order dendritic branches (data not shown). These observations indicate that spine-like protrusions on some DA neurons are unique actin-rich fine dendritic structures.

Figure 1.

Morphological and molecular differences between dendritic branches and spine-like protrusions. a: A ddaA neuron with numerous spine-like protrusions on dendritic shafts. b: A vpda neuron that elaborates smooth dendritic branches. c: A ddaE neuron that belongs to the same class as vpda neurons. d: Spine-like protrusions of ddaA neurons are labeled by membrane-bound mCD8-GFP (arrows). e: A dendritic branch of a ddaA neuron is labeled by microtubule-associated Tau-GFP. Note that the spine-like protrusions are not labeled by Tau-GFP. f: Spine-like protrusions of ddaA neurons are highly enriched with actin as labeled by actin-GFP. Bar = 50 μm for A–C and 15 μm for D–F.

Ena Promotes the Formation of Both Dendritic Branches and Spine-Like Protrusions

Using DA neurons as a model system, a genetic screen was carried out that led to the identification of 12 loci that control different aspects of dendritic morphogenesis (Gao et al., 1999). One of the genes is ena, which encodes a protein belonging to an evolutionarily conserved family of actin regulatory proteins that bind to profilin and barbed ends of actin filaments and prevent filament capping (Krause et al., 2003). In ena46 mutant embryos at late stage, fewer lateral branches extend from the dorsal dendrites (Gao et al., 1999; Fig. 2A). To further analyze this phenotype, we examined ena46 homozygous mutant embryos at different stages. At 16 hr AEL, the extension of dorsal dendrites toward the dorsal midline appears to be more or less normal; however, mutants had significantly fewer dynamic transient lateral fine branches, suggesting a defect in the initiation of new branches (Fig. 2A). At 18–20 hr AEL stage, the number of lateral dendrites that extend toward or reach the adjacent segment boundaries is reduced in ena46 mutant embryos compared with wild-type embryos (Fig. 2A), indicating a major defect in dendritic branching. A similar phenotype was observed in ena46/Df(2R)P34 mutants and was indistinguishable from ena46 homozygous mutants, suggesting that ena46 is a genetically null allele (data not shown).

Figure 2.

Abnormal dendritic development of DA sensory neurons in ena mutant embryos. A: Dendritic arborization of dorsal cluster DA neurons was imaged at 16, 18, and 20 hr after egg laying (AEL). DA neurons were labeled by GFP under the control of Gal4 109(2)80. Scale bars = 20 μm. B: Genomic organization of the ena locus. A single nucleotide mutation (C to T) generated a stop codon in the coding region (black boxes). Boxes indicate exons. C: Western blot analysis indicates the absence of Ena protein in mutant embryos. The arrowhead indicates the location of Ena protein band on the blot. D: Ena expression in the nervous system as shown by immunostaining of a wild-type embryo. E: Ena expression is absent in an ena46 mutant embryo. F: A GFP-labeled DA sensory neuron. G: Subcellular localization of Ena in the same neuron in E as shown by immunostaining. White arrows indicate dendrites, and blue arrowheads indicate axons.

The ena gene contains 11 exons (Fig. 2B). To characterize the mutation in ena46 mutants, we cloned and sequenced the genomic DNA corresponding to the ena locus and found a point mutation in exon 7 (C to T) that changed the glutamine 489 to a stop codon. After separation of wild-type lysates on a polyacrylamide gel, an Ena-specific monoclonal antibody (5G2) recognized a single band with an apparent molecular mass of 90 kDa (Fig. 2C). This band is absent in lysates of ena46 mutant embryos, consistent with the presence of the point mutation described above. After a long exposure of the Western blot, a faint band near 50 kDa could be seen, suggesting that the Ena fragment generated by the point mutation is highly unstable and was only present at a very low level, or the ena mRNA becomes less stable. The absence of Ena in mutant embryos was also confirmed by immunostaining (Fig. 2D,E). Ena is localized in dendrites, cell bodies, and axons of DA neurons (Fig. 2F,G), consistent with its roles in axon guidance (Wills et al., 1999; Bashaw et al., 2000) and dendritic morphogenesis (Gao et al., 1999; Fig. 2).

To study specific cell-autonomous functions of Ena in controlling dendritic branching of DA sensory neurons, we used the MARCM technique (Lee and Luo, 1999) to generate mutant single neuron clones in larval PNS as described (Sweeney et al., 2002). In the dorsal cluster, six of the eight MD neurons, named as dendritic arborization (DA) neurons, elaborate highly branched dendritic trees. Individual DA neurons (ddaA, ddaB, ddaC, ddaD, ddaE, and ddaF) were named according to their relative positions along the dorsal-ventral and anterio-posterial axis, with the most dorsal one as ddaF and the most ventral one as ddaA (Merritt and Whitington, 1995; Sweeney et al., 2002). Each DA neuron exhibits a distinctive dendritic branching pattern and a specific dendritic field, presumably reflecting their diverse physiological functions (Sweeney et al., 2002).

DdaC neurons elaborate highly branched dendritic trees that cover a large area from the anterior to the posterior segment boundaries (Fig. 3A). Loss of ena decreased the number of dendritic ends of ddaC neurons (424.2 ± 21.0, n = 17 versus 616.6 ± 44.4, n = 13, P = 0.0011) (Fig. 3A–C). Unlike other dorsal cluster DA neurons, ddaA neurons exhibit numerous spine-like protrusions on their major dendritic shafts (Sweeney et al., 2002; Fig. 3D). MARCM analysis indicated that loss of ena activity in this assay system decreased the number of spine-like protrusions on ddaA neurons (283.0 ± 9.2, n = 8 versus 327.9 ± 18.1, n = 16, P = 0.039) (Fig. 3D–F). Surprisingly, the effect of the ena46 mutation on ddaA neurons (Fig. 3F) and ddaD neurons (data not shown) was less severe than that on ddaC neurons, which could be explained by a differential requirement of Ena for different dendritic fine structures or by differential perdurance of Ena in different DA sensory neurons during the MARCM analysis (Lee and Luo, 1999; Sweeney et al., 2002). We also examined ddaB neurons that exhibited a significant decrease in the number of dendritic ends (19.3 ± 1.7, n = 11 versus 43.1 ± 3.8, n = 23, P = 4e-06) (Fig. 3G–I).

Figure 3.

Ena has a cell-autonomous function in controlling dendritic morphogenesis of DA neurons in Drosophila. A: A wild-type (WT) ddaC neuron. B: An ena mutant ddaC neuron. C: Quantification of dendritic defects of ddaC neurons caused by the ena mutation. D: WT ddaA neuron. E: An ena mutant ddaA neuron. F: Quantification of reduction in spine-like protrusions caused by the ena mutation. G: WT ddaB neuron. H: An ena mutant ddaB neuron. I: Quantification of dendritic defects of ddaB neurons caused by the ena mutation. *P < 0.05 vs. WT; ***P < 0.001 vs. WT. Scale bar = 50 μm.

In the dorsal cluster, ddaE and ddaF (as named in Sweeney et al., 2002) extend smooth dendritic branches toward either posterior or anterior segment boundaries, separately. The ena46 mutation decreased the number of dendritic ends of ddaE neurons (19.4 ± 1.2, n = 9 versus 32.9 ± 1.5, n = 17, P = 2e-7) (Fig. 4A–C) and ddaF neurons (data not shown). To further confirm this dendritic phenotype, we examined another ena allele: enaGC1, using the MARCM technique. The enaGC1 mutation was caused by a gamma ray–generated inversion and used previously to investigate ena function in axon guidance (Bashaw et al., 2000). We observed a similar dendritic phenotype on enaGC1 mutant ddaE neurons (20.0 ± 1.9, n = 6 versus 32.9 ± 1.5, n = 17, P = 4e-7) (Fig. 4C) and other dorsal cluster DA neurons (data not shown), as well as vpda neurons in the ventral cluster (21.4 ± 1.2, n = 9 versus 39.7 ± 0.7, n = 15, P = 2e-9) (Fig. 4D–F). Like ddaE neurons, vpda neurons extend smooth dendritic branches. Because the dendritic trees of ddaE and vpda neurons are relatively simple and easy to trace and analyze, we used only these two neurons for further genetic analysis in the following studies.

Figure 4.

Effects of different ena alleles on dendritic branching. A: A wild-type (WT) ddaE neuron. B: A mutant ddaE neuron homozygous for the ena46 allele. C: Quantification of dendritic defects of ddaE neurons that are homozygous for the ena46 or enaGC1 alleles. D: WT vpda neuron. E: A mutant vpda neuron homozygous for the enaGC1 allele. F: Reduction in the number of dendritic ends caused by the enaGC1 mutation. ***P < 0.001 vs. WT. Bar = 50 μm.

Ena Mutations Exert More Severe Effect on Higher-Order Dendritic Branches

To further analyze the effect of ena mutations on dendritic branching, we used ddaE neurons as a case study. We counted different orders of dendritic branches using the centrifugal method as in one of our previous studies (Li et al., 2004). Ena mutations had more severe effects on higher-order dendritic branches. For instance, there was a 63% decrease in tertiary branches of ddaE neurons, 31% decrease in secondary branches, and no difference in the number of primary branches (Fig. 5). This finding is consistent with our analysis of ena46 mutant embryos, which showed normal extension of dorsal dendrites at 14–16 h AEL (Fig. 1).

Figure 5.

Quantitative analysis of the effects of ena on different orders of dendritic branches of ddaE neurons. Different orders of dendritic branches were counted using the centrifugal method. Primary branches were not affected by the ena mutation, while higher-order dendritic branches were more severely affected. ***P < 0.001 vs. wild-type.

Abl Limits Dendritic Branching and the Formation of Spine-Like Protrusions

Ena was first identified as a dosage-sensitive suppressor of abl mutations and encodes a protein that can be phosphorylated by Abl (Gertler et al., 1995; Comer et al., 1998). Abl is expressed predominantly in the axons of the central nervous system (CNS) during Drosophila embryogenesis (Gertler et al., 1989). It is also detected in other tissues, including muscles, epidermis, and PNS neurons in embryonic and larval stages (Bennett and Hoffmann, 1992). To study the cell-autonomous function of Abl in controlling dendritic morphogenesis, we used the MARCM technique to generate GFP-labeled postmitotic mutant DA neurons in Drosophila larvae as we did for ena. Two abl mutant alleles are available: Abl4 is a protein-null allele, while abl1 produces a truncated protein at a level similar to that of the wild-type allele (Grevengoed et al., 2001). Therefore, for this study, we used the abl4 allele to avoid potential complications in interpreting results from the abl1 allele. The loss of abl activity increased the number of dendritic ends of ddaE neurons (38.2 ± 2.0, n = 13 versus 32.9 ± 1.5, n = 17; P = 0.014) (Fig. 4C), a dendritic phenotype opposite to that caused by ena mutations.

To further test the opposing effects of Ena and Abl on dendritic branching of DA sensory neurons, we overexpressed the proteins Ena or Abl in vpda neurons using the driver Gal4 109(2)80, which targets gene expression in all DA neurons (Gao et al., 1999). For this gain-of-function study, all the larvae including wild-type controls express mCD8-GFP in DA neurons under the control of Gal4 109(2)80. We found that the number of dendritic ends was increased by overexpression of Ena on the wild-type background (31.2 ± 0.7, n =30 versus 23.1 ± 0.5, n = 40; P = 4e-14) and decreased by overexpression of Abl (19.6 ± 0.4, n = 40 versus 23.1 ± 0.5, n = 40, P = 1.7e-6) (Fig. 6). Similar results were obtained when Abl was overexpressed from an EP line (20.9 ± 0.5, n = 30 versus 23.1 ± 0.5, n = 40, P = 0.0046) (Fig. 6). These findings suggest that Abl is normally required to limit dendritic branching. To determine if the kinase activity of Abl is required for the effects on dendrite development, we expressed Abl containing a mutant kinase domain (Wills et al., 1999b). No effect on dendritic morphogenesis was observed (Fig. 6), suggesting that the kinase activity is essential for Abl to function in this regulatory process. These studies demonstrate that both Ena and Abl function cell-autonomously to control dendritic branching but with opposing effects.

Figure 6.

Gain-of-function studies showing that Ena and Abl have opposite effects on dendritic branching. Different proteins were expressed under the control of Gal4 109(2)80 and their effects on dendritic branching were analyzed. Because we traced and counted the dendritic ends of dorsal dendrites of vpda neurons only, the number is lower than that in Figure 4. **P < 0.01 vs. wild-type; ***P < 0.001 vs. wild-type.

Ena Genetically Interacts With P120ctn

To further examine the function of the Ena/Abl pathway on dendritic morphogenesis, we studied the role of p120ctn, the single member of p120ctn family in flies (Myster et al., 2003; Pacquelet et al., 2003). One of the mammalian homologs of fly p120ctn, δ-catenin, was reported to be a substrate of Abl (Lu et al., 2002). Abl kinase activity is also essential for phosphorylation of δ-catenin (Rhee et al., 2002). Since the p120ctn-null mutants used in this study, p120308, were viable (Myster et al., 2003), we examined dendritic morphology of DA neurons in third instar larvae. We used the driver Gal4 109(2)80 to label all the DA neurons and traced lateral part of dendritic trees of vpda neurons. The loss of p120ctn did not affect significantly the number of dendritic ends of vpda neurons (21.7 ± 0.5, n = 30 versus 23.1 ± 0.5, n = 40; P = 0.07) (Fig. 7). We also expressed a p120ctn RNAi construct (Pacquelet et al., 2003) in DA neurons and found no change in the number of dendritic ends of vpda neurons (23.0 ± 0.6, n = 40 versus 23.1 ± 0.5, n = 40; P = 0.87). Similarly, overexpression of p120ctn had no effect on the number of dendritic branches of vpda neurons (23.4 ± 0.6, n = 30 versus 23.1 ± 0.5, n = 40; P = 0.78). These results suggest that p120ctn, unlike ena, is not essential for dendritic branching of vpda neurons.

Figure 7.

Ena interacts genetically with p120ctn in controlling dendritic branches. Transgenes were expressed in DA neurons under the control of Gal4 109(2)80. Only dendritic branches extending from the main dorsal dendrites of vpda neurons were counted. *P < 0.05 vs. wild-type (WT); **P < 0.01 vs. WT; ***P < 0.001 vs. WT.

However, we observed genetic interactions between ena and p120ctn. ena46/+ heterozygous larvae exhibited a slight decrease in the number of dendritic ends of vpda neurons (21.3 ± 0.4, n = 40 versus 23.1 ± 0.5, n = 40; P = 0.01). This dendritic branching phenotype in ena46/+ heterozygous larvae was enhanced by the loss of p120ctn (18.8 ± 0.4, n = 40 versus 21.3 ± 0.4, n = 40; P = 0.006) (Fig. 7). Moreover, p120ctn was required for Ena to exert its overextension effect on dendritic branching of vpda neurons: overexpression of Ena on the wild-type background significantly increased the number of dendritic ends of vpda neurons, but failed to do so on the p120ctn mutant background (Fig. 7). These findings revealed a supporting role for p120ctn in dendritic branching that was not evident when p120ctn alone was mutated.

P120ctn Primarily Controls the Formation of Spine-Like Protrusions

Although p120ctn mutations alone did not significantly affect dendritic branching of vpda neurons that extend smooth dendrites, p120ctn did affect the formation of spine-like protrusions on other neurons. We labeled all DA neurons with GFP using Gal4 109(2)80 and counted the number of spine-like protrusions on v'pda neurons in third instar larvae. We found that the number (data not shown) and the density of spine-like protrusions (13.8 ± 0.8/100 μm, n = 40 versus 18.6 ± 0.9/100 μm, n = 40; P = 0.013) were reduced (Fig. 8). Because p120ctn is close to the centromere, MARCM analysis cannot be utilized. Therefore, to address the cell-autonomous function of p120ctn in neurons, we expressed a p120ctn RNAi construct whose effectiveness had been demonstrated before in Drosophila embryos (Pacquelet et al., 2003). Expression of this RNAi construct reduced the density of spine-like protrusions on v'pda neurons (13.1 ± 0.8/100 μm, n = 40 versus 18.6 ± 0.9/100 μm, n = 40; P = 2.0e-5) (Fig. 8). Conversely, overexpression of p120ctn increased the density of spine-like protrusions (22.7 ± 1.7/100 μm, n = 30 versus 18.6 ± 0.9/100 μm, n = 40; P = 0.002) (Fig. 8). These findings suggest that a major function of p120ctn is to promote the formation of spine-like protrusions on some DA sensory neurons in Drosophila larvae.

Figure 8.

P120ctn plays a role in the formation of spine-like protrusions. A: Spine-like protrusions on a dendritic branch of a wild-type (WT) v'pda neuron. B: Spine-like protrusions on a v'pda neuron that expresses a p120ctn RNAi construct. C: Statistic analysis of the effects of p120ctn on the formation of spine-like protrusions. The measurement of density of spine-like protrusions is described in the Experimental Procedures section. **P < 0.02 vs. WT; ***P < 0.001 vs. WT.

DISCUSSION

To study how the regulators of the actin cytoskeleton control dendritic morphogenesis, we used the MARCM technique to visualize single wild-type or mutant PNS neurons in living Drosophila larvae. We found that Ena is required to promote dendritic branching while Abl exerts an opposite effect. We further show that ena genetically interacts with p120ctn and provide the first evidence to demonstrate a role for p120ctn in dendritic morphogenesis.

Ena Is a Key Regulator of Dendritic Morphogenesis

Previous studies implicated Ena in the actin-dependent process of axon guidance (reviewed in Krause et al., 2003). Ena, one of the founding members of Ena/VASP family proteins, is present at the leading edge of lamellipodia and at the tips of filopodia and directly binds to the actin monomer–binding protein profilin (Krause et al., 2003). In cultured hippocampal neurons, Ena/VASP activity is required for the normal formation of filopodia on growth cones and neurite shafts (Lebrand et al., 2004). Here using both loss-of-function and gain-of-function approaches, we demonstrated that Ena regulates the dendritic branching of different subtypes of DA sensory neurons in Drosophila. Dendritic filopodia may be involved in dendritic growth and branching (e.g., Portera-Cailliau et al., 2003). As shown by time-lapse analysis, numerous filopodia-like thin processes at the tips of lateral dendrites of DA neurons undergo rapid extension and retraction during Drosophila embryogenesis. Some are stabilized and eventually become lateral branches (Gao et al., 1999). Our immunostaining studies showed that Ena was localized in dendrites of DA sensory neurons and that the number of dendritic filopodia was significantly reduced in ena mutant embryos. Therefore, it is highly likely that Ena plays analogous roles in dendrites to control the formation of filopodia, precursors of stable dendritic branches.

The Abl Kinase Limits Dendritic Branching in a Way Opposite to That of Ena

Here we show that Abl, a nonreceptor tyrosine kinase, plays a key role in several developmental processes including axon guidance and epithelial morphogenesis (Hernandez et al., 2004) and has a cell-autonomous function in limiting dendritic branching, opposite to that of Ena. Loss of Abl activity resulted in an increased number of dendritic branches and spine-like protrusions, while overexpression of Abl inhibited the formation of dendritic branches and spine-like protrusions. A mutant Abl construct that abolishes its kinase activity has no effect on dendritic development, suggesting that phosphorylation of its substrates is required for its activity in this process. Thus, the extent of dendritic branching appears to be controlled, in part, by the balance between the activities of Ena and Abl.

Several lines of evidence indicate that Abl also interacts with cadherin complexes. Abl mutations in Drosophila significantly enhance the axonal defects found in armadillo (Arm, the β-catenin homologue) mutants (Loureiro and Peifer, 1998). Drosophila E-cadherin interacts with Abl in controlling epithelial morphogenesis (Grevengoed et al., 2001). Furthermore, Abl forms a complex with the axon guidance receptor Robo and N-cadherin, and the kinase activity of Abl is essential for phosphorylation of β-catenin (Rhee et al., 2002). Abl also interacts with and phosphorylates δ-catenin, a member of the p120ctn subfamily in mammals (Lu et al., 2002). Several components in the N-cadherin complex regulate various aspects of neuronal morphogenesis (Lu et al., 2002; Togashi et al., 2002; Elul et al., 2003; Martinez et al., 2003; Yu and Malenka, 2003; Bamji et al., 2003).

P120ctn Plays a Role in the Formation of Spine-Like Protrusions

The potential interaction between Abl and p120ctn prompted us to examine the role of p120ctn in dendritic branching. Drosophila p120ctn mutant embryos do not show defects in adherens junctions, and p120ctn is not required for DE-cadherin function in vivo (Myster et al., 2003; Pacquelet et al., 2003). However, studies in mammalian systems indicate that p120ctn plays a key role in maintaining normal levels of cadherin (reviewed in Peifer and Yap, 2003). In this study, we described the morphological and molecular differences between dendritic branches and spine-like protrusions on some DA neurons (Fig. 1). Although those spine-like protrusions do not make synaptic connections with other neurons, they do share some similarities with mammalian spines. Most notably, these processes are more or less perpendicular to dendritic shafts and are highly enriched in actin (Fig. 1). Here we provide the first evidence that p120ctn has a function in neural development. We found that loss of p120ctn activity reduced the number of spine-like protrusions on dendrites of some DA sensory neurons, while overexpression of p120ctn promotes the formation of these fine dendritic structures. Since overexpression of α-catenin, another cadherin-associated protein, increases spine density in cultured mammalian neurons (Abe et al., 2004), it remains possible that p120ctn regulates actin cytoskeleton dynamics through modulating the cadherin complex. Interestingly, ena and p120ctn interacted genetically, revealing a supporting role for p120ctn in modulating dendritic branching of a subset of DA neurons, which was not obvious when P120ctn was mutated alone. Taken together, these genetic analyses suggest that different regulators of the actin cytoskeleton exert their specific effects on dendritic morphogenesis through interactive molecular pathways.

EXPERIMENTAL PROCEDURES

Fly Genetics

All the genetic crosses were carried out at 25°C with standard food medium. For overexpression studies, the Gal4 109(2)80 fly line was crossed with various UAS lines or an abl EP line to drive target gene expression in all MD neurons. To examine dendritic phenotypes in p120ctn mutant larvae, ena46 heterozygous larvae, or in larvae expressing p120ctn RNAi constructs (Pacquelet et al., 2003), Gal4 109(2)80 and UAS-mCD8::GFP elements present on the second chromosome were used to visualize dendrites.

Single-Neuron MARCM

The cell-autonomous functions of abl and ena in single DA neurons were analyzed as described (Li et al., 2004). Briefly, the ena46 mutations were recombined onto the chromosome containing FRT42B. ena46, FRT42B/CyO male flies were crossed with Gal4C155, UAS-mCD8::GFP, hs-FLP1/FM7 virgin flies. Then, Gal4C155, UAS-mCD8::GFP, hs-FLP1; ena46, FRT42B/+ male flies were crossed with Gal4C155, UAS- mCD8::GFP, hs-FLP1; tubP-Gal80, FRT42B/CyO virgin flies. Embryos from this cross were collected on grape agar plates for 3 hr in a 25°C incubator. The embryos were aged for 3 hr and then heat-shocked in a 37°C water bath for 40 min to induce mitotic recombination. The embryos were then kept in a moisture chamber at 25°C and allowed to develop for 3–4 days. Third instar larvae were collected, and those containing a single mCD8::GFP-labeled dorsal cluster PNS neuron were selected under a Nikon fluorescence dissection microscope. Images of dendritic morphology of single DA neurons were recorded with a Nikon confocal microscope (D-Eclipse C1). To quantify dendritic branching, the total numbers of dendritic ends of different DA neurons were counted. Similar analyses were carried out for the enaGC1 allele (Gertler et al., 1995) and the abl1 allele (Grevengoed et al., 2001).

Identification of Mutations in ena46

Two background lethal mutations in genes other than ena were eliminated through recombination. To identify ena mutations, we isolated genomic DNA from homozygous ena46 mutant embryos. Primers were designed corresponding to the ena genomic sequence, and independent polymerase chain reactions were carried out to clone and sequence both strands of the genomic DNA fragment.

Quantitative Analysis of DA Neuron Dendrites

Embryos of the desired genotypes were collected and kept at 25°C for 4 days. Third instar larvae were collected, and images of GFP-labeled dorsal and ventral cluster DA neurons and MARCM-generated single neurons were obtained with a Nikon confocal microscope. The total numbers of the dendritic ends were counted and analyzed to reflect the branching complexity. For technical convenience, in overexpression studies, the dorsal part of vpda neuron dendrites or several specific dendritic branches of v'pda neurons were traced. Therefore, the number of dendritic ends in Figure 6 is lower that that in Figure 4. In some cases, different orders of dendritic branches were also counted with the centrifugal method (Uylings et al., 1975). According to this method, branches extending directly from the cell body are defined as primary dendrites, and branches extending from primary dendrites are defined as secondary dendrites, and so on. All statistic analyses were done using Student's t-tests.

Western Blot Analysis

Ena expression in Drosophila embryos was analyzed by Western blot according to the standard protocol provided by Bio-Rad (Richmond, CA). Wild-type or ena mutant embryos were used to prepare protein extracts. Mouse anti-Ena monoclonal antibody (Developmental Studies Hybridoma Bank, 1:200) was used as the primary antibody, and horseradish peroxidase–conjugated donkey anti-mouse IgG antibody (Jackson ImmunoResearch Laboratories, 1:1,000) as the secondary antibody.

Immunostaining

For antibody immunostaining of DA neurons in embryos or dissected third instar larvae, monoclonal antibody against Ena (1:50) was used as the primary antibody and Cy3-conjugated goat anti-mouse IgG (Jackson ImmunoResearch Laboratories, West Grove, PA; 1:200) as the secondary antibody. The embryos or dissected larvae were mounted in 90% glycerol in PBS, and confocal images were obtained with a confocal microscope.

Acknowledgements

We thank T. Kornberg, M. Peifer, P. Rorth, D. Van Vactor, and Bloomington Stock Center for fly lines, S. Ordway and G. Howard for editorial assistance, F. Wang for assistance with the MARCM analysis, and M. Ruettinger for manuscript preparation. We also thank F. Gertler, L. Reichardt, and Gao lab members for earlier comments on this work.

Ancillary