Increased expression of Down Syndrome Cell Adhesion Molecule (Dscam) is implicated in the pathogenesis of brain disorders such as Down syndrome (DS) and fragile X syndrome (FXS). Here, we show that the cellular defects caused by dysregulated Dscam levels can be ameliorated by genetic and pharmacological inhibition of Abelson kinase (Abl) both in Dscam-overexpressing neurons and in a Drosophila model of fragile X syndrome. This study offers Abl as a potential therapeutic target for treating brain disorders associated with dysregulated Dscam expression.

DOI: http://dx.doi.org/10.7554/eLife.05196.001

eLife digest

Information is transmitted through the brain by cells called neurons, which are connected into specific circuits and networks. As the brain develops, several different signaling molecules control how the connections between neurons develop. If these signals occur at the wrong time or wrong place, or in the wrong amount, the neurons may not connect in the right way; this is the cause of several brain disorders.

One of the signaling molecules that helps neural circuits to form in the developing brain is the Dscam protein. Having too much Dscam has been linked to neurons with enlarged presynaptic terminals. Presynaptic terminals are the parts of each neuron that send information on to the next cell, and when they are enlarged it results in the neuron not being able to communicate with other neurons in a targeted way. People with brain disorders including Down syndrome, epilepsy and possibly fragile X syndrome often have excessive amounts of Dscam.

It was not known precisely how Dscam signals within neurons. Sterne, Kim and Ye have now investigated this by exploring the effects of Dscam on a group of well-known neurons in the larvae of the fruit fly species Drosophila. The presynaptic terminals of single neurons in this group were labeled in the larvae using a genetic marker. This revealed that the neurons of larvae that had been engineered to produce too much Dscam had larger presynaptic terminals than normal.

Further investigation showed that for Dscam to influence how a presynaptic terminal grows, it must interact with another signaling protein called Abelson tyrosine kinase (or Abl for short). Therefore, the larger presynaptic terminals seen in larvae that produce too much Dscam are a result of the Dscam protein activating too much Abl.

There are several drugs that are approved for use in humans that suppress the activity of Abl. Sterne, Kim and Ye used two of these to treat fruit fly larvae, and found that they reversed the detrimental effects of extra Dscam on the larvae's neural circuit. Furthermore, the drugs fixed neural defects in a fruit fly model designed to reproduce the symptoms of fragile X syndrome.

Overall, the results presented by Sterne, Kim and Ye suggest that suppressing the abnormally high activity of the Abl protein could be a way of treating the brain disorders caused by having excessive amounts of the Dscam protein. The next step is to test whether Dscam and Abl interact in the same way in mammals and whether the proposed treatment is effective in treating mammalian models of disorders that involve dysregulated Dscam signaling.

DOI: http://dx.doi.org/10.7554/eLife.05196.002

Main text


Dscam levels are increased in the brains of human patients with DS and in mouse models of DS (Saito et al., 2000; Alves-Sampaio et al., 2010). Recent research also suggests that fragile X mental retardation protein (FMRP) binds directly to the mRNAs of Dscam from mouse brain (Brown et al., 2001; Darnell et al., 2011), and studies in Drosophila neurons further confirmed that FMRP suppresses Dscam translation (Cvetkovska et al., 2013; Kim et al., 2013). In the dendritic arborization (da) neurons in Drosophila larva, Dscam expression level is instructive for presynaptic terminal growth (Kim et al., 2013). Consistent with this, increased Dscam in Drosophila FXS models results in enlarged presynaptic arbors (Kim et al., 2013). These findings indicate the importance of proper Dscam levels in normal development and in the pathogenesis of brain disorders.

Because of the link between increased Dscam expression and neuronal defects in DS and FXS models, targeting Dscam or its signaling mechanism might prove therapeutic for these disorders. Currently, neither methods for targeting Dscam proteins nor those for targeting the signaling pathway activated by dysregulated Dscam are available, impeding the development of such therapies. In fact, very little is known about how Dscam signaling is transduced in vivo. In Drosophila, Dscam has previously been shown to bind to Dock (Schmucker et al., 2000), while in mammals it has been shown to associate with Uncoordinated-5C, Focal adhesion kinase (FAK), Fyn kinase, and PAK1 (Li and Guan, 2004; Purohit et al., 2012). In addition, studies suggest possible genetic interactions between Dscam and the Abelson tyrosine kinase (Abl) in neurite development in the central nervous system (CNS) of Drosophila embryos (Andrews et al., 2008; Yu et al., 2009). However, evidence demonstrating the requirement of these potential interactors for the defects that arise from increased Dscam expression is lacking. Moreover, whether pharmacologically targeting these molecules in vivo might alleviate the effects of increased Dscam expression is unknown.

The evolutionarily conserved Abl kinase transduces extracellular cues into cytoskeletal rearrangements that affect cell motility and shape (Bradley and Koleske, 2009) and is implicated in axonal development, including axon guidance and extension (Wills et al., 1999a; Wills et al., 1999b; Wills et al., 2002; Hsouna et al., 2003; Lee et al., 2004; Forsthoefel et al., 2005). Overexpression of Abl causes increased axon growth in the Drosophila CNS (Leyssen et al., 2005), which is reminiscent of the effect caused by Dscam overexpression in C4da neurons (Kim et al., 2013). In addition, previous studies in Drosophila have indicated that abl mutations have an additive effect with Dscam mutations, such that abl/Dscam double mutant embryos have more severe axon midline crossing defects than either abl or Dscam mutants alone (Andrews et al., 2008; Yu et al., 2009). However, the molecular nature of this interaction, that is, whether or not Dscam acts through Abl, and particularly whether inhibition of Abl mitigates neuronal defects caused by dysregulated Dscam, is unknown.

Here we show that Dscam activates Abl through its cytoplasmic domain, which is required for the presynaptic arbor enlargement caused by dysregulated Dscam expression in vivo. Importantly, we demonstrate that the pharmacological inhibition of Abl ameliorates exuberant presynaptic arbor growth both in flies overexpressing Dscam and in a fly model of FXS.

Results and discussion

We took advantage of the Drosophila larval class IV dendritic arborization (C4da) neurons to delineate the molecular mechanism of Dscam signaling in presynaptic arbor development, because the presynaptic terminal growth of these neurons is highly sensitive to Dscam levels in a linear fashion (Kim et al., 2013). For example, loss of Dscam causes C4da presynaptic terminals to fail to grow while increased Dscam levels lead to increased presynaptic terminal growth (Kim et al., 2013). From tests of candidate genes that potentially mediate Dscam function, including FAK, Fyn, PAK, RhoA, and Abl, we identified Abl as a key molecule mediating Dscam's functions in presynaptic terminal growth. We first asked whether Abl is sufficient to promote presynaptic terminal growth in C4da neurons. Consistent with a previous study performed in Drosophila adult CNS neurons (Leyssen et al., 2005), overexpression of Abl in C4da neurons caused significant overgrowth of the presynaptic terminals (Figure 1A,B,E). Since Abl is known to have both kinase-dependent and kinase-independent functions (Henkemeyer et al., 1990; Schwartzberg et al., 1991; Tybulewicz et al., 1991), we tested whether expression of a kinase-dead form of Abl, Abl-K417N (Henkemeyer et al., 1990; Wills et al., 1999b), could promote presynaptic terminal growth. We found that C4da presynaptic terminals overexpressing Abl-K417N were indistinguishable from wild-type (Figure 1D,E), indicating that Abl kinase activity is required. Consistent with the idea that Abl kinase activation is important, expression of a constitutively active form of Abl, BCR-Abl, led to extremely exuberant overgrowth (Figure 1C,E). Taken together, these results suggest that Abl is sufficient to promote presynaptic terminal growth and that the extent to which Abl instructs presynaptic terminal growth is related to Abl kinase activation.

Since overexpression of Abl increases presynaptic terminal growth, similar to Dscam, we next tested whether Dscam requires Abl to instruct presynaptic terminal growth. For this, we used the mosaic analysis with a repressible cell marker (MARCM) technique to overexpress Dscam in abl1 mutant C4da neurons (Lee and Luo, 2001) and assessed presynaptic terminal length. We found that although Dscam overexpression led to significantly (150%) longer presynaptic terminals than wild-type clones (Figure 1F,G,O), abl1 mutant clones that overexpressed Dscam did not differ in length from abl1 mutant clones (Figure 1H,I,O). Presynaptic terminal length was also subtly but significantly shorter in abl1 mutant clones compared to wild-type controls (Figure 1I,O). A different loss-of-function allele of abl, abl4, exhibited similar effects on the presynaptic overgrowth caused by Dscam overexpression (Figure 1J,K,O), confirming that loss of abl function is responsible for blocking the presynaptic phenotypes caused by increased Dscam levels. As a control, abl loss-of-function mutations did not affect the expression of the Dscam transgenes in the C4da cell body or presynaptic terminals (Figure 1—figure supplement 1).

As a further proof-of-concept, we asked whether loss of abl could mitigate the effects of dysregulated Dscam levels without utilizing Dscam transgenes. FXS is caused by an absence of FMRP (Kremer et al., 1991), and is modeled in Drosophila using loss-of-function mutants for the Drosophila homolog of FMR1, dFMRP (Zhang et al., 2001; Dockendorff et al., 2002). It has previously been shown that FMRP binds to Dscam mRNA in both mammals and Drosophila (Darnell et al., 2011; Cvetkovska et al., 2013; Kim et al., 2013) and that dFMRP represses Dscam expression to control presynaptic terminal growth, so that dFMRP mutants exhibit increased presynaptic terminal length in C4da neurons (Kim et al., 2013). Strikingly, loss of only a single copy of abl significantly rescued presynaptic terminal length to wild-type levels (Figure 1L–N,P). These results suggest that Abl is required for Dscam to instruct presynaptic terminal growth.

An important function of Dscam in neuronal development is to mediate self-avoidance between neurites of the same neuron (Zipursky and Grueber, 2013). Abl does not seem to be required by Dscam for either dendrite growth (Figure 1—figure supplement 2) or for dendritic self-avoidance in C4da neurons. Loss of abl did not compromise the ectopic avoidance caused by overexpressing Dscam in distinct types of da neurons (Hattori et al., 2007; Hughes et al., 2007; Matthews et al., 2007) (Figure 1—figure supplement 3). This suggests a divergence in Dscam signaling for the development of presynaptic terminals and dendritic branches. Taken together, these results indicate that Abl is specifically required for Dscam-mediated presynaptic terminal growth.

Next, we asked how Abl might mediate Dscam signaling. Abl can be activated by binding to specific proteins, such as the cytoplasmic domains of membrane receptors (Bradley and Koleske, 2009). In contrast to the exuberant presynaptic terminal overgrowth caused by Dscam overexpression in C4da neurons (Figure 2A, middle), overexpressing a mutant form of Dscam that lacked most of the cytoplasmic domain (DscamΔCyto) did not cause presynaptic terminal overgrowth (Figure 2A, bottom). DscamΔCyto was trafficked to the axon terminals and expressed at a similar level to full-length Dscam (Figure 2—figure supplement 1). These results suggest that the cytoplasmic domain is required for Dscam to instruct presynaptic terminal growth.

We then asked whether Dscam and Abl physically interact through the Dscam cytoplasmic domain. We found that Dscam and Abl proteins co-immunoprecipitated from transfected Drosophila Schneider 2 (S2) cells expressing these two proteins (Figure 2B, second lane from right). In contrast, DscamΔCyto did not co-immunoprecipitate Abl (Figure 2B, furthest right lane). These results suggest that Dscam and Abl proteins form a complex through Dscam's cytoplasmic domain. Next, to test the in vivo interaction of Dscam and Abl in presynaptic terminals specifically, we determined whether Abl localization in presynaptic terminals was altered by the expression of Dscam or DscamΔCyto (Figure 2C). When expressed alone or with DscamΔCyto::GFP, Abl::Myc was diffusely distributed in the presynaptic terminals, with little colocalization with DscamΔCyto::GFP (Figure 2C, middle and bottom). However, when expressed with Dscam::GFP, Abl::Myc became more punctate and clearly colocalized with Dscam::GFP (Figure 2C, top). We used Manders' Correlation Coefficients to quantify the colocalization of Dscam::GFP and Abl::Myc. Colocalization analysis revealed a significant increase in both M1 and M2 (Figure 2C, bottom right) when Abl::Myc was coexpressed with Dscam::GFP as compared to when Abl::Myc was coexpressed with DscamΔCyto::GFP, where M1 represents the fraction of Abl that overlaps with Dscam, and M2 represents the fraction of Dscam that overlaps with Abl. These findings support the idea that Abl and Dscam interact in presynaptic terminals in vivo.

Do increased Dscam levels activate Abl kinase? In mammals, autophosphorylation of Abl at tyrosines 245 and 412 (Y245 and Y412) stabilizes the active conformation of the kinase (Brasher and Van Etten, 2000; Tanis et al., 2003). As a result, phospho-specific antibodies raised against Y412 have been employed to detect active Abl kinases (Brasher and Van Etten, 2000). This approach has been used successfully to recognize the phosphorylation of the corresponding tyrosines (Y539/522) in Drosophila as an assay for Abl kinase activation (Stevens et al., 2008). Since the ability of Abl to instruct presynaptic terminal growth relies on Abl kinase activity, we tested whether Dscam activates Abl using a phosho-Y412-Abl (p-Abl) antibody. We found that Abl kinase activation was significantly increased (2.6 fold) when Abl and Dscam were co-expressed in S2 cells (Figure 3A). Furthermore, unlike wild-type Dscam, DscamΔCyto did not increase Abl kinase activation. In fact, it appears to act as a dominant-negative, as Abl activity was significantly decreased from control (Figure 3A, right). As a negative control, no signal was detected when the kinase-dead Abl-K417N was blotted with p-Abl antibody in the same assay, suggesting that our assay specifically reported Abl activation (Figure 3—figure supplement 1). These results suggest that Dscam enhances Abl kinase activity. To investigate whether the same is true in presynaptic terminals in vivo, we devised a novel method of reporting Abl activation specifically in C4da presynaptic terminals. To achieve this, we used a previously reported probe that reports Abl activity, Pickles2.31 (Mizutani et al., 2010). Pickles2.31 is composed of a fragment of a characteristic Abl substrate, CrkL, sandwiched between the fluorescent proteins Venus and enhanced CFP (ECFP) (Figure 3B). It has previously been reported that activated Abl phosphorylates Pickles2.31 on the Y207 residue of the CrkL fragment, which can be detected with an antibody against CrkL-phospho-Y207 (p-CrkL) (Mizutani et al., 2010). After expressing Pickles2.31 specifically in C4da neurons with the ppk-Gal4 driver, we dissected the larval CNS and immunoprecipitated Pickles2.31 from the lysates. Since the cell bodies of C4da neurons reside in the body wall, using only the larval CNS allowed us to monitor Pickles2.31 phosphorylation only in the C4da neuron presynaptic terminals (Figure 3C). We found that overexpression of Dscam in C4da neurons led to an increase in Y207 phosphorylation of Pickles2.31 in the presynaptic terminals, while overexpression of DscamΔCyto was indistinguishable from control (mCD8-mRFP) (Figure 3D). Consistent with the notion that Pickles2.31 is an Abl activity indicator, overexpression of BCR-Abl led to a robust increase in phospho-Y207 levels as compared to the control. These results suggest that Dscam activates Abl both in culture and in C4da presynaptic terminals in vivo, and that this activation requires the cytoplasmic domain of Dscam.

These results raised the interesting possibility that targeting Abl might be a viable therapy for brain disorders caused by increased Dscam expression. Abl is a well-established target for treating chronic myeloid leukemia, and there are multiple Abl inhibitors that are approved by the US Food and Drug Administration (FDA). As a proof-of-concept experiment, we attempted to rescue the developmental defects caused by Dscam overexpression using Abl inhibitors. We first tested nilotinib, which is a FDA-approved second-generation Abl kinase inhibitor that can cross the blood–brain barrier (Weisberg et al., 2005; Hebron et al., 2013). Using cultured S2 cells overexpressing Abl, we found that nilotinib robustly inhibited Drosophila Abl (Figure 4A). Based on these results, we tested whether administration of nilotinib to developing larvae could rescue the effects of increased Dscam expression in C4da presynaptic terminals in vivo. To do this, we performed MARCM to visualize single C4da neurons in animals fed nilotinib or vehicle and assessed presynaptic terminal length. While overexpression of Dscam caused increased (152%) presynaptic terminal length in animals fed vehicle (Figure 4B–D), the effect was significantly rescued (to 115% of control) by feeding the animals with nilotinib (Figure 4B,E). Consistent with the idea that these effects were due to inhibition of Abl activity rather than a reduction in Dscam expression, nilotinib did not change the expression of the Dscam transgene (Figure 4—figure supplement 1A).

Administration of nilotinib to developing larvae did not lead to adverse effects on overall development and neuronal growth. At the dose we used, nilotinib did not cause a change in presynaptic terminal growth (Figure 4F) or dendritic growth (Figure 4—figure supplement 2A,B) in wild-type larvae. Moreover, it did not impact the number of adults that eclosed or the dynamics of eclosion when compared to vehicle-fed flies (Figure 4—figure supplement 2C,D).

Although frequently used to inhibit pathological increases in Abl activity in patients, nilotinib is known to have several off-targets, including c-Kit, PDGFR, Arg, NQ02, and DDR1 (Hantschel et al., 2008). Consistent with the idea that nilotinib acts on Abl rather than on an off-target molecule to rescue presynaptic terminal growth, administering nilotinib to larvae overexpressing Dscam in abl1 clones did not lead to a further decrease in presynaptic terminal length when compared to vehicle-fed control (Figure 4—figure supplement 3C,D,G). To further rule out the possibility that the observed rescue of presynaptic terminal length by nilotinib was the result of an off-target effect, we tested bafetinib, another Abl inhibitor with non-overlapping off-targets, Fyn and Lyn (Kimura et al., 2005). Bafetinib has also been shown to cross the blood brain barrier (Santos et al., 2010). Like nilotinib, administration of bafetinib to Dscam-overexpressing larvae led to a significant decrease in presynaptic terminal length (Figure 4—figure supplement 3A,B,F,G) without changing the expression of the Dscam transgene (Figure 4—figure supplement 1B). Bafetinib alone did not change presynaptic terminal length in wild-type larvae when compared to wild-type larvae fed vehicle (Figure 4—figure supplement 3E,G). Taken together, these results suggest that pharmacological inhibition of Abl mitigates the consequences of increased Dscam signaling in vivo.

We next sought to test the efficacy of nilotinib treatment in a model of a disease associated with dysregulated Dscam expression, FXS. Thus, we tested whether administration of nilotinib could rescue the presynaptic overgrowth caused by increased Dscam expression in dFMRP mutants. We found that, while dFMRP mutants fed vehicle showed a significant increase (130%) in presynaptic terminal length (Figure 4B,G), administration of nilotinib to dFMRP mutants almost completely rescued (to 103% of control) the exuberant presynaptic terminal growth to wild-type levels (Figure 4B,H). These results suggest that pharmacological inhibition of Abl kinase is effective for mitigating the effects of increased Dscam level in an in vivo model of FXS.

In this study, we show that Dscam requires Abl to promote presynaptic terminal growth in vivo and that the binding of Abl to the Dscam cytoplasmic domain leads to Abl kinase activation. Furthermore, we show that treating larvae with Abl inhibitors rescues the developmental defects caused by increased Dscam levels in vivo in both Dscam-overexpressing neurons and disease-relevant models. Taken together, these results suggest that Abl is a potential drug target for the treatment of brain disorders associated with dysregulated Dscam expression, including DS and FXS.

Materials and methods

Fly strains

abl1 (Gertler et al., 1989), abl4 (Bennett and Hoffmann, 1992), ppk-Gal4 (Kuo et al., 2005), UAS-Dscam[]::GFP (Yu et al., 2009), UAS-Abl, UAS-BCR-Abl, UAS-Abl-K417N (Wills et al., 1999b), and dFMRPΔ50M (Zhang et al., 2001) were used in this study.

DNA constructs and generation of transgenic flies

To generate pUASTattB-Abl::Myc for expression in S2 cells, the coding region of Abl was recovered from UAS-Abl transgenic flies by PCR, subcloned into pUASTattB-Myc by using the InFusion cloning system following manufacturer's protocol (Clontech, Mountain View, California). We generated pUASTattB-Abl-K417N::Myc by PCR mutagenesis as previously described (O'Donnell and Bashaw, 2013) from pUASTattB-Abl::Myc. UAS-Dscam[]::GFP was previously generated as described (Kim et al., 2013). To generate UAS-DscamΔCyto, the Dscam coding region was digested with SstI and ligated with the GFP cDNA. Pickles2.31 was generously provided by Dr Yusuke Ohba at RIKEN Brain Science Institute (Mizutani et al., 2010). To generate UAS-Pickles2.31, the Pickles2.31 coding region was subcloned from pCAGGS-Pickles2.31 into pUASTattB using the InFusion cloning system following the manufacturer's protocol (Clontech). Transgenic flies carrying UAS-DscamΔCyto, UAS-Abl::Myc, and UAS-Pickles2.31 were generated by germline transformation with support from BestGene, Inc.

Labeling presynaptic terminals using MARCM

The MARCM technique was used to visualize single neurons homozygous for abl1, abl4, or dFMRPΔ50, and overexpressing Dscam[]::GFP as previously described (Kim et al., 2013).

Immunostaining and imaging

Immunostaining of third-instar larvae was accomplished as previously described (Ye et al., 2011). Antibodies used include chicken anti-GFP (Aves, Tigard, Oregon) and rabbit anti-RFP (Rockland, Limerick, Pennsylvania). Samples were dehydrated and mounted with DPX mounting media (Electron Microscopy Sciences, Hatfield, Pennsylvania). Confocal imaging was completed with a Leica SP5 confocal system equipped with a resonant scanner and 63× oil-immersion lens (NA = 1.40). Images were collected and quantified as previously described (Kim et al., 2013).

S2 cell culture and transfection

Drosophila S2 cells were maintained in Drosophila Schneider's medium supplemented with 10% fetal bovine serum at 25°C in a humidified chamber. Cells were transfected with indicated DNA constructs together with tubulin-Gal4 (Lee and Luo, 2001) by using Lipofectamine 2000 (Life Technologies, Grand Island, New York) according to manufacturer's protocol.

Co-immunoprecipitation and Western blotting

To perform co-immunoprecipitation, transfected S2 cells were harvested and lysed on ice with lysis buffer (50 mM Tris-HCl/pH 7.4, 150 mM NaCl, 2 mM sodium vanadate, 10 mM sodium fluoride, 1% Triton X-100, 10% glycerol, 10 mM imidazole and 0.5 mM phenylmethylsulfonyl fluoride). Lysates were centrifuged for 15 min at 20,000×g, 4°C and the resulting supernatant was incubated with Protein A/G PLUS-Agarose beads (Santa Cruz Biotechnology, Paso Robles, California) conjugated to mouse monoclonal anti-GFP clone 20 (Sigma-Aldrich, St. Louis, Missouri) for 4 hr at 4°C. After washing once with lysis buffer, twice with lysis buffer containing 0.1% deoxycholate, and 3 times with lysis buffer lacking Triton X-100, the immunoprecipitates and total lysates were resolved on 7.5% SDS-PAGE gels followed by western blot analysis as previously described (Kim et al., 2013).

Primary antibodies used in western blotting were mouse monoclonal anti-tubulin (Sigma), mouse anti-Myc (Sigma-Aldrich), mouse monoclonal anti-Aequorea Victoria GFP JL-8 (Clontech), and rabbit anti-phospho-Tyr412-c-Abl (Cell Signaling, Beverly, Massachusetts).

In vivo Abl activity assay with Pickles2.31

To assay in vivo Abl activation, UAS-Pickles2.31 was expressed specifically in C4da neurons using ppk-Gal4 along with other UAS transgenes. The CNS was dissected from third-instar larvae into ice-cold PBS with 2 mM sodium vanadate (∼100 per experimental condition). After a brief centrifugation, larval CNSs were transferred into lysis buffer as described above in immunoprecipitation and western blotting. Cells were disrupted using a pestle followed by brief sonication. Immunoprecipitation and western blotting of Pickles2.31 was then accomplished as described above. Primary antibodies used were rabbit anti-eGFP (a gift from Dr Yang Hong) and rabbit anti-phospho-Tyr 207-CrkL (Cell Signaling).

Drug treatment of Drosophila larvae and S2 cells

Nilotinib (Abcam, United Kingdom) and bafetinib (ApexBio Technology, Houston, Texas) were dissolved in dimethyl sulfoxide (DMSO) at 94 mM and 50 mM, respectively, as stock solutions before adding to S2 cells or fly food. S2 cells transfected with Abl::Myc were treated with either 5 μM nilotinib or the same volume of DMSO as a vehicle control for 6 hr before harvested and subjected to western blot analysis.

Nilotinib and bafetinib were administered to larvae by rearing the larvae on standard corn meal food containing different concentrations of the drugs. The highest concentrations that did not affect overall larval development were used. Fly viability on nilotinib treatment was performed by counting the number of adult flies. Seven virgin female and seven male flies were crossed and transferred to standard corn meal food containing either 380 μM nilotinib or the same volume of DMSO (0.4% final concentration). Embryos were collected for 24 hr and allowed to develop. Eclosed adult flies were counted on a daily basis.

The MARCM technique was used to generate and visualize mutant single C4da neurons as described above except that Drosophila embryos were collected and raised for 4 days on standard corn meal food containing either 380 μM nilotinib, 125 μM Bafetinib, or 0.4% DMSO. Sample preparation, imaging, and quantification were then completed as described above.

Colocalization analysis

Colocalization of Dscam and Abl was quantified with Manders' Correlation Coefficients using the Just Another Colocalization Plugin (JACoP) (Bolte and Cordelieres, 2006) in ImageJ. Images were analyzed in three dimensions. Manders' Correlation Coefficients vary between 0 and 1, with 0 representing no overlap between images and 1 representing complete colocalization. M1 and M2 describe the overlap of each channel with the other (Bolte and Cordelieres, 2006). M1 presents a measure of the fraction of Abl::Myc that overlaps Dscam(ΔCyto)::GFP, while M2 presents a measure of the fraction of Dscam(ΔCyto)::GFP that overlaps Abl::myc.

Statistical analysis

Two-way student's t- test was used for statistical analysis. *: p < 0.05; **: p < 0.01; ***: p < 0.001; ****: p < 0.0001; ns: not significant.



We thank Dr Tzumin Lee for Dscam fly stocks, Dr Yusuke Ohba for the Pickles2.31 cDNA, Dr Yang Hong for the rabbit anti-GFP antibody, and Drs Cheng-yu Lee and Hideyuki Komori for the pUASTattB-Myc vector. This work was supported by grants from NIH (R01MH091186), Protein Folding Disease Initiative of the University of Michigan, and the Pew Scholars Program in the Biological Sciences to BY.

Decision letter

Eunjoon Kim, Reviewing editor, Korea Advanced Institute of Science and Technology, Republic of Korea

eLife posts the editorial decision letter and author response on a selection of the published articles (subject to the approval of the authors). An edited version of the letter sent to the authors after peer review is shown, indicating the substantive concerns or comments; minor concerns are not usually shown. Reviewers have the opportunity to discuss the decision before the letter is sent (see review process). Similarly, the author response typically shows only responses to the major concerns raised by the reviewers.

Thank you for sending your work entitled “Dysregulated Dscam levels act through Abelson tyrosine kinase to enlarge presynaptic arbors” for consideration at eLife. Your article has been favorably evaluated by K VijayRaghavan (Senior editor), a Reviewing editor, and three reviewers.

The Reviewing editor and the reviewers discussed their comments before we reached this decision, and the Reviewing editor has assembled the following comments to help you prepare a revised submission.

We note key findings of the paper which show that loss of Abl kinase can reduce presynaptic terminal overgrowth caused by DsCAM overexpression, DsCAM intracellular domain can interact and activate Abl kinase and in particular that an FDA approved Abl inhibitor nilotinib can reduce presynaptic overgrowth caused by DsCAM overexpression or in fly dFMRP mutants, which also show increased Dscam expression. An added significance of the work therefore is the potential translational application of Abl inhibitors in treatment of neurological disorders.

General assessment:

CNS phenotypes for Dscam loss or misexpression had been explored in prior work. Andrews et al. (2008) showed that loss of both Dscam and its paralog Dscam3 result in CNS midline crossing defects, and that overexpression of Dscam results in ectopic midline crossings. Yu et al. (2009) showed that RNAi knockdown of Dscam resulted in morphogenesis defects as well as commissureless phenotypes. However, the effects of Dscam loss at axon terminals growth does not appear to have been demonstrated. Overall, this is a well-written and interesting study with some exciting results linking Dscam to a key axonal signalling molecule, even if some aspects of the work are at a preliminary stage.

The conclusion that Abl kinase activity is a relevant output of Dscam in determining presynaptic arbor size is supported by a convincing set of experiments in Figure 1 (and supplements), and this is a real strength of the manuscript. Given how little is known about signal transduction downstream of Dscam, this is significant. Further, the observation that Dscam signaling in regulating presynaptic arbor size and dendrite self-avoidance involves different molecular mechanisms is an interesting discovery. The studies demonstrating that Dscam can interact with Abl and promote Abl activation in S2 cells support the model that Dscam modulates Abl activity in neurons, but demonstrating the interaction in neurons would substantially improve the manuscript. Finally, the authors convincingly show that the Abl pharmacological inhibitor nilotinib can suppress presynaptic arbor overgrowth induced by Dscam or FMRP mutation, but the mechanism of action is unclear. The following substantive concerns need to be addressed.

Substantive concerns:

1a) A principal concern is that data for in vivo specificity of nilotinib on Abl are not convincing. Figure 3B is not interpretable without controls, such as abl(lf) and abl(CA) or DsCAM overexpression. Figure 3C-H, the analyses on the neuronal phenotypes in flies fed with nilotinib are surprisingly limited or selective. The value of using a fly model to test potential drug targets is that we can gain a full understanding of any possible off-target or side effects. Thus, it is equally important that they show other aspects of the neurons and animal health etc. are, or are not, affected by the drug treatment. Moreover, one can't find any data showing how the dose of nilotinib treatment, which would be highly relevant for any further interests in translational application, was decided.

1b) The specificity of the inhibitor is not discussed (or examined). Nilotinib inhibits a number of tyrosine kinases and appears to broadly inhibit tyrosine phosphorylation in larvae. The assertion that the effects on presynaptic arbor size reflect inhibition of Abl activation is not convincing. Do other small molecule inhibitors of Abl exist that have a more targeted effect (i.e., they don't broadly block tyrosine kinases) and if so, do they similarly affect presynaptic arbor growth?

1c) Figure 3. The specificity of nilotinib to Abl in Drosophila is not verified. The decrease in phospho-tyrosine in 3b demonstrates a general decrease in the CNS, and does not necessarily show a specific targeting of Abl. This must be addressed. To demonstrate specificity, a possible experiment is to repeat the experiment shown in 3B in flies that are mutated in the Abl phosphorylation site Tyr412, or simply null for Abl. If nilotinib is specific to blocking phospho-Abl, there should be no difference in p-Tyr levels whether or not nilotinib is added, if Abl phospho-sites are blocked or Abl is missing. However, if addition of nilotinib still results in a decrease in p-Tyr even when Abl phospho-sites are blocked, this suggests nilotinib is having an effect on the catalytic activity of other kinases.

2a) Figure 3. What is the effect of nilotinib treatment in wild-type flies? The results of wild type + nilotinib can be shown in addition to the experiments in 3c-g. It would be very reassuring to see that nilotinib elicits abl-like phenotypes in other parts of the nervous system or elsewhere (e.g. epithelial development in the embryo)—if the drug is selectively blocking Abl, it should phenocopy abl mutants, but not mutants in other kinases.

2b) What is the effect of drug treatment on presynaptic arbor size in a wild type background? It's impossible to evaluate the experimental results without this control.

3a) The data presented suggest that Dscam can associate with Abl and promote Abl phosphorylation, but this is not demonstrated in vivo.

3b) Figure 2. The results shown in panel 2B and 2C demonstrate binding of Abl to the cytoplasmic domain of Dscam in S2 cells using transfected constructs and levels of expression that may contribute to detection of non-physiological interaction. This is important because the fraction of the input pool that IPs with Dscam in cells overexpressing both proteins is rather low. More importantly, they do not show that this binding occurs in vivo, and that even if it does occur in vivo, that this binding is important to axon terminal growth in particular. To determine the role of Abl-Dscam binding in vivo the group should consider performing fluorescence tagging and/or immunolocalization of Abl in the three genetic backgrounds shown in panel 2A (WT, Dscam::GFP OE, Dscam cyto::GFP OE). If there is a bona fide interaction with Dscam at the terminal, co-localization of Abl with Dscam::GFP should be observed. Moreover, if this interaction requires Dscam cytoplasmic domain, expression of Dscam cyto::GFP instead of full- length Dscam should result in a re-distribution of Abl.

4) Figure 1B. The text (second paragraph of the Results section) and the figure legend state that Abl overexpression leads to a pre-synaptic overgrowth phenotype. However, while a defect in the axons is apparent in panel 1B, it is not clear whether this is due to increased growth necessarily. For example, the gross defects could result from fasciculation defects between axons as they enter the CNS neuropil. Indeed, the apparent decrease in mRFP signal intensity in the Abl OE flies raise the possibility that it is not overgrowth, but defasiculation, that is observed in 1B. This is important because Abl has been linked to many receptors/adhesion molecules that may also contribute to the normal neuropil structure for sensory terminals. Abl nulls are not shown to confirm that the phenotypes are only selective to overexpression.

5) Figure 1, general comment. All axon terminal phenotypes shown are for Dscam OE. While Dscam loss phenotypes have been shown in other systems (e.g. midline crossing), it has not been demonstrated for terminal growth. It would be useful to include data on the effects of Dscam loss at the synaptic terminal, as well as genetic experiments on the interaction between Dscam and abl loss of function alleles (as has been done in other systems). If the two genes are in one pathway for sensory terminal growth, then the double null should be indistinguishable from the single mutants. It is curious that Abl mutant terminals look relatively normal. Does this mean that Abl is not required for other receptors or DSCAM under normal expression levels?

6) Figure 1–figure supplement 1A. Dscam-GFP levels in the neuronal soma is shown to be unchanged in abl mutants. However, the observations in the cell body do not rule out the possibility that at the synapse, abl loss could have effects on Dscam levels, perhaps due to changes in Dscam protein trafficking (Abl has been implicated in axon transport by Bill Saxton). Just because abl nulls revert the Dscam OE phenotype, this doesn't prove a downstream function. Abl can be required upstream of Dscam at the synapse. To test this, a parallel analysis of Dscam-GFP could be done at axon terminals to verify that the effect of abl is the same both in the soma and at the presynaptic terminal.

7) The abl C4da dendritic phenotype analysis should be added to the supplemental figures.

8) Figure 2C. There are no data confirming the stability of the Dscam cytoplasmic domain deletion. In fact the recovery in the IP (Figure 2B) is very poor and raises the concern that this receptor mutant is not capable of accumulating to the same concentration in the OE experiment, which could explain the lack of an in vivo OE phenotype in a way that does not support requirement of the Cyto domain. Moreover, why is full-length Dscam detected as a doublet? Can this be explained by cleavage, phosphorylation, etc.?

DOI: http://dx.doi.org/10.7554/eLife.05196.015

Author response


If your username is different from your full name, we require you to identify yourself within the comment itself. Comments are checked by a moderator (and/or an eLife editor) before they appear. Comments should be constructive, relevant to the article, conform to our terms and conditions, and include any pertinent competing interests.