Jelly Belly Trans-Synaptic Signaling to Anaplastic Lymphoma Kinase Regulates Neurotransmission Strength and Synapse Architecture

Abstract

In Drosophila the secreted signaling molecule Jelly Belly (Jeb) activates Anaplastic Lymphoma Kinase (Alk), a receptor tyrosine kinase, in multiple developmental and adult contexts. We have shown previously that Jeb and Alk are highly enriched at Drosophila synapses within the CNS neuropil and neuromuscular junction (NMJ) and postulated a conserved intercellular signaling function. At the embryonic and larval NMJ Jeb is localized in the motor neuron presynaptic terminal whereas Alk is concentrated in the muscle postsynaptic domain surrounding boutons, consistent with anterograde trans-synaptic signaling. Here, we show by functional inhibition of Jeb-Alk signaling that neurotransmission is regulated by Jeb secretion. Jeb is a novel negative regulator of neuromuscular transmission. Reduction or inhibtion of Alk function results in enhanced synaptic transmission. Activation of Alk conversely inhibits synaptic transmission. Restoration of wildtype postsynaptic Alk expression in Alk partial loss-of-function mutants rescues NMJ transmission phenotypes and confirms that postsynaptic Alk regulates NMJ transmission. The effects of impaired Alk signaling on neurotransmission are observed in the absence of associated changes in NMJ structure. Complete removal of Jeb in motor neurons, however, disrupts both presynaptic bouton architecture and postsynaptic differentiation. Non-physiologic activation of Alk signaling also negatively regulates NMJ growth. Activation of Jeb-Alk signaling triggers the Ras-MAP kinase cascade in both pre- and postsynaptic compartments. These novel roles for Jeb-Alk signaling in the modulation of synaptic function and structure have potential implications for recently reported Alk functions in human addiction, retention of spatial memory, cognitive dysfunction in neurofibromatosis and the pathogenesis of amyotrophic lateral sclerosis.

INTRODUCTION

Anaplastic lymphoma kinase (Alk) was identified as a human proto-oncogene (Morris, Kirstein et al. 1994) and subsequently implicated in neuroblastoma and non-small cell lung cancers (Azarova, Gautam et al.; Soda, Choi et al. 2007; Caren, Abel et al. 2008; George, Sanda et al. 2008; Janoueix-Lerosey, Lequin et al. 2008; Koivunen, Mermel et al. 2008; Mosse, Laudenslager et al. 2008). The normal function of mammalian Alk is not well established, but Alk is expresssed primarily in the nervous system (Iwahara, Fujimoto et al. 1997; Hurley, Clary et al. 2006; Vernersson, Khoo et al. 2006; Degoutin, Brunet-de Carvalho et al. 2009). In mice loss of Alk function alters behavioral responses to ethanol and cocaine (Lasek, Gesch et al. 2011; Lasek, Lim et al. 2011), enhances spatial memory and novel object recognition, and reduces anxiety and depression (Bilsland, Wheeldon et al. 2008; Weiss, Xue et al. 2011). Human Alk variants are associated with increased risk for spontaneous amyotrophic lateral sclerosis (ALS) (Dunckley, Huentelman et al. 2007). The cellular mechanisms of Alk function related to these phenotypes are not known.

Alk’s cellular functions have been best characterized in C. elegans and Drosophila. In C. elegans, SCD-2/Alk interacts with a FSN-1, an SCF-like ubiquitin ligase, to negatively regulate neuromuscular junction (NMJ) morphological differentiation during development (Liao, Hung et al. 2004). HEN-1/Jeb to SCD-2/Alk signaling also integrates with insulin and TGF-beta signaling to control a binary choice between dormant versus reproductive developmental programs in response to thermal and nutrient stress (Reiner, Ailion et al. 2008). In adult worms, HEN-1/Jeb and SCD-2/Alk are required non-developmentally for integration of conflicting sensory stimuli (Shinkai, Yamamoto et al.; Ishihara, Iino et al. 2002).

In Drosophila Jeb’s earliest function is to provide positional information during embryonic visceral muscle development (Loren, Scully et al. 2001; Weiss, Suyama et al. 2001; Englund, Loren et al. 2003; Lee, Norris et al. 2003). At later larval stages Alk signaling in neuroblasts insulates neurogenesis from nutrient stress (Cheng, Bailey et al.). During adult eye development photoreceptors secrete Jeb as an anterograde signal to Alk expressing neurons in the optic lamina to ensure correct synaptic targeting (Bazigou, Apitz et al. 2007). Alk has also been found to interact genetically with the tumor suppressor Neurofibromin-1 (Nf1) in late larval and adult Drosophila to negatively regulate both adult body size and associative olfactory learning (Gouzi, Moressis et al. 2011). Remarkably both body size and learning defects caused by dNf1 mutations can be rescued by Alk inhibition.

We previously reported Jeb and Alk localization at Drosophila synapses in a pattern supporting anterograde signaling (Rohrbough and Broadie 2011). jeb and alk null mutants show compromised locomotion and loss of patterned NMJ transmission activity but we were unable to define the underlying defects (Rohrbough and Broadie 2011). Here we have employed a variety of conditional genetic techniques to investigate the role of Jeb-Alk signaling in the larval NMJ. Using transgenic RNAi and partial loss of function, termperature-sensitive mutants, we find that neurotransmission is negatively regulated by anterograde Jeb-Alk signaling. Using a tissue mosaic approach and two gain of function strategies, we find that Jeb regulates synaptic architecture. We also show that Jeb-Alk signaling drives phosphorylated ERK both pre- and postsynaptically. Taken together, these data reveal Jeb-Alk as a signaling pathway that regulates neurotransmission and synaptic architecture.

METHODS

Drosophila Stocks

The jeb^ko5644 and alk¹ null mutations have been fully described by us and others (Loren, Scully et al. 2001; Weiss, Suyama et al. 2001), and were analyzed in our recent study (Rohrbough and Broadie 2011). Homozygous mutant genotypes were determined by lack of GFP-marked balancers. To obtain MARCM mosaic larvae with GFP labeled jeb^ko5644 mutant motor neurons, the yw, C155-GAL4, UAS-mCD8GFP, hsFlp; FRTG13,GAL 80 line was crossed to yw; G13 jeb^ko5644/CyO. Eggs were collected after laying overnight on apple juice-agar plates and heat-shocked for 1 hr at 37°C. Animals were then incubated at 25°C until wandering third instar stage. Larvae screened by epifluorescence for GFP expression were dissected, fixed and imaged using confocal microscopy. For overexpression of Jeb, Alk or constituitively activated Alk (Act-Alk), UAS transgenes (Lee, Norris et al. 2003) were driven with neuron- and muscle-specific GAL4 lines (e.g. neuronal elav, C380, C57 and muscle 24B (Davis, Schuster et al. 1997; Koh, Popova et al. 1999)). For knockdown of Jeb and Alk in neurons or muscles, elav- (neuronal), 24B- (muscle) and sg30- (mesodermal (Azpiazu, Lawrence et al. 1996)) GAL4 lines were crossed to either a UAS-Jeb RNAi or a UAS-Alk RNAi lines, or a UAS-Alk^RPTP dominant negative construct (Yang, Eriksson et al. 2007). Transgenic RNAi constructs were derived from the pWIZ vector (Lee and Carthew 2003). An AvrII to NheI fragment consisting in 3141 bp of coding sequence from the dAlk cDNA was inserted into the pWIZ vector. The fragment was inserted into NheI and AvrII sites to direct hairpin RNA synthesis, producing snapback dsRNA when the UAS promoter is activated.

We identified a temperature-sensitive alk allele from a screen of single-nucleotide missense polymorphisms in the Alk kinase domain (Winkler, Schwabedissen et al. 2005). This alk^ts allele consists in a missense mutation that alters a highly conserved asparagine at position 1245 to aspartic acid. We confirmed the mutations by sequencing of the genomic DNA (data not shown). At 29°C alk^ts fails to complement alk¹ null lethality, and exhibits an alk mutant midgut phenotype. At 25°C, alk^ts exhibits limited adult viability, while alk^ts/alk¹ transheterozygotes exhibit larval lethality. Both alk^ts/alk¹ and alk^ts can be effectively maintained to 3^rd instar/pupal stages by initial rearing 18°C for 48–72 hrs from egg laying, followed by subsequent rearing at 25°C to the 3^rd instar stage. The alk^ts fully complements alk¹at 18°C.

Immunocytochemistry

Wandering third instar larvae were dissected and then fixed in 4% paraformaldehyde for 20–30 mins, and processed according to standard immunohistochemical protocols. Affinity-purified rabbit and guinea pig anti-Jeb antisera have been previously described (Weiss, Suyama et al. 2001; Englund, Loren et al. 2003; Rohrbough and Broadie 2011). Specificity has been previously demonstrated for rabbit and guinea pig anti-Alk antisera (kindly provided by Dr. Ruth Palmer) (Loren, Scully et al. 2001; Loren, Englund et al. 2003; Rohrbough and Broadie 2011). Antisera were used at the following dilutions: anti-Jeb (guinea pig or rabbit), 1:1000; anti-Alk (guinea pig or rabbit), 1:1000; anti-horse radish peroxidase (HRP; rabbit; Sigma), 1:200; Cy5-conjugated anti-HRP (goat; Jackson Laboratories, West Grove, PA), 1:100; 1:8; anti-cysteine string protein (Csp; mouse, supplied by Doris Kretszchmar), 1:100; anti-synaptotagmin (dsyt2; rabbit; (Littleton, Bellen et al. 1993)), 1:1000; anti-bruchpilot (Brp; nc82; mouse (Wucherpfennig, Wilsch-Brauninger et al. 2003), supplied by Doris Kretszchmar), 1:100; anti-discs large (Dlg; MAb4f3, Developmental Studies Hybridoma Bank (DSHB)), 1:800; anti-Dlg (rabbit; pdz2, (Thomas, Ebitsch et al. 2000)), 1:40,000; anti-GluRIIC (rabbit, supplied by Aaron DiAntonio), 1:5,000; and anti-diphosphoERK (ppERK; mouse; Sigma; supplied by Andrea Page-McCaw), 1:1000. For GluRIIC staining, larvae were dissected in HL-3 saline (Stewart, Atwood et al. 1994) and fixed in 3.7% formaldehyde for 5 min (Marrus and DiAntonio 2004). Secondary antibodies Alexa Fluor-555, Fluor-546, Fluor-488 and Fluor 647 (Molecular Probes) were used at 1:500–1:1000.

Microscopy and morphometry

Epifluorescence imaging was done with a Zeiss Axioskop2. Confocal imaging was done with both a Bio-Rad 2100 attached to a Nikon Eclipse E800 microscope, and a Zeiss 510 Meta laser scanning confocal microscope. Individual neuromuscular junction (NMJ) images were collected by sectioning through and projecting the entire depth of the termini. Ventral nerve cord (VNC) images were collected in a series of 30–40 µm depth, and projecting 15–20 µm segments through the neuropil. Z-stack projections were merged using NIH ImageJ (http://rsb.info.nih.gov) or Zeiss LSM software. For NMJ morphological analyses, preparations were triple-labeled with anti-Syt, anti-HRP and anti-Dlg to image muscles 6/7 in segments A2 and A3. Low magnification (10x) DIC images were used to measure dorsal surface area of both muscles using the rectangle function of ImageJ, and NMJ bouton number was normalized to this surface area (Schuster, Davis et al. 1996). NMJ epifluorescence images were obtained using a 25× oil immersion lens and deconvolved using the iterative algorithm function of AxioVision 4.3 software, with NMJ area and bouton numbers measured using the rectangle and cross-hair functions of ImageJ. For quantification of Jeb staining intensity in control and Alk-RNAi NMJs staining was performed with guinea pig anti-Jeb and rabbit anti-HRP-Cy5. Secondary antibody to visualize anti-Jeb was congugated with Alexa 546. Serial sections were obtained on laser scanning microscope at 0.5 micron intervals through 1b NMJs on muscles 12 and 13 at 60× magnification. The Z-stack projections were merged with ImageJ. All measurements were done blind to genotype, and two-tailed t-tests with a 95% confidence interval were performed using Prism 4 (Graphpad Software, Inc.). For Figure 3,6,7 analyses, synaptic fluorescence intensities were quantified on projected confocal Z-stacks using ImageJ. NMJ terminals and VNC neuropil regions were outlined by thresholding fluorescence signals in selected synaptic domains. Anti-Jeb and anti-dpERK were quantified within HRP-defined synaptic regions as well as nonsynaptic muscle domains. Fluorescence was normalized to control values in the same experiments. Matched images for all genotypes in each experiment were exported to Adobe photoshop for display.

Electrophysiology

Two-electrode voltage-clamp (TEVC) recordings of excitatory junctional currents (EJCs) were made from muscle 6 (segment A3) in wandering third instar larvae at 18°C, as previously described (Rohrbough, Pinto et al. 1999; Rohrbough, Grotewiel et al. 2000). Briefly, larvae were dissected dorsally in recording saline containing (in mM): 128 NaCl, 2 KCl, 0.2 CaCl₂, 4 MgCl₂, 70 sucrose, 5 trehalose (pH 7.1), and secured flat to sylgard-coated cover slips with Vetbond histoacryl glue. Segmental nerves were cut near the VNC, and preparations transferred to fresh recording saline containing 0.5 mM or 0.3 mM CaCl₂. Voltage recording and current injection electrodes were filled with 3M KCl and 3:1 K Acetate:KCl (3M), respectively, and had resistances of 12–20 MΩ. EJCs were evoked by supratheshold stimulation of the segmental nerve (0.3–0.5 ms; Grass Instruments S-88 stimulator and SIU) sufficient to recruit both motor inputs, using a fire-polished glass suction electrode filled with recording saline. Recordings were made at a holding potential of −60 mV using an Axon Instruments Axoclamp 2B amplifier, Digidata computer interface, and PClamp 9 acquisition and analysis software. Records were filtered at 2 kHz and digitized at 10 kHz. Mean peak EJC amplitudes were determined from ten consecutive responses evoked at 0.2 Hz. In some animals, recordings were made from both A3 hemisegments, with EJC amplitude was averaged between hemisegments for each N=1. Intracellular voltage recordings of evoked excitatory junctional potentials (EJPs) were made in HL-3 saline (Stewart, Atwood et al. 1994) containing 1mM Ca²⁺. Mean amplitude was determined from 20–30 consecutive compound EJPs. Only muscles with a resting potential greater than −50 mV were used for analysis. Data are presented as mean ± S.E.M. Statistical comparisons were made using 1-way ANOVA and Dunnet multiple comparison test (Kalaidagraph). Presentation current traces were exported and plotted using Excel and Kaleidagraph software, and trace images displayed in Adobe Photoshop.

RESULTS

Presynaptic Jeb and postsynaptic Alk suggests anterograde signaling pathway

We have previously described the presynaptic localization of the secreted ligand, Jeb, and postsynaptic localization of the transmembrane receptor, Alk, at the NMJ, consistent with anterograde Jeb-to-Alk trans-synaptic signaling (Rohrbough and Broadie 2011). Both ligand and receptor are similarly strongly concentrated within the central synaptic neuropil from late embryonic stages, with overlapping patterns of expression. To lay the foundation for the study of this pathway during postembryonic synaptic development, we first thoroughly re-examined Jeb and Alk synaptic expression at the larval NMJ, which undergoes dramatic postembryonic growth and functional strengthening (Gramates and Budnik 1999; Koh, Gramates et al. 2000). We labeled NMJs with anti-sera against Jeb or Alk, co-labeling for a range of presynaptic and postsynaptic proteins. The specificity of the anti-Jeb and anti-Alk antisera have been demonstrated in prior publications and confirmed here in the third instar larval NMJ by experiemnts that correlate staining with overexperession and transgenic RNAi targeting against Jeb and Alk (Loren, Scully et al. 2001; Weiss, Suyama et al. 2001; Loren, Englund et al. 2003). Jeb colocalizes in the presynaptic compartment with vesicle markers, such as Cysteine string protein (Csp) and Synaptotagmin (Syt). This presynaptic localization is confined within the neuronal membrane defined by anti-horse radish peroxidase (HRP; Fig. 1A, C). Jeb distribution within presynaptic boutons is strikingly punctate, consistent with concentration in secretory vesicles (Fig. 1C, F). Secreted Jeb is also detected in the extracellular space beyond the HRP-defined neuronal plasma membrane, as previously shown in non-permeablized preparations (Fig. 1C) (Rohrbough and Broadie 2011). Extraneuronal Jeb may also derive from Alk-mediated endocytosis in muscle (Weiss, Suyama et al. 2001). Jeb is present at all morphological classes of synaptic boutons (Johansen, Halpern et al. 1989; Hoang and Chiba 2001). In type Ib (big) and type Is (small) boutons, the major excitatory glutamatergic terminals, Jeb is more highly concentrated at Is boutons (Fig. 1A,F). Jeb is also strongly concentrated in the smaller, non-glutamatergic type II and type III boutons, which have neuromodulatory roles (Monastirioti, Gorczyca et al. 1995).

Fig. 1. Jeb secreted from presynaptic boutons and Alk concentrated postsynaptically.

Representative images of Jeb and Alk labeling at the wandering third instar muscle 6/7 NMJ. A, The presynaptic marker vesicle-associated anti-Csp (green; left panel) with anti-Jeb (red). Right: Anti-Jeb labeling is shown alone. Jeb is localized to large (1b) boutons and more strongly to smaller (1s) boutons, co-localized with Csp. B, Presynaptic anti-Csp (green; left panel) with anti-Alk (red) and; Right: Anti-Alk labeling is shown alone. Alk is localized at both 1b and 1s boutons (arrows), but more more strongly localized to 1b boutons. Alk is concentrated in the postsynaptic compartment surrounding Csp labeling. C, Single sections of individual synaptic boutons. Jeb punctae (red) are contained within presynaptic boutons labeled for synaptic vesicle protein anti-Syt (green) and neuronal membrane marker anti-HRP (blue). Jeb labeling beyond bouton margins is secreted extracellular protein. D, Z-series projections of individual boutons labeled for Alk (red) and presynaptic Csp (green). Alk surrounds Csp with concentration in the muscle postsynaptic SSR. E, Double-labeling for the predominantly SSR protein Dlg (green) and Alk (red) shows colocalization at 1b and 1s bouton subtypes (arrows), with strongest expression at 1b boutons which extensive SSR. Inset: Higher-magnification single section of one type 1b bouton; Alk strongly overlaps with Dlg, with slightly broader peripheral localization. F, Triple labeling for anti-Jeb (green), Alk (red) and HRP (blue). Jeb is largely localized within presynaptic boutons (arrows) labeled by HRP. Inset: Higher-magnification single section through a type 1b bouton. Jeb puncta are clearly revealed within the HRP-labeled bouton membrane, with Alk concentrated in the surrounding postsynaptic domain. Scale: 10 µm (A, B, E, F), 5 µm (C–D, and insets in E–F).

Alk is postsynaptically localized at both subclasses of type I boutons. These are characterized by extensive postsynaptic subsynaptic reticulum (SSR; (Atwood, Govind et al. 1993; Jia, Gorczyca et al. 1993)), which harbors the postsynaptic scaffold Discs Large (Dlg) (Lahey, Gorczyca et al. 1994) and glutamate receptors (GluRs) (Fig. 1B,E). Alk expression broadly encircles the anti-HRP labeled presynaptic membrane and anti-Csp labeled presynaptic boutons (Fig. 1B,D,F). Alk extensively overlaps with Dlg in the SSR, but Alk displays a slightly broader expression in the SSR periphery, extending beyond Dlg (Fig. 1E). Alk expression is distinctly stronger in type Ib boutons, consistent with the more extensive type Ib SSR (Fig. 1E,F). Alk is not detectible at type II or III boutons, which do not have associated SSR (data not shown) (Atwood, Govind et al. 1993; Jia, Gorczyca et al. 1993). It is possible that Alk is present below detection levels at these modulatory terminals, or that a second unknown Jeb receptor exists at these sites, or that Jeb secretion from these terminals activates Alk elsewhere on the muscle. We note that Alk labeling is present at much lower levels throughout the muscle, and thus is not exclusively localized to postsynaptic domains. Based on the localization of Jeb to presynaptic boutons and Alk to postsynaptic muscles, we propose that Jeb signals to Alk primarily in an anterograde trans-synaptic fashion at the NMJ, as it does in the developing visual system (Bazigou, Apitz et al. 2007).

Alk signaling in muscle negatively regulates NMJ synaptic transmission

Our previous studies revealed severe larval locomotion and endogenous neurotransmission defects in jeb and alk null mutants, as well as in animals with targeted neuronal or muscle Alk overexpression (Rohrbough and Broadie 2011). Surviving adult flies that overexpress Jeb or Alk are weak and uncoordinated, display abnormally elevated wing positions and typically live no more than several days after eclosion. To assess the role of Jeb-Alk pathway inhibition and activation in functional synaptic transmission, we recorded evoked excitatory junctional currents (EJC) and excitatory junctional potentials (EJP) at the larval NMJ.

We first examined the consequence of Alk hyperactivation by targeted Jeb expression 1) ectopically in muscle; 2) specifically in motor neurons and; 3) pan-neuronally (Figs. 2&3). Ectopic Jeb expression in muscle (24B-GAL4>UAS-Jeb) significantly reduced EJC amplitudes by 30% compared to controls (control: 176 ± 17 nA; 24B>Jeb: 123 ± 12 nA; P=0.04; Fig. 2B). The same result is obtained when EJP is measured (Fig. 3). We find a 42% reduction in evoked EJP amplitude when Alk is activated by ectopic expression of Jeb in muscle (control: 31.71 ± 2.4 mV; 24B-GAL4>UAS-Jeb: 18.27 ± 1.2 mV; P< 0.001). Similary motor neuron specific overexpression of Jeb reduced evoked EJP amplitude by 13% (control: 37.85 ± 1.3 mV; c380-GAL4>Jeb: 33.0 ± 1.5 mV; P=0.02) (Fig. 3). In contrast, mean EJC amplitude was unchanged by pan-neuronal Jeb overexpression (elav-GAL4>UAS Jeb) (elav/+ control: 168 ± 10 nA; elav>Jeb: 168 ± 17 nA; Fig. 2B). We hypothesize that neuronal Jeb secretion is regulated to account for the relatively modest phenotype associated with elevated expression of Jeb specifically in motor neurons compared to ectopic expression in muscles. Augmented expression of Jeb throughout the nervous system with elav-GAL4 is predicted to have effects both centrally and peripherally which are hard to sum. The level and timing of augmented Jeb experssion under the control of elav-GAL4 is also hard to compare to either 24B-GAL4 or C380-GAL4. Ectopic expression of Jeb in muscle is not regulated and is, therefore, a potent autocrine activator of Alk as shown previously in the embryonic visceral mesoderm (Weiss, Suyama et al. 2001). These results correlate well with analysis of Ras/MAPK/ERK activation as described below.

Fig 2. Jeb-Alk signaling regulates evoked synaptic transmission strength.

Evoked excitatory junctional current (EJC) records with five consecutive responses shown for each example genotype. A, Representative EJCs with Jeb overexpression in neurons (elav>Jeb) or muscle (24B>Jeb) (0.5 mM external Ca²⁺). B, Quantified EJC amplitudes normalized to elav/+ and 24B/+ controls. Ectopic muscle Jeb expression reduces transmission strength by 30% (P=0.04 compared to elav/+ control; 8–10 animals/genotype). C, EJCs recorded in temperature-sensitive Alk loss-of-function alleles (alk^ts, alk^ts/alk¹ and alk^ts/f01491), compared to wildtype control (left; 0.3 mM external Ca²⁺). Right-most EJCs were recorded from alk^ts/f01491 with muscle-targeted Alk expression (24B>alk^ts/f01491). D, Quantified normalized transmission amplitude is elevated by 79% in alk^ts, 210% in alk^ts/alk¹ and 250% in alk^ts/f01491 (P<0.002 for each allele vs. control). Transmission level is significantly rescued in alk^ts/f01491 by 24B-targeted Alk (P<0.04 vs. alk^ts/f01491; P>0.19 vs. control). N: 5–9 animals per genotype.

Figure 3. Reduced Alk synaptic expression in Alk loss-of-function hypomorphs.

A–B: Confocal images of larval NMJ 6/7 co-labeled with anti-HRP (left panels) and anti-Alk (right panels). A, sg30-GAL4/+ control. B, Larva with mesoderm-targeted UAS AlkRNAi (sg30>Alk RNAi), showing reduction in Alk muscle immunofluorescence (right), but normal NMJ size and elaboration (left). C, Quantified bouton density (NMJ area/muscle area) at NMJ 6/7 is not altered by expression of Alk RNAi in postsynaptic muscle. D, Evoked excitatory junction potentials (EJPs) at muscle 6 in sg30 control and sg30>Alk RNAi larvae, showing increased EJP amplitude for sg30>Alk RNAi. E, Quantified mean EJP amplitude is increased by 20% for sg30>Alk RNAi (**; P<0.003 vs. control), indicating increased NMJ transmission. F–I Larval NMJ 6/7 stained with anti-Alk in wild-type (C), alk^ts/alk hypomorph (D), alk^ts/f01491 hypomorph (E), alk^ts/f01491 with muscle-targeted UAS Alk expression (24B>alk^ts/f01491). Synaptic Alk expression is markedly reduced in LOF hypomorphs, both of which show altered NMJ transmission (Fig. 2). 24B GAL4-targeted Alk expression in the alk^ts/f01491 hypomorph rescues synaptic Alk expression in addition to significantly rescuing NMJ transmission level (Fig. 2). Images in A-B and F-I were acquired in separate experiments and imaged using BioRad 2100 and Zeiss 510 confocal microsopes, and processed using Axiovision and LSM software, respectively. Scale: 10 µm. J–L Larval NMJ 6/7 stained with anti-Jeb in sg30 control (J) and sg30>Alk RNAi (K). Inhibition of Alk expression increases Jeb accumulation in the NMJ. Quantified Jeb immunofluorescence normalized to HRP in sg30 control and sg30>Alk RNAi (L) demonstrates 2.1 fold increase in Jeb accumulation (*;P<0.04).

Concordantly Alk inhibtion either genetically or by transgenic RNAi enhances neuromuscular transmission. When Alk is inhibited by tissue specific transgenic RNAi driven by the mesodermal sg30-GAL4, ~1/3 of newly-hatched larvae display gut defects characteristic of null alk mutants, while viable third-instar larvae exhibit a pronounced, 45 % reduction in NMJ and muscle anti-Alk labeling intensity (sg30>Alk RNAi; Fig. 3A, B). NMJ terminal size and structure appeared similar to controls, showing neither disrupted bouton morphology, nor altered NMJ architecture (Fig. 3C). However, functional evoked EJP recordings demonstrated a significant, 20% increase in mean EJP amplitude for sg30>Alk RNAi (control: 31.46 ± 1.1 mV; sg30>Alk RNAi: 36.57 ± 1.7 mV; P=0.02; Fig. 3D). This demonstrates strengthened functional transmission as a consequence of reduced postsynaptic Alk activity.

To test the effect of reduced Alk function on synaptic transmission more rigorously we identified a temperature-sensitive alk allele from a screen of single-nucleotide missense polymorphisms in the Alk kinase domain (see Methods; (Winkler, Schwabedissen et al. 2005)). This alk^ts allele alters a highly conserved asparagine at position 1245 to aspartic acid, which we confirmed by sequencing of the genomic DNA (data not shown). At the 29°C restrictive temperature, alk^ts fails to complement alk¹ null lethality, and likewise exhibits alk null phenotypic gut defects. At 25°C, alk^ts exhibits limited adult viability, while alk^ts/alk¹ transheterozygotes exhibit larval lethality, but both genotypes can be effectively maintained to 3^rd instar/pupal stages by initial rearing 18°C for 48–72 hrs from egg laying, followed by subsequent rearing at 25°C to the 3^rd instar stage. EJC recordings were made in reduced (0.3 mM) external Ca²⁺ to more clearly reveal a predicted elevation in NMJ transmission. Consistent with this prediction, EJC amplitude was strongly elevated by 79% in alk^ts mutants, and increased by >2-fold in alk^ts/alk¹ transheterozygotes, compared to wildtype controls (control: 40 ± 5 nA; alk^ts: 72 ± 5 nA; alk^ts/alk: 85 ± 10 nA; P<0.002 vs. control for both alk^ts and alk^ts/alk; Fig. 2 C,D). Similar results were obtained for measured EJP (data not shown).

To confirm that Alk functions in muscle to regulate synaptic transmission we took advantage of the recently reported alk^f01491 allele, which contains a P-element insertion in the 5’ UTR. Alk levels are reduced by ~70% in adult heterozygotes (Lasek, Lim et al. 2011). This insert also contains UAS sites for GAL4-driven Alk expression and rescue of the mutation. We therefore compared alk^ts/f01491 transheterozygotes to alk^ts/f01491 with muscle-targeted Alk expression (24B-GAL4> alk^ts/f01491) (Fig. 2D), rearing animals for 48 hrs at 18°C and the subsequently shifting to 25°C through the wandering third instar stage. Synaptic transmission in alk^ts/f01491 animals displayed strongly elevated (2.4-fold) mean EJC amplitudes compared to controls, which is a larger increase than alk^ts/alk¹ (99 ± 18 nA; P<0.002 vs. control; Fig. 2C,D). Muscle-specific Alk expression rescued the alk^ts/f01491 transmission to near wildtype levels (52 ± 6 nA; P>0.19 vs. control; Fig. 2C, far right and 2D). We confirmed that Alk NMJ synaptic expression is substantially reduced in alk^ts/alk1 and alk^ts/f01491 mutant larvae (Fig. 3F–H), consistent with a stronger hypomorphic LOF condition. Alk NMJ synaptic expression is robustly rescued by GAL4 driven muscle-targeted Alk (Fig. 3I), in parallel with the rescue of transmission amplitude. The results clearly demonstrate that Alk is required in the muscle to limit neurotransmission strength.

Reduced Alk expresssion in the NMJ results in increased Jeb accumulaiton

Though Jeb and Alk as a ligand-receptor pair function together in multiple contexts in Drosophila, including early mesoderm development, axon targeting in the visual system and neurogenesis, it is possible that, in spite of their colocalization at the NMJ, they are not functionally paired in this context. We sought to test the functional link between Jeb and Alk in the NMJ by evaluating the effect of reduced Alk expression in muscle on synaptic Jeb levels. To achieve this we specifically inhibited Alk expression by sg30>Alk RNAi in muscle as described above. Targeting of Alk by this method produces a 45% reduction in Alk expression (Fig. 3 A,B). We stained these NMJs with anti-Jeb and quantified the synaptic levels by normalization to HRP. As shown in Fig. 3 J–L Jeb accumulation in the NMJ is increased approximately twofold by inhibition of Alk expression in muscle (Jeb intensity normalized to HRP 2.1X greater in sg30>Alk RNAi, p=0.038). As above this result supports both an anterograde signaling model and conservation of the Jeb/Alk, ligand/receptor, functional interaction in the larval NMJ.

Loss of Jeb in motor neurons is associated with aberrant NMJ structure and postsynaptic assembly

Embryonic NMJs are structurally normal in jeb and alk null mutants (Rohrbough and Broadie 2011). In C. elegans, however, NMJ architecture is regulated by FSN-1 and Scd-2/Alk (Liao, Hung et al. 2004). We therefore sought to determine if the much more extensive growth and synaptic patterning that occurs during postembryonic stages of NMJ development is altered either by loss of Jeb function or by Alk activation. To circumvent the late embryonic lethality of jeb null mutations (Weiss, Suyama et al. 2001), we employed the Mosaic Analysis with Repressible Cell Marker (MARCM) system of somatic recombination (Lee and Luo 2001) to generate jeb null mutant motor neurons in a mosaic larva. The MARCM technique is subject to several limitations. One limitation is that it generates random mutant clones at a low frequency. This necessitates systematic screening to identify rare mutant motor neurons marked by GFP expression. Inadequate numbers of mutant motor neurons are generated for quantified analysis of any single NMJ class, or for functional investigations. A second limitation is that null mutant muscles cannot be generated, because mature muscles are multinucleate syncitia formed by the fusion of multiple myoblasts (Bate 1990). A third limitation is that one arm of a chromosome is rendered homozygous by the random recombination event. Linked second site mutations on that chromosome arm are difficult to exclude as possible causes for the observed phenotype. Fig. 4 illustrates a range of NMJ terminals formed by motor neurons in which Jeb function has been genetically removed by MARCM, and the jeb-null neurons labelled with a cell surface-targeted GFP. All jeb mutant motor neurons examined (N=18) display extensively disrupted NMJ architecture and bouton morphology compared to the corresponding wildtype control NMJs in contralateral or adjacent hemisegments of the same animal (Fig. 4 C,E,H,J,L).

Fig 4. Null jeb motor neurons display disrupted synaptic architecture and postsynaptic assembly at the NMJ.

Representative images of MARCM clone jeb null mutant NMJs compared to wildtype control terminals. A, Left: MARCM GFP expression in a jeb null mutant NMJ 12 terminal. Right: anti-HRP (green) and anti-Dlg (red) labeling of the same NMJ. jeb null terminal (arrow) has irregularly shaped and spaced boutons, and loss of anti-Dlg labeling, compared to the adjacent wildtype (WT) boutons (arrowhead) formed by the type 1b motor neuron input. B, Wildtype terminals on the contralateral NMJ 12 in the same animal, illustrating type 1b/1s boutons (image orientation reversed for comparison). C, GFP-expressing jeb null mutant NMJ 6/7 terminal labeled with anti-Dlg (red). Mutant terminals form irregularly shaped and spaced boutons (arrows) and extraneous processes (small arrows), and lack boutons at the distal ends of the terminal. Wildtype terminals from a second motor neuron are indicated (arrowhead). D, GFP-expressing jeb null NMJ 6/7 in another preparation labeled with anti-Dlg (red), showing irregular bouton formation proximally (large arrows) and distally (small arrows). E, Wildtype NMJ 6/7 terminal in a control preparation expressing GFP in all neurons and labeled with anti-Dlg. Regular morphologies of type 1s and 1b boutons are indicated. F, Low magnification image of MARCM jeb null NMJ 6/7, showing numerous thin branches that fail to form regular boutons. G, Highlighted area in (F) shown at higher magnification, labeled with anti-GluRIIC (green) and anti-Dlg (red). The jeb terminal shows thin branches with weak anti-Dlg bouton localization (large arrows), and abnormally clustered glutamate receptors along branches (small arrows). H, Wildtype NMJ 6/7 on the contralateral side of the same animal shows regular bouton morphology with concentrated Dlg expression (red) and distributed GluRIIC punctae (green) localized within Dlg domain. I, Low magnification MARCM GFP in a jeb null mutant NMJ 1 terminal. J, Highlighted region in (I) shown at higher magnification, labeled with anti-HRP (green) and anti-Brp (red). Brp-labeled active zones are abundant within irregularly shaped and spaced boutons (arrows). A wildtype terminal is also present (arrowheads). K, Wildtype NMJ 1 terminal in the adjacent anterior segment displays regular bouton profiles by comparison. L, Low magnification MARCM GFP in a jeb mutant NMJ 15 terminal. M, Highlighted region in (L) shown at higher magnification, labeled with presynaptic anti-Syt (green) and postsynaptic anti-Dlg (red). In the jeb terminal, Syt is localized within irregularly shaped and spaced boutons (arrow) along the muscle. Wild-type boutons on adjacent NMJ 16 are indicated (arrowheads). N, Wildtype NMJ 15 in the contralateral segment has normally shaped and spaced boutons. Scale: 50 µm in low-magnification panels (F,I,K); 10 µm in other panels.

Null jeb mutant motor neuron terminals lack well-defined presynaptic boutons, and also display severely disrupted postsynaptic differentiation. Discrete bouton structures are poorly recognizable, preventing identification of bouton class on the basis of the hallmark architectural differences defining specific NMJ terminals. In the instances where neuron identity can be inferred based on co-inervation and stereotypical patterns we observe strong structural disruptions. For example, the jeb null terminal shown at NMJ 12 (Fig. 4A) contains malformed boutons that could be type Is, II or III, all of which are normally present. A wildtype type 1b terminal at the same NMJ is readily identifiable (WT, arrowhead; Fig. 4A). By default this would identify the mutant neuron as, most likely, 1s. The wildtype, contralateral 1s neuron shown in Fig. 4B by contrast has well formed boutons. In Fig. 4C, a type 1b wildtype terminal is present at NMJ 6/7 (arrowhead), by which the jeb mutant terminal at the same NMJ is identified as type 1s. However, the morphological irregularity of the mutant terminal structure prevents the definition of discrete boutons. In Fig. 4D we judge the mutant neuron to be 1b based on the absence of other detectable neurons with associated Dlg staining. Muscles 6 and 7 are typically innervated by a branched 1b neuron (Hoang and Chiba 2001). Compared to the 1b wildtype neuron in Fig. 4E, again, bouton structure is aberrant. Similarly the Jeb mutant neuron shown in Fig. 4F,G is likley to be 1b, dually innervating muscles 6 and 7, based on no detectable GluRIIC associated with another, wild type neuron on this muscle. We therefore provide a 1b NMJ for comparison. Though this identification is not beyond quesion the very disorganized pre- and postsynaptic structure is very abnormal for either 1b or 1s neurons. Additional examples of jeb null NMJ terminals are illustrated with poorly defined boutons, irregular proportions and abnormal bouton spacing along the presynaptic axonal branches (Fig. 4 F,G,I,J,L,M). In each mosaic, wildtype terminals on the same muscle and control hemisegments appear morphologically normal, confirming a cell-autonomous Jeb requirement to construct normal NMJ architecture.

In parallel, we examined the molecular differentiation of the jeb null NMJ synapses. Postsynaptically, the membrane associated scaffold Dlg is disorganized or absent in the absence of Jeb function (Fig. 4A–H). Associated with most jeb null motor neurons, Dlg fails to form a well-defined concentrated postsynaptic domain encircling boutons, with Dlg completely undetectable at some mutant boutons (Figs. 4A,G). In wildtype NMJs on the same muscle, Dlg intensity and localization is consistently strong, with a postsynaptic halo of expression completely surrounding the presynaptic boutons (Fig. 4A–E,H). In contrast, concentrated glutamate receptor (GluR) puncta are found at jeb null mutant terminals, but GluR puncta appear spatially disorganized and loosely localized at jeb terminal branches and boutons (Fig. 4G vs. 4H). At wildtype NMJs, on the same muscle or in contralateral control segments, GluRs are consistently and tightly localized to punctate domains opposing and surrounding presynaptic boutons (Marrus and DiAntonio 2004) (Fig. 4H). Presynaptically, we examined the active zone component Bruchpilot (Brp/nc82) (Wucherpfennig, Wilsch-Brauninger et al. 2003) and the synaptic vesicle protein Synaptotagmin (Littleton, Bellen et al. 1993) (Fig. 4I–N). In jeb null mutant terminals, clusters of Brp-marked presynaptic active zones appear correctly localized, albeit within morphologically abnormal synaptic varicosities (Fig. 4J). Synaptotagmin also appears correctly localized within jeb null boutons/varicosities (Fig. 4M). Presynaptic active zones and synaptic vesicles are appropriately localized, despite abnormal bouton shape, size and spacing, in jeb mutant neurons. In constrast, loss of presynaptic Jeb clearly and strongly impacts postsynaptic molecular assembly, consistent with an anterograde Jeb trans-synaptic signaling function.

Constitutive Jeb-Alk signaling in the muscle alters NMJ growth and morphology

To complement the above presynaptic jeb loss-of-function studies, we next analyzed structural consequences of postsynaptic Alk activation. Hyperactivation of muscle Alk signaling results in the compaction of NMJ terminals into a smaller muscle area (Fig. 5). Similar results were obtained by muscle expression of truncated, constitutively activate Alk (Lee, Norris et al. 2003) (Fig. 5A–C) and by ectopically expressing jeb in muscle, where it can activate Alk signaling in an autocrine fashion as it does in the developing visceral mesoderm (Weiss, Suyama et al. 2001) (Fig. 5D–F). When Jeb is ectopically expressed in muscle NMJ terminals have normal numbers of boutons, but are less extensively elaborated and reduced in size, with boutons densely packed near the secondary branch point of the nerve axon (Fig. 5A,C,D,F). Quantification of bouton number and distribution compared to GAL4 controls confirmed that Alk hyperactivation resulted in signficantly smaller NMJ areas (P<0.0003; Fig. 5H), with resulting increased bouton density (P<0.0001; Fig. 5G). Expression of activated Alk in muscle has a qualitatively similar structural effect.

Fig 5. Jeb overexpression and Alk receptor activation compacts NMJ bouton distribution.

Representative images of wandering third instar muscle 6/7 NMJs co-labeled with anti-HRP (red) and anti-Dlg (green). A, C57>UAS-Activated (Act) Alk. B, C57-GAL4/+ alone control. C, Terminals in A and B shown in their entirety at lower magnification, showing compacted NMJ morphology of c57>Act Alk terminal. D, Muscle Jeb overexpression with 24B>UAS-Jeb. E, 24B-GAL4/+ control. F, Terminals in D and E shown in their entirety, showing compacted NMJ morphology of 24B>Jeb terminal. Scale: 20 µm (A, B, D, E), 50 µm (C, F). G–I. Quantification of bouton number and distribution in the above genotypes. Horizontal lines indicate the median; boxed areas indicate 25^th–75^th percentiles; whiskers indicate maximum and minimum values. Jeb overexpression increases bouton number (p<0.0001) and reduces NMJ area (p<0.0003; asterisks). Total bouton number normalized to muscle surface area does not differ significantly from controls (p = 0.38).

Increased presynaptic Jeb secretion would be expected to hyperactivate muscle Alk via anterograde synaptic signaling. To test whether this prediction, we overexpressed Jeb in motor neurons (c380-GAL4>UAS-Jeb) or pan-neuronally (elav-GAL4>UAS-Jeb). Jeb neuronal overexpression does not appear to produce smaller NMJs with more densely compacted boutons, as seen for direct Alk activation in muscle. Anti-Jeb labeling shows Jeb to be markedly elevated throughout NMJ terminals, and highly concentrated in boutons (Fig. 6B,D and data not shown). We did not, however, detect increased Jeb in the bouton perisynaptic region or nonsynaptic muscle areas region (Fig. 6D, and data not shown). The lack of a similar compaction of NMJ structural growth associated with neuronal Jeb overexpression suggests that postsynaptic Alk activation is not as strongly induced by neuronal overexpression compared to ectopic expression. This correlates with the comparatively modest neurotransmission phenotypes observed in the same experimental contexts. As noted above we hypothesize that Jeb secretion in the NMJ is physiological regulated.

Fig 6. Jeb-Alk anterograde signaling activates synaptic Ras/MAPK/ERK cascade.

A, Activated diphospho-ERK (dpERK, red) and Jeb (Green) labeling in the ventral nerve cord (15 µm projections). dpERK is enriched within the synaptic neuropil (np) of normal control larvae (left). ERK signaling is clearly elevated by Alk pathway hyperactivation by neuronal Jeb overexpression (elav>Jeb; middle). dpERK is depressed by Alk pathway inhibition by neuronal Jeb LOF (elav>Jeb RNAi; right). See text for quantification. B, dpERK and Jeb expression at NMJs (large arrowheads) and surrounding musclature (muscles 7, 6, 13, 12). In controls (24B/+; left), activated ERK is evident in the presynaptic nerve (n), and weakly expressed within NMJ terminals and in postsynaptic muscle (m). Presynaptic Jeb overexpression (elav>Jeb; middle) greatly elevates dpERK within the presynaptic nerve (n) and throughout NMJ terminals (arrowheads), as well as throughout the postsynaptic muscle, where dpERK is observed concentrated near muscle nuclei. Postsynaptic Alk inhibition (24B>Alk RNAi/alk; right) slightly depresses muscle ERK levels. C, Magnified panels showing anti-dpERK in postsynaptic muscle 6 (m6) adjacent to the NMJ, for the respective genotypes shown in (B). Blue: HRP. N: nucleus. Muscle dpERK is clearly elevated by presynaptic Jeb expression (middle panel). D, dpERK (red) at individual HRP-labeled NMJs (blue) and in postsynaptic muscle (muscle 4) in control, elav>Jeb, and 24B>Alk RNAi/alk larvae. Smaller panels to right show boutons (arrows) at 2X magnification, with anti-Jeb signal (green) included in lower panel. In control, dpERK is concentrated at higher levels within presynaptic nerve axon (n), and localized to some synaptic boutons, and distributed throughout the muscle (left). Presynaptic Jeb overexpression (elav>Jeb) dramatically elevates dpERK both at NMJ boutons and throughout surrounding muscle. Compared to control, postsynaptic Alk inhibition (right) has little effect on neuronal and NMJ dpERK, but depresses muscle ERK activity. Scale: 20 µm (A, B), 10 µm (C, D), 5 µm (D, detailed panels).

Anterograde Jeb-Alk trans-synaptic signaling activates the Ras/MAPK/ERK pathway

Jeb-Alk intercellular signaling in the embryonic mesoderm activates the Ras/MAPK/ERK pathway (Englund, Loren et al. 2003; Lee, Norris et al. 2003). We therefore hypothesized that anterograde Jeb-Alk trans-synaptic signaling likewise drives Ras/MAPK/ERK pathway activation at the NMJ. To test this hypothesis, we assessed MAPK activity in response to both activated or inhibited Jeb-Alk signaling, using anti-diphospho(dp)-ERK as the MAPK pathway reporter (Loren, Scully et al. 2001; Lee, Norris et al. 2003; Tsai, Kao et al. 2008; Wairkar, Toda et al. 2009). We first assessed dpERK levels in the nervous system within the ventral nerve cord (VNC) synaptic neuropil (Fig. 6A). The neuropil shows significant levels of dpERK signaling in control (elav/+) larvae, representing a basal level of Ras/MAPK/ERK pathway activation. Central Alk hyperactivation by neuronally targeted Jeb expression (elav>Jeb; Fig 6A, center) strongly increases Ras/MAPK/ERK activation throughout the VNC, and notably within the synaptic neuropil (97% increase in dpERK intensity over control; P<0.0001; Fig. 7E). In contrast, Jeb signaling inhibition by neuronally-targeted Jeb RNAi (elav>Jeb RNAi, Fig. 6A, right) effectively depresses the level of activated synaptic Ras/MAPK/ERK signaling (50% of control dpERK intensity; P<0.01; Fig. 7E).

Figure 7. Elevated Ras/MAPK/ERK signaling by hyperactivated Jeb/Alk synaptic signaling.

A, Ventral muscles 7, 6, 13, 12 and NMJs (arrowheads) in control larva stained for activated ERK (dpERK, red) and HRP (blue). dpERK is detected in nerve axons and NMJ terminals. Boxed NMJ area is shown at right with merged HRP, dpERK, and Jeb (green) labeling (top), Jeb alone (middle), and dpERK alone (lower). B, Ventral muscles 7, 6, 13, 12 and NMJs (arrowheads) in larva with muscle-targeted Jeb overexpression (24B>Jeb), stained for dpERK, HRP and Jeb as in (A). Panels to right illustrate merged (top), Jeb (middle), and dpERK labeling. dpERK is elevated throughout the muscle, and dramatically concentrated around muscle nuclei in conjunction with Jeb. C, Control NMJ muscle 4 stained for dpERK, HRP, and Jeb (left panel). Right panel shows dpERK labeling visible at terminal and muscle nucleus (N; arrow). Arrowhead marks proximal NMJ bouton. D, NMJ muscle 4 in 24B>Jeb overexpression larva stained for dpERK, HRP, and Jeb as in (C). Arrowhead marks proximal NMJ bouton. Arrows indicate muscle nuclei, showing greatly increased dpERK levels. Scale: 20 µm (A, B), 10 µm (C, D). E, Quantified activated ERK fluorescence intensity within the ventral nerve cord synaptic neuropil for neuronal Jeb overexpression (elav>Jeb) and neuronal LOF (elav>Jeb RNAi) conditions vs. controls. F, Quantified activated ERK fluorescence intensity at NMJs (black bars) and in nonsynaptic muscle regions (gray bars). Presynaptic Jeb elevation dramatically increases activated ERK at NMJs (>3X), and significantly elevates (1.7X) activated ERK throughout postsynaptic muscle, consistent with neuronal Jeb activation of Alk receptors in muscle. Ectopic exrpession of Jeb in muscle dramatically activates ERK within the muscle. Alk inhibition by neuronal Jeb RNAi or muscle Alk RNAi significantly depresses ERK as indicated (* P<0.05. ** P<0.01, *** P <0.001).

We next examined how Jeb-Alk signaling activity regulates the Ras/MAPK/ERK pathway at the NMJ. In control animals, dpERK activity is detected in peripheral segmental nerve branches, as well as within presynaptic terminal axon branches (Fig. 6B,D). As previously reported (Tsai, Kao et al. 2008; Wairkar, Toda et al. 2009), punctate dpERK is clearly evident in a subset of presynaptic boutons, with scattered dpERK activity also present throughout the postsynaptic muscle (Fig. 6C,D). Neuronally targeted Jeb dramatically elevates (>4.5-fold) dpERK signaling at the NMJ (Fig. 6B,D; center panels). Within the NMJ domains dpERK labeling appears either within, or closely associated with, the presynaptic terminal. However, presynaptically augmented Jeb expression also results in an increase in dpERK activity throughout the postsynaptic muscle, notably in the vicinity of muscle nuclei (72% increase in dpERK intensity over control; P<0.0001; Fig. 6B–D, Fig. 7F), This result shows that presynaptically secreted Jeb activates postsynaptic Alk and Ras/MAPK/ERK pathway signaling. Moreover, inhibition of NMJ Jeb/Alk signaling by presynaptically targeted Jeb RNAi or postsynaptically targeted Alk RNAi depresses dpERK intensity. Presynaptic Jeb RNAi significantly depresses NMJ synaptic dpERK (72% of control; P<0.0001; Fig. 7F), while postsynaptic Alk reduction significantly depresses dpERK both at synaptic boutons (80% of control; P<0.003), as well as throughout the nonsynaptic muscle, relative to control NMJs (82% of control; P<0.02; Fig. 7F). Presumably because the muscle expresses a low basal level of activated ERK, Alk inhibition produces only modest (~20%) further decreases in dpERK. In contrast, activated Alk signaling has a much stronger positive effect on dpERK detected at the NMJ. These results support the model that postsynaptic ERK activation is positively regulated by presynaptic Jeb secretion with nerve-to-muscle Jeb-Alk anterograde signaling. Finally, we examined the downstream Ras/MAPK/ERK pathway response to ectopic Jeb overexpression in the muscle (24B-GAL4>UAS-Jeb). Under this condition, Jeb is dramatically elevated throughoutout the larval musclature, most strongly concentrated around muscle nuclei (24B>Jeb; Fig. 7B,D). dpERK is likewise dramatically upregulated in muscle, with a similar pattern of highly concentrated expression surrounding muscle nuclei (Fig. 7B,D). Quantified dpERK fluorescence is elevated 2.5-fold at NMJ terminals. and >3-fold in non-synaptic muscle regions (P<0.0001 vs. controls; Fig. 7F). Ras/MAPK/ERK activation in the muscle therefore is strongly correlated to the ectopic expression of the activating Jeb ligand. With respect to activation of the Ras/MAPK/Erk pathway in muscle we note that overexpression of Jeb in neurons, though active, is not nearly as potent as ectopic expression of Jeb in muscle. The increased activation by neuronal overexpression is less than 2-fold (72%) while ectopic expression of Jeb in muscle produces a greater than 3-fold increase in dpERK. The relative efficacy of these two methods for activating Alk corresponds to the morphologic and electrophysiologic phenotypes produce by these two methods.

In summary, Jeb-Alk signaling positively regulates Ras/MAPK/ERK pathway activation in both the central nervous system synaptic neuropil and at NMJ synapses. Increased anterograde, trans-synaptic Jeb-Alk signaling (elav>Jeb) elevates dpERK within and surrounding NMJ terminals. With ectopic muscle Jeb expression (24B>Jeb), dpERK levels and distribution are dramatically altered, with >3-fold higher overall muscle levels, and strong dpERK perinuclear accumulation. We propose that Ras/MAPK/ERK hyperactivation is responsible for both the architectural and transmission synaptic phenotypes reported here.

DISCUSSION

Jeb-Alk is an anterograde signaling pathway at the NMJ

Our results support an anterograde signaling model in which presynaptically secreted Jeb activates postsynaptic Alk. The data to support this hypothesis derives from multiple tests. First, immunolabeling shows Jeb is concentrated within presynaptic boutons, while Alk is present in the surrounding postsynaptic SSR (Rohrbough and Broadie 2011). Second, targeted postsynaptic Alk expression in Alk LOF mutants is sufficient to rescue synaptic transmission defects, a strong demonstration that Alk is required in the postsynaptic muscle to regulate neurotransmission. Third, post-synaptic inhibtion of Alk by tissue specific RNAi results in 2-fold increased accumulation of perisynaptic Jeb. Fourth, the MARCM clonal approach demonstrates Jeb may be required within presynaptic motor neurons to regulate postsynaptic molecular assembly. Fifth, elevated presynaptic Jeb expression activates postsynaptic Ras/MAPK/ERK activation, while inhibition of postsynaptic Alk reduces Ras/MAPK/ERK activitation.

Jeb-Alk signaling regulates postembryonic NMJ neurotransmission strength

In structurally normal NMJs we find strong effects on neurotransmission as a consequence of perturbations in Jeb-Alk signaling. Our clearest, most consistent results derive from techniques that activate or inhibit Jeb-Alk signaling postsynaptically. Postsynaptic hyperactivation of Alk weakens NMJ synaptic transmission (Fig. 2). This functional phenotype parallels the negative regulation of synaptic growth by postsynaptic Alk activation. Consistent with the inhibitory effect of Alk activation on neurotransmission, we observe enhanced neurotransmission as a consequence of muscle specific reductions in Alk levels by transgenic RNAi (Fig. 3). Additional confirmation for Alk-dependent inhibition of neurotransmission is provided by analysis of a hypomorphic temperature sensitive allele of Alk. Partial loss of Alk function results in strongly increased NMJ neurotransmission (Fig. 2). The implication is that Alk activity limits or negatively regulates synaptic strength. We also show that muscle-specific Alk expression in the strongest alk^ts/alk^f01491 partial loss of function genotype rescues reduced neurotransmission to near wild-type levels, a conclusive demonstration that postsynaptic Alk function negatively regulates the strength of NMJ neurotransmission. This function is novel as, to our knowledge, Jeb-Alk transynaptic signaling is the only known negative regulator of synaptic transmission.

Presynaptic manipulation of Jeb yields less strong though still consistent results. Transmission is uneffected by increased pan-neuronal Jeb expression, though this activates Ras/MAPK/ERK both centrally and presynaptically at the NMJ and, to a lesser degree, within the postsynaptic muscle. Motor neuron electrical activity activates neuronal Ras/MAPK/ERK signaling, and this presynaptic Ras/MAPK/ERK activation is positively linked to both structural and functional NMJ synaptic remodeling (Koh, Ruiz-Canada et al. 2002; Hoeffer, Sanyal et al. 2003; Marrus and DiAntonio 2004; Freeman, Bowers et al. 2010). Motor neuron specific over expression of Jeb does produce a modest but statistically significant reduction in neuromuscular transmission. Ectopic expression of Jeb in muscle results in substantial inhibiton of neuromuscular transmission. One hypothesis that may account for the diffence between pan-neuronal and motor neuron or muscle specific manipulation of Jeb-Alk signalling is that the effects of manipulating pan-neuronal Jeb represent a composite of central and peripheral effects on the motor neuron. In first instar larvae we find that both jeb and alk mutants display impaired central output to motor neurons most consistent with a central synaptic defect (Rohrbough and Broadie 2011). The integrated physiologic function subserved by Jeb-Alk signaling in the NMJ, which has yet to be determined, will provide the essential context for interpretting these results. The novel inhibitory role of Jeb-Alk signaling in NMJ transmission implies that it is part of a transynaptic regulatory network that integrates neuronal activity and responses with other homeostatic mechanisms. We provide indirect evidence that Jeb secretion is regulated. The physiologic regulation of Jeb secretion is a critical missing component of understanding how Jeb-Alk signaling fits into the regulation of synaptic plasticity.

Jeb-Alk signaling regulates postembryonic NMJ synaptic growth and patterning

Jeb-Alk signaling is not required for embryonic NMJ synaptogenesis or differentiation, although jeb and alk null mutants display impaired locomotion and reduced NMJ transmission (Rohrbough and Broadie 2011). At later developmental stages, removing Jeb in motor neurons strongly disrupts late larval NMJ synaptic terminal architecture and bouton morphology. Postsynaptic Dlg scaffolding and GluR clustering are strongly perturbed in association with jeb mutant terminals. The mosaic analysis supports a cell-autonomous, anterograde signaling function for Jeb. One mechanistic hypothesis is that Jeb-Alk nerve-to-muscle signaling regulates NMJ morphogenesis by recruiting or regulating cell adhesion molecules (CAMs). In the developing adult visual system, anterograde Jeb-Alk signaling induces the expression of post-synaptic adhesion molecules Dumbfounded/Kirre, Roughest/IrreC and Flamingo to shape the optic neuropil target environment (Bazigou, Apitz et al. 2007). At the larval NMJ, adhesion molecules such as fasciclins and integrins regulate activity-dependent synaptic growth and structural remodeling (Schuster, Davis et al. 1996; Schuster, Davis et al. 1996; Beumer, Rohrbough et al. 1999; Beumer, Matthies et al. 2002; Ashley, Packard et al. 2005). Our results imply that Jeb-Alk signaling either directly regulates Dlg localization or indirectly drives Dlg-dependent postsynaptic differentiation. Dlg has demonstrated roles in NMJ morphogenesis and GluR expression and field regulation (Bachmann, Timmer et al. 2004; Chen and Featherstone 2005), and directly binds and regulates fasciclin II and βPS integrin (Beumer, Matthies et al. 2002). Future work will test the hypothesis that Jeb-Alk signaling organizes or regulates adhesion receptors and postsynaptic scaffolding to control bouton differentiation and shape functional synaptic architecture.

The structural phenotypes observed in the Jeb mutant NMJs produced by the MARCM technique have not been replicated by an alternative method so we wish to be clear about the limitations of the available data. One caveat is that the structural defects observed in the context of MARCM derived tissue mosaics could derive from a linked, second site mutations on the same chromosome arm. We also note that we have not observed these structural phenotypes in the context of partial loss of Alk function produced by either transgenic RNAi or a hypomorphic alk allele. The discrepancy between these observatons may also derive from technical limitations. As noted above the MARCM technique is not applicable to somatic muscles as these cells are multinucleate syncitia. Randomly induced recombination events are exceeding unlikely to generate Alk mutations in all of the nuclei in any one muscle. We cannot therefore employ the MARCM technique to produce Alk null mutant muscles to confirm our Jeb mosaic analysis. We have instead employed partial loss of Alk function produced by trangenic RNAi or a hypomorphic Alk allele to investigate Alk function in muscle. The strength of the loss of function obtained by either transgenic RNAi or hypomorphic Alk alleles is, however, limited by the requirement for viability to late larval stages. The MARCM system, in contrast, produces a strong, null, loss of Jeb function. We hypothesize that the structural phenotypes may only be observable with a strong loss of function that is not compatable with viability as produced by either trangenic RNAi or hypomorphic Alk alleles.

An alternative, though we think unlikely, explanation for the lack of symmetry between the Jeb null MARCM phenotypes and the Alk hypomorphic phenotypes is that Jeb signals via an alternative receptor in the NMJ. We note that Jeb and Alk, as a ligand-receptor pair, have been consistently observed to act together not just in the developing mesoderm but also in the developing visual system, larval neurogenesis and the embryonic neuropil in Drosophila (Englund, Loren et al. 2003; Lee, Norris et al. 2003; Cheng, Bailey et al.; Bazigou, Apitz et al. 2007; Rohrbough and Broadie 2011). The same pairing has been observed in C. elegans in the response to environmental stress and integration of conflicting sensory inputs (Reiner, Ailion et al. 2008; Shinkai, Yamamoto et al.; Ishihara, Iino et al. 2002). In light of the consistent association of Jeb and Alk in multiple species and contexts along with their colocalization at the larval NMJ, we propose that Jeb signals through Alk in the NMJ just as it does elsewhere. As evidence to support the Jeb-Alk ligand-receptor interaction in the NMJ we show that a 45% reduction of Alk expression in muscle results in 2-fold increased Jeb accumulation in the perisynaptic region (Fig. 3 J–L). Of course, these considerations do not exclude the possibility that Jeb signals through an alternative receptor to regulate synaptic architecture.

The credibility of the structural phenotypes observed in jeb mutant tissue mosaics is additionally supported by distortions of NMJ architecture that result from hyperactivated postsynaptic Alk signaling (Fig. 5). NMJs maintain a normal number of boutons, but the terminal is spatially compacted into a smaller area. Hyperactivated postsynaptic Alk signaling dramatically elevates activated ERK in muscle nucleii and perinuclear regions. This postsynaptic hyperactivation of Alk and Ras/MAPK/ERK restricts NMJ expansion and elaboration. We propose therefore that trans-synaptic signaling by secreted neuronal Jeb acts through postsynaptic Alk to limit terminal growth and expansion. Jeb-Alk signaling additionally promotes postsynaptic organization (e.g. Dlg and GluR).

Although Alk functions in muscle, regulation of presynaptic structure by Jeb may occur by linked retrograde signals. For example, Glass bottom boat (Gbb), a BMP ligand and retrograde signal (McCabe, Marques et al. 2003; Goold and Davis 2007), regulates postembryonic NMJ growth and function, with gbb mutants showing fewer NMJ boutons and reduced transmission. Determining how other known trans-synaptic signaling pathways integrate with Jeb-Alk signaling will be critical to our understanding of synaptic growth and function.

A conserved role for Jeb-Alk signaling in synaptic mechanisms

In other systems, Jeb-Alk signaling has been studied primarily at the level of behavior. In C. elegans, the Jeb homolog Hen-1 was identified in a forward genetic behavioral screen for impaired ability to integrate conflicting sensory input (Ishihara, Iino et al. 2002). The Hen-1 phenotype is non-developmental and can be rescued only by adult Hen-1 expression. There is no uniquely identified mammalian Jeb/Hen-1 homolog (Palmer, Vernersson et al. 2009), but ALK is expressed in the mammalian nervous system during development and at maturity (Iwahara, Fujimoto et al. 1997; Vernersson, Khoo et al. 2006). Alk is expressed in the mouse hippocampus and Alk loss of function enhances behavioral performance in tests dependent on hippocampal function (Squire 2004; Bilsland, Wheeldon et al. 2008; Weiss, Xue et al. 2011). Similarly, Drosophila learning has shown a dependence on the Ras/MAPK/ERK cascade, which is activated by Jeb-Alk signaling and is probably inhibited by Drosophila neurofibromin (dNf1) (Gouzi, Moressis et al. 2011). Genetic or pharmacolic inhibtion of Jeb-Alk signaling enhances associative learning while increased Jeb-Alk signaling or loss of dNf1 impairs learning. Inhibition of Alk rescues dNf1 mutant learning deficits (Gouzi, Moressis et al. 2011). These studies suggest that the Jeb-Alk trans-synaptic pathway acts in concert with other, negative regulators of Ras/MAPK/ERK signaling to balance developmental and learning-related synaptic structural and functional changes. Strikingly, a whole-genome association study recently identified human ALK as one of a small number of genes associated with sporadic amyotrophic lateral sclerosis (ALS), a devestating neurodegerative disease of central motor units (Dunckley, Huentelman et al. 2007). If Alk has a conserved inhibitory role in synaptic physiological regulation, hypofunctional human Alk variants may result in augmented motor unit activity and contribute to excitotoxicity and progressive motor unit degeneration in ALS. Pharmacologic activation of Alk has already been hypothesized to have therapeutic benefit in treating ALS (Dunckley, Huentelman et al. 2007). We hope to gain further insight from future studies into the mechanism by which the Jeb-Alk signaling pathway regulates synaptic adaptivity in both normal and pathological states.