Anterograde Jelly belly ligand to Alk receptor signaling at developing synapses is regulated by Mind the gap

Synaptogenesis is driven by an overlapping cascade of anterograde and retrograde trans-synaptic signals. At vertebrate neuromuscular junction (NMJ) synapses, presynaptically secreted signals, including Agrin and Wnt glycoproteins, regulate the assembly of postsynaptic domains (Wu et al., 2010; Kummer et al., 2006; Van Vactor et al., 2006). Presynaptic growth and functional differentiation likewise rely on secreted retrograde signals, including Wnts and TGFβ proteins (Packard et al., 2003; Salinas, 2003; Salinas, 2005). At Drosophila glutamatergic NMJs, presynaptically secreted Wnts [Wingless (Wg), Wnt5] act through postsynaptic Frizzled (Fz2) and Derailed receptors to regulate postsynaptic differentiation (Packard et al., 2002; Mathew et al., 2005; Ataman et al., 2008; Liebl et al., 2008; Korkut et al., 2009). Wg and Wnt5 also act bidirectionally on the presynaptic terminal through Fz2, via a pathway that is dependent on Shaggy (also known as Gsk3β) (Ataman et al., 2008; Miech et al., 2008). Additionally, a retrograde signaling pathway mediated by the TGFβ ligand Glass bottom boat (Gbb) acts on the Wishful thinking (Wit), Thick veins (Tkv), and Saxophone (Sax) receptors to regulate presynaptic growth and function at both NMJ and central synapses (Wu et al., 2010; Aberle et al., 2002; McCabe et al., 2003; Baines, 2004; McCabe et al., 2004).

Secreted trans-synaptic signaling ligands necessarily transverse and interact within the extracellular matrix (ECM), a complex milieu of secreted glycoproteins, proteoglycans and growth factors located within the synaptic cleft and adjacent perisynaptic domains (Dityatev and Schachner, 2003; Pavlov et al., 2004; Dityatev and Schachner, 2006). For example, the delivery of presynaptically secreted Wg and Wg-binding Evenness interrupted (Evi; Wntless – FlyBase) protein to postsynaptic Fz2 receptors occurs via large exosomes extruded through the cleft ECM domain (Korkut et al., 2009). The molecular assembly and dynamic maintenance of this `synaptomatrix' (Vautrin, 2010) is therefore likely to influence many forms of intercellular synaptic signaling. We recently discovered and characterized Mind the gap (Mtg), a presynaptically secreted glycoprotein that is predicted to bind synaptic cleft ECM glycosaminoglycans (Rohrbough et al., 2007; Rushton et al., 2009). Loss of Mtg abrogates synaptic cleft ECM formation at developing synapses (Rohrbough et al., 2007) and impairs the synaptic localization of integrin ECM receptors (Rushton et al., 2009), showing that Mtg is required to establish the synaptomatrix. mtg null embryos display severely disrupted postsynaptic domains, including loss of localized glutamate receptors and disrupted functional transmission, indicating a crucial link between synaptic ECM and synaptic induction (Rohrbough et al., 2007; Rushton et al., 2009).

The signaling ligand Jelly belly (Jeb) and its receptor Anaplastic lymphoma kinase (Alk; also known as dAlk), a receptor tyrosine kinase (RTK), have recently emerged as a potential new trans-synaptic signaling pathway (Bazigou et al., 2007; Palmer et al., 2009). Alk is proposed to function in nervous system development, based on its expression in mammalian brain (reviewed by Palmer et al., 2009; Vernersson et al., 2006). However, Alk function in vivo has been examined most extensively in Drosophila. Jeb secreted from developing somatic mesoderm activates Alk to drive embryonic visceral muscle formation (Loren et al., 2001; Englund et al., 2003; Lee et al., 2003; Loren et al., 2003). Jeb-Alk signaling activates Erk (Rolled – FlyBase) and the transcription of downstream genes, including those that encode cell adhesion proteins and the TGFβ signaling molecule Dpp (Loren et al., 2001; Englund et al., 2003; Lee et al., 2003; Varshney and Palmer, 2006; Shirinian et al., 2007). Intriguingly, Jeb and Alk expression in late embryonic CNS is consistent with a developmental signaling requirement in neurons (Loren et al., 2001; Weiss et al., 2001). Of key interest is the finding that anterograde Jeb-Alk signaling mediates retinal axon targeting and circuit formation in the visual medulla by regulating neuronal cell adhesion proteins (Bazigou et al., 2007).

Here we test the prediction that Jeb-Alk signaling functions during embryonic synaptogenesis, and the hypothesis that this signaling requires synaptic cleft ECM integrity. Jeb localizes presynaptically and Alk postsynaptically to developing synapses, establishing an anterograde synaptic signaling pathway. Synaptic expression of Jeb and Alk is strongly modified by loss of Mtg, indicating that secreted Jeb and Mtg pathways intersect in the synaptic ECM environment. In jeb and Alk null mutant embryos, NMJ synapses form correctly, assemble pre- and postsynaptic domains, and respond normally to exogenously applied neurotransmitter. However, endogenous synaptic transmission frequency and strength and larval locomotion are strongly impaired, comparable to similar impairments in mtg mutants. Together, these findings indicate that Jeb-Alk signaling represents an anterograde inductive pathway that is regulated by Mtg and required to establish functional embryonic synapses.

Flies were maintained on standard agar/yeast/molasses/cornmeal food at 25°C, and staged as hours after egg laying (AEL) following a 2- to 4-hour timed laying period. Larvae were reared at low density in food tubes to wandering L3 stage. Homozygous and rescued mutants were selected by the absence of GFP balancers (CyO Kr-Gal4, UAS-GFP or TM3 Kr-Gal4, UAS-GFP). The mtg alleles included the point-mutant null mtg¹/Df(CA3) (Rohrbough et al., 2007), transgenically rescued mtg¹ (UAS-mtg::GFP 14.2/+; Uh1-Gal4, mtg¹/mtg¹) (Rushton et al., 2009) and elav-Gal4/+; UAS-mtg::GFP 9.3/+. The brec^P2 GluRIID mutant removes 80-90% of functional glutamate receptors (GluRs) (Featherstone et al., 2005). jeb^k05644 (Bloomington I0576) contains a P-element that interrupts transcription within a large central intron; the resulting product lacks the LDLa domain and is not internalized by target cells (Weiss et al., 2001). Alk¹ is truncated at the first extracellular MAM domain, creating a functional null; Alk⁸ is truncated just after the transmembrane domain (Loren et al., 2003). The Alk⁸-encoded protein localizes to the membrane and binds Jeb, but lacks the required intracellular RTK signaling domain (Englund et al., 2003; Lee et al., 2003). UAS-Jeb and UAS-Alk were driven with elav-Gal4 and 24B-Gal4 in wild-type and mutant backgrounds. Controls included UAS-mtg::GFP 14.1/+ (without a Gal4 driver) for mtg mutant comparisons, and wild-type (Oregon-R) controls for jeb and Alk mutant comparisons. We observed no differences between control genotypes in any assays.

Imaging of staged embryos, L1 and L3 wandering larvae was as previously described (Rohrbough et al., 2004; Rohrbough et al., 2007; Rushton et al., 2009). Staged embryos were dechorionated in bleach, washed in distilled H₂O and transferred to saline (135 mM NaCl, 5 mM KCl, 4 mM MgCl₂, 0.2 mM CaCl₂, 5 mM TES, 75 mM sucrose, 2 mM NaOH, pH 7.2). Embryos were glued to Sylgard-coated coverslips, dissected dorsally and glued flat (Featherstone et al., 2009), fixed for 10-15 minutes in 4% paraformaldehyde, and washed in PBS (Invitrogen, Carlsbad, CA, USA) containing 0.1% Triton X-100 (Fisher) for at least 1 hour. Detergent was omitted for detergent-free staining. Control and mutant animals were processed and imaged on the same coverslips under identical conditions.

Primary antibodies were incubated overnight at 4°C at the following dilutions: mouse anti-Bruchpilot (Brp), 1:100 (supplied by Hugo Bellen, Baylor College of Medicine, Houston, TX, USA); rabbit anti-Synaptotagmin (Syt), 1:500 (supplied by Hugo Bellen); mouse anti-Synapsin (Syn), 1:100 [Developmental Studies Hybridoma Bank (DSHB), Iowa]; rabbit anti-GluRIIC, 1:500 (supplied by Aaron DiAntonio, Washington University, St Louis, MI, USA); mouse anti-Discs large (Dlg) mAb 4F3, 1:500 (DSHB); rabbit anti-Pak, 1:500 (supplied by Lawrence Zipursky, University of California, Los Angeles, CA, USA); guinea pig anti-Jeb and guinea pig/rabbit anti-Alk, 1:1000 (supplied by Ruth Palmer, Umea Institute, Sweden). Fluorescent Alexa-conjugated secondary antibodies (Molecular Probes; 1:500) and anti-horseradish peroxidase (HRP; TxR- or Cy5-conjugated; 1:50 to 1:100; Jackson laboratories, West Grove, PA, USA) were applied for 1.5-2.5 hours at room temperature. Mtg::GFP was visualized with FITC-conjugated anti-GFP (1:500; ab6662, Abcam, Cambridge, MA, USA).

Images were acquired on a Zeiss 510 confocal microscope (40-63× oil objectives) using LSM acquisition software, at effective magnifications of 100-200×. We examined embryonic NMJs 6/7, 13 and 12 in segments A2-A4, and larval NMJs 6/7, 13 and 4 in segment A3. Synaptic fluorescence intensities were measured on projected confocal z-stacks using ImageJ (NIH). Series depth was always kept constant for comparisons, with mutant values normalized to controls in the same experiment. Outlined NMJ and neuropil regions were created by thresholding synaptic antibody signals to determine mean synaptic fluorescence/area and score synaptic puncta. Matched control and mutant stacks in each experiment were exported to Adobe Photoshop. Any pixel level adjustments were performed on the entire image set.

Movement was assessed in early larvae (less than 2 hours post-hatching) on agar plates after a 2- to 3-hour staged laying period. Animals were first lightly touched at the head and tail to confirm touch responsivity and movement capability. Peristaltic full-body contractions were scored during two to four successive 120-second intervals, and averaged for each animal.

Whole-cell patch clamp recordings were made from muscle 6 in A2-A4 at –60 mV holding potential using an Axopatch 1D amplifier and PClamp7 software (Axon Instruments) (Rohrbough et al., 2007). Dissected 20- to 22-hour embryos or newly hatched larvae were exposed for 45-90 seconds to collagenase (1.0 mg/ml, Sigma type IV in 0.2 mM Ca²⁺ saline), and then washed with recording saline containing 1.8 mM Ca²⁺ (Featherstone et al., 2009). Endogenous NMJ excitatory junctional currents (EJCs) were recorded for 1-2 minutes in the absence of stimulation, and analyzed using a Clampfit 7 event detection template. Amplitudes were measured for clear, non-superimposed EJCs. All distinct EJCs were counted for frequency measurements. Evoked EJCs were elicited by CNS stimulation (1 msecond, 20-40 V applied through a Grass Instruments stimulus isolation unit), using a patch pipette (3-5 μm tip diameter) at the ventral nerve cord midline. Glutamate responses were elicited by iontophoretic application of 100 mM l-glutamate (pH 9-10 in distilled H₂O) from sharp microelectrodes, positioned between muscles 6 and 7 to obtain maximal response (Rohrbough et al., 2007). This method largely samples GluR localized to postsynaptic domains (Broadie and Bate, 1993; Rohrbough et al., 2007). Iontophoretically applied glutamate elicited faster/larger control and mtg mutant responses than pressure-applied glutamate (Rohrbough et al., 2007); this was likely to be due to faster application revealing a larger number of functional receptors. All responses were recorded using the same range of stimulus strength and duration (–50 to –60 V, 4-5 mseconds). Neuronal recordings were made from dorsal ventral ganglion neurons exposed by focal protease treatment, as previously described (Featherstone et al., 2005). Endogenous sustained excitatory currents and potentials were recorded for 2-8 minutes (–70 mV) to determine average activity frequency (per minute) and amplitude. Currents were sampled at 10 kHz and filtered at 1-2 kHz. Data records were exported for display using Excel (Microsoft) and KaleidaGraph (Synergy Software) spreadsheet and graphics software.

One-way ANOVA and Dunnett multiple comparisons tests (KaleidaGraph) were used for all statistical comparisons between genotypes. Reported values are mean ± s.e.m.

Secreted Jeb activates the Alk receptor in developing embryonic visceral muscle (Loren et al., 2001; Weiss et al., 2001; Englund et al., 2003; Lee et al., 2003; Loren et al., 2003). Jeb and Alk are similarly expressed in embryonic CNS (Loren et al., 2001; Weiss et al., 2001), but a signaling role in the developing nervous system has not been demonstrated. The neuronally secreted protein Mtg is required for assembly of the synaptic cleft ECM and functional postsynaptic domain induction during embryonic synaptogenesis (Rohrbough et al., 2007; Rushton et al., 2009). We have hypothesized that Mtg shapes the synaptomatrix environment to enable and modulate trans-synaptic signaling. We therefore investigated the potentially intersecting function of Mtg and a predicted Jeb-Alk anterograde synaptic signaling pathway.

We first examined Jeb and Alk expression and localization at larval glutamatergic NMJ synapses, using fluorescent confocal imaging of anti-Alk and anti-Jeb labeling. Each body wall muscle is stereotypically innervated by two to four motoneurons, which form synaptic boutons of characterized size/morphology and function (types Is/b, II, III; Fig. 1). Alk was strongly expressed at large type Ib boutons (Fig. 1A), which are surrounded by extensively folded subsynaptic reticulum (SSR) that is concentrated with postsynaptic proteins. In z-series projections, Alk superimposed extensively with the anti-HRP-labeled presynaptic membrane, but had a much broader domain and was clearly localized to the surrounding postsynaptic muscle (Fig. 1A′). Orthogonal and rotated bouton projections further confirmed postsynaptic Alk localization and distribution, resembling that of the SSR scaffold protein Dlg. At some boutons, Alk appeared concentrated in punctate domains (Fig. 1A). Alk was more weakly localized at smaller type Is boutons, which have less extensive SSR (Fig. 1B), and was not detectably expressed at type II boutons, which lack SSR. By contrast, Jeb was primarily presynaptic, and exhibited a more localized, punctate bouton expression pattern (Fig. 1B). Jeb was strongly concentrated in smaller bouton subtypes (Is, II, III), and more heterogeneously and weakly expressed in type 1b boutons, a localization pattern reciprocal to that observed for Alk. Jeb was typically localized to numerous small puncta completely contained within, and surrounded by, Alk in colabeled type I boutons (Fig. 1B′). Jeb synaptic puncta were also contained within HRP-labeled boutons, as well as within peripheral nerve branches and presynaptic NMJ axons, consistent with presynaptic localization at the NMJ.

To visualize secreted extracellular Jeb, NMJs were stained using detergent-free conditions to prevent cell permeabilization (Rushton et al., 2009). Extracellular Jeb is localized at all bouton subtypes, predominantly closely associated with the HRP-labeled presynaptic membrane (Fig. 1C), placing secreted Jeb within the previously visualized Alk receptor domain. We also observed concentrated Jeb puncta a short distance from a synaptic bouton (Fig. 1C′). These Jeb accumulations are likely to be neuronally derived, but might represent protein secreted by muscles or other cells (e.g. glia) in proximity to muscle.

We next focused on roles for the Jeb-Alk signaling pathway during synaptogenesis, examining mature embryos (20-22 hours AEL) and newly hatched first instar (L1) larvae (22-24 hours AEL). This developmental period coincides with the jeb and Alk null mutant lethal stage (Englund et al., 2003; Loren et al., 2003). In the embryonic CNS, Jeb and Alk are highly enriched within the synaptic neuropil domain, as defined by the presynaptic active zone marker Brp (Fig. 2A,B). Jeb exhibited dense and extensive punctate localization along longitudinal neuropil axon tracts, and was also present in neuronal somata and segmental nerve trunks (Fig. 2A). Alk exhibited extensive overlap with Brp (Fig. 2B) and Jeb (see below). However, Alk exhibited a more homogenous neuropil distribution than Brp or Jeb, more closely resembling the distribution of postsynaptic Dlg (see Fig. S3 in the supplementary material). Alk appeared to be only weakly expressed in CNS neuronal somata. Thus, Jeb and Alk are primarily localized to developing CNS synapses, with distinctions in localization consistent with pre- and postsynaptic expression, respectively.

Mtg and Jeb synaptic localizations showed striking similarities (Fig. 2C). Mtg is specifically targeted to CNS and NMJ synapses and is secreted from presynaptic terminals (Rushton et al., 2009). In mtg null mutants rescued by Mtg::GFP (mtg¹; Uh1:UAS-Mtg::GFP), Mtg also exhibits characteristic punctate localization in CNS neurons and processes, but is less extensively distributed and is concentrated prominently along a ventral subset of longitudinal axonal tracts within the neuropil (Rushton et al., 2009). Although our confocal analysis could not resolve protein co-expression at individual central synapses, Mtg and Jeb showed partially overlapping expression domains in the neuropil, with Jeb domains often appearing to contact Mtg puncta (data not shown). Jeb and Mtg exhibited extensive colocalization at NMJ synapses. Both proteins were strongly expressed in smaller boutons (type Is, II, III) and colocalized more diffusely in overlapping distinct puncta in type Ib boutons (Fig. 2C). These data indicate that Mtg and Jeb are similarly secreted at synapses.

We next examined Jeb and Alk expression at developing NMJ synapses in embryos and newly hatched larvae. Jeb was localized to puncta of variable size and intensity that were contained within, or closely associated with, presynaptic boutons, and often appeared partially or completely colocalized with presynaptic Brp (Fig. 3A). Larger concentrated aggregates that incompletely superimposed with Brp were likely to represent accumulations of secreted Jeb. Jeb puncta also localized in proximity to postsynaptic glutamate receptor (GluR) domains, but showed little colocalization (Fig. 3C). Jeb was also clearly concentrated within embryonic segmental nerve branches and presynaptic axons. We conclude that presynaptic neurons and nerve terminals are the primary source of NMJ-localized Jeb. However, Jeb aggregates were also present on embryonic perisynaptic and nonsynaptic muscle (Fig. 3C). Since Jeb is expressed in developing somatic mesoderm, such sites might represent patches of muscle-derived Jeb.

The Alk receptor exhibited more uniform expression at embryonic NMJs (Fig. 3B). Alk distribution appeared primarily or exclusively postsynaptic, localizing to boutons surrounding HRP-labeled neuronal membrane. Postsynaptically, Alk overlapped with Pak kinase (also known as dPak) (Fig. 3B) and GluR domains (not shown), and was particularly aligned with Dlg (Fig. 3D). Like other postsynaptic components, Alk was also weakly expressed in muscle, including muscle attachment sites. We conclude that Alk localizes to postsynaptic domains opposite presynaptic terminals that are releasing Jeb, consistent with anterograde signaling during embryonic synaptogenesis.

Loss or mislocalization of Alk receptor in target cells prevents Jeb uptake, causing accumulation of secreted Jeb (Englund et al., 2003; Lee et al., 2003; Stute et al., 2004). We tested whether Jeb synaptic expression and localization are similarly disrupted in Alk truncation mutants and whether Alk localization is altered in jeb^k0564 mutants. In the CNS neuropil of jeb mutants, anti-Jeb fluorescence was strongly reduced (to 45% of control; P<0.0001 versus control; n=6), whereas anti-Alk neuropil fluorescence was unaltered (97% of control) (Fig. 4A,B,D), indicating that Jeb is unnecessary for Alk synaptic localization. In the CNS neuropil of Alk¹ mutants, Alk labeling was strongly reduced (38% of control; P<0.0001; n=3), with weak residual localization (Fig. 4B). In Alk⁸ mutants, neuropil Alk labeling was reduced (63% of control; P<0.02; n=3; Fig. 4B), consistent with antibody recognition of the N-terminal region (Loren et al., 2003) and reported plasma membrane localization of the mutant protein (Englund et al., 2003). Jeb neuropil levels were significantly increased in Alk mutants (25% over control; P<0.003; n=6; Fig. 4A,D). At NMJs, Jeb expression was nearly eliminated in jeb mutants (Fig. 4C). In Alk¹ and Alk⁸ mutants, Jeb NMJ labeling intensity was elevated (23% over control; P<0.0001), and the number of Jeb-positive synaptic puncta increased by 62% (92±7 puncta for Alk versus 57±6 puncta for control; NMJs 6/7, 12 and 13; P<0.003) (Fig. 4C,E,F). We conclude that Jeb accumulation at Alk mutant synapses is likely to represent presynaptically secreted protein that cannot be internalized in the absence of postsynaptic Alk.

Jeb ligand and Alk receptor necessarily interact intimately with the ECM at developing synapses. Jeb and Mtg occupy overlapping synaptic domains, and Alk colocalizes within the postsynaptic domain regulated by Mtg (Rohrbough et al., 2007; Rushton et al., 2009). To test our hypothesis that Jeb-Alk and Mtg functions intersect, we examined Jeb and Alk at mtg null embryonic NMJ synapses. The labeling intensity and area of Alk synaptic expression were reduced to 64% (P<0.0001) and 55% of control levels (P<0.02), respectively, at NMJs 6/7, 12 and 13 (Fig. 5A,D,E). These changes were significantly restored in rescued mtg mutants (Fig. 5D,E). By contrast, Jeb synaptic labeling intensity was increased by 29% (P<0.0002), and the area of Jeb NMJ expression was dramatically increased by more than 2-fold (105%) over the control (P<0.0002) (Fig. 5B,C,E). The number of mtg mutant Jeb synaptic puncta was increased by 75% overall (P<0.002; NMJs 6/7, 12 and 13; Fig. 5F). The Jeb expression area and puncta were significantly restored in rescued mtg mutants (Fig. 5E,F). Increased Jeb expression occurred in combination with reduced and mislocalized GluRIIC expression, which was substantially restored in rescued mutants (Fig. 5C) (Rohrbough et al., 2007; Rushton et al., 2009).

We further tested whether the expression of Jeb and Alk at larval NMJs is modulated by Mtg gain-of-function. Neuronally targeted Mtg overexpression (elav; UAS-Mtg) increased Jeb NMJ expression by 23% over the control (P<0.02), without significantly altering Alk levels (see Fig. S1 in the supplementary material). By contrast, postsynaptically targeted Mtg (24B-Gal4; UAS-Mtg) significantly depressed Jeb NMJ levels (67% of control; P<0.0002) and insignificantly (P=0.07) decreased Alk expression (see Fig. S1 in the supplementary material). In the central neuropil of mtg null embryos, in contrast to the expression changes at the NMJ, the Alk labeling intensity was insignificantly altered, whereas the Jeb level was reduced to 71% of that of the control (P<0.003; see Fig. S2 in the supplementary material).

An accumulation of synaptic Jeb at mtg mutant NMJs could be a secondary result of strongly downregulated Alk expression, as observed at Alk mutant synapses. However, the increase in Jeb staining level, expression area and Jeb puncta number at mtg mutant NMJs were all more pronounced than for Alk mutants (Fig. 4C-E, Fig. 5B,D-F), indicating that Mtg regulates Jeb independently of changes in Alk. Together, these results indicate that Mtg regulates postsynaptic Alk expression, and additionally exerts a direct effect upon Jeb expression during synaptogenesis at the glutamatergic NMJ.

Jeb and Alk expression at developing synapses suggests a role in regulating synaptic connectivity and/or functional differentiation. We examined features of CNS and NMJ synapse differentiation in jeb and Alk embryonic mutants for structural and morphological defects. In mutants, the overall CNS morphology appeared normal in terms of segmental nerve branching and segmental muscle patterning. Anti-FasII labeling in the mutant CNS neuropil revealed distinct, well-ordered longitudinal axonal tracts, exhibiting proper dorsal-ventral and medial-lateral spacing with no evidence of aberrant projections or pathfinding errors (see Fig. S3A,B in the supplementary material). Central synapse differentiation was assayed with antibodies against presynaptic Brp and Syt and postsynaptic Dlg proteins (see Fig. S3C-E in the supplementary material). In jeb and Alk mutants, neuropil synaptic differentiation appeared normal, with comparable labeling intensity, density and distribution of presynaptic and postsynaptic labels.

We likewise examined jeb and Alk mutant NMJs for presynaptic (Brp, Syt, Syn) and postsynaptic (Dlg, GluRIIC) proteins. Mutant NMJs (6/7, 12 and 13) appeared correctly and stereotypically formed and elaborated, and no examples of muscle innervation failure or synaptic targeting errors were detected (Fig. 6A). Mutant presynaptic active zones appeared normally formed, showing Brp puncta distribution and presynaptic vesicle/vesicle-associated protein (Syt, Syn) expression that were comparable to controls (Fig. 6A,B). jeb mutants showed only a small increase (P<0.05) in Brp puncta number, whereas Alk mutants were indistinguishable from controls at NMJs 6/7, 12 and 13 (see Fig. S4A in the supplementary material). Quantification of NMJ synaptic area based on Syt labeling showed no difference between control and jeb, Alk¹ or Alk⁸ mutants (see Fig. S4B in the supplementary material). Similarly, the postsynaptic Dlg and GluR synaptic expression levels and distribution at mutant NMJs were also comparable to those of the control, including clear colocalization of GluR puncta within the Dlg domain (Fig. 6C,D). These results indicate that jeb and Alk mutants show apparently normal structural synaptogenesis, and develop normal pre- and postsynaptic specializations in the CNS and at the NMJ.

Homozygous jeb and Alk mutant embryos showed reduced hatching, and mutant larval movement was typically sluggish and highly limited. To quantify movement defects, we scored peristaltic body contractions in newly hatched (under 2 hours) larvae. Locomotory movement in jeb and Alk mutants was decreased to 20-35% of the control level (P<0.0001 versus controls; Fig. 7A), characterized by slower contractions and extended pauses. Severely impaired and motionless larvae remained capable of briefly resuming movement when stimulated, suggesting a defect in the central circuit output driving locomotion. To address neuronal versus muscle Jeb and Alk function in locomotion, we first assayed wild-type animals with muscle (24B-Gal4) or neuronally targeted (elav-Gal4) UAS-Jeb and UAS-Alk overexpression. Alk muscle overexpression impaired movement as strongly as loss of Alk, whereas neuronal overexpression reduced movement by more than 30% (P<0.005 versus control). By contrast, neither muscle nor neuronal Jeb overexpression altered movement (P>0.2 versus control; Fig. 7A). Somatic muscle defects are clearly associated with loss of Jeb-Alk signaling (Stute et al., 2004), and potentially contribute to defective locomotion. To test this hypothesis, we introduced neuron- or muscle-targeted UAS-Jeb into the jeb null background. Movement in jeb mutant larvae was significantly rescued by neuronal Jeb expression (jeb; elav>jeb; P<0.003 versus jeb), but not by muscle expression (Fig. 7A), indicating a neuronal Jeb requirement for normal locomotion.

We next carried out NMJ electrophysiological assays in mature embryos to test the functional basis of mutant synaptic defects. Postsynaptic GluR function was assessed with glutamate-gated current responses in mtg, jeb and Alk mutants (Fig. 7B-E). Control NMJs exhibited visible contractions to single iontophoretic glutamate pulses, and in whole-cell recordings displayed robust 2000-3000 pA responses (2611±120 pA; n=19; Fig. 7B,E). As a GluR loss-of-function control, we assayed a strong GluRIID (KaiRIA – FlyBase) mutant, brec^P2 (Featherstone et al., 2005), which exhibited only small, noisy glutamate responses (380±88 pA; n=4; Fig. 7B,E). Null mtg mutants (mtg¹ and mtg¹/Df) showed weak/limited glutamate-driven muscle contractions, with a 55% reduction in mean response amplitude (1165±199 pA; n=10; P<0.0001 versus control; Fig. 7C,E). By contrast, jeb and Alk mutant NMJs exhibited visible glutamate-driven muscle contractions, and consistently large and robust postsynaptic current amplitudes comparable to controls (jeb, 2230±88 pA; Alk¹, 2998±256 pA; n=4 per mutant; P>0.6 versus control; Fig. 7D,E). To further test NMJ function, we recorded action potential-dependent excitatory junction currents (EJCs) evoked by central stimulation. Control and jeb mutant NMJs displayed comparable, robust EJCs with amplitudes of greater than 1 nA (see Fig. S5A,B in the supplementary material). We conclude from these assays that jeb and Alk mutants have functional NMJs, without detectable impairment of either presynaptic release capability or postsynaptic responsiveness. Therefore, the primary mutant impairment appears to be a defective generation of central excitatory output.

To test embryonic motor circuit differentiation, we recorded endogenous NMJ synaptic transmission driven by central synaptic motor output (Fig. 8). Functionally mature neurotransmission is characterized by large (greater than 1 nA) EJCs occurring in patterned episodic bursts (Broadie et al., 1997), which are absent when centrally generated action potentials are blocked (Featherstone et al., 2001; Rohrbough et al., 2007). For control recordings, 70% (12/17) exhibited patterned and/or large (exceeding 1 nA) activity-dependent EJCs. Control EJC amplitude histograms revealed a majority component of 100-300 pA amplitudes, with a significant second component (32% of EJCs) of amplitudes exceeding 500 pA, which correspond to activity-driven patterned EJCs (Fig. 8A). In mtg mutants (mtg¹ and mtg¹/Df), no defined EJCs were detected in 5/7 (over 70%) records. Small isolated EJCs were observed in 1/7 embryos, and large amplitude EJCs were recorded in only 1/7 embryos (Fig. 8B). This result clearly shows that loss of Mtg greatly reduces overall activity-dependent transmission.

jeb and Alk mutants similarly exhibited severely reduced endogenous neurotransmission. In jeb mutants, no large (exceeding 500 pA) or patterned EJCs were recorded (maximum amplitude of 439 pA; Fig. 8C). No EJCs were detected in 3/5 jeb recordings, and overall EJC frequency was below 1 Hz in 2/5 active cells (89 EJCs during 486 seconds; Fig. 8C). In Alk mutants, no large or patterned EJCs were recorded (maximum amplitude of 429 pA), and overall EJC frequency was below 1 Hz in the 4/5 active cells (88 EJCs during 463 seconds; Fig. 8D). In summary, EJCs exceeding 400-500 pA were infrequent or effectively absent in mtg, jeb and Alk mutants, with overall transmission frequency reduced by at least 90% from normal levels (control, 3.4±4.1 Hz; mtg, 0.35±0.78 Hz; jeb, 0.17±0.25 Hz; Alk¹, 0.20±0.21 Hz; P≤0.01 versus control for all mutant genotypes) (Fig. 8E). Cumulative EJC amplitude in mutants was reduced by 50-75% (control, 385±8 pA; mtg¹, 187±14 pA; jeb, 92±9 pA; Alk¹, 131±10 pA; P≤0.0001 versus control for all mutant genotypes) (Fig. 8F). The loss of endogenous activity-dependent and patterned transmission suggests an impairment in functional central synapse connectivity in the absence of Jeb-Alk signaling.

To directly assay central synaptic connectivity in early larvae, we recorded synaptic activity in the dorsal central motoneurons that drive glutamatergic NMJ transmission and episodic locomotion (Baines et al., 2002; Baines, 2003) (see Fig. S5C in the supplementary material). Control neurons displayed characteristic sustained excitatory currents supporting action potentials, with a mean event frequency of 4.5/minute (n=6), as we reported previously (Featherstone et al., 2005). By contrast, jeb, Alk and mtg mutant recordings showed more variable levels of synaptically driven activity, including examples lacking activity. Overall excitatory event frequencies in mutants were reduced by 30-50% (jeb, 3.0/minute; Alk¹, 3.3/minute; mtg¹, 2.3/minute; n=4-5; see Fig. S5D in the supplementary material); however, owing to heightened variability, these reductions were not statistically significant from controls (P>0.4). Excitatory current amplitudes were comparable for all genotypes (not shown). These results indicate functional interneuron-motoneuron connectivity in jeb, Alk¹ and mtg¹ mutants, and suggest that mutant locomotory impairments might be due to upstream synaptic defects.

Jeb and Alk are localized to pre- and postsynaptic junctions during embryonic synaptogenesis, predicting an inductive anterograde synaptic signaling role. Jeb-Alk RTK signaling at embryonic somatic-visceral mesoderm junctions similarly directs visceral muscle specification and differentiation (Loren et al., 2001; Englund et al., 2003; Lee et al., 2003; Loren et al., 2003). Jeb is the only identified Alk ligand, and Alk is the only identified Jeb receptor. It was recently shown that the C. elegans Alk ortholog SCD-2 is similarly neuronally expressed and activated by a Jeb-like secreted ligand, HEN-1, which contains an LDLa domain (Reiner et al., 2008). Jeb-Alk anterograde signaling has recently been shown to regulate circuit formation in the Drosophila developing optic lobe (Bazigou et al., 2007).

Jeb-Alk NMJ and neuropil expression patterns indicate that anterograde signaling occurs at both peripheral and central synapses. Jeb localizes to NMJ presynaptic terminals and is secreted extracellularly, whereas Alk localizes to opposing postsynaptic membranes. The Jeb neuronal expression/trafficking profile suggests transport to the NMJ, rather than neuronal Jeb uptake from muscle, as previously suggested (Weiss et al., 2001). Jeb and Alk display reciprocal expression levels at NMJ synapses, with lower Jeb levels at boutons expressing highest postsynaptic Alk levels. Jeb is also strongly increased at Alk mutant synapses, suggesting that internalization of secreted Jeb in postsynaptic cells requires Alk receptor function. This predicted synaptic signaling cascade therefore parallels the mechanism in mesoderm development (Loren et al., 2001; Englund et al., 2003; Lee et al., 2003; Loren et al., 2003).

Our working hypothesis predicts that the ECM environment modulates trans-synaptic ligand-receptor interactions. A key finding, therefore, is that the Jeb-Alk pathway is regulated by Mtg, a presynaptically secreted glycoprotein crucial for synaptic cleft ECM formation (Rohrbough et al., 2007; Rushton et al., 2009). In the absence of Mtg, postsynaptic Alk is strongly reduced and secreted Jeb is dramatically accumulated at NMJ synapses. Maintenance of Alk might be part of a larger role for Mtg in postsynaptic differentiation, as numerous postsynaptic components are lost/mislocalized in mtg mutants (Rohrbough et al., 2007). Alternatively, Mtg might more directly regulate Alk, possibly by ECM tethering/anchoring of the Alk receptor. The Jeb upregulation should be partly attributable to the Mtg-dependent reduction in postsynaptic Alk. However, synaptic Jeb is upregulated to a much greater degree, despite a less severe downregulation of Alk, in mtg than in Alk null mutants. Jeb NMJ expression is also modulated independently of Alk by targeted neuronal or muscle Mtg overexpression, indicating that Mtg regulates Jeb independently of Alk. We conclude that Mtg expression and function are highly likely to regulate developmental Jeb-Alk synaptic signaling. However, this interpretation must be verified in future studies by demonstrating a regulatory function for Mtg in previously established Jeb-Alk RTK molecular signaling pathways (Englund et al., 2003; Lee et al., 2003; Bazigou et al., 2007).

Mtg and Jeb are co-expressed in developing NMJ presynaptic boutons, and are secreted to occupy largely overlapping domains within the synaptomatrix. Our findings suggest that Mtg normally acts at NMJ synapses to restrict localized Jeb accumulation within the synaptomatrix. We suggest that the Mtg-dependent ECM might function as a barrier to maintain localized Jeb pools and/or as a scaffold that is required to appropriately present or proteolytically remove Jeb in the extracellular signaling space. It is presently unclear whether Mtg has a parallel regulatory role at developing central synapses, where Mtg is expressed in a more limited neuronal subset. Changes in central Jeb/Alk expression might be indirectly related to Mtg loss or overexpression in the CNS. Alternatively, changes in neuronal Mtg level might have greater effects on Jeb/Alk NMJ expression. Mammalian Alk candidate ligands, such as pleiotrophin, heparin affinity regulatory peptide (HARP), heparin-binding neurotrophic factor (HBNF), and midkine, are heparin-binding growth factors (Palmer et al., 2009), further highlighting that Alk activation occurs via ligands that function within the complex and dynamic glycomatrix. We propose that Mtg-dependent modulation of extracellular space is critical for the signaling activity of multiple trans-synaptic signals.

The Jeb-Alk pathway is not detectably required for embryonic axonal pathfinding, synapse morphogenesis or molecular assembly during synaptogenesis, including the proper localized expression of pre- and postsynaptic proteins. Likewise, Jeb-Alk function is not required for establishing functional NMJ synapses, including postsynaptic GluR domains. Jeb-Alk signaling is likely to have a role(s) during postembryonic NMJ development. The Alk receptor is required for expression and signaling of the TGFβ signaling component Dpp in developing endoderm (Shirinian et al., 2007), and Alk is similarly suggested to modulate a TGFβ pathway in C. elegans (Reiner et al., 2008). Therefore, Alk potentially regulates the TGFβ-dependent retrograde signaling pathway(s) involved in synaptic plasticity and function during larval NMJ development (Aberle et al., 2002; Haghighi et al., 2003; McCabe et al., 2003; McCabe et al., 2004).

Our results indicate that Jeb and Alk have a role in the development of locomotion behavior. Jeb-Alk signaling regulates somatic as well as visceral muscle differentiation, with similar defects resulting from Alk removal or ectopic overexpression in muscle (Stute et al., 2004). Likewise, we find that either muscle or neuronal Alk overexpression impairs locomotion and results in early larval lethality. However, jeb and Alk mutant muscle responds to direct stimulation and evoked NMJ transmission is normal, indicating that the primary locomotory impairment is not defective muscle or NMJ function. Moreover, jeb mutant locomotion is significantly rescued by neuronal, but not muscle, Jeb expression, consistent with a requirement for Jeb signaling from central neurons. Importantly, loss of Jeb-Alk signaling severely reduces endogenous NMJ neurotransmission by effectively reducing the occurrence of centrally generated, patterned synaptic output to the NMJ (Broadie et al., 1997). The underlying excitatory synaptic drive onto motoneurons parallels the development of locomotion behavior (Baines et al., 1999; Baines, 2003). Central neuron recordings show functional excitatory synaptic input to jeb/Alk and mtg mutant motoneurons, which surprisingly show no significant loss of activity that might be suggested by the severe locomotion impairments. CNS dissection/recording conditions may effectively re-excite depressed motor activity (Carhan et al., 2004), similar to the effect of direct stimulation in provoking mutant movement.

Our results indicate that anterograde Jeb-Alk synaptic signaling is crucial for the maturation of locomotory behavior, and that Mtg regulatory activity intersects with the Jeb-Alk pathway during NMJ synaptic differentiation. We propose that Jeb-Alk signaling is essential for the functional differentiation of the central synaptic connections that drive motor circuit activity. Loss of Jeb-Alk signaling function impairs central excitatory synaptic transmission, resulting in a loss of endogenous central pattern generator activity driving motor output to the NMJ. Future studies will be directed towards dissecting the intersecting roles of Mtg and Jeb secreted signals in the functional differentiation of central motor circuits.

We particularly thank Ruth Palmer (Umea Institute, Sweden) for the generous gift of jeb and Alk mutant lines and Jeb and Alk antibodies. We also thank Hugo Bellen (Baylor College of Medicine), Aaron DiAntonio (Washington University) and Larry Zipursky (UCLA) for antibodies, and Emma Rushton for many discussions regarding this work. This work was supported by NIH grant GM54544 to K.B. Deposited in PMC for release after 12 months.