The Fak56 protein is highly expressed in the ventral nerve cord during embryonic stages 2225. To examine whether Fak56 has a role in NMJ formation, we dissected late third instar larvae from a transheterozygous Fak56^N30/K24 mutant that deletes the Fak56 gene and lacks Fak56 mRNA expression (Additional file 1A, B, and Additional file 1 legend for the generation of Fak56 mutants). This Fak56 null mutant was immunostained with horseradish peroxidase (HRP) in order to label axonal processes 30, and phalloidin (Pha) to label muscle fibers. No abnormality of motor axonal tracts could be detected, and the pattern and size of muscles were normal, in agreement with earlier observations 28. However, a more detailed examination revealed that Fak56 null mutant NMJs were overelaborated in comparison to wild-type ones (Figure 1A, B). NMJs innervating muscles 6 and 7 (NMJ 6/7s) of abdominal segment 3 (A3) were analyzed by immunostaining for HRP and synaptotagmin (Syt) to label presynaptic boutons 31. Altered branching patterns and ectopic synaptic boutons were observed, with increases in both Ib and Is boutons (arrows and arrowheads, respectively, in Figure 1C, D). Quantitatively, the number of synaptic boutons was increased by 44% and the total branch length increased by 22% when normalized to the total area of muscles 6 and 7 (quantified in Figure 1E). The Fak56 activity was not limited to NMJ 6/7s since NMJ 4s also displayed overgrowth phenotypes in both total branch length (62% increase) and bouton number (101% increase) (Figure 1F–H). Furthermore, previous reported Fak56^CG1 null mutants 28 also displayed a significant NMJ overgrowth phenotype when compared to wild-type control (Additional file 1C–E).

When scored for NMJ 6/7s, the altered branching pattern in Fak56^N30/K24 mutants showed secondary branch reduction by 21% but increases in higher-order branches (73% for tertiary branches and 424% for beyond tertiary; Figure 1I). The increase in higher-order branches was not caused by extension of multiple branches from single boutons, since a normal bifurcating pattern was observed.

To confirm that NMJ overgrowth phenotypes in the Fak56^N30/K24 mutant are due to the absence of Fak56 activity, a UAS-Fak56 transgene 25 was introduced. We found that neuronal expression of Fak56 with elav-GAL4 in the Fak56^N30/K24 mutant completely suppressed the NMJ phenotypes, as shown in assays for total branch length and bouton number of NMJ 6/7s. In contrast, Fak56 expression with the muscle-specific MHC-GAL4 failed to rescue Fak56 mutant phenotypes (Figure 1E). Taken together, these results suggest that Fak56 is specifically required in presynaptic neurons but not postsynaptic muscles to restrict NMJ growth. The exuberant NMJs in Fak56 null mutants were constructed normally, since molecular markers for various synaptic structures were expressed in a wild-type pattern (Additional file 2). Synaptic ultrastructure analyzed by transmission electron microscopy revealed no significant alternations in pre- and post-synaptic structures (Additional file 3).

To examine whether the enlarged NMJ in Fak56 null mutants is associated with functional changes in transmitter release, postsynaptic currents were recorded. In the null Fak56^N30/K24 mutant, no alteration was observed in the amplitude of spontaneous release of neurotransmitter or miniature junctional potentials (mEJPs) at a low Ca²⁺ concentration (0.2 mM), as shown in the cumulative frequency plot (Figure 1J). Similar skews of distributions were measured for wild type and Fak56^N30/K24 (1.5 ± 0.1 in wild type and 1.7 ± 0.2 in Fak56^N30/K24, 0.25 <p < 0.5 by Kruskal-Wallis h test). The variance/mean of mEJP amplitudes were also similar (0.28 ± 0.04 in wild type and 0.25 ± 0.04 in Fak56^N30/K24, 0.25 <p < 0.5 by Kruskal-Wallis H test). The frequency of mEJP was not changed significantly (1.2 ± 0.2 Hz in wild type and 1.8 ± 0.3 Hz in Fak56^N30/K24, p = 0.16, Student's t-test). Resting membrane potentials were similar in these measurements (-69.1 ± 1.9 mV in wild-type and -66.6 ± 1.5 mV in Fak56^N30/K24, p = 0.32 by Student's t-test). However, the mean amplitude of nerve-evoked EJPs was significantly enhanced at Fak56 mutant NMJs compared to wild type (p = 0.026 by Student's t-test, Figure 1K; measurements were also performed at 1 mM [Ca²⁺]; Additional file 4). These data demonstrate a role of Fak56 in modulating the electrophysiological behavior of Drosophila NMJs.

We then tested whether Fak56 mediates specific integrin activities at NMJs by genetic analysis. Integrin receptors are composed of heterodimeric α and β subunits 32. In the Drosophila genome, there are five α subunits: αPS1 (encoded by multiple edematous wings, mew), αPS2 (inflated, if), αPS3 (Vol or scb), αPS4 and αPS5 (both αPS4 and αPS5 uncharacterized), and two β subunits (βPS (myospheroid, mys) and βν(βν)) 333435363738. We tested for possible genetic interactions between the available mutant alleles of integrin subunits and Fak56. In Fak56^N30/KG hypomorphic animals, expression of Fak56 mRNA was reduced, but the NMJ appeared phenotypically normal (Additional file 1A, B; Figure 2A). However, when single mutant alleles of scb² and βν¹ were introduced into the Fak56^N30/KG background, significant NMJ overgrowth was induced (Figure 2B, C). This overgrowth phenotype was not detected when mew¹, if^k27e and mys¹ were introduced (quantified in Figure 2G). As controls, larvae that were heterozygous for the scb² or βν¹ mutant alleles displayed normal NMJ bouton number and length (quantified in Figure 2G). These results suggest that compromised αPS3 or βν integrin signaling demands the full-strength of Fak56 activity to constrain NMJ growth. Since NMJ overgrowth has been observed for αPS3 but not βν mutants 9, we examined NMJ phenotypes in the viable βν^1/2 mutant. Strikingly, significant increases in both branch length and bouton number were detected, similar to those observed in Fak56 null mutant larvae (Figure 2D). In summary, these genetic analyses suggest that αPS3 and βν are the main integrin subunits in regulating Fak56 activity during NMJ growth.

The laminins are ECM components composed of heterotrimers of α,β and γ subunits, and are major signals for integrin receptors 39. In Drosophila, LanA and wing blister (wb) encode two different α chains. We performed genetic interaction for both α chain mutants to test their involvement in Fak56 activity. Introducing one mutant allele of LanA^9–32 but not wb^4Y18 into the Fak56^N30/KG hypomorphic background promoted a significant increase in the number of synaptic boutons (Figure 2E, G). The total NMJ length was also increased, although it was not significant (p = 0.37). While the hypomorphic LanA^9–32/216 mutant displayed normal NMJ phenotypes, transheterozygous βν^1/+;LanA^9–32/+ displayed strong overgrowth phenotypes, with 61% increase in the bouton number and 32% increase in the total length compared to wild-type NMJs (Figure 2F, G). These results are consistent with a role for the α subunit LanA as a component of laminins to signal integrins during NMJ growth.

Activated FAK forms a complex with Src, and the dual FAK-Src kinase complex induces downstream signaling 40. To test whether Src is involved in Fak56-regulated NMJ growth, we performed genetic interactions between Fak56 and the Drosophila Src genes Src42A and Src64B. Reducing one gene dosage of either Src42A (Src42A^E1) or Src64B (Src64B^PI) in the Fak56^KG/N30 background displayed significant NMJ overgrowth, as scored for total branch length and bouton number (Figure 3A, B, E). Controls of Src42A^E1/+;Src64B^+/+ and Src42A^E1/+; Src64B^PI/+ in a wild-type background displayed no significant NMJ overgrowth (quantified in Figure 3E), suggesting that the efficiency of Src signaling at NMJs is dependant upon Fak56 activity in a dose-dependant manner. These results are consistent with a role for a FAK-Src complex in the restriction of NMJ growth.

We then tested whether severe Src mutants display NMJ growth defects. In the viable Src42A^E1/+; Src64B^PI/PI mutant that generates the least Src activity 41, the number of boutons was significantly increased and the total branch length was slightly enhanced (Figure 3C, E). To test whether Src has any contribution in the complete absence of Fak56 activity, we generated the combinatorial mutant Src42A^E1/+Fak56^N30/K24;Src64B^PI/+ and found that reducing the gene dosage of Src further increased the number of boutons in the Fak56 null mutant by 21%. In comparison to wild-type animal controls, Src42A^E1/+Fak56^N30/K24;Src64B^PI/+ mutants displayed an 80% increase in the bouton number and 25% increase in total branch length (Figure 3D). In summary, these genetic analyses suggest that Fak56 and Src have overlapping and distinct contributions in inhibiting NMJ growth.

In mammals, activation of FAK and the FAK homolog Pyk2 proceeds with an auto-phosphorylation step at the conserved Y397 of FAK and Y402 of Pyk2 164042, which corresponds to Y430 in Fak56 222543. To examine the activation of Fak56 at NMJs, we immunostained larval NMJs with the anti-FAK [pY³⁹⁷] antibody, which detects Fak56 activation at muscle attachment sites 28. As shown for NMJ 12/13, phospho-FAK (pFAK) was expressed strongly in Ib boutons (white arrows in Figure 4A1) and weakly in Is boutons (white arrowheads). Expression at NMJ 4 was also prominent (Figure 4B1). In co-staining for HRP, the pFAK signals could be found within boutons and inter-bouton tracks (Figure 4A1, B1), suggesting a presynaptic activation of Fak56. Strong pFAK expression was also detected within the incoming axons that were co-labeled by HRP (yellow arrowhead and inset image in Figure 4B1). Cytosolic punctate staining was also present in muscles. In Fak56^N30/K24 null mutants, pFAK signals in axons, presynapses and muscles were completely absent (Figure 4C1, C2), confirming the specificity of the anti-pFAK antibody in detecting Fak56 activation signals.

In the βν^1/1 integrin mutant, the pFAK staining in presynapses was dramatically reduced while the muscle punctate staining pattern was still retained (Figure 4D1, D2), indicating that integrin signaling mediated by the βν subunit is required for Fak56 activation in presynapses of NMJs. Taken together with the requirement of βν in restricting NMJ growth, these results suggest the presynaptic activation of Fak56 in restricting NMJ growth. To test this, the autoactivation site Y430 in Fak56 was mutated to phenylalanine to generate the UAS-Fak56^Y430F transgene. When ectopically expressed in neurons by elav-GAL4, Fak56^Y430F induced significant NMJ overgrowth phenotypes (Figure 4F). As a control, the wild-type Fak56 transgene caused slight but no significant reduction in NMJ growth (quantified in Figure 4G). This dominant negative effect by the Fak56^Y430F mutant suggests that phosphorylation at Y430 in the presynapse is critical for normal Fak56 function to constrain NMJ growth.

To further investigate the role of Fak56 at the presynapse, we generated an RNAi transgene to deplete Fak56 expression (see Materials and methods and Additional file 1F). Expression of the Fak56RNAi transgene in presynapses (elav>Fak56RNAi) resulted in an increase in both total branch length and bouton number of NMJs compared to the elav>LacZ control (Figure 5E, G). In contrast, Fak56 depletion in muscles using MHC-GAL4 retained normal NMJ phenotypes (not shown).

It has been shown that presynaptic ERK activation promotes larval NMJ growth 20. We then tested whether Fak56 had an effect on ERK activation at NMJs, which can be monitored by immunostaining for diphospho-ERK (dpERK) 44. The expression of dpERK was detected in punctate patterns in some but not all boutons (Figure 5A1, A2) 20.

We then examined whether dpERK expression at NMJs was altered by presynaptic depletion of Fak56 using RNA interference (RNAi). In elav>Fak56RNAi, dpERK expression was highly enriched in almost all boutons at the enlarged NMJ (Figure 5B, B1). To quantify the difference among wild-type and Fak56 mutants, the level of dpERK immuno-reactivity within the presynaptic region was normalized to that of co-stained HRP. We found that in elav>Fak56RNAi the ratio was increased by 3.3-fold when compared to that in elav>lacZ. Consistently, neuronal expression of the dominant-negative Fak56^Y430F also resulted in strongly enhanced dpERK expression to 3.1-fold (Figure 5C). The enhancement in dpERK expression levels in both approaches to block Fak56 function suggests that Fak56 activation suppresses ERK signaling in presynaptic boutons.

To test whether NMJ overgrowth phenotypes in Fak56 mutants were caused by the increased ERK activity, one wild-type allele of the ERK gene rolled (rl) 45 was replaced with the null allele rl^EMS698 46 in elav>Fak56RNAi larvae. The control heterozygous rl^EMS698/+ larvae displayed normal NMJ phenotypes. However, reduction of ERK gene dosage by 50% completely suppressed the NMJ overgrowth phenotypes observed in elav>Fak56RNAi (Figure 5D–F). The BMP/Gbb signaling pathway also promotes NMJ growth 47. We then tested whether the BMP/Gbb pathway would have a similar regulation in Fak56 mutant NMJs. Three mutants in the BMP/Gbb signaling pathway components were tested for potential genetic interactions with Fak56 but failed to significantly modify NMJ phenotypes in elav>Fak56RNAi larvae (Additional file 5). Taken together, these results suggest that Fak56 specifically downregulates the growth-promoting ERK signaling during NMJ growth.

It has been shown that ERK signaling regulates NMJ growth through the modulation of the protein levels of the cell-adhesion protein FasII 20. At NMJs, FasII protein levels are inversely correlated with ERK activation. In elav>Fak56RNAi mutants, the NMJ FasII level was reduced (Figure 6B1). Using the elav>LacZ as the reference, a 33.5% reduction in the ratio of the FasII level to the HRP level was detected (Figure 6A1, D). Comparison of FasII expression between wild type and Fak56^N30/K24 also revealed a 26.6% reduction in the Fak56 mutant (images not shown). Analyses of these two mutants suggest that Fak56 activity in presynapses is required for the full expression of FasII at NMJs. To examine whether Fak56-regulated FasII expression is mediated through ERK, the FasII protein level was examined in elav>Fak56RNAi;rl^EMS698/+. We found that the FasII protein level at NMJs of elav>Fak56RNAi was significantly restored by introducing the rl^EMS698 allele, with only 14.9% reduction compared to elav>LacZ (Figure 6C1, D). These results suggest that Fak56 regulation of FasII expression at NMJs is at least partially mediated by ERK.

Flies were reared at 25°C except where specifically indicated. Wild-type flies used in this study were the w¹¹¹⁸ strain. Mutant alleles Fak56^KG00304, mew¹, if^k27e, scb², mys¹, Src42A^E1, Src64B^PI and rl^EMS698 were obtained from the Bloomington stock center. βν¹, βν² 34, LanA^9–32, LanA²¹⁶ 63 and wb^4Y18 64 have been previously described. The various Fak56 alleles used in this study are described in detail in Additional file 1. The transgenic lines elav-GAL4 (X) (used in neuronal Fak56 knockdown and overexpression), elav-GAL4 (III) (used in neuronal Fak56 rescue), and UAS-LacZ were obtained from the Bloomington stock center.UAS-Fak56 28 and MHC-GAL4 65 have been described previously. The pUAST-Fak56RNAi construct was generated by subcloning two inverted Fak56 cDNA fragments (base pairs 629–1177) into the pUAST vector and the knockdown effect was examined (Additional file 1F).pUAST-Fak56^Y430F flies were generated from pUAST-Fak56 by PCR based site-directed mutagenesis. To enhance the Fak56RNAi transgene expression, embryos from the elav-GAL4 (X) and pUAST-Fak56RNAi cross were collected for 6 hours, kept at 25°C for 45 hours and shifted to 30°C until late third instar.

In all experiments, wandering late third instar larvae were dissected for analysis of NMJ phenotypes. After dissection, tissues were incubated in fixative solution (4% formaldehyde in 1× phosphate-buffered saline) for 20 minutes. For immunostaining, primary antibodies used were against synaptotagmin (mouse, 1:25; DHSB, Iowa City, IA, USA), HRP conjugated with TRITC (rabbit, 1:100; Jackson ImmunoResearch, West Grove, PA, USA), FAK [pY³⁹⁷] (rabbit, 1:50; Biosource-Invitrogen, Carlsbad, CA, USA), FasII (1D4, 1:100; DHSB) and dp-ERK-1/2 (mouse, 1:20; Sigma-Aldrich, St. Louis, MO, USA). Alexa 488-, Cy3- and Cy5-conjugated secondary antibodies and TRITC-phalloidin were used (Jackson ImmunoResearch).

Confocal images were acquired using a Zeiss LSM 510 Meta and processed using Adobe Photoshop CS. Images for quantification of NMJ branch length and bouton number were from a projection of 10 z-sections of 6.5–8 μm in total. To quantify the NMJ length and muscle area, the images were analyzed by a measurement tool in Zeiss LSM Image Examiner. For quantification of signal intensity at NMJs, images were acquired under the same scanning parameters. NMJs were outlined and the signal intensity was calculated by histogram analysis in Adobe Photoshop CS.

For sample preparation, dissected larval body walls (including the central nervous system and motor axons) were exposed in cold (4°C) HL3.1 Ca²⁺ free saline (70 mM NaCl, 5 mM KCl, 4 mM MgCl₂, 10 mM NaHCO₃, 5 mM trehalose, 115 mM sucrose, 5 mM HEPES pH 7.2) 66. Experiments were performed on muscle 6 of segment A3 in late third instar larvae. The segmental nerve was cut near the ventral ganglion. Preparations were then incubated in HL3.1 saline containing 0.2 or 1 mM CaCl₂ for electrophysiological experiments at room temperature (22°C). For stimulation and recording, a glass microelectrode (30–50 MO in resistance) filled with 3 M KCl was impaled in the sixth muscle of the third abdominal segment to record the EJPs. The mEJPs occurring in the background within 200 seconds were obtained without any stimulation on the segmental nerve. To evoke an EJP, the segmental nerve was stimulated every 30 seconds through the cut end with a suction electrode with 0.1 ms of pulse duration at 2 times the threshold voltage. Once the threshold voltage was reached, the size of EJPs remained unchanged despite the increase in stimulating voltage. Signals were digitized at 64 KHz by a PCI-6221 data-acquisition card (National Instrument, Austin, Texas, USA), and saved on an IBM compatible PC for analysis.