Antibodies against p75 neurotrophic receptor (Cat No; sc-6188) were obtained from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Antibodies to Gab1 (Cat No; 3232S), horseradish peroxidase (HRP)-linked anti-rabbit IgG (Cat No; 7074), HRP-linked anti-mouse IgG (Cat No; 7076S) were obtained from Cell Signaling Technology (Boston, MA, USA). Antibody against S100 (Cat No; Z0311) was obtained from Dako (Carpinteria, CA, USA). Anti-neurofilament (NF) antibody (Cat No; N4142) were purchased from Sigma-Aldrich (St. Louis, MO, USA). FluoroLink Cy3-labelled goat anti-rabbit (Cat No; PA43004) and anti-mouse IgG (Cat No; PA43002) were purchased from Amersham Biosciences/GE Healthcare (Piscataway, NJ, USA). Alexa Fluor 488-labeled donkey anti-rat (Cat No; A21208) and anti-rabbit IgG (Cat No; A21206), and Alexa Fluor 488 conjugated alpha-bungarotoxin (Cat No; B13422, α-BT) were purchased from Molecular Probes (Carlsbad, CA, USA). All other undesignated reagents were purchased from Sigma-Aldrich.

Genotyping for the Gab1 locus, myelin protein zero (MPZ)-Cre transgene, and Gab2 knockout mice were performed by PCR analysis of tail genomic DNA as we described previously [8]. Myogenic factor 5-Cre (Myf5-Cre) transgenic mice (B6.129S4-Myf5tm3(cre) Sor/J; Stock No: 007893B6) were purchased from Jackson Laboratory (Bar Harbor, ME, USA). All surgical procedures were performed according to protocols approved by Dong-A University Committee on animal research, which follows the guidelines for animal experiments that were established by the Korean Academy of Medical Sciences.

For Western blot analysis, muscles were harvested and transferred to tubes containing lysing matrix beads and a lysis buffer [8]. Tubes were placed on TissueLyser LT (Qiagen) and then subjected to homogenization for 2~3 min. The lysates were centrifuged at 8,000 g for 10 min at 4℃, and the supernatant was collected. Protein (25~35 µg) was separated by SDS-PAGE, and then transferred onto a nitrocellulose membrane (Amersham Biosciences). After blocking with 0.1% Tween-20 and 5% nonfat dry milk in Tris buffered saline (pH. 7.2, TBS) at room temperature for 1 hr, the membranes were incubated with primary antibodies (1:500~1,000) in TBST containing 2% nonfat dry milk at 4℃ overnight. After three 15 min washes with TBST, the membranes were incubated with a horseradish peroxidase-conjugated secondary antibody (1:3,000) for 1 hr at room temperature. The signals were detected using the ECL system (ECL Advance kit, Amersham Biosciences).

Whole mount staining for NMJ analysis was performed according to the previous article with minor modifications [18]. Briefly, dissected hind-limbs (neonates) or extensor digitorum longus (EDL) muscles were fixed with 4% PFA, and EDL muscles were teased apart into four pieces under a stereomicroscope. Teased EDL muscle fibers were blocked with blocking buffer (5% BSA, 0.5% Triton X-100/PBS) and were then incubated with primary antibodies (S100 or NF) for 1~3 days at 4℃. EDL muscles were rinsed three times (2 hr per wash) with 0.5% Triton X-100 in PBS and were incubated with Alexa Fluor 488-conjugated α-BT (1:1,500), together with Cy3-labeled secondary antibodies (1:4,000 dilution), for 24 hr at 4℃. Samples were washed and whole-mounted on glass slides with mounting medium. Z serial sections were collected with a Zeiss confocal laser-scanning microscope (LSM700, Carl Zeiss) and projected into a single image. AChR staining with α-BT reveals the typical pretzel-like fluorescent cluster which normally consists of two or four AChR islands or strips. The area of each AChR cluster (an endplate) was determined by measuring the total areas of AChR islands in a cluster using Image J software. At least 200 stacked NMJ images were captured from 3~6 animals for analyzing AChR morphology.

Polyinnervation is observed in the immature state, wherein NMJs are innervated by more than one motor axon [19]. To count the polyinnervated endplates, at least 100 stacked NMJ images were captured from 3~6 animals at P10 and in adulthood. To count tSC at NMJ, Hoechst staining was used in combination with S100 labeling to mark tSCs at NMJs in EDL muscles at P10 and approximately 100 stacked images were used (n=3).

Primary SCs were isolated from the sciatic nerves of 4~5 day old Sprague-Dawley pups as we previously reported [8]. When the cell density reached 70% confluency, lentiviral particles expressing Gab1-shRNAi or scrambled shRNAi were administrated at various concentrations for 18~24 hr. Then, the medium was replaced with fresh DMEM containing 10% FBS and cultured for another 24 hr. The medium was then replaced with the growing medium (DMEM containing 1% FBS, NRG1 (30 ng/ml), forskolin (5 µM)) for one day and the morphology of SCs was analyzed following immunostaining against SC markers such as S100 or p75 neurotrophin receptor (p75). The length of SCs was measured from captured image using Image J software. Approximately 200 SCs from each of shRNAi-infected SCs (n=3) were analyzed.

To analyze the role of Gab1 in SC proliferation, BrdU incorporation assay were performed. Scrambled or Gab1-shRNAi lentivirus-infected SCs were plated on PDL-coated-96 well plates at a density of 8×10⁴ cells/well, and cells were cultured overnight in DMEM containing 1% FBS and 50 ng/ml NRG1. For the BrdU assay, infected SCs were plated on a coverslip coated with PDL at a density of 1×10⁵ cells/well and cultured in DMEM containing 0.5% FBS and 200 ng/ml NRG overnight. Cells were incubated with 20 µM BrdU for 6 hr and then fixed with 4% PFA for 10 min. BrdU immunostaining was performed as we described previously [20], and the number of BrdU-positive cells per coverslip was counted and three independent experiments were performed for quantitative analysis.

Statistical analysis was performed by using GraphPad Prism software (GraphPad, San Diego, CA). p-values were from Student's two-tailed test, and results were expressed as mean and SEM.

In Gab1(-/-) embryos, the migration of muscle precursor cells into limbs is hampered, and this genotype is embryonic lethal [21], precluding the determination of the role of Gab1 in the postnatal development of the NMJ. To characterize this process, skeletal muscle-specific Gab1 conditional knockout (Gab1-MSKO) mice were established by removing floxed Gab1 (Gab1^f/f) with Cre recombinase under the myogenic factor 5 (Myf5) promoter, which begins to be active at approximately E10 [22]. Gab2 KO mice, that is viable [23], were also used to examine the role of Gab2 in postnatal NMJ development. Gab1-MSKO mice showed smaller body size compared to the control mice (Gab1^f/f) at birth and it was maintained throughout postnatal life as we recently reported [24]. Western blot analysis showed that the expression levels of Gab1 in skeletal muscles, but not in the brain, were significantly reduced in adult Gab1-MSKO mice compared to their control mice and that the expression of Gab2 both in the brain and skeletal muscles was not detected in Gab2 KO mice (Fig. 1B). The low levels of Gab1 expression in the skeletal muscles of Gab1-MSKO mice might represent the expression of Gab1 in non-muscle tissues such as connective tissue coverings and vessels within muscles.

We next examined the role of muscular Gab1 and Gab2 in NMJ development by analyzing the morphological characteristics of AChR clustering using fluorescein-conjugated α-BT. EDL muscles from adult Gab1^f/f and Gab1-MSKO mice (8~10 weeks old) were whole-mount stained with α-BT and an anti-neurofilament antibody to label motor nerve terminals. As shown in Fig. 1C, normal AChR staining in Gab1^f/f mice revealed the typical pretzel-like endplate structure (a NMJ, an AChR cluster) that contains two to four AChR islands [25], whereas AChR clustering in Gab1-MSKO mice exhibited abnormally fragmented endplate morphology with a higher number of α-BT-labeled AChR islands in a cluster (2.7±0.06 vs. 6.6±0.19, p<0.01) (Fig. 1C, D). We also measured the area of α-BT-labeled clusters, and found that the mean area of AChR clusters in Gab1-MSKO mice was smaller than that of Gab1^f/f mice (284.4±9.6 µm² vs. 352.1±5.7 µm², p<0.01) (Fig. 1E). In contrast, the number of islands and the mean area of AChR clusters in Gab2 KO mice were not different from those of Gab2 (+/-) mice (Fig. 1G~I).

We investigated the developmental maturation of AChR clusters in Gab1^f/f and Gab1-MSKO mice by classifying the maturation of endplate morphology at P1 and P14. Based on previous reports [19,26], the maturation of endplate morphology was classified into ovoid plaque, perforated and branched morphologies in the order of maturation and the branched form shows AChR islands in a cluster (Fig. 2A). At P1, approximately 75% and 25% of endplates were ovoid plaque and perforated types, respectively, in control mice, whereas in Gab1-MSKO mice, only 10% of endplates showed perforated morphology (Fig. 2B, C). At P14, most endplates (~60%) belonged to perforated type morphology in control mice. However, in Gab1-MSKO mice, there was a marked delay in the maturation of endplate morphology, with the immature ovoid plaque type being the most frequent type of endplate morphology even at P14 (Fig. 2B, D).

An immature NMJ is innervated by more than one motor axon (polyneuronal innervation), and synaptic elimination during the second postnatal maturation period results in mononeuronal innervation in adulthood [13]. It has been believed that tSCs play an essential role in synaptic elimination [13], possibly via receiving axonal NRG1-type III signals [16]. To determine the role of Gab1 in tSCs in synaptic elimination, we thus examined the polyinnervation states of motor endplates in mice in which the gab1 gene was excised in tSCs using MPZ-cre transgenic (MPZ is consistently expressed in tSCs [27]) (Gab1-SCKO) during postnatal development. We immunostained using an anti-neurofilament (NF) antibody and α-BT in the EDL muscle to count NMJs containing more than two axons. At P10, 8.25% and 17.54% of motor endplates were polyneuronally innervated in control and Gab1-SCKO mice, respectively (p<0.05, n=3) (Fig. 3A, C). In adults, synaptic elimination was completed in control mice (fewer than 1% were polyneuronally innervated) whereas 8.07% of endplates in Gab1-SCKO mice still received polyneuronal innervation (p<0.01, n=3) (Fig. 3B, C).

Because synaptic elimination affects the maturation of endplate morphology [13], we analyzed the morphology of endplates in adult Gab1-SCKO mice. We found that postsynaptic AChR aggregates were fragmented into small islands, a pattern similar to that found in adult Gab1-MSKO mice, in Gab1-SCKO mice (Fig. 3D). However, the mean area of AChR clusters was not significantly different between control and Gab1-SCKO mice (Fig. 3E). Finally, we investigated the developmental maturation of AChR clusters in Gab1^f/f and Gab1-SCKO mice at P14 as we done in Gab1-MSKO analysis. Approximately 20% of endplates belonged to immature ovoid type morphology in control mice whereas ~50% of endplates remained as ovoid type in Gab1-SCKO mice (Fig. 3F). Thus, Gab1 deficiency in tSCs resulted in abnormal development not only in motor innervation but also in endplate maturation.

It was previously shown that synaptic elimination by tSCs requires an adequate number of tSCs and tSC process elongation [13,16,26]. Furthermore, it has been reported that the number of tSCs during postnatal development would be regulated by axonal NRG1-type III [16]. We thus counted the number of tSCs in endplates of Gab1-SCKO mice at P10 after triple fluorescent staining with an anti-S100 antibody, α-BX and Hoechst staining. The mean number of tSCs in endplates of EDL muscle was 3.1 at P10 in control mice, whereas it decreased to 2.54 (p<0.01) in Gab1-SCKO mice (Fig. 4A, B), suggesting that Gab1 is required for the generation of adequate number of tSCs during postnatal development.

To examine whether Gab1 is required for the proliferation of NRG1-induced SCs, we cultured primary rat SCs and investigated the effect of Gab1 silencing in NRG1-induced SCs proliferation using BrdU incorporation analysis. One day after lentiviral Gab1 shRNAi or scrambled shRNAi infection, cells were cultured in the presence of recombinant NRG1 (200 ng/ml) for one day and then proliferating cells were determined using BrdU incorporation assay. We found that the number of BrdU-positive cells was significantly reduced in Gab1-silenced primary SCs, compared to scrambled shRNAi-infected SCs (Fig. 4C, D), suggesting the implication of Gab1 in NRG1-induced SC proliferation.

We next investigated whether Gab1 mediates the process-forming activity of NRG1 in cultured SCs. To address this question, primary SCs, that were infected with Gab1 or scrambled shRNAi lentivirus, were grown in the presence of NRG1 for one day, and the length of SCs was measured after immunostaining against a SC marker protein, p75. Gab1-silenced SCs exhibited shorter processes compared to control cells which were infected with scrambled shRNAi (Fig. 4E, F). Together, our results indicate that proper development of tSCs via Gab1 may regulates the rate of redundant motor axon pruning at developing NMJ.