Efficient synaptic transmission requires a high local specialization of pre- and postsynaptic cells. At the neuromuscular junction (NMJ), these specializations include aggregates of acetylcholine receptors (AChRs). Proteins of the postsynaptic apparatus implicated in the aggregation of AChRs include utrophin, a synapse-specific homolog of dystrophin (Ohlendieck et al., 1991; Bewick et al., 1992; Tinsley et al., 1992, 1994), α- and ß-dystroglycan (Ibraghimov-Beskrovnaya et al. 1992; Fallon and Hall, 1994), and rapsyn (Frail et al., 1988; Apel et al., 1995), thought to link AChRs to the cytoskeleton. Agrin, a heparansulfate proteoglycan that is synthesized by motor neurons and deposited into the synaptic basal lamina, was shown to trigger redistribution of AChRs to form postsynaptic aggregates (McMahan, 1990; Wallace, 1996; Ruegg and Bixby, 1998). It is now clear that agrin organizes postsynaptic differentiation by stimulating MuSK, a receptor tyrosine kinase that is expressed selectively at in skeletal muscle (Jennings et al., 1993; Valenzuela et al., 1995; Glass et al., 1996). Agrin and MuSK are essential for synapse formation, as mice lacking agrin or MuSK fail to form neuromuscular synapses and consequently die at birth because of a failure to move or breathe (Gautam et al., 1996, DeChiara et al., 1996). Nevertheless, the mechanisms by which agrin activates MuSK are poorly understood. Agrin stimulates the rapid tyrosine phosphorylation of MuSK in myotubes, but, if transiently expressed in fibroblast or myoblasts, is not phosphorylated by agrin (Glass et., al 1996). These data thus indicate that activation of MuSK depends on at least one additional component expressed in myotubes but not in myoblasts. The current hypothesis predicts that this component, which was termed muscle-associated specificity component (MASC; Glass et al., 1996), together with MuSK form an agrin receptor complex. In this thesis, we investigated of how agrin activates MuSK and of how the signal is transmitted further downstream leading to the accumulation of AChRs at the synapse. In a first part, we aimed to identify proteins that are associated with MuSK using the membrane bound split-ubiquitin system, a method that is based on the yeast two-hybrid (YTH) system. In contrast to the original YTH system, this novel method allows to screen for proteins that pass the membrane or are associated with it. In a first step, we showed that a bait and pray both containing a constitutively active form of MuSK, activates the reporter genes by self dimerization indicating that this system is a valuable method for identifying components not intact with MuSK. Moreover, we also demonstrated that the bait MuSK and a soluble form of agrin used in the YTH-screen are correctly expressed indicating that the system does not generate false-positive signals and can be used for screening. After these validation experiments, a total number of 3x107 clones were screened resulting in >5’000 putative candidates. However, none of them could be could not be reconfirmed in bait dependency tests. After several attempts to improve the method per se and to decrease the number of candidates, we were forced to drop the project as it turned out that the system had many intrinsic problems that could not be solved in a useful time window. In a second project, we therefore concentrated to on mapping sites in agrin important for its MuSK phosphorylation and α-dystroglycan binding property. As agrin’s AChR clustering activity is mediated by the most C-terminal laminin G-like domain we concentrated on this region. Moreover, one particular splice variant containing an exon of 8 amino acids in length within the B/z splice site of the LG3 domain was shown to be the most potent isoform, whereas splice variants lacking an insert in the B/z splice site are not active at all (Gesemann et al., 1996). Guided by the crystal structure of the LG3 domain derived from different splice variants (Stetefeld et al., 2004), we analyzed the contribution of single amino acids within the B/z-8 exon of agrin in activate MuSK and show that the activity resides mostly within the side chains of a three-peptide motif ‘Asn-Glu-Ile’, which is also highly conserved between species. In addition, we demonstrate that amino acids flanking the B/z splice site also strongly contribute to agrin’s activity. Finally, we demonstrate that binding affinity to α-DG positively affects its MuSK phosphorylation activity. Based on these data we propose a model where α-DG plays an auxiliary role in capturing agrin at the muscle surface and thus efficiently presents the molecule to the agrin receptor complex. In summary, the results reported in this thesis are a further step to elucidate the detailed mechanism of how agrin instructs the muscle to form postsynaptic structures. As similar mechanisms are also of work in the formation of synapses in the brain, these results are likely to be also important for furthering the understanding of how these structures are formed and altered during development and in process of learning and memory.