Testing the Agrin Hypothesis

the Agrin Hypothesis and providing important new infor-Neuroscience Program mation on reciprocal signaling mechanisms that regu-Department of Physiology and late synaptic differentiation (illustrated in Figure 1). Gau-Howard Hughes Medical Institute tam et al. (1996) describe the phenotype of a mouse in University of California, San Francisco which the differentially spliced exons required for the San Francisco, California 94143–0724 AChR clustering activity of agrin are deleted. DeChiara et al. (1996) describe the targeted mutation of the puta-The mature neuromuscular junction provides an ex-tive receptor tyrosine kinase MuSK, which parallels the treme example of reciprocal subcellular differentiation. agrin mutant phenotype very closely. Finally, Glass et During development of this interface, small regions of al. (1996) demonstrate that MuSK is a component of the nerve, muscle, and the interposed basal lamina form a receptor complex that mediates agrin signaling. morphologically complex structure evolved to maximize Gautam et al. (1996) provide a critical test of the agrin the speed and reliability of synaptic transmission (reviewed in Hall and Sanes, 1993). Almost 20 years ago, McMahan and colleagues demonstrated that information sufficient to direct differentiation of either nerve terminals or muscle fibers resides in the morphologically specialized basal lamina at the differentiated neuromus-cular junction. An early step in differentiation is accumulation under nerve terminals of preexisting acetylcholine receptors (AChR) on myotube surfaces. Receptor density is also enhanced by local synthesis of new receptors. Agrin was purified as an activity present in synaptic basal lamina which promotes AChR clustering on cultured myotubes. Agrin is a 200 kDa protein, synthesized in embryonic motor neurons, that is released from nerve terminals and deposited in the synaptic basal lamina (Ferns and Hall, 1992). Agrin is a large basal lamina glycoprotein with many distinct domains. The C-terminal region, which includes EGF repeats and laminin G-like domains, contains the AChR clustering activity. Alternative splicing at three sites within this region generate agrin iso-forms with variable activities in receptor clustering assays. Of particular importance, motor neuron agrin contains an 8 amino acid insert at the most C-terminal of these sites and is the most active isoform in receptor clustering assays. In contrast, agrin is also expressed by skeletal muscle, but muscle-derived agrin lacks this insert and is therefore not active in AChR clustering assays. The Agrin Hypothesis (McMahan, 1990) predicts that neural agrin is transported to nerve terminals where it is released and incorporated into the nascent synaptic basal lamina. By interacting …