Rodent nerve-muscle cell culture system for studies of neuromuscular junction development: refinements and applications.

Understanding of vertebrate neuromuscular junction (NMJ) development has been advanced by experimentation with cultures of dissociated embryonic nerve and skeletal muscle cells, particularly those derived from Xenopus and chick embryos. We previously developed a rodent (rat) nerve-muscle coculture system that is characterized by extensive induction of acetylcholine receptor (AChR) aggregation at sites of axonal contact with myotubes (Dutton et al., 1995). In this article, we report modifications of this culture system and examples of its application to the study of NMJ development: (1) We describe improved methods for the enrichment of myoblasts to give higher yields of myotubes with equal or greater purity. (2) We demonstrate lipophilic dye labeling of axons in cocultures by injection of dye into neuron aggregates and show the feasibility of studying the growth of living axons on myotubes during synapse formation. (3) We describe the preparation of a better-defined coculture system containing myotubes with purified rat motoneurons and characterize the system with respect to axon-induced AChR aggregation. (4) We demonstrate dependence of the pattern of axon-induced AChR aggregation on muscle cell species, by the use of chick-rat chimeric co-cultures. (5) We provide evidence for the role of alternatively-spliced agrin isoforms in synapse formation by using single cell RT-PCR with neurons collected from co-cultures after observation of axon-induced AChR aggregation. Microsc. Res. Tech. 49:26-37, 2000. Published 2000 Wiley-Liss, Inc.

Cytoskeleton-membrane interactions at the postsynaptic density of Xenopus neuromuscular junctions.

Acetylcholine receptors of mature muscle fibres are concentrated in the postsynaptic membrane by mechanisms that are not yet understood. As one possibility, receptors might be anchored to cytoskeletal elements in the postsynaptic density that is located beneath the membrane where receptors are concentrated. To address this possibility, we examined the cytoskeleton at the postsynaptic density and determined the organization of cytoskeletal filaments relative to clustered acetylcholine receptors (AChR). Xenopus nerve-muscle co-cultures were sheared to expose the cytoplasmic membrane surface, then quick-frozen, deep-etched, and rotary-replicated. Areas with a high concentration of AChR had aggregates of particles protruding from the cytoplasmic surface of the membrane, with actin microfilaments attached to and cross-linking the aggregates. Microfilaments contacted only a few of the particles in an aggregate. These findings suggest that short-range interactions may bind individual AChR into small aggregates, while microfilaments tie these aggregates together at the nerve-muscle junction.

Microfilament-membrane interactions in Xenopus myocytes.

We used quick-freeze, deep-etch, rotary-replication transmission electron microscopy to determine at molecular resolution the organization of microfilaments at the cytoplasmic surface of the sarcolemma of Xenopus myocytes. We demonstrate that actin microfilaments interact with the sarcolemma in two distinct ways. In one, which resembled focal contacts in Xenopus fibroblasts [Samuelsson et al., 1993: J. Cell Biol. 122:485-496], bundles of microfilaments approached the sarcolemma at sites containing aggregates of membrane-associated particles. Immunogold cytochemistry showed that these particle aggregates contained vinculin, talin and beta 1-integrin. In the second, which covered most of the cytoplasmic surface of the sarcolemma, individual actin microfilaments formed an extensive, lattice-like array. Particle aggregates associated with this array of actin microfilaments also labeled with antibodies to vinculin, talin and beta 1-integrin. The unique, lattice-like association of actin microfilaments with the membrane in Xenopus myocytes suggests that the organization of actin filaments over most of the sarcolemma is distinct from focal contacts, mediating widespread associations of the actin cytoskeleton with the cytoplasmic membrane face.

Acetylcholine receptor aggregation at nerve-muscle contacts in mammalian cultures: induction by ventral spinal cord neurons is specific to axons.

We used a novel mammalian coculture system to study ACh receptor (AChR) redistribution and synaptic structure at nerve-muscle contacts. Ventral spinal cord (VSC) neurons were plated on cultures containing extensive myotubes but few fibroblasts. Neurite-induced redistribution of AChRs occurred within 6 hr after plating neurons and was maximal between 36-48 hr. This AChR redistribution appeared in two patterns: (1) AChR density at sites directly apposed to the neurite where neurites crossed preexisting AChR patches was sharply reduced, (2) Newly aggregated AChRs formed swaths lateral to the neurite path. VSC neurons induced more AChR aggregation than hippocampal, superior cervical ganglion and dorsal root ganglion neurons. The 43 and 58 kDa postsynaptic proteins were colocalized with AChR-enriched domains in all VSC neurite-induced aggregates whereas the colocalization of laminin was variable. Electron microscopy of regions with neurite-induced AChR aggregation showed postsynaptic membrane specializations characteristic of developing synapses and, in older cultures, features of more mature synaptic structure. Thus, the coculture system is useful for studying early stages of neuromuscular junction (NMJ) formation. Neurites in these cocultures were identified as axons or dendrites by morphological criteria and by their immunoreactivity for synaptophysin and phosphorylated heavy neurofilament subunits or for microtubule associated protein 2 (MAP2), respectively. Axons showed a 10-fold higher induction of AChR aggregation than did dendrites. Thus, at least one essential signaling molecule necessary for the induction of AChR aggregation at sites of interaction with muscle appears to be expressed in a polarized fashion in developing VSC neurons.

Clustered acetylcholine receptors have two levels of organization in Xenopus muscle cells.

We studied the organization of acetylcholine receptor (AChR) clusters by shearing cultured Xenopus muscle cells with a stream of buffer, and preparing rotary replicas of the exposed cytoplasmic surface of the sarcolemma. AChR clusters contained numerous particles that protruded from the sarcolemma and formed an irregular array composed of discrete aggregates. AChR were located within these particle aggregates, as shown by comparison of the replicas to labeling by fluorescent alpha-bungarotoxin, and by immunogold cytochemistry with antibodies specific for the receptor. The aggregates were cross-linked by a dense network of 7 nm filaments that replicated with the banded pattern characteristic of actin microfilaments. The organization of receptors into the small aggregates was independent of the organization of these aggregates into clusters, as alkaline extraction removed the microfilament network and disrupted the irregular array of particle aggregates, but did not disperse individual receptors from the aggregates. We conclude that two levels of interactions organize AChR clusters in Xenopus muscle cells: short-range interactions that assemble individual AChR into small aggregates, and long-range interactions, perhaps mediated by actin microfilaments, that anchor the aggregates into larger clusters.

Structures linking microfilament bundles to the membrane at focal contacts.

We used quick-freeze, deep-etch, rotary replication and immunogold cytochemistry to identify a new structure at focal contacts. In Xenopus fibroblasts, elongated aggregates of particles project from the membrane to contact bundles of actin microfilaments. Before terminating, a single bundle of microfilaments interacts with several aggregates that appear intermittently over a distance of several microns. Aggregates are enriched in proteins believed to mediate actin-membrane interactions at focal contacts, including beta 1-integrin, vinculin, and talin, but they appear to contain less alpha-actinin and filamin. We also identified a second, smaller class of aggregates of membrane particles that contained beta 1-integrin but not vinculin or talin and that were not associated with actin microfilaments. Our results indicate that vinculin, talin, and beta 1-integrin are assembled into distinctive structures that mediate multiple lateral interactions between microfilaments and the membrane at focal contacts.

Quick-freeze, deep-etch replication of cells in monolayers.

We have made several technical improvements for quick-freeze, deep-etch replication of monolayers of cells grown on, or attached to, glass coverslips. Cells studied include muscle cells of rat and Xenopus cultured on glass coverslips, and erythrocytes attached to coverslips coated with poly-L-lysine. We describe methods for identifying particular areas of cultures, e.g., clusters of acetylcholine receptors on muscle cells, by light microscopy and then relocating these areas after replication. For good preservation of structure by quick-freezing, it is necessary to ensure that the surface to be frozen is covered by a minimum depth of water (less than 10 microns). Insufficient or excess water left on the sample during freezing causes recognizable artifacts in its replica. We describe two ways to control the water table--one by improving visual control of water removal, the other by blowing excess water off the sample surface with a jet of nitrogen applied during its descent to the freezing block. Finally, we describe a new specimen holder that allows us to etch and replicate six samples at once with good thermal contact between the stage and samples.

Novel peptide neuronal system in rat brain and pituitary.

Immunohistofluorescence studies of the rat central nervous system with antibodies to Phe-Met-Arg-Phe-NH2 (molluskan cardioexcitatory peptide) revealed a widespread neuronal system in the brain, spinal cord, and posterior pituitary. Immunoreactive axons and cell bodies were mainly located in cortical, limbic, and hypothalamic areas. Immunostaining of serial sections of the brain and pituitary showed that the Phe-Met-Arg-Phe-NH2 immunoreactive neurons were different from neurons labeled by antibodies to either Met-enkephalin or the putative Met-enkephalin precursor Tyr-Gly-Gly-Phe-Met-Arg-Phe, which is structurally related to Phe-Met-Arg-Phe-NH2. Control staining by antiserum absorption and radioimmunoassay indicated that the antibodies that caused the specific immunofluorescence recognized peptides with an amidated Arg-Phe sequence at the carboxyl terminus.