Introduction Skeletal muscle fibres are permanent multinucleated, non-mitotic cells. Consequently, there is considerable interest in how such a unique cell type is formed. Muscle fibres are derived from multinu-cleated myotubes which are themselves formed, during embryonic development, by the fusion of mononucleated myoblasts. Myoblast fusion provides a naturally occurring process in which mechanistic proposals from the study of fusion in moder fusion systems can be tested. Conversely, proposals derived from myo-blast fusion studies can be profitably applied to appropriate vesicular model systems. In vivo, the normal development of muscle has been observed in many species. Although the transition from mono-to predominantly multi-nucleated muscle cells can be observed in sections of embryonic muscle, this is asynchronous and poses almost insurmountable difficulties for the biochemical analysis of discrete stages. Tissue culture techniques dramatically simplify this analysis. Holtzer et al.(1958) were the first to show that the fusion of myoblasts occurred in primary culture, and since then fusion studies have emphasized the impor-tance of this powerful model in vitro. Muscle cell fusion can be studied using both primary cultures and identified cell lines. Primary cells are prepared from embryonic tissues by mechanical or enzymic disaggregation (see Konigsberg, 1978) and cultured in medium supplemented with serum (horse serum or foetal calf serum) and embryo extract. Primary muscle cells can be prepared from a variety of insect, avian and mammalian species. For fusion the most commonly used are embryonic chick or quail muscle cells. Cell lines which also undergo some stages of myogenesis, including fusion, have been isolated. The most extensively studied are the L6 and L8 lines isolated from rat muscle tissue (Yaffe, 1968; Yaffe & Saxel, 1977). Whilst several interesting reports involving the use of cell lines exist and are quoted in this Review, it should be remembered that these cells are trans-formed and they may not be truly representative of a physiological situation. For example, the fusion of L6 myoblasts does not have the absolute requirement of primary cultures for embryo extract. Some common features are observed in all systems. Myoblasts proliferate until they withdraw from the cell cycle. They align both during and after proliferation, and then fuse. Fig. 1 shows the transition from proliferating myoblasts, to aligned myoblasts, to fused myotubes. The cells that fuse are terminally differentiated, post-mitotic myo-blasts which are generated from actively proliferating precursors. Myoblast fusion was last exten-sively reviewed by Bischoff (1978) and readers are referred to that review for background information not covered in the present article. Many reports exist in which additions have been made to the culture media that result in less fusion or a lack of fusion being observed 24h, or more, later. Such studies are, in general, uninformative about the actual process of membrane union and thus have not been quoted in this Review. Instead, I have attempted to review selectively those reports that add to our knowledge of the actual process of myoblast fusion.

Alignment and recognition Myoblast fusion is very cell specific. Although heterotypic fusion between, for example, rabbit and rat myoblasts has been demonstrated, fusion between rat heart or kidney cells and myoblasts does not occur (see Bischoff, 1978). It is possible that the specificity of fusion is a reflection of the process of recognition and adherence of the cells during the process of alignment (seebelow). Alignment is the parallel apposition ofthe long axes of the myoblasts. The …