Myosins, actin-based motor proteins, are expressed as multiple isoforms in all eukaryotic cells. They are oligomers consisting of one or two heavy chains to which one or more light chains are non-covalently attached. Myosins have been classified into 18 families based on the amino acid sequence differences in the N-terminal head domains, which contain highly conserved regions including actin- and nucleotide-binding sites 12. The tail of myosin is the most variable domain and seems to be responsible for the specific role myosin plays in the cell.

Functional activities of most myosins such as actin-dependent ATPase activity or ability to move actin filaments in vitro are regulated in several ways, mainly by phosphorylation of the regulatory light chain, Ca²⁺-binding, or phosphorylation of the heavy chain 13 It has been previously claimed that the myosin heavy chain (MHC)undergoes tyrosine phosphorylation during insulin-mediated skeletal muscle differentiation, thus linking signal transduction to highly ordered myosin assembly 4. Insulin modulates an association of myosin with C-terminal SRC kinase (Csk), a tyrosine kinase signalling molecule, and these interactions are fundamental in skeletal muscle differentiation. Although the claims of tyrosine phosphorylation of MHC in vivo remain somewhat controversial, tyrosine phosphorylation of non-muscle MHC IIa has also been implicated as an early event in human platelet activation 5. To settle this controversy -and establish the role, if any, of MHC tyrosine phosphorylation it is important to identify sites at which such phosphorylation events may occur.

We have mapped potential phosphorylation sites on the skeletal muscle myosin heavy chain utilizing an integrated bioinformatics approach, supporting web-based motif predictions with evolutionary and structural data. Of all the sites analyzed in the bioinformatics approach, the data suggest Y163 and Y1856 as the leading candidates for insulin-mediated tyrosine phosphorylation.

In total, twenty-three myosin heavy chain sequences were used for tyrosine phosphorylation motif prediction, which were divided into five groups of orthologous sequences (Figure 1). Important motifs are likely to be conserved during evolution and so we considered only those sites that were predicted to be phosphorylation motifs in all the sequences of at least one orthologous group. Because phosphorylation site predictors have a tendency to over-predict, we increased stringency by accepting only those motifs that received a NetPhos score of 0.8 or higher, or were predicted by both NetPhos and Scansite, in at least one human adult skeletal myosin heavy chain. This yielded fourteen putative sites (Table 1). Of these, six were predicted by both methods, including two motifs that were conserved across all sequences (MYH1_HUMAN Y163 and Y1856).

To be phosphorylated, tyrosine residues must be accessible on the surface of the protein. Although the three-dimensional conformations of homologous myosin molecules will not be identical, the high degree of sequence conservation between human adult skeletal muscle myosin heavy chains and the three myosin sequences present in PDB allowed the inference of solvent accessibility. This was confirmed by the generally good agreement in surface accessibility measures both between models and between the different myosin chains of 1BR2 (data not shown). From these data, two sites (Y286 and Y435) were buried while a further two (Y313 and Y504) had very low solvent accessibility (Table 1). 3D data were not available for the four tyrosines in the C-terminal of the protein.

If tyrosine phosphorylation of MMHC-II is part of a signalling cascade, it is likely that some other protein will interact with the phosphotyrosine. We used Scansite to look for phosphotyrosine interaction motifs and found three SH2 domain recognition motifs that matched potentially exposed phosphorylation sites. Because Scansite also identifies the interacting protein, we interrogated the Gene Cards 20 entry for each kinase and SH2 domain for expression patterns. Only four kinases and two SH2 domains had evidence from UniGene 21 or SAGE 22 of expression in skeletal muscle, while only one kinase (INSR) and one SH2 domain (PIK3R1) had evidence from both (Table 2). Interestingly, both of the latter pair were predicted to interact with the same, totally conserved, motif (MYH1_HUMAN Y163). Furthermore, both are involved in insulin-mediated pathways (see Discussion). Two kinases expressed in skeletal muscle were predicted to interact with Y1856. These were an SRC kinase and ABL1, which interacts with SORBS1 following insulin stimulation 23.

Bioinformatics alone cannot identify a functional motif; supporting experiments will always be needed for conclusive evidence. Nevertheless, while other sites cannot be categorically excluded, the combined data presented here identify Y163 and Y1856 as the most likely sites for tyrosine phosphorylation events in skeletal muscle. Both NetPhos and Scansite predicted these motifs for all mammalian adult skeletal myosin heavy chain sequences analysed, indicating strong evolutionary conservation (Figure 1). A kinase predicted to be responsible for phosphorylation of each site is expressed in skeletal muscle, as was an SH2 domain protein that was predicted to interact with a phosphotyrosine at Y163. Analysis of predicted secondary structures and homologous 3D structures indicates that these sites may be accessible on the protein surface. The position of Y163 as part of an alpha helix within the globular myosin head domain does raise concerns that it is potentially difficult to phosphorylate, even though it is on the surface of the domain. Nevertheless, depending on the relative conformations of the solved in vitro chicken myosin structures compared to in vivo human myosin, Y163 might still be available for phosphorylation. Y1856 is in region of low predicted secondary structure (Table 1), indicative of a flexible loop region more usually associated with phosphorylation sites.

Myosin heavy chain (MHC) undergoes tyrosine phosphorylation during insulin-mediated differentiation in skeletal muscles and the degree of phosphorylation increases in line with differentiation 4. Interestingly, for the strongest candidate tyrosine phosphorylation site, Y163, both the kinase and interacting SH2 domain predicted by Scansite are involved in insulin-mediated pathways. The kinase INSR is a transmembrane receptor that binds insulin 24 while the SH2 domain protein PIK3R1 is necessary for the insulin-stimulated increase in glucose uptake and glycogen synthesis in insulin-sensitive tissues 25. We can therefore conclude that Y163 remains a strong candidate site for insulin-mediated tyrosine phosphorylation of myosin heavy chain, despite concerns over accessibility.

As phosphorylation sites are often in the tails of proteins, the tyrosines outside the main globular domains, namely Y1379, Y1492 and Y1856, are also potential candidates for phosphorylation. The strongest of these is Y1856, which is both C-terminal and predicted to be phosphorylated by the kinases SRC and ABL1, which are found in skeletal muscle (Table 2). Perhaps of most interest is ABL1, a protein known to be associated with "Sorbin and SH3 domain containing 1" (SORBS1) during insulin signalling in other cell lines 23. SORBS1 is highly expressed in skeletal muscle (data not shown) and is involved in formation of actin stress fibres and focal adhesions; its orthologue, CAP, has been identified as an important adaptor during insulin signalling in mice 262728. Furthermore, the SORBS1 gene has been implicated in the pathogenesis of human disorders with insulin resistance 29.

Csk has been shown to be associated with the hormone 1, 25(OH)₂-vitamin D₃ resulting in the stimulation of the growth-related mitogen-activated protein kinase (MAPK). The phosphorylated form of MAPK is then translocated to the nucleus where it induces the expression of c-myc oncoprotein associated with skeletal muscle proliferation 30. In addition, Csk has also been implicated in the regulation of integrins and the control of cell attachment and shape 31 Goel et al. showed that insulin can phosphorylate myosin, leading to an association with Csk and thus to a decrease in c-Src activity. This has also been shown in fibroblast cell lines following stimulation of the insulin-like growth factor-I receptor 32. This demonstrates the potential for skeletal muscle differentiation after phosphorylation of Y163 and/or Y1856 of the MHC.

Harney et al. have shown that non-muscle myosin heavy chain type IIA in platelets undergoes tyrosine phosphorylation and subsequent dephosphorylation in a time-dependent manner 5. In common with other cells, the cytoskeleton of platelets comprises actin filaments, microtubules and myosin molecules. Myosins form rings within the platelet that maintain a spherical shape and several lines of evidence suggest that these rings reorient following platelet activation to permit spreading 3334. While myosin function therefore appears critical to platelet spreading, studies using cytoskeleton inhibitors have shown that at least the early events of platelet activation are not dependent on the cytoskeletal changes 35. Our work is suggestive that tyrosine phosphorylation of myosin heavy chain, whether in skeletal muscle or in platelets, is a significant event that may initiate cytoskeletal reorganization of muscle cells and platelets. Our studies provide a good starting point for further functional analysis of MHC phosphor-signalling events within different cells.

The author(s) declare that they have no competing interests.