Implications for MstnPP and latent complex function in vivo

5.2.1 β-Aggregation by MstnPP

CD thermal denaturation profiles show a marked transition to a β-sheet rich spectrum at high temperatures, suggestive of β-aggregation, a characteristic of amyloid formation (Benjwal, Verma et al. 2006). What implications does this have for the myostatin precursor protein and latent complex in vivo? For the native protein, β- aggregation is not expected to occur, as both the MstnPP and latent complex spectra showed stability at physiological temperature. However, recent studies have suggested a role for aggregated MstnPP in the amyloid disease sporadic inclusion body myositis (sIBM) (Wojcik, Engel et al. 2005; Wojcik, Nogalska et al. 2007; Askanas and Engel 2008). An investigation into amyloid formation by MstnPP is presented in Chapters 8 and 9.

5.2.2 The propeptide as a chaperone

The myostatin precursor protein is the translated form of myostatin and evidence suggests that the propeptide region in this form plays a role in the correct folding of the myostatin growth factor (Jin, Dunn et al. 2004; Funkenstein and Rebhan 2007). Refolding of the human myostatin precursor was successful in a protocol designed for zebrafish precursor and not in that used for the porcine protein (Chapter 3) despite a higher sequence similarity between human MstnPP and the latter. If sequence differences are important, residue 144 (Fig. 5.3, circled in yellow), which is lysine in pig and glutamic acid in human and zebrafish, may contribute to refolding differences and therefore have a role in chaperone activity. Glu 144 is within a region of the MstnPP that is relatively scarce in both predicted secondary structure and disorder (Fig. 5.3). If this residue is involved in folding of the growth factor, the loop-location coupled with an absence of disorder would impart flexibility, yet stability. Consistent with this hypothesis, Glu 144 is in the C-terminus of the myostatin propeptide domain, suggested previously to be important for stabilisation (Jiang, Liang et al.

2004).

Figure 5.3 The propeptide as a chaperone.

Section of sequence analysis taken from Figure 5.1. Disorder, green; α-helix, blue; β-sheet, red. The glutamic acid residue that may be involved in propeptide-assisted growth factor folding is circled in yellow.

A second hypothesis is that the predicted disorder may play a role in the chaperone- like activity of the propeptide domain by imparting flexibility and allowing regions to

undergo numerous different noncovalent interactions with the growth factor as it folds. The challenge of TGF-β folding lies predominantly in the structure of the cysteine knot and therefore, the formation of native disulphides. Flexible regions of the myostatin propeptide may be able to interact transiently with specific growth factor regions during the folding process to enable the formation of one disulphide over another, or to promote the function of disulphide isomerases in the endoplasmic reticulum. Furthermore, the disorder-prone regions contain a high proportion of charged residues which may be able to stabilise charges in the growth factor domain during folding. As the growth factor also has a high proportion of hydrophobic residues, another possibility is that charged regions of the propeptide shield these hydrophobic regions from the environment to prevent aggregation during folding.

Primary sequence analysis shows that the myostatin propeptide contains an EF-hand calcium-binding domain (Table 4.1). This has not been documented for any other TGF-β family member and NPS@ searches on TGF-β 1-3, BMP-2, inhibin-α, activin A and GDF-11 do not show this motif. The reasons for this motif in the myostatin sequence, if any, are unclear. Calcium plays an important role in ER processes; since chaperone-assisted folding occurs in the ER, one possibility is that the propeptide uses calcium as a cofactor for chaperone activity or that calcium binding may be involved in the recruitment of other chaperones.

5.2.3 Export of myostatin

As a member of the TGF-β family, the myostatin precursor is most likely processed in the trans-Golgi by furin convertase and exported as a latent complex; complex formation has been suggested to be important for export (Gray and Mason 1990; McMahon, Dignam et al. 1996; Jiang, Liang et al. 2004; Walton, Makanji et al.

2009).

The N-terminus of the myostatin propeptide contains one conserved N-glycosylation motif (NISK, Fig. 1.2). As glycosylation plays an important role in the export and subsequent bioactivity of TGF-β1 and TGF-β2 (Brunner, Lioubin et al. 1992), it may perform a similar function for myostatin; there are no glycosylation motifs in the growth factor domain supporting a necessity for the propeptide.

Evidence suggests that the myostatin precursor protein is also secreted. In contrast to the latent complex which is secreted into the serum, MstnPP is found extracellularly in skeletal muscle, suggesting a local role (Anderson, Goldberg et al. 2008). Therefore, while formation of the latent complex may not be necessary for the export of myostatin, the propeptide domain probably is.

Residues 99-266 of the propeptide have been suggested to be involved in stabilization of myostatin (Jiang, Liang et al. 2004), which may be necessary for export of both the complex and precursor protein. Although evidence supports some level of structural rearrangement on latent complex formation, a proportion of the propeptide/growth factor interactions involved in stabilization may be conserved between the precursor and the latent complex. Residues 99-266 are predicted to be β-sheet-predominant and contain less random coil relative to the N-terminus; an increase in structure may provide the proposed stability. However, limited proteolysis results suggest decreased structural stability of this region relative to the N-terminus (Fig. 4.6). Therefore, further investigation is necessary.

5.2.4 Inhibition by the myostatin propeptide

The role of the propeptide in binding and inhibition of growth factor activity is well- documented (Hill, Davies et al. 2002; Jiang, Liang et al. 2004; Yang and Zhao 2006). One possibility is that the propeptide blocks the receptor-binding site of the growth factor in a manner similar to follistatin (Cash, Rejon et al. 2009) and/or BMP-7 (Sengle, Ono et al. 2008); however, no structural evidence to support this has been published.

Deletion mutagenesis has shown residues 42-115 of the propeptide to be responsible for inhibition (Jiang, Liang et al. 2004). This region is predicted to consist of two N- terminal α-helices and a C-terminal β-sheet (Fig. 4.1), separated by a region of predicted intrinsic disorder (Fig. 5.1). Interestingly, the disordered region contains the putative metalloproteinase cleavage site involved in post-secretion activation of the mature growth factor (Wolfman, McPherron et al. 2003; Lee 2008). The predicted intrinsic disorder may therefore be functional, allowing protrusion of the sequence for cleavage and growth factor activation.

After cleavage, structuring of disordered regions may result in enhanced interactions of the propeptide with the growth factor dimer, for high affinity and specificity binding and inhibition.

How might cleavage result in release of the myostatin growth factor? In vitro, heat and acid treatment can be used to activate latent myostatin (Zimmers, Davies et al. 2002; Funkenstein and Rebhan 2007). Acid treatment is most likely to have a similar effect as in vivo metalloproteinases, with cleavage at one of the many aspartate residues present in the inhibitory domain (Smith 2002). Heat-induced release probably stems from denaturation of the propeptide, especially if some regions of propeptide are protruding and relatively flexible.