Methods | GDF8-WTa | GDF8-Mutant-1a | GDF8-Mutant-2a |
|---|
DeepGOPlus | Fail | Fail | Success |
PFmulDL | Fail | Fail | Success |
NetGO3 | Success | Success | Fail |
AnnoPRO | Success | Success | Success |
- ‘Success’ denotes that the gain/loss-of-function is successfully predicted by the corresponding method, while ‘Fail’ indicates that it is incorrectly predicted. As demonstrated, significant functional variations among GDF8-WT, GDF8-Mutant-1, and GDF8-Mutant-2 can only be “successfully” captured by our newly developed AnnoPRO
- aWild type GDF8 (GDF8-WT) is a growth differentiation factor of 375 amino acids. There are two GDF8 mutants (GDF8-Mutant-1 and GDF8-Mutant-2). GDF8-Mutant-1 contained eight mutations (D267N, F268L, T277S, E312Q, H328Q, G355D, E357Q, and A366G) which locate far away from the binding interface between GDF8 and follistatin-288 (FS288). The interaction between GDF8-WT and FS288 formed a protein complex to further bind to heparin. This is the molecular mechanism underlying GDF8-WT’s key GO term: ‘heparin binding’ (GO:0008201). Because all eight mutations were far away from the binding interface between GDF8 and FS288, it is expected that the ‘heparin binding’ function remains in GDF8-Mutant-1 [55]. Meanwhile, GDF8-Mutant-2 contains three mutations (F315Y, V316M, and L318M, on the binding surface between GDF8 and FS288) which are reported as the key residues indicating protein’s ‘heparin binding’ function [55]. In other words, it is expected that GDF8-Mutant-2 loses its wild type’s ‘heparin binding’ function [55]. All in all, there is gain-of-function of ‘heparin binding’ in both GDF8-WT and GDF8-Mutant-1, while there is loss-of-function in GDF8-Mutant-2
