Do Newly Born Orphan Proteins Resemble Never Born Proteins? A Study using Deep Learning Algorithms

Abstract

Newly Born proteins, devoid of detectable homology to any other proteins, known as orphan proteins, occur in a single species or within a taxonomically restricted gene family. They are generated by expression of novel Open Reading Frames, and appear throughout evolution. We used the recently developed programs for predicting protein structures, RoseTTAFold and AlphaFold2, to compare such Newly Born proteins to random polypeptides generated by shuffling sequences of native proteins, which have been called ‘ Never Born’ proteins. The two programs were used to compare the structures of two sets of four Never Born proteins, one set that had been expressed and shown to be intrinsically disordered, and a second set that had been shown experimentally to possess substantial secondary structure. Since the programs rely to a large extent on multisequence alignment, the models generated were scored as being of low quality. However, a significant pattern emerged when the models generated by RoseTTAFold were examined. Specifically, all four members of Group 1 were shown to be very extended, as would be expected for intrinsically disordered proteins. In contrast, all four members of Group 2 appeared to be compact, and possessed substantial secondary structure. As a further control, both programs predicted unfolded structures for three well characterized intrinsically disordered proteins. The two programs were used to predict the structures of two orphan proteins whose crystal structures have been solved, both of which display novel folds. RoseTTAFold predicted both structures very well, whereas AlphaFold2 predicted only one well. The two programs were used to predict the structures of five orphan proteins with well-identified biological functions, one of which is predicted to be intrinsically disordered, and four to be folded. Both programs displayed the intrinsically disordered protein as an unfolded structure. RoseTTAFold displayed all four of those predicted to be folded as compact folded structures, with apparent novel folds, as determined by Dali and Foldseek. It is plausible that new biological functions may be implemented by orphan proteins due to their novel folds.

Assessing structure and disorder prediction tools for de novo emerged proteins in the age of machine learning

“…Also, entirely new structures are highly unlikely to derive from an ancestral protein homologous in sequence but structurally different [ Illergård et al, 2009 ; Chothia and Lesk, 1986 ]. Novel folds were also rarely identified within new experimentally solved structures [ Tóth-Petróczy and Tawfik, 2014 ] but recent advancements will increase dramatically the structural coverage of the known sequence space and could lead to identification and definition of new protein folds and families [ Liu et al, 2022 , Varadi et al, 2021 , Bordin et al, 2023 ]. These advancements also provide new opportunities to search for structural homology of de novo proteins on a larger set of protein structures with popular structure homology algorithms already including predictions [ van Kempen et al, 2022 ; La et al, 2009 ; Holm, 2022 ; Aderinwale et al, 2022 ].…”