Tracing the function expansion for a primordial protein fold in the era of fold-based function prediction: beta-trefoil

The incredibly narrow protein fold bottleneck, which separates the billions of unique proteins on one side to deliver diverse biological functions on the other, arises from folds tolerating mutations during evolution. One such fold, called beta-trefoil, is present in functionally diverse proteins including cytokines involved in the immune system. The unrecognizable sequence-level diversity even among paralogs of interleukin-1, one of the cytokines within the same chromosomal locus, suggests the resilience of this fold to mutational on-slaught. Furthermore, beta-trefoil containing-proteins are known to exist along with other domains allowing them to bind to diverse macromolecular types in achieving the observed functional diversity. In this study, we use the beta-trefoil, a fold prevalent across kingdoms, to assess the reach and limitations of function prediction using fold-fold comparison. We found thirty-two distinct functions for proteins containing at least one trefoil domain by mining both the PDB and AlphaFold databases. Among the novel domain architecture including trefoil domain, we find domains such as chitinase, lipase, beta-glucosidase, protein kinase, peptidoglycan-binding + peptidase matrixin, glycosyl hydrolases family 3 + PA14 + fibronectin type- III, alpha galactosidase A, PhoD-like phosphatase, insecticidal crystal toxin, trypsin, alginate lyase and two unknown domains. The limitations of structure-structure comparison tools may arise from converging evolution in folds. This was addressed by creating phylogenetic tree from structure-based multiple sequence alignment of trefoil domains from diverse kingdoms and functions. The clades by kingdoms as opposed to functions may support the hypothesis that beta-trefoil evolved independently, perhaps creating a twilight zone in function prediction using fold-fold comparison.