Phosphate binding: Founding or niche function? Only phospho-ligand binders were considered for this analysis (as either a core domain or a codomain); the interaction cutoff was 3. For X-groups, the instance cutoff was set to 2; for F-groups, the instance cutoff was set to 1, due to their smaller size. (A) A cumulative distribution of the F-group count for X-groups that bind phospho-ligands versus those that do not. (B) The fraction of F-groups that interact with at least one phospho-ligand (Fp). The upper right hand corner of the plot represents X-groups with a high total number of F-groups, as well as a relatively high fraction of F-groups that bind phospho-ligands. Phosphate binding is likely the founding function of those X-groups, which have also enjoyed a relatively long evolutionary history. For example, the P-loop domains-like X-group is a fundamental phosphate binder, where the “P” stands for phosphate and phospho-ligand binding is a hallmark functionality. On the other end (bottom right), the niche functionality scenario is represented by Ig-like β-sandwich, a highly diverse X-group where only about 1% of the F-groups are phospho-ligand binders. Fp is robust to the interaction cutoff (SI Appendix, Fig. S3A). (C) The number of different phospho-ligands that bind a given X-group. Instance cutoff = 2; both core domain and codomain binding events considered. While 87% of X-groups bind fewer than six phospho-ligands, most of the fundamental phosphate binder X-groups bind more than six different phospho-ligands. Ligand binding statistics are robust to the interaction cutoff (SI Appendix, Fig. S3B).

N-helix binding mode dominates the PDB but not evolutionary emergences. (A) The distribution of phospho-ligand binding modes in the PDB. About half of core phospho-ligand binding events adopt the N-helix binding mode (interaction cutoff = 3; alternative cutoffs yielded highly similar results) (SI Appendix, Fig. S5A). (B) The estimated number of emergences of each binding mode in the context of a core domain using an interaction cutoff of 3 and an instance cutoff of 1 (estimated emergences across a range of interaction cutoffs were highly similar; see SI Appendix, Fig. S5B). In contrast to the PDB dataset, along 4 billon y of evolution, the side-chain binding mode emerged most readily, approximately twice as frequently as the N-helix mode. (C) The distribution of phospho-ligand binding modes for the ancient phosphate binders indicates the dominance of the N-helix binding mode (the TIM β/α-barrel is an exception due to the unique features of its N-helix binding site; see main text for more details). The overrepresentation of the ancient binders in the PDB also explains the dominance of N-helix binding in the PDB (A). (D) Canonical N-helix binding modes for each of the ancient phosphate-binding X-groups. Most of the ancient phosphate binders use a simple strand-loop-helix motif to bind phosphate, while the TIM β/α-barrel incorporates an additional short helix oriented to interact with ligands bound in its central tunnel. PDB identifiers, from left to right: 1J21, 3OJW, 6BVE, 1N1M, 1KYI, 4WNI.

Patterns of coincidental, promiscuous sulfate binding. Overall, 5,232 core, nonredundant sulfate binding events in which three residues belonging to the same chain bind the sulfate ion were detected in the PDB. (A) When all of these sulfate binding events are considered, the N-helix binding mode is the least common (consistent with emergences across all X-groups) (Fig. 3B). However, when only short, contiguous sequences with prebiotic amino acid composition are considered, the N-helix binding mode becomes the preferred solution. (B) A cumulative distribution of all sulfate binding events that interact with a single chain indicates that N-helix binding sites are realized with shorter sequence spans compared to any other binding mode: Nearly 60% of binding events are comprised of three consecutive residues (i.e., sequence span = 3; all sulfate binding events included). (C) N-helix binding sites are preferentially realized with prebiotic sequences when all single-chain sulfate binding events are considered. The prebiotic amino acids were taken to be Gly, Ala, Ser, Thr, Asp, Glu, Val, Leu, Ile, and Pro. (D) The basic amino acids Arg and Lys are near-essential for side-chain binding (present in ∼90% of all binding events) but not for N-helix binding (present in ∼40% of all binding events).