FUNCTIONAL INNOVATION OF SUBSTRATE SPECIFICITY ALONG THE EVOLUTION OF THE VITAMIN KINASE FAMILY: A MOLECULAR AND PROTEIN STRUCTURE APPROACH

The ATP-dependent vitamin kinase family belongs to the Ribokinase superfamily and is involved in vitamin B1 and B6 biosynthesis by phosphorylating different substrates on a primary alcohol group (R-OH). Within this family, two enzymes can perform the phosphorylation of hydroxymethyl-pyrimidine (HMP) to hydroxymethyl-pyrimidine phosphate (HMP-P). First, a bifunctional pyridoxal kinase encoded by a pdxK-like-HMPPK gene (PdxK-PLK/HMPPK) that can phosphorylate HMP to HMP-P and can also phosphorylate pyridoxal (PL) to pyridoxal-5-phosphate (PLP). Second, a hydroxymethyl-pyrimidine phosphate kinase encoded by the thiD gene (ThiD-HMPPK) that phosphorylates HMP to HMP-P and consecutively phosphorylates HMP-P to produce hydroxymethyl-pyrimidine diphosphate (HMP-PP). This second phosphorylation reaction is unique in the family of vitamin kinases and the Ribokinase superfamily, as it is performed on a methyl phosphate group (R-CH2-PO4) and not on a primary alcohol. There are very few types of catalytic mechanisms describing the phosphorylation of a phosphate group. Interestingly, phylogenetic analyses indicate that PdxK-PLK/HMPPK enzymes diverged from ThiD-HMPPKs, suggesting that the HMP-P kinase activity would have been lost during evolution, while the PLK activity appeared as an evolutionary novelty in the PdxK-PLK/HMPPK group.
In this work, we reconstructed ancestral sequences from these enzyme families, expressed them in recombinant form, purified and characterized. We crystallized an ancestral ThiD-HMPPK from the order Enterobacteriales (AncEnHMPPK) in complex with HMP-P and non-hydrolyzable analogs of ATP. We identified crucial residues for HMP-P binding and catalysis. To determine how HMP-P phosphorylation evolved towards pyridoxal phosphorylation in this family, we analyzed the conservation of key residues for HMP-P phosphorylation and traced them in the ancestral sequences. This allows us to identified point mutations that would have occurred during the divergence of PdxK-PLK/HMPPK from ThiD-HMPPK enzymes These analyses indicate that mutations A110C, H179A, and T211A (AncEnHMPPK numeration) would have been the main changes at the active site of ThiD-HMPPK during the evolution towards the PdxK-PLK/HMPPK group.

Then, we characterized the last common ancestor of ThiD-HMPPK, which showed that it can phosphorylate HMP and HMP-P, but lacks the capability to phosphorylate PL. Interestingly, along the evolutionary pathway, an intermediate ancestor between ThiD-HMPPK and PdxK-PLK/HMPPK enzymes (named ancC) demonstrates a broadestsubstrate specificity, showing proficiency in phosphorylating HMP, HMP-P, and PL. Among these activities, HMP phosphorylation stands out with the highest catalytic efficiency, while PL phosphorylation exhibits the least efficiency. Similarly, the last common ancestor of PdxK-PLK/HMPPK enzymes can phosphorylate HMP, PL, and HMP-P, but there is a remarkably change in substrate specificity, as PL phosphorylation becomes the enzymatic activity with the highest catalytic efficiency, while the phosphorylation of HMP-P becomes the activity with the lowest efficiency. These results suggest that the functional optimization of PL phosphorylation was concomitant with the loss of HMP-P phosphorylation and highlight how a functional evolutionary novelty co-existed with the ancestral function during the evolutionary pathway of an enzyme family.