Analysis of 6328 dereplicated PQQ dehydrogenase (PQQ-DH) genes from Tara Ocean metagenomes and metagenome assembled genomes (MAGs). (A) Approximately maximum-likelihood tree of PQQ-DH from this study, reference sequences are pruned. For full tree including reference sequences, refer to Fig. S2 and Supplementary Data S1. Scale bar represents 1 substitution per site. (B) Pie chart representing the relative abundance of unique PQQ-DHs represented in each clade. (C) Legend for A and B. Horizontal bars represent the fraction of sequences in each clade containing lanthanide, calcium or unknown binding domains. PQQ-DH groups with experimentally confirmed representatives are illustrated with a star beside the legend and the fill colour representing their experimentally observed metal dependence: Lanthanide (blue) and/or calcium (yellow). Note, most PQQ-DHs have relatively high levels of substrate promiscuity and overlapping specificities (summarized in Fig. S4). (D-M) the two aspartate (ASP) residues required for Ln coordination in PQQ glucose dehydrogenase (GDH) and sorbose dehydrogenase (SDH) are positioned in the active site for Ln interaction. Predicted tertiary structure of (D) GDH and (E) SDH and their characteristic beta barrel structure (green). Confidence of (F) GDH and (G) SDH predicted structures as measured by pLDDT. (H) GDH aligned with its most similar structure from PDB, PedH, a Ln-dep alcohol dehydrogenase (E-value = 3.35e⁻⁵⁸, PDB: 6ZCW, in cream). (I) Zoomed-in view of the active site showing the ASP residues of GDH and their confidence (blue). The PQQ and residues of 6ZCW are in cream, the green sphere is praseodymium (Pr³⁺), the charges are visualized as negative (red) and positive (blue), and interactions are visualized as purple dashed lines. (J) Predicted active site of GDH binding a Ln³⁺ metal cofactor based on its alignment with 6ZCW, black dashed lines represent the interaction distances between ASP residues (cream) and the lanthanide cofactor. The PQQ and metal of 6ZCW are grayed out. (K) SDH aligned with its most similar structure from PDB (E-value = 1.74e⁻³⁴, PDB: 4CVB, in cream). (L) Zoomed-in view of the active site illustrating the SDH aspartate residues and their confidence (blue), PQQ (cream) and Ca²⁺ (green sphere) of 4CVB. (M) Predicted active site of SDH binding a Ln³⁺ metal cofactor based on its alignment with 4CVB, black dotted lines represent the interaction distances between ASP residues of SDH (cream) and Ln³⁺. The PQQ and metal of 4CVB are grayed out.