Fig 1.
Minimal Rossmann-like motif (RLM) definition.
(A) RLM SSEs adapted from 5-formly-3-hydroxy-2-methylpyridine 4-carboxylic acid (FHMPC) 5-dehydrogenase (PDB: 4OM8) are numbered and colored in rainbow, with magenta catalytic loop between first β-strand—element I (β1) and first α-helix—element II (α1). The second α-helix—element IV (α2) forms crossover between second β-strand—element III (β2) and third β-strand—element V (β3). The crossover loop is unstructured loop at the N-terminal part of α2. Element IV can be α-helix, β-strand, or loop. The unlabeled SSEs (colored in slate) are considered as an insertion to the RLM, which can occur between element III (β2) and element IV (α2) or in any of the loops connecting the RLM SSEs. (B) An interaction matrix defines RLM search strategy using ProSMoS program [22]. Interaction type “T” considers the angle between vectors corresponding to particular RLM elements. (C) RLM scheme with average AL2CO positional conservation index [23] among family level representatives. RLM bins are colored according to conservation index from blue (not conserved) to red (highly conserved), non-RLM elements are shown in gray. Left side of all SSE corresponds to N-terminus, right side to the C-terminus.
Table 1.
Top three ECOD Family groups with largest number of unique EC numbers.
Fig 2.
(A) The ratio of observed and expected frequencies of metabolic reactions defines over (ratio>1) and under (ratio<1) represented pathway categories for RLM enzymes. Asterisks denote significant values according to Fisher’s exact test (P < 0.05). (B) Reaction categories of ECOD RLM families are depicted as a pie chart: families without any assigned EC reactions (Null), families with a single EC reaction type (Singleton), families with similar chemistries, but different substrates (Heterogeneous substrate), and families with different chemistries (Heterogeneous reaction) (C) Distribution of EC reaction classes among all RLM ECOD families that have an EC number assignment.
Fig 3.
Heterogeneous reactions catalyzed by the Rhodanese F-group (EC numbers in bold) through Cys residue intermediates: (A) Rhodanese reactions transfer sulfur groups from thiosulfate to cyanide (EC: 2.8.1.1) or from mercaptopyruvate to oxidize thioredoxin (EC: 2.8.1.2). (B) GlpE rhodanese domain from PDB: 1GMX (gray cartoon) with RLM (rainbow) and active site with an additional conserved motif (magenta stick). (C) Sulfur group transfer to the C-terminal ampylated Gly residue of sulfur carrier protein (SCP) in a bi-functional MOCS3 (Molybdenum Cofactor Synthesis 3) enzyme (EC: 2.8.1.11 and EC: 2.7.7.80). (D) CDC25 phosphate hydrolysis from protein Phospho-Tyr residue (EC: 3.1.3.48). (E) CDC25 phosphatase rhodanese-like domain from PDB: 1QB0 with RLM domain (rainbow) retains similar active site (magenta stick). (B, E) Note that the β-strand (β2; green) can be considered as being barely existent (or vestigial) among the universe of RLMs being considered in this work.
Fig 4.
RLM protein phosphatases arose multiple times in evolution.
Cartoon depictions are labeled at the termini. (A) HAD domain-like phosphatase (pink cartoon) adopts the parallel β-strand topology order indicated below and binds Mg2+ cofactor (green sphere) using the active site motif (magenta) from the RLM (rainbow). (B) PGM phosphatases (light blue cartoon) adopt the mixed β-strand topology order indicated below and catalyze metal-independent hydrolysis using a composite active site motif (magenta) that is both within and outside of the RLM (rainbow). Flavodoxin-like protein phosphatases (light orange cartoon, rainbow RLM, and magenta active site) fall into different ECOD T-groups with the typical flavodoxin-like β-strand order indicated below. The * denotes alternate T-group topologies for β-strand 5 (missing in panel C, permuted and antiparallel in panel D, and permuted in panel E). (C) Phosphotyrosine protein phosphatases I-like (D) (Phosphotyrosine protein) phosphatases II (E) rhodanese/cell cycle control phosphatase.
Fig 5.
Ligands from RLM catalyzed reactions.
(A) ClassyFire superclasses KEGG-defined compounds associated with RLM EC numbers reflect substrates and products of the reaction. Ligands populate 18 out of 31 ClassyFire superclasses and one additional unclassified group (marked by *). (B) UniProt KB and KEGG distinguishes RLM associated cofactors (41 organic and inorganic cofactor compounds. Top 19 compounds are shown).
Table 2.
Major binding modes of the 10 most populated RLM H-groups in ECOD.
Fig 6.
(A) Ligands from RLM catalyzed reactions. Combined KEGG compounds and UniProt cofactors (colored and classified according to ClassyFire superclass in legend) by assigned EC reaction count (Y-axis, cutoff at 100, with total number for EC>100 indicated) are distributed across top 23 ECOD homology groups (X-axis). (B-H) Superpositions of representative domains from largest H-groups. Secondary structure shown only for one representative. RLM is colored in rainbow. Ligands shown as spheres. Each sphere corresponds to one atom of particular ligand. Spheres are colored according ligands superclasses in legend of panel A.
Fig 7.
Diverse binding modes of inorganic metal cofactors.
RLM SSEs are colored in rainbow with yellow crossover loop and magenta catalytic loop. (A) Mg2+ (green sphere) bound to 4-methyl-5-beta-hydroxyethylthiazole kinase (PDB: 1ESQ) Rossmann-like X-group domain (slate cartoon) coordinates ATP substrate (black stick, colored by element). (B) Mg2+ (green sphere) bound to shikimate kinase (PDB: 2SHK) P-loop domains-like X-group (pink cartoon) coordinates ADP (black stick, colored by element). (C) Mg2+ (green sphere) bound to sphingosine kinase (PDB: 3VZD) flavodoxin-like X-group domain (green cartoon) coordinates pyrophosphate (orange stick). (D) Ni2+ bound to trimer of ornithine transcarbamylase (PDB: 2W37) Rossmann-like X-group domain (slate cartoon) mediates trimerization (white and gray cartoon chains) through coordinating residues (stick). (E) MnmE G-domain (PDB: 2GJ8) from P-loop domains-like X-group (pink cartoon) binds K+ (violet sphere) near the transition state analog Mg2+ (green sphere)-GDP-AlF (black stick, colored by element). (F) Co2+ (pink spheres) bound to active site of cobalt-activated peptidase TET1 (PDB: 2CF4) phosphorylase/hydrolase-like X-group domain (wheat cartoon).
Fig 8.
(A) Distribution of ligands from “Nucleotide/nucleoside” superclass for H-groups with more than 10 EC reactions. KEGG compounds (colored and classified according to classes of “Nucleotide/nucleoside” superclass in legend) by assigned fraction of EC reactions in particular homology groups (Y-axis) are distributed across ECOD H-groups with more than 10 unique EC numbers (X-axis). Values in parenthesis show the number of unique EC numbers per H-group. (B-E) 5’-deoxyribonucleosides reveal different binding modes in different H-groups. RLM (SSEs colored rainbow, with yellow crossover loop) bind adenine nucleotide (magenta stick) of SAH/MTA. 5’-deoxyribonucleosides-binding Rossmann-fold domain (slate cartoon) from (B) Mouse nicotinamide N-methyltransferase (PDB: 2I62, EC: 2.1.1.1) binds SAH, the Asp/Glu motif is shown by the red sphere (C) methyltransferase (PDB: 2CX8, EC: 2.1.1.193) binds SAH, the “knot” is formed by the blue β-strand, magenta loop, cyan α-helix and red β-strand with the following loop, which is located under the magenta loop, (D) MTA/AdoHcy nucleosidase (PDB: 1Z5O, EC: 3.2.2.9) binds 5'-methylthioadenosine (MTA), (E) 5'-fluoro-5'-deoxyadenosine synthase (PDB: 1RQP, EC: 2.5.1.63) binds SAM.
Table 3.
Binding modes of 5’-deoxyribonucleosides class for major H-groups.
Fig 9.
Binding site divergence in RLM homologs.
RLM (SSEs colored rainbow, with yellow crossover loop) from homologs bind adenine nucleotide (magenta stick) and diphosphate (orange stick) components of NAD (black nicotinamide ring) from different substrates (stick, colored elements) using similar binding modes. NAD(P)-binding Rossmann-fold domain (slate cartoon) from (A) myo-inositol dehydrogenase (PDB: 4MIN) binds NAD and from (B) succinyl-CoA Synthetase (PDB: 2NU8) binds CoA (gray cysteamine). HUP domains (salmon cartoon) from (C) nicotinamide mononucleotide adenylyltransferase (NMNATase) (PDB: 1EJ2) binds NAD and from (D) bi-functional ATP Sulfurylase-APS Kinase (PDB: 2GKS) binds ADP.
Fig 10.
Nucleotides/Nucleosides allosteric binding.
Examples of RLMs that do not mediate binding of ligands are shown in rainbow, with typical binding site loops in magenta. (A) ADP (black stick, colored by element) binds to bovine glutamate dehydrogenase (PDB: 1NQT) in the inter-chain interface (second chain in white). N-terminal flavodoxin-like domain is in light pink and C-terminal Rossmann-like domain is in light blue. Pivot helix is in purple. R459 shown by purple sticks. (B) Two SAMs (black stick, colored by element) bind to threonine synthase (PDB: 2C2B) interdomain interface distant from the typical Rossmann-like domain (N-terminal domain, light blue) active site marked by the PLP cofactor (magenta stick). C-terminal RLM domain is in light pink. The rigid block is in purple.
Fig 11.
Fe-S clusters bound to RLM enzymes.
RLM SSEs (colored in rainbow, with crossover loop in yellow and catalytic loop in magenta) folds bind Fe-S clusters (yellow/orange sticks, colored by element) differently. (A) Ferrochelatase (PDB: 2HRE) flavodoxin-like fold (light green cartoon) binds [2Fe-2S] cluster using Cys residues (sticks) from C-terminal extension to the RLM and binds protoporphyrin IX (black sticks, colored by element) using a catalytic loop insertion. (B) Respiratory Complex I NDUFS 7 (PDB: 5LC5, chain B) from a domain in the X-group other Rossmann-like fold with the crossover binds [4Fe-4S] using Cys residues (stick) from RLM α1 and from the C-terminal extension. (C) CODH/ACS (PDB: 1OAO) prismane 3rd domain from the same X-group as in B (gray cartoon) binds [4Fe-4S] using residues (stick) from RLM catalytic loop, crossover, and loop C-terminal to β1.
Table 4.
Characteristics of Fe-S cluster RLM enzymes.
Fig 12.
ECOD homology groups reveal different methotrexate binding modes in RLM enzymes.
Domains with RLM SSEs (colored in rainbow, with crossover loop in yellow and catalytic loop in magenta) bind methotrexate (shown in thick sticks; pteridine ring Cα atoms are colored by magenta, rest by yellow) differently. (A) DHFR (PDB: 1U72, EC: 1.5.1.3, ECOD H-gr: 2111.5) binds methotrexate and NADH (green sticks, colored by element). (B) Pteridine reductase (PDB: 1E7W, EC: 1.5.1.33, ECOD: 2003.1) binds methotrexate and NADH (sticks, colored by element). (C) Gamma-glutamyl hydrolase (PDB: 4L8W, EC: 3.4.19.9, ECOD: 2007.1) binds methotrexate and glutamic acid (sticks, colored by element).