Synthetic active-site design reveals hidden catalysis and metabolic connections in natural proteomes

Even proteins with established biological functions may harbour catalytic architectures that escape conventional annotation, concealing biochemical reactions and metabolic connections. We previously engineered Ser–His–Asp/Glu catalytic triads into natural Escherichia coli proteins through a few substitutions, conferring hydrolytic activity. Here, we repurposed these synthetic sites as molecular probes to uncover hidden catalysis in natural proteomes. Combining sequence-based homology searches, structural analysis, molecular simulations, and AI-assisted pocket characterization, we identified cryptic catalytic triads in iron-uptake-associated EfeO-I and EfeO-II proteins, spatially distinct from their canonical metal-binding sites. Experimental characterization revealed ester- and amide-hydrolytic activities enabling sequential enterobactin hydrolysis and L-serine release. Native EfeO-containing bacteria grew with Fe(III)-enterobactin under carbon-limited conditions, whereas metabolic models reproduced this phenotype only when both reactions were incorporated. These findings demonstrate that synthetic active-site design can reveal previously unrecognized catalytic functions and metabolic connections, exemplified here by the link between siderophore processing and central carbon metabolism. All files associated with this dataset can be downloaded directly from the Zenodo repository. Spreadsheet files can be opened with Microsoft Excel, LibreOffice Calc, or compatible software. File names and numbering correspond to the supplementary materials cited in the associated publication. Table S1 | Structural features of Ser–His–Asp/Glu catalytic centres in naturally occurring serine α/β-hydrolases, non-α/β-hydrolase proteins engineered with synthetic catalytic centres, and natural non-α/β-hydrolase proteins harbouring equivalent catalytic architectures (A) Summary of catalytic triad and oxyanion-hole assignments for all the proteins analysed. (B) Geometric parameters of catalytic Ser–His–Asp/Glu triads. (C) Identification and structural evaluation of oxyanion-hole residues. (D) Planarity and spatial organization of catalytic architectures. (E) Complete structural dataset and geometric descriptors used for catalytic architecture screening. (F) Amino acid sequences and accession information for all proteins included in the analyses. (G) Sequence-based pairwise distance matrix. (H) Catalytic centre-based distance matrix. Table S3 | Identification and characterization of EfeO-Im homologues harbouring catalytic architectures equivalent to the synthetic EfeO-Im architecture (A) UniProt homologues. The 1,000 homologous proteins were retrieved from UniProt using EfeO-Im as a query, including sequence identity, amino acid sequence, conservation of residues corresponding to the synthetic catalytic architecture, and taxonomic classification based on the Genome Taxonomy Database (GTDB)66. Sixty homologues naturally harbour the complete set of Ser–His–Asp/Glu residues corresponding to the synthetic EfeO-Im catalytic architecture. (B) MarDB homologues. EfeO-Im homologues identified through sequence- and structure-based screening of the MarDB database. The dataset provides sequence identifiers, taxonomic classification, alignment of the catalytic region, sequence identity relative to EfeO-Im, and protein validation status. Information on genomic origin (isolate or metagenome-assembled genome), geographical location, and environmental isolation source is also included where available. Table S5 | Sequence and expression construct information for the native and engineered proteins analysed in this study Amino acid sequences of the 7 natural EfeO-Im homologues selected for experimental characterization, including predicted signal peptides, synthetic gene constructs cloned into pET-45b(+) expression vectors, and calculated molecular weights. The corresponding catalytically inactive variants contain Ser-to-Ala substitutions at the nucleophilic serine residues (residue numbering is provided in Table S5). Table S9 | Growth of native EfeO-I and II-containing bacterial strains under varying carbon and iron conditions Experimental conditions, carbon-source calculations, growth curves, and growth parameters for D. soli, A. pini, and P. bourgognense cultured under different carbon availabilities and with iron supplied as FeCl3 or Fe(III)-enterobactin. The table includes the estimated carbon contribution from enterobactin hydrolysis and summarizes growth across the 5 experimental conditions. Table S10 | Genome annotations used for reconstruction of genome-scale metabolic models. Genome annotation data for D. soli LMG 29193 (GCA_034044715.1), A. pini LMG 27745 (GCA_900187945), and P. bourgognense LMG 22837 (GCA_003337575) used to curate the corresponding genome-scale metabolic models. The table includes CDS-level gene and product annotations, database cross-references, EC numbers, KEGG orthology assignments, and sequence information. Models were manually curated for EfeUOB-dependent iron uptake and Fe(III)-enterobactin utilization. Because canonical FepA/FepBCDG homologues were not identified, Fe(III)-enterobactin uptake was assigned to TonB-dependent iron-siderophore receptors annotated in each genome. Their specificity for Fe(III)-enterobactin was inferred from annotation and represents a modelling assumption, supported by the experimentally observed utilization of extracellular Fe(III)-enterobactin. Expanded models additionally contained the 2 experimentally demonstrated EfeO-mediated hydrolysis reactions. Table S11 | Manual curation of iron transport and siderophore-associated reactions in genome-scale metabolic models of three EfeO-containing bacteria. Reactions added, removed, or modified to represent iron uptake, Fe(III)-enterobactin transport, and transport of hydrolysis products. The table provides reaction identifiers, stoichiometries, gene–protein–reaction associations, available ModelSEED identifiers, and supporting evidence or modelling assumptions. Table S12 | EfeO-mediated ester- and amide-hydrolytic reactions incorporated into genome-scale metabolic models. Reaction identifiers, stoichiometries, subcellular localization, gene–protein–reaction associations, and supporting evidence for sequential Fe(III)-enterobactin and DHB–Ser hydrolysis. Reactions were assigned to the characterized EfeO-I and EfeO-II proteins from D. soli (MDY0882902_1), A. pini (SNB69647_1), and P. bourgognense (RCW82739_1). An ester-hydrolysis-only model was included to assess the contribution of the subsequent amide-hydrolytic step. Table S13 | Experimental and model-predicted growth rates of three EfeO-containing bacteria under different iron and carbon availability conditions. Experimental growth rates (μ ± SD) and model predictions for D. soli, A. pini, and P. bourgognense under five carbon- and iron-availability conditions. Predictions compare baseline models with models incorporating EfeO-mediated ester hydrolysis alone or sequential ester and amide hydrolysis. Model predictions are intended to compare metabolic capabilities qualitatively rather than reproduce experimental growth rates quantitatively. All growth rates are expressed in h-1. Data S1 | Raw data for in vitro and in vivo ester and polyester activity assays (A) Ester hydrolysis activities of engineered EfeO-Im and 7 natural EfeO-Im homologues against the 96-member ester library, including raw absorbance measurements and calculated specific activities. Native E. coli EfeO-I and the corresponding Ser-to-Ala variants of the 7 natural EfeO-Im homologues were assayed in parallel as negative catalytic controls. None showed hydrolytic activity above background levels under the assay conditions and are therefore not shown in the dataset.A.1, Specific activities (µmol min⁻¹ mg⁻¹) of engineered EfeO-Im and 7 natural EfeO-Im homologues determined against the 96-member ester library under the standard assay conditions described in STAR★METHODS. A.2, Raw absorbance measurements at 550 nm recorded during ester hydrolysis following enzyme incubation at 30 °C for the indicated times. Abs/min values were calculated from the linear region of each kinetic trace and subsequently converted into specific activities as described in STAR★METHODS. Reactions were performed in 384-well microplates; the identity and plate position of each ester are provided in the worksheet A.2. Each plate contained triplicate measurements for each enzyme assayed against the 96 ester substrates, together with enzyme-free controls for nonenzymatic substrate hydrolysis. Related to Fig. 5. (B) In vitro polyester hydrolysis activities of purified EfeO-Im and natural EfeO-Im homologues towards nPET and pPET at 30, 40, and 60 °C, including raw and processed values. Datasets include the concentrations of PET degradation products and the corresponding specific activity values obtained after incubation of nPET and pPET with purified enzymes at 30, 40 and 60 °C. Product quantification was performed by HPLC using calibration curves; raw chromatograms are provided in Data S2. Data are provided in triplicate and include both raw and processed values, together with all applied dilution and concentration factors. Related to Table S6. (C) In vivo polyester hydrolysis activities of whole-cell biocatalysts towards nPET, including raw and processed values. Datasets include the concentrations of soluble PET degradation products obtained after incubation of nPET with native bacterial strains harbouring validated EfeO-Im orthologues at 30 °C for 48 h. Product quantification was performed by HPLC using calibration curves; raw chromatograms are provided in Data S2. Data are provided for three independent biological replicates and include both raw and processed values, together with all applied dilution and concentration factors. Related to Table S7. The workbook contains raw and processed datasets, including replicate measurements and calculated values used for data analysis. Individual worksheets are organized according to the experimental workflow described in the manuscript, enabling reproduction of the enzymatic activity calculations and the corresponding figure and tables. Data S2 | Unprocessed HPLC chromatograms. Original HPLC chromatograms used for the quantification of PET hydrolysis products reported in this study. Files are provided in plain-text (.txt) format as exported from Varian ProStar software (Varian Inc., Palo Alto, CA, USA). The dataset includes chromatograms underlying the in vitro PET-hydrolysis assays with purified enzymes and the in vivo nPET-depolymerization assays with native bacterial strains reported in Data S1B and Data S1C, respectively related to Table S6 and S7. Data S3 | Uncropped SDS–PAGE gel image underlying Fig. S3. Uncropped scans of the 12% Tris–glycine SDS–polyacrylamide gel used for SDS–PAGE analysis of the purified EfeO-Im homologues are shown in Fig. S3. Data S4 | Raw UHPLC–MS data for siderophore hydrolysis assays. Raw UHPLC–MS data from enzymatic reactions with engineered EfeO-Im and 7 natural EfeO-Im homologues, including the corresponding enzyme-free controls shown in Fig. S4. Native E. coli EfeO-I and the corresponding Ser-to-Ala variants of the 7 natural EfeO-Im homologues were assayed in parallel under identical conditions and showed no detectable siderophore hydrolysis above background levels; these negative-control data are therefore not shown.