Sugarcane bagasse derived xylooligosaccharides produced by an arabinofuranosidase/xylobiohydrolase from Bifidobacterium longum in synergism with xylanases.

Arabinoxylan is a major hemicellulose in the sugarcane plant cell wall with arabinose decorations that impose steric restrictions on the activity of xylanases against this substrate. Enzymatic removal of the decorations by arabinofuranosidases can allow a more efficient arabinoxylan degradation by xylanases. Here we produced and characterized a recombinant Bifidobacterium longum arabinofuranosidase from glycoside hydrolase family 43 (BlAbf43) and applied it, together with GH10 and GH11 xylanases, to produce xylooligosaccharides (XOS) from wheat arabinoxylan and alkali pretreated sugarcane bagasse. The enzyme synergistically enhanced XOS production by GH10 and GH11 xylanases, being particularly efficient in combination with the latter family of enzymes, with a degree of synergism of 1.7. We also demonstrated that the enzyme is capable of not only removing arabinose decorations from the arabinoxylan and from the non-reducing end of the oligomeric substrates, but also hydrolyzing the xylan backbone yielding mostly xylobiose and xylose in particular cases. Structural studies of BlAbf43 shed light on the molecular basis of the substrate recognition and allowed hypothesizing on the structural reasons of its multifunctionality.

Characterization and functional evidence for Orf2 of Streptomyces sp. 139 as a novel dipeptidase E.

Aspartyl dipeptidase (dipeptidase E) can hydrolyze Asp-X dipeptides (where X is any amino acid), and the enzyme plays a key role in the degradation of peptides as nutrient sources. Dipeptidase E remains uncharacterized in Streptomyces. Orf2 from Streptomyces sp. 139 is located in the exopolysaccharide biosynthesis gene cluster, which may be a novel dipeptidase E with "S134-H170-D198" catalytic triad by sequence and structure comparison. Herein, recombinant Orf2 was expressed in E. coli and characterized dipeptidase E activity using the Asp-ρNA substrate. The optimal pH and temperature for Orf2 are 7.5 and 40 ℃; Vmax and Km of Orf2 are 0.0787 mM·min-1 and 1.709 mM, respectively. Orf2 exhibits significant degradation activities to Asp-Gly-Gly, Asp-Leu, Asp-His, and isoAsp-Leu and minimal activities to Asp-Pro and Asp-Ala. Orf2 contains a Ser-His-Asp catalytic triad characterized by point mutation. In addition, the Asp147 residue of Orf2 is also proven to be critical for the enzyme's activity through molecular docking and point mutation. Transcriptome analysis reveals the upregulation of genes associated with ribosomes, amino acid biosynthesis, and aminoacyl-tRNA biosynthesis in the orf2 mutant strain. Compared with the orf2 mutant strain and WT, the yield of crude polysaccharide does not change significantly. However, crude polysaccharides from the orf2 mutant strain exhibit a wider range of molecular weight distribution. The results indicate that the Orf2 links nutrient stress to secondary metabolism as a novel dipeptidase E. KEY POINTS: • A novel dipeptidase E with a Ser-His-Asp catalytic triad was characterized from Streptomyces sp. 139. • Orf2 was involved in peptide metabolism both in vitro and in vivo. • Orf2 linked nutrient stress to mycelia formation and secondary metabolism in Streptomyces.

Two extracellular α-arabinofuranosidases are required for cereal-derived arabinoxylan metabolism by Bifidobacterium longum subsp. longum.

Members of the genus Bifidobacterium are commonly found in the human gut and are known to utilize complex carbohydrates that are indigestible by the human host. Members of the Bifidobacterium longum subsp. longum taxon can metabolize various plant-derived carbohydrates common to the human diet. To metabolize such polysaccharides, which include arabinoxylan, bifidobacteria need to encode appropriate carbohydrate-active enzymes in their genome. In the current study, we describe two GH43 family enzymes, denoted here as AxuA and AxuB, which are encoded by B. longum subsp. longum NCIMB 8809 and are shown to be required for cereal-derived arabinoxylan metabolism by this strain. Based on the observed hydrolytic activity of AxuA and AxuB, assessed by employing various synthetic and natural substrates, and based on in silico analyses, it is proposed that both AxuA and AxuB represent extracellular α-L-arabinofuranosidases with distinct substrate preferences. The variable presence of the axuA and axuB genes and other genes previously described to be involved in the metabolism of arabinose-containing glycans can in the majority cases explain the (in)ability of individual B. longum subsp. longum strains to grow on cereal-derived arabinoxylans and arabinan.

Heteroxylan hydrolysis by a recombinant cellulase-free GH10 xylanase from the alkaliphilic bacterium Halalkalibacterium halodurans C-125.

The search for affordable enzymes with exceptional characteristics is fundamental to overcoming industrial and environmental constraints. In this study, a recombinant GH10 xylanase (Xyn10-HB) from the extremely alkaliphilic bacterium Halalkalibacterium halodurans C-125 cultivated at pH 10 was cloned and expressed in E. coli BL21(DE3). Removal of the signal peptide improved the expression, and an overall activity of 8 U/mL was obtained in the cell-free supernatant. The molecular weight of purified Xyn10-HB was estimated to be 42.6 kDa by SDS-PAGE. The enzyme was active across a wide pH range (5-10) with optimal activity recorded at pH 8.5 and 60 °C. It also presented good stability with a half-life of 3 h under these conditions. Substrate specificity studies showed that Xyn10-HB is a cellulase-free enzyme that conventionally hydrolyse birchwood and oat spelts xylans (Apparent Km of 0.46 mg/mL and 0.54 mg/mL, respectively). HPLC analysis showed that both xylans hydrolysis produced xylooligosaccharides (XOS) with a degree of polymerization (DP) ranging from 2 to 9. The conversion yield was 77% after 24 h with xylobiose and xylotriose as the main end-reaction products. When assayed on alkali-extracted wheat straw heteroxylan, the Xyn10-HB produced active XOS with antioxidant activity determined by the DPPH radical scavenging method (IC50 of 0.54 mg/mL after 4 h). Owing to its various characteristics, Xyn10-HB xylanase is a promising candidate for multiple biotechnological applications.

Non-canonical substrate recognition by the human WDR26-CTLH E3 ligase regulates prodrug metabolism.

The yeast glucose-induced degradation-deficient (GID) E3 ubiquitin ligase forms a suite of complexes with interchangeable receptors that selectively recruit N-terminal degron motifs of metabolic enzyme substrates. The orthologous higher eukaryotic C-terminal to LisH (CTLH) E3 complex has been proposed to also recognize substrates through an alternative subunit, WDR26, which promotes the formation of supramolecular CTLH E3 assemblies. Here, we discover that human WDR26 binds the metabolic enzyme nicotinamide/nicotinic-acid-mononucleotide-adenylyltransferase 1 (NMNAT1) and mediates its CTLH E3-dependent ubiquitylation independently of canonical GID/CTLH E3-family substrate receptors. The CTLH subunit YPEL5 inhibits NMNAT1 ubiquitylation and cellular turnover by WDR26-CTLH E3, thereby affecting NMNAT1-mediated metabolic activation and cytotoxicity of the prodrug tiazofurin. Cryoelectron microscopy (cryo-EM) structures of NMNAT1- and YPEL5-bound WDR26-CTLH E3 complexes reveal an internal basic degron motif of NMNAT1 essential for targeting by WDR26-CTLH E3 and degron mimicry by YPEL5's N terminus antagonizing substrate binding. Thus, our data provide a mechanistic understanding of how YPEL5-WDR26-CTLH E3 acts as a modulator of NMNAT1-dependent metabolism.

Alkaliphilic and thermostable lipase production by leaf litter fungus Leptosphaerulina trifolii A SMR-2011.

Fungi that inhabit fire-prone forests have to be adapted to harsh conditions and fungi affiliated to Ascomycota recovered from foliar litter samples were used for bioprospecting of molecules such as enzymes. Agni's fungi isolated from leaf litter, whose spores are capable of tolerating 110 oC were screened for thermostable lipases. One of the isolates, Leptosphaerulina trifolii A SMR-2011 exhibited high positive lipase activity than other isolates while screening through agar plate assay using Tween 20 in the medium. Maximum lipase activity (173.2 U/mg) of L. trifolii was observed at six days of inoculation and decreased thereafter. Among different oils used, the maximum lipase activity was attained by soybean oil (940.1 U/mg) followed by sunflower oil (917.1 U/mg), and then by mustard oil (884.8 U/mg), showing its specificity towards unsaturated fatty acids. Among the various organic nitrogen sources tested, soybean meal showed maximum lipase activity (985.4 U/mg). The partially purified enzyme was active over a wide range of pH from 8 to 12 with a pH optimum of 11.0 (728.1 U/mg) and a temperature range of 60-80 oC with an optimal temperature of 70 oC (779.1 U/mg). The results showed that lipase produced by L. trifolii is alkali stable and retained 85% of its activity at pH 11.0. This enzyme also showed high thermal stability retaining more than 50% of activity when incubated at 60 oC to 90 °C for 2 h. The ions Ca2+ and Mn2+ induced the lipase activity, while Cu2+ and Zn2+ ions lowered the activity compared to control. These results suggests that the leaf litter fungus L. trifolii serves as a potential source for the production of alkali-tolerant and thermostable lipase.

Cyanobacterial α-carboxysome carbonic anhydrase is allosterically regulated by the Rubisco substrate RuBP.

Cyanobacterial CO2 concentrating mechanisms (CCMs) sequester a globally consequential proportion of carbon into the biosphere. Proteinaceous microcompartments, called carboxysomes, play a critical role in CCM function, housing two enzymes to enhance CO2 fixation: carbonic anhydrase (CA) and Rubisco. Despite its importance, our current understanding of the carboxysomal CAs found in α-cyanobacteria, CsoSCA, remains limited, particularly regarding the regulation of its activity. Here, we present a structural and biochemical study of CsoSCA from the cyanobacterium Cyanobium sp. PCC7001. Our results show that the Cyanobium CsoSCA is allosterically activated by the Rubisco substrate ribulose-1,5-bisphosphate and forms a hexameric trimer of dimers. Comprehensive phylogenetic and mutational analyses are consistent with this regulation appearing exclusively in cyanobacterial α-carboxysome CAs. These findings clarify the biologically relevant oligomeric state of α-carboxysomal CAs and advance our understanding of the regulation of photosynthesis in this globally dominant lineage.

Structure, mechanism, and evolution of the last step in vitamin C biosynthesis.

Photosynthetic organisms, fungi, and animals comprise distinct pathways for vitamin C biosynthesis. Besides this diversity, the final biosynthetic step consistently involves an oxidation reaction carried out by the aldonolactone oxidoreductases. Here, we study the origin and evolution of the diversified activities and substrate preferences featured by these flavoenzymes using molecular phylogeny, kinetics, mutagenesis, and crystallographic experiments. We find clear evidence that they share a common ancestor. A flavin-interacting amino acid modulates the reactivity with the electron acceptors, including oxygen, and determines whether an enzyme functions as an oxidase or a dehydrogenase. We show that a few side chains in the catalytic cavity impart the reaction stereoselectivity. Ancestral sequence reconstruction outlines how these critical positions were affixed to specific amino acids along the evolution of the major eukaryotic clades. During Eukarya evolution, the aldonolactone oxidoreductases adapted to the varying metabolic demands while retaining their overarching vitamin C-generating function.

Adaptive evolution: Eukaryotic enzyme's specificity shift to a bacterial substrate.

Prolyl-tRNA synthetase (ProRS), belonging to the family of aminoacyl-tRNA synthetases responsible for pairing specific amino acids with their respective tRNAs, is categorized into two distinct types: the eukaryote/archaeon-like type (E-type) and the prokaryote-like type (P-type). Notably, these types are specific to their corresponding cognate tRNAs. In an intriguing paradox, Thermus thermophilus ProRS (TtProRS) aligns with the E-type ProRS but selectively charges the P-type tRNAPro, featuring the bacterium-specific acceptor-stem elements G72 and A73. This investigation reveals TtProRS's notable resilience to the inhibitor halofuginone, a synthetic derivative of febrifugine emulating Pro-A76, resembling the characteristics of the P-type ProRS. Furthermore, akin to the P-type ProRS, TtProRS identifies its cognate tRNA through recognition of the acceptor-stem elements G72/A73, along with the anticodon elements G35/G36. However, in contrast to the P-type ProRS, which relies on a strictly conserved R residue within the bacterium-like motif 2 loop for recognizing G72/A73, TtProRS achieves this through a non-conserved sequence, RTR, within the otherwise non-interacting eukaryote-like motif 2 loop. This investigation sheds light on the adaptive capacity of a typically conserved housekeeping enzyme to accommodate a novel substrate.

Structural and mechanistic insights into Quinolone Synthase to address its functional promiscuity.

Quinolone synthase from Aegle marmelos (AmQNS) is a type III polyketide synthase that yields therapeutically effective quinolone and acridone compounds. Addressing the structural and molecular underpinnings of AmQNS and its substrate interaction in terms of its high selectivity and specificity can aid in the development of numerous novel compounds. This paper presents a high-resolution AmQNS crystal structure and explains its mechanistic role in synthetic selectivity. Additionally, we provide a model framework to comprehend structural constraints on ketide insertion and postulate that AmQNS's steric and electrostatic selectivity plays a role in its ability to bind to various core substrates, resulting in its synthetic diversity. AmQNS prefers quinolone synthesis and can accommodate large substrates because of its wide active site entrance. However, our research suggests that acridone is exclusively synthesized in the presence of high malonyl-CoA concentrations. Potential implications of functionally relevant residue mutations were also investigated, which will assist in harnessing the benefits of mutations for targeted polyketide production. The pharmaceutical industry stands to gain from these findings as they expand the pool of potential drug candidates, and these methodologies can also be applied to additional promising enzymes.

A salt bridge of the C-terminal carboxyl group regulates PHPT1 substrate affinity and catalytic activity.

PHPT1 is a histidine phosphatase that modulates signaling in eukaryotes through its catalytic activity. Here, we present an analysis of the structure and dynamics of PHPT1 through a combination of solution NMR, molecular dynamics, and biochemical experiments. We identify a salt bridge formed between the R78 guanidinium moiety and the C-terminal carboxyl group on Y125 that is critical for ligand binding. Disruption of the salt bridge by appending a glycine residue at the C-terminus (G126) leads to a decrease in catalytic activity and binding affinity for the pseudo substrate, para-nitrophenylphosphate (pNPP), as well as the active site inhibitor, phenylphosphonic acid (PPA). We show through NMR chemical shift, 15N relaxation measurements, and analysis of molecular dynamics trajectories, that removal of this salt bridge results in an active site that is altered both structurally and dynamically thereby significantly impacting enzymatic function and confirming the importance of this electrostatic interaction.

Exploration of three Dyadobacter fermentans enzymes uncovers molecular activity determinants in CE15.

Glucuronoyl esterases (GEs) are serine-type hydrolase enzymes belonging to carbohydrate esterase family 15 (CE15), and they play a central role in the reduction of recalcitrance in plant cell walls by cleaving ester linkages between glucuronoxylan and lignin in lignocellulose. Recent studies have suggested that bacterial CE15 enzymes are more heterogeneous in terms of sequence, structure, and substrate preferences than their fungal counterparts. However, the sequence space of bacterial GEs has still not been fully explored, and further studies on diverse enzymes could provide novel insights into new catalysts of biotechnological interest. To expand our knowledge on this family of enzymes, we investigated three unique CE15 members encoded by Dyadobacter fermentans NS114T, a Gram-negative bacterium found endophytically in maize/corn (Zea mays). The enzymes are dissimilar, sharing ≤ 39% sequence identity to each other' and were considerably different in their activities towards synthetic substrates. Combined analysis of their primary sequences and structural predictions aided in establishing hypotheses regarding specificity determinants within CE15, and these were tested using enzyme variants attempting to shift the activity profiles. Together, the results expand our existing knowledge of CE15, shed light into the molecular determinants defining specificity, and support the recent thesis that diverse GEs encoded by a single microorganism may have evolved to fulfil different physiological functions. KEY POINTS: • D. fermentans encodes three CE15 enzymes with diverse sequences and specificities • The Region 2 inserts in bacterial GEs may directly influence enzyme activity • Rational amino acid substitutions improved the poor activity of the DfCE15A enzyme.

Substrate uptake patterns shape niche separation in marine prokaryotic microbiome.

Marine heterotrophic prokaryotes primarily take up ambient substrates using transporters. The patterns of transporters targeting particular substrates shape the ecological role of heterotrophic prokaryotes in marine organic matter cycles. Here, we report a size-fractionated pattern in the expression of prokaryotic transporters throughout the oceanic water column due to taxonomic variations, revealed by a multi-"omics" approach targeting ATP-binding cassette (ABC) transporters and TonB-dependent transporters (TBDTs). Substrate specificity analyses showed that marine SAR11, Rhodobacterales, and Oceanospirillales use ABC transporters to take up organic nitrogenous compounds in the free-living fraction, while Alteromonadales, Bacteroidetes, and Sphingomonadales use TBDTs for carbon-rich organic matter and metal chelates on particles. The expression of transporter proteins also supports distinct lifestyles of deep-sea prokaryotes. Our results suggest that transporter divergency in organic matter assimilation reflects a pronounced niche separation in the prokaryote-mediated organic matter cycles.

Cloning, Expression, Characterization and Immobilization of a Recombinant Carboxylesterase from the Halophilic Archaeon, Halobacterium salinarum NCR-1.

Only a few halophilic archaea producing carboxylesterases have been reported. The limited research on biocatalytic characteristics of archaeal esterases is primarily due to their very low production in native organisms. A gene encoding carboxylesterase from Halobacterium salinarum NRC-1 was cloned and successfully expressed in Haloferax volcanii. The recombinant carboxylesterase (rHsEst) was purified by affinity chromatography with a yield of 81%, and its molecular weight was estimated by SDS-PAGE (33 kDa). The best kinetic parameters of rHsEst were achieved using p-nitrophenyl valerate as substrate (KM = 78 µM, kcat = 0.67 s-1). rHsEst exhibited great stability to most metal ions tested and some solvents (diethyl ether, n-hexane, n-heptane). Purified rHsEst was effectively immobilized using Celite 545. Esterase activities of rHsEst were confirmed by substrate specificity studies. The presence of a serine residue in rHsEst active site was revealed through inhibition with PMSF. The pH for optimal activity of free rHsEst was 8, while for immobilized rHsEst, maximal activity was at a pH range between 8 to 10. Immobilization of rHsEst increased its thermostability, halophilicity and protection against inhibitors such as EDTA, BME and PMSF. Remarkably, immobilized rHsEst was stable and active in NaCl concentrations as high as 5M. These biochemical characteristics of immobilized rHsEst reveal its potential as a biocatalyst for industrial applications.

Transcriptomic analysis of polysaccharide utilization loci reveals substrate preferences in ruminal generalists Segatella bryantii TF1-3 and Xylanibacter ruminicola KHP1.

Bacteria of the genera Xylanibacter and Segatella are among the most dominant groups in the rumen microbiota. They are characterized by the ability to utilize different hemicelluloses and pectin of plant cell-wall as well as plant energy storage polysaccharides. The degradation is possible with the use of cell envelope bound multiprotein apparatuses coded in polysaccharide utilization loci (PULs), which have been shown to be substrate specific. The knowledge of PUL presence in rumen Xylanibacter and Segatella based on bioinformatic analyses is already established and transcriptomic and genetic approaches confirmed predicted PULs for a limited number of substrates. In this study, we transcriptomically identified additional different PULs in Xylanibacter ruminicola KHP1 and Segatella bryantii TF1-3. We also identified substrate preferences and found that specific growth rate and extent of growth impacted the choice of substrates preferentially used for degradation. These preferred substrates were used by both strains simultaneously as judged by their PUL upregulation. Lastly, ß-glucan and xyloglucan were used by these strains in the absence of bioinformatically and transcriptomically identifiable PUL systems.

Biochemical Characterization of a Novel Thermostable Ulvan Lyase from Tamlana fucoidanivorans CW2-9.

Ulvan is a complex sulfated polysaccharide extracted from Ulva, and ulvan lyases can degrade ulvan through a ß-elimination mechanism to obtain oligosaccharides. In this study, a new ulvan lyase, EPL15085, which belongs to the polysaccharide lyase (PL) 28 family from Tamlana fucoidanivorans CW2-9, was characterized in detail. The optimal pH and salinity are 9.0 and 0.4 M NaCl, respectively. The Km and Vmax of recombinant EPL15085 toward ulvan are 0.80 mg·mL-1 and 11.22 µmol·min -1 mg-1·mL-1, respectively. Unexpectedly, it is very resistant to high temperatures. After treatment at 100 °C, EPL15085 maintained its ability to degrade ulvan. Molecular dynamics simulation analysis and site-directed mutagenesis analysis indicated that the strong rigidity of the disulfide bond between Cys74-Cys102 in the N-terminus is related to its thermostability. In addition, oligosaccharides with disaccharides and tetrasaccharides were the end products of EPL15085. Based on molecular docking and site-directed mutagenesis analysis, Tyr177 and Leu134 are considered to be the crucial residues for enzyme activity. In conclusion, our study identified a new PL28 family of ulvan lyases, EPL15085, with excellent heat resistance that can expand the database of ulvan lyases and provide the possibility to make full use of ulvan.

Structural insights into catalytic promiscuity of chalcone synthase from Glycine max (L.) Merr.: Coenzyme A-induced alteration of product specificity.

Catalytic promiscuity of enzymes plays a pivotal role in driving the evolution of plant specialized metabolism. Chalcone synthase (CHS) catalyzes the production of 2',4,4',6'-tetrahydroxychalcone (THC), a common precursor of plant flavonoids, from p-coumaroyl-coenzyme A (-CoA) and three malonyl-CoA molecules. CHS has promiscuous product specificity, producing a significant amount of p-coumaroyltriacetic lactone (CTAL) in vitro. However, mechanistic aspects of this CHS promiscuity remain to be clarified. Here, we show that the product specificity of soybean CHS (GmCHS1) is altered by CoA, a reaction product, which selectively inhibits THC production (IC50, 67 µM) and enhances CTAL production. We determined the structure of a ternary GmCHS1/CoA/naringenin complex, in which CoA is bound to the CoA-binding tunnel via interactions with Lys55, Arg58, and Lys268. Replacement of these residues by alanine resulted in an enhanced THC/CTAL production ratio, suggesting the role of these residues in the CoA-mediated alteration of product specificity. In the ternary complex, a mobile loop ("the K-loop"), which contains Lys268, was in a "closed conformation" placing over the CoA-binding tunnel, whereas in the apo and binary complex structures, the K-loop was in an "open conformation" and remote from the tunnel. We propose that the production of THC involves a transition of the K-loop conformation between the open and closed states, whereas synthesis of CTAL is independent of it. In the presence of CoA, an enzyme conformer with the closed K-loop conformation becomes increasingly dominant, hampering the transition of K-loop conformations to result in decreased THC production and increased CTAL production.

Structure-Function Analysis of the Essential Mycobacterium tuberculosis P450 Drug Target, CYP121A1.

Cytochrome P450 CYP121A1 is a well-known drug target against Mycobacterium tuberculosis, the human pathogen that causes the deadly disease tuberculosis (TB). CYP121A1 is a unique P450 enzyme because it uses classical and non-classical P450 catalytic processes and has distinct structural features among P450s. However, a detailed investigation of CYP121A1 protein structures in terms of active site cavity dynamics and key amino acids interacting with bound ligands has yet to be undertaken. To address this research knowledge gap, 53 CYP121A1 crystal structures were investigated in this study. Critical amino acids required for CYP121A1's overall activity were identified and highlighted this enzyme's rigid architecture and substrate selectivity. The CYP121A1-fluconazole crystal structure revealed a novel azole drug-P450 binding mode in which azole heme coordination was facilitated by a water molecule. Fragment-based inhibitor approaches revealed that CYP121A1 can be inhibited by molecules that block the substrate channel or by directly interacting with the P450 heme. This study serves as a reference for the precise understanding of CYP121A1 interactions with different ligands and the structure-function analysis of P450 enzymes in general. Our findings provide critical information for the synthesis of more specific CYP121A1 inhibitors and their development as novel anti-TB drugs.

Genetic and functional diversity of ß-N-acetylgalactosamine-targeting glycosidases expanded by deep-sea metagenome analysis.

ß-N-Acetylgalactosamine-containing glycans play essential roles in several biological processes, including cell adhesion, signal transduction, and immune responses. ß-N-Acetylgalactosaminidases hydrolyze ß-N-acetylgalactosamine linkages of various glycoconjugates. However, their biological significance remains ambiguous, primarily because only one type of enzyme, exo-ß-N-acetylgalactosaminidases that specifically act on ß-N-acetylgalactosamine residues, has been documented to date. In this study, we identify four groups distributed among all three domains of life and characterize eight ß-N-acetylgalactosaminidases and ß-N-acetylhexosaminidase through sequence-based screening of deep-sea metagenomes and subsequent searching of public protein databases. Despite low sequence similarity, the crystal structures of these enzymes demonstrate that all enzymes share a prototype structure and have diversified their substrate specificities (oligosaccharide-releasing, oligosaccharide/monosaccharide-releasing, and monosaccharide-releasing) through the accumulation of mutations and insertional amino acid sequences. The diverse ß-N-acetylgalactosaminidases reported in this study could facilitate the comprehension of their structures and functions and present evolutionary pathways for expanding their substrate specificity.

A multimodal Transformer Network for protein-small molecule interactions enhances predictions of kinase inhibition and enzyme-substrate relationships.

The activities of most enzymes and drugs depend on interactions between proteins and small molecules. Accurate prediction of these interactions could greatly accelerate pharmaceutical and biotechnological research. Current machine learning models designed for this task have a limited ability to generalize beyond the proteins used for training. This limitation is likely due to a lack of information exchange between the protein and the small molecule during the generation of the required numerical representations. Here, we introduce ProSmith, a machine learning framework that employs a multimodal Transformer Network to simultaneously process protein amino acid sequences and small molecule strings in the same input. This approach facilitates the exchange of all relevant information between the two molecule types during the computation of their numerical representations, allowing the model to account for their structural and functional interactions. Our final model combines gradient boosting predictions based on the resulting multimodal Transformer Network with independent predictions based on separate deep learning representations of the proteins and small molecules. The resulting predictions outperform recently published state-of-the-art models for predicting protein-small molecule interactions across three diverse tasks: predicting kinase inhibitions; inferring potential substrates for enzymes; and predicting Michaelis constants KM. The Python code provided can be used to easily implement and improve machine learning predictions involving arbitrary protein-small molecule interactions.