Macrocyclization and Backbone Modification in RiPP Biosynthesis.

The past decade has seen impressive advances in understanding the biosynthesis of ribosomally synthesized and posttranslationally modified peptides (RiPPs). One of the most common modifications found in these natural products is macrocyclization, a strategy also used by medicinal chemists to improve metabolic stability and target affinity and specificity. Another tool of the peptide chemist, modification of the amides in a peptide backbone, has also been observed in RiPPs. This review discusses the molecular mechanisms of biosynthesis of a subset of macrocyclic RiPP families, chosen because of the unusual biochemistry involved: the five classes of lanthipeptides (thioether cyclization by Michael-type addition), sactipeptides and ranthipeptides (thioether cyclization by radical chemistry), thiopeptides (cyclization by [4+2] cycloaddition), and streptide (cyclization by radical C-C bond formation). In addition, the mechanisms of backbone amide methylation, backbone epimerization, and backbone thioamide formation are discussed, as well as an unusual route to small molecules by posttranslational modification.

LanCLs add glutathione to dehydroamino acids generated at phosphorylated sites in the proteome.

Enzyme-mediated damage repair or mitigation, while common for nucleic acids, is rare for proteins. Examples of protein damage are elimination of phosphorylated Ser/Thr to dehydroalanine/dehydrobutyrine (Dha/Dhb) in pathogenesis and aging. Bacterial LanC enzymes use Dha/Dhb to form carbon-sulfur linkages in antimicrobial peptides, but the functions of eukaryotic LanC-like (LanCL) counterparts are unknown. We show that LanCLs catalyze the addition of glutathione to Dha/Dhb in proteins, driving irreversible C-glutathionylation. Chemo-enzymatic methods were developed to site-selectively incorporate Dha/Dhb at phospho-regulated sites in kinases. In human MAPK-MEK1, such "elimination damage" generated aberrantly activated kinases, which were deactivated by LanCL-mediated C-glutathionylation. Surveys of endogenous proteins bearing damage from elimination (the eliminylome) also suggest it is a source of electrophilic reactivity. LanCLs thus remove these reactive electrophiles and their potentially dysregulatory effects from the proteome. As knockout of LanCL in mice can result in premature death, repair of this kind of protein damage appears important physiologically.

Discovery, biosynthesis, and engineering of lantipeptides.

Aided by genome-mining strategies, knowledge of the prevalence and diversity of ribosomally synthesized natural products (RNPs) is rapidly increasing. Among these are the lantipeptides, posttranslationally modified peptides containing characteristic thioether cross-links imperative for bioactivity and stability. Though this family was once thought to be a limited class of antimicrobial compounds produced by gram-positive bacteria, new insights have revealed a much larger diversity of activity, structure, biosynthetic machinery, and producing organisms than previously appreciated. Detailed investigation of the enzymes responsible for installing the posttranslational modifications has resulted in improved in vivo and in vitro engineering systems focusing on enhancement of the therapeutic potential of these compounds. Although dozens of new lantipeptides have been isolated in recent years, bioinformatic analyses indicate that many hundreds more await discovery owing to the widespread frequency of lantipeptide biosynthetic machinery in bacterial genomes.

The mechanism of thia-Michael addition catalyzed by LanC enzymes.

In both eukarya and bacteria, the addition of Cys to dehydroalanine (Dha) and dehydrobutyrine (Dhb) occurs in various biological processes. In bacteria, intramolecular thia-Michael addition catalyzed by lanthipeptide cyclases (LanC) proteins or protein domains gives rise to a class of natural products called lanthipeptides. In eukarya, dehydroamino acids in signaling proteins are introduced by effector proteins produced by pathogens like Salmonella to dysregulate host defense mechanisms. A eukaryotic LanC-like (LanCL) enzyme catalyzes the addition of Cys in glutathione to Dha/Dhb to protect the cellular proteome from unwanted chemical and biological activity. To date, the mechanism of the enzyme-catalyzed thia-Michael addition has remained elusive. We report here the crystal structures of the human LanCL1 enzyme complexed with different ligands, including the product of thia-Michael addition of glutathione to a Dhb-containing peptide that represents the activation loop of Erk. The structures show that a zinc ion activates the Cys thiolate for nucleophilic attack and that a conserved His is poised to protonate the enolate intermediate to achieve a net anti-addition. A second His hydrogen bonds to the carbonyl oxygen of the former Dhb and may stabilize the negative charge that builds up on this oxygen atom in the enolate intermediate. Surprisingly, the latter His is not conserved in orthologous enzymes that catalyze thia-Michael addition to Dha/Dhb. Eukaryotic LanCLs contain a His, whereas bacterial stand-alone LanCs have a Tyr residue, and LanM enzymes that have LanC-like domains have a Lys, Asn, or His residue. Mutational and binding studies support the importance of these residues for catalysis.

Bacteriophage targeting of gut bacterium attenuates alcoholic liver disease.

Chronic liver disease due to alcohol-use disorder contributes markedly to the global burden of disease and mortality1-3. Alcoholic hepatitis is a severe and life-threatening form of alcohol-associated liver disease. The gut microbiota promotes ethanol-induced liver disease in mice4, but little is known about the microbial factors that are responsible for this process. Here we identify cytolysin-a two-subunit exotoxin that is secreted by Enterococcus faecalis5,6-as a cause of hepatocyte death and liver injury. Compared with non-alcoholic individuals or patients with alcohol-use disorder, patients with alcoholic hepatitis have increased faecal numbers of E. faecalis. The presence of cytolysin-positive (cytolytic) E. faecalis correlated with the severity of liver disease and with mortality in patients with alcoholic hepatitis. Using humanized mice that were colonized with bacteria from the faeces of patients with alcoholic hepatitis, we investigated the therapeutic effects of bacteriophages that target cytolytic E. faecalis. We found that these bacteriophages decrease cytolysin in the liver and abolish ethanol-induced liver disease in humanized mice. Our findings link cytolytic E. faecalis with more severe clinical outcomes and increased mortality in patients with alcoholic hepatitis. We show that bacteriophages can specifically target cytolytic E. faecalis, which provides a method for precisely editing the intestinal microbiota. A clinical trial with a larger cohort is required to validate the relevance of our findings in humans, and to test whether this therapeutic approach is effective for patients with alcoholic hepatitis.

Biosynthesis of 3-thia-α-amino acids on a carrier peptide.

A subset of natural products, such as polyketides and nonribosomal peptides, is biosynthesized while tethered to a carrier peptide via a thioester linkage. Recently, we reported that the biosyntheses of 3-thiaglutamate and ammosamide, single amino acid-derived natural products, employ a very different type of carrier peptide to which the biosynthetic intermediates are bound via an amide linkage. During their biosyntheses, a peptide aminoacyl-transfer ribonucleic acid (tRNA) ligase (PEARL) first loads an amino acid to the C terminus of the carrier peptide for subsequent modification by other enzymes. Proteolytic removal of the modified C-terminal amino acid yields the mature product. We termed natural products that are biosynthesized using such pathways pearlins. To investigate the diversity of pearlins, in this study we experimentally characterized another PEARL-encoding biosynthetic gene cluster (BGC) from Tistrella mobilis (tmo). The enzymes encoded in the tmo BGC transformed cysteine into 3-thiahomoleucine both in vitro and in Escherichia coli. During this process, a cobalamin-dependent radical S-adenosylmethionine (SAM) enzyme catalyzes C-isopropylation. This work illustrates that the biosynthesis of amino acid-derived natural products on a carrier peptide is a widespread strategy in nature and expands the spectrum of thiahemiaminal analogs of amino acids that may serve a broader, currently unknown function.

Biosynthesis of phosphonic and phosphinic acid natural products.

Natural products containing carbon-phosphorus bonds (phosphonic and phosphinic acids) have found widespread use in medicine and agriculture. Recent years have seen a renewed interest in the biochemistry and biology of these compounds with the cloning of the biosynthetic gene clusters for several family members. This review discusses the commonalities and differences in the molecular logic that lie behind the biosynthesis of these compounds. The current knowledge regarding the metabolic pathways and enzymes involved in the production of a number of natural products, including the approved antibiotic fosfomycin, the widely used herbicide phosphinothricin (PT), and the clinical candidate for treatment of malaria FR-900098, is presented. Many of the enzymes involved in the biosynthesis of these compounds catalyze chemically and biologically unprecedented transformations, and a wealth of new biochemistry has been revealed through their study. These investigations have also suggested new strategies for natural product discovery.

Unexpected Transformations during Pyrroloiminoquinone Biosynthesis.

Pyrroloiminoquinone-containing natural products have long been known for their biological activities. They are derived from tryptophan, but their biosynthetic pathways have remained elusive. Studies on the biosynthetic gene cluster (BGC) that produces the ammosamides revealed that the first step is attachment of Trp to the C-terminus of a scaffold peptide in an ATP- and tRNA-dependent manner catalyzed by a PEptide Aminoacyl-tRNA Ligase (PEARL). The indole of Trp is then oxidized to a hydroxyquinone. We previously proposed a chemically plausible and streamlined pathway for converting this intermediate to the ammosamides using additional enzymes encoded in the BGC. In this study, we report the activity of four additional enzymes from two gene clusters, which show that the previously proposed pathway is incorrect and that Nature's route toward pyrroloiminoquinones is much more complicated. We demonstrate that, surprisingly, amino groups in pyrroloiminoquinones are derived from (at least) three different sources, glycine, asparagine, and leucine, all introduced in a tRNA-dependent manner. We also show that an FAD-dependent putative glycine oxidase (Amm14) is required for the process that incorporates the nitrogens from glycine and leucine and that a quinone reductase is required for the incorporation of asparagine. Additionally, we provide the first insights into the evolutionary origin of the PEARLs as well as related enzymes, such as the glutamyl-tRNA-dependent dehydratases involved in the biosynthesis of lanthipeptides and thiopeptides. These enzymes appear to all have descended from the ATP-GRASP protein family.

Facile Method for Determining Lanthipeptide Stereochemistry.

Lanthipeptides make up a large group of natural products that belong to the ribosomally synthesized and post-translationally modified peptides (RiPPs). Lanthipeptides contain lanthionine and methyllanthionine bis-amino acids that have varying stereochemistry. The stereochemistry of new lanthipeptides is often not determined because current methods require equipment that is not standard in most laboratories. In this study, we developed a facile, efficient, and user-friendly method for detecting lanthipeptide stereochemistry, utilizing advanced Marfey's analysis with detection by liquid chromatography coupled with mass spectrometry (LC-MS). Under optimized conditions, 0.05 mg of peptide is sufficient to characterize the stereochemistry of five (methyl)lanthionines of different stereochemistry using a simple liquid chromatography setup, which is a much lower detection limit than current methods. In addition, we describe methods to readily access standards of the three different methyllanthionine stereoisomers and two different lanthionine stereoisomers that have been reported in known lanthipeptides. The developed workflow uses a commonly used nonchiral column system and offers a scalable platform to assist antimicrobial discovery. We illustrate its utility with an example of a lanthipeptide discovered by genome mining.

Mechanism of Action of Ribosomally Synthesized and Post-Translationally Modified Peptides.

Ribosomally synthesized and post-translationally modified peptides (RiPPs) are a natural product class that has undergone significant expansion due to the rapid growth in genome sequencing data and recognition that they are made by biosynthetic pathways that share many characteristic features. Their mode of actions cover a wide range of biological processes and include binding to membranes, receptors, enzymes, lipids, RNA, and metals as well as use as cofactors and signaling molecules. This review covers the currently known modes of action (MOA) of RiPPs. In turn, the mechanisms by which these molecules interact with their natural targets provide a rich set of molecular paradigms that can be used for the design or evolution of new or improved activities given the relative ease of engineering RiPPs. In this review, coverage is limited to RiPPs originating from bacteria.

Biosynthesis of fosfomycin in pseudomonads reveals an unexpected enzymatic activity in the metallohydrolase superfamily.

The epoxide-containing phosphonate natural product fosfomycin is a broad-spectrum antibiotic used in the treatment of cystitis. Fosfomycin is produced by both the plant pathogen Pseudomonas syringae and soil-dwelling streptomycetes. While the streptomycete pathway has recently been fully elucidated, the pseudomonad pathway is still mostly elusive. Through a systematic evaluation of heterologous expression of putative biosynthetic enzymes, we identified the central enzyme responsible for completing the biosynthetic pathway in pseudomonads. The missing transformation involves the oxidative decarboxylation of the intermediate 2-phosphonomethylmalate to a new intermediate, 3-oxo-4-phosphonobutanoate, by PsfC. Crystallographic studies reveal that PsfC unexpectedly belongs to a new class of diiron metalloenzymes that are part of the polymerase and histidinol phosphatase superfamily.

A Novel Pathway for Biosynthesis of the Herbicidal Phosphonate Natural Product Phosphonothrixin Is Widespread in Actinobacteria.

Phosphonothrixin is an herbicidal phosphonate natural product with an unusual, branched carbon skeleton. Bioinformatic analyses of the ftx gene cluster, which is responsible for synthesis of the compound, suggest that early steps of the biosynthetic pathway, up to production of the intermediate 2,3-dihydroxypropylphosphonic acid (DHPPA) are identical to those of the unrelated phosphonate natural product valinophos. This conclusion was strongly supported by the observation of biosynthetic intermediates from the shared pathway in spent media from two phosphonothrixin producing strains. Biochemical characterization of ftx-encoded proteins confirmed these early steps, as well as subsequent steps involving the oxidation of DHPPA to 3-hydroxy-2-oxopropylphosphonate and its conversion to phosphonothrixin by the combined action of an unusual heterodimeric, thiamine-pyrophosphate (TPP)-dependent ketotransferase and a TPP-dependent acetolactate synthase. The frequent observation of ftx-like gene clusters within actinobacteria suggests that production of compounds related to phosphonothrixin is common within these bacteria. IMPORTANCE Phosphonic acid natural products, such as phosphonothrixin, have great potential for biomedical and agricultural applications; however, discovery and development of these compounds requires detailed knowledge of the metabolism involved in their biosynthesis. The studies reported here reveal the biochemical pathway phosphonothrixin production, which enhances our ability to design strains that overproduce this potentially useful herbicide. This knowledge also improves our ability to predict the products of related biosynthetic gene clusters and the functions of homologous enzymes.

Substrate Specificity of the Flavoenzyme BhaC1 That Converts a C-Terminal Trp to a Hydroxyquinone.

The preparation of protein-protein, protein-peptide, and protein-small molecule conjugates is important for a variety of applications, such as vaccine production, immunotherapies, preparation of antibody-drug conjugates, and targeted delivery of therapeutics. To achieve site-selective conjugation, selective chemical or enzymatic functionalization of proteins is required. We have recently reported biosynthetic pathways in which small, catalytic scaffold peptides are utilized for the generation of amino acid-derived natural products called pearlins. In these systems, peptide amino-acyl tRNA ligases (PEARLs) append amino acids to the C-terminus of a scaffold peptide, and tailoring enzymes encoded in the biosynthetic gene clusters modify the PEARL-appended amino acid to generate a variety of natural products. Herein, we investigate the substrate selectivity of one such tailoring enzyme, BhaC1, that participates in pyrroloiminoquinone biosynthesis. BhaC1 converts the indole of a C-terminal tryptophan into an o-hydroxy-p-quinone, a promising moiety for site-selective bioconjugation. Our studies demonstrate that BhaC1 requires a 20-amino acid peptide for substrate recognition. When this peptide was appended at the C-terminus of proteins, the C-terminal Trp was modified by BhaC1. The enzyme is sufficiently selective that only small changes to the sequence of the peptide are tolerated. An AlphaFold model for substrate recognition explains the selectivity of the enzyme, which may be used to install a reactive handle onto the C-terminus of proteins.

Unexpected Methyllanthionine Stereochemistry in the Morphogenetic Lanthipeptide SapT.

Lanthipeptides are polycyclic peptides characterized by the presence of lanthionine (Lan) and/or methyllanthionine (MeLan). They are members of the ribosomally synthesized and post-translationally modified peptides (RiPPs). The stereochemical configuration of (Me)Lan cross-links is important for the bioactivity of lanthipeptides. To date, MeLan residues in characterized lanthipeptides have either the 2S,3S or 2R,3R stereochemistry. Herein, we reconstituted in Escherichia coli the biosynthetic pathway toward SapT, a class I lanthipeptide that exhibits morphogenetic activity. Through the synthesis of standards, the heterologously produced peptide was shown to possess three MeLan residues with the 2S,3R stereochemistry (d-allo-l-MeLan), the first time such stereochemistry has been observed in a lanthipeptide. Bioinformatic analysis of the biosynthetic enzymes suggests this stereochemistry may also be present in other lanthipeptides. Analysis of another gene cluster in Streptomyces coelicolor that is widespread in actinobacteria confirmed another example of d-allo-l-MeLan and verified the bioinformatic prediction. We propose a mechanism for the origin of the unexpected stereochemistry and provide support using site-directed mutagenesis.

Class V Lanthipeptide Cyclase Directs the Biosynthesis of a Stapled Peptide Natural Product.

Lanthipeptides are a class of cyclic peptides characterized by the presence of one or more lanthionine (Lan) or methyllanthionine (MeLan) thioether rings. These cross-links are produced by α,ß-unsaturation of Ser or Thr residues in peptide substrates by dehydration, followed by a Michael-type conjugate addition of Cys residues onto the dehydroamino acids. Lanthipeptides may be broadly classified into at least five different classes, and the biosynthesis of classes I-IV lanthipeptides requires catalysis by LanC cyclases that control both the site-specificity and the stereochemistry of the conjugate addition. In contrast, there are no current examples of LanCs that occur in class V biosynthetic clusters, despite the presence of lanthionine rings in these compounds. In this work, bioinformatics-guided co-occurrence analysis identifies more than 240 putative class V lanthipeptide clusters that contain a LanC cyclase. Reconstitution studies demonstrate that the cyclase-catalyzed product is notably distinct from the product formed spontaneously. Stereochemical analysis shows that the cyclase diverts the final product to a configuration that is distinct from one that is energetically favored. Structural characterization of the final product by multi-dimensional NMR spectroscopy reveals that it forms a helical stapled peptide. Mutational analysis identified a plausible order for cyclization and suggests that enzymatic rerouting to the final structure is largely directed by the construction of the first lanthionine ring. These studies show that lanthipeptide cyclases are needed for the biosynthesis of some constrained peptides, the formations of which would otherwise be energetically unfavored.

Accessing Diverse Pyridine-Based Macrocyclic Peptides by a Two-Site Recognition Pathway.

Macrocyclic peptides are sought-after molecular scaffolds for drug discovery, and new methods to access diverse libraries are of increasing interest. Here, we report the enzymatic synthesis of pyridine-based macrocyclic peptides (pyritides) from linear precursor peptides. Pyritides are a recently described class of ribosomally synthesized and post-translationally modified peptides (RiPPs) and are related to the long-known thiopeptide natural products. RiPP precursors typically contain an N-terminal leader region that is physically engaged by the biosynthetic proteins that catalyze modification of the C-terminal core region of the precursor peptide. We demonstrate that pyritide-forming enzymes recognize both the leader region and a C-terminal tripeptide motif, with each contributing to site-selective substrate modification. Substitutions in the core region were well-tolerated and facilitated the generation of a wide range of pyritide analogues, with variations in macrocycle sequence and size. A combination of the pyritide biosynthetic pathway with azole-forming enzymes was utilized to generate a thiazole-containing pyritide (historically known as a thiopeptide) with no similarity in sequence and macrocycle size to the naturally encoded pyritides. The broad substrate scope of the pyritide biosynthetic enzymes serves as a future platform for macrocyclic peptide lead discovery and optimization.

Mechanism of Radical S-Adenosyl-l-methionine Adenosylation: Radical Intermediates and the Catalytic Competence of the 5'-Deoxyadenosyl Radical.

Radical S-adenosyl-l-methionine (SAM) enzymes employ a [4Fe-4S] cluster and SAM to initiate diverse radical reactions via either H-atom abstraction or substrate adenosylation. Here we use freeze-quench techniques together with electron paramagnetic resonance (EPR) spectroscopy to provide snapshots of the reaction pathway in an adenosylation reaction catalyzed by the radical SAM enzyme pyruvate formate-lyase activating enzyme on a peptide substrate containing a dehydroalanine residue in place of the target glycine. The reaction proceeds via the initial formation of the organometallic intermediate Ω, as evidenced by the characteristic EPR signal with g∥ = 2.035 and g⊥ = 2.004 observed when the reaction is freeze-quenched at 500 ms. Thermal annealing of frozen Ω converts it into a second paramagnetic species centered at giso = 2.004; this second species was generated directly using freeze-quench at intermediate times (∼8 s) and unequivocally identified via isotopic labeling and EPR spectroscopy as the tertiary peptide radical resulting from adenosylation of the peptide substrate. An additional paramagnetic species observed in samples quenched at intermediate times was revealed through thermal annealing while frozen and spectral subtraction as the SAM-derived 5'-deoxyadenosyl radical (5'-dAdo•). The time course of the 5'-dAdo• and tertiary peptide radical EPR signals reveals that the former generates the latter. These results thus support a mechanism in which Ω liberates 5'-dAdo• by Fe-C5' bond homolysis, and the 5'-dAdo• attacks the dehydroalanine residue of the peptide substrate to form the adenosylated peptide radical species. The results thus provide a picture of a catalytically competent 5'-dAdo• intermediate trapped just prior to reaction with the substrate.

Characterization of glutamyl-tRNA-dependent dehydratases using nonreactive substrate mimics.

The peptide natural product nisin has been used as a food preservative for 6 decades with minimal development of resistance. Nisin contains the unusual amino acids dehydroalanine and dehydrobutyrine, which are posttranslationally installed by class I lanthipeptide dehydratases (LanBs) on a linear peptide substrate through an unusual glutamyl-tRNA-dependent dehydration of Ser and Thr. To date, little is known about how LanBs catalyze the transfer of glutamate from charged tRNAGlu to the peptide substrate, or how they carry out the subsequent elimination of the peptide-glutamyl adducts to afford dehydro amino acids. Here, we describe the synthesis of inert analogs that mimic substrate glutamyl-tRNAGlu and the glutamylated peptide intermediate, and determine the crystal structures of 2 LanBs in complex with each of these compounds. Mutational studies were used to characterize the function of the glutamylation and glutamate elimination active-site residues identified through the structural analysis. These combined studies provide insights into the mechanisms of substrate recognition, glutamylation, and glutamate elimination by LanBs to effect a net dehydration reaction of Ser and Thr.

Overall Retention of Methyl Stereochemistry during B12-Dependent Radical SAM Methyl Transfer in Fosfomycin Biosynthesis.

Methylcobalamin-dependent radical S-adenosylmethionine (SAM) enzymes methylate non-nucleophilic atoms in a range of substrates. The mechanism of the methyl transfer from cobalt to the receiving atom is still mostly unresolved. Here we determine the stereochemical course of this process at the methyl group during the biosynthesis of the clinically used antibiotic fosfomycin. In vitro reaction of the methyltransferase Fom3 using SAM labeled with 1H, 2H, and 3H in a stereochemically defined manner, followed by chemoenzymatic conversion of the Fom3 product to acetate and subsequent stereochemical analysis, shows that the overall reaction occurs with retention of configuration. This outcome is consistent with a double-inversion process, first in the SN2 reaction of cob(I)alamin with SAM to form methylcobalamin and again in a radical transfer of the methyl group from methylcobalamin to the substrate. The methods developed during this study allow high-yield in situ generation of labeled SAM and recombinant expression and purification of the malate synthase needed for chiral methyl analysis. These methods facilitate the broader use of in vitro chiral methyl analysis techniques to investigate the mechanisms of other novel enzymes.

Substrate Sequence Controls Regioselectivity of Lanthionine Formation by ProcM.

Lanthipeptides belong to the family of ribosomally synthesized and post-translationally modified peptides (RiPPs). The (methyl)lanthionine cross-links characteristic to lanthipeptides are essential for their stability and bioactivities. In most bacteria, lanthipeptides are maturated from single precursor peptides encoded in the corresponding biosynthetic gene clusters. However, cyanobacteria engage in combinatorial biosynthesis and encode as many as 80 substrate peptides with highly diverse sequences that are modified by a single lanthionine synthetase into lanthipeptides of different lengths and ring patterns. It is puzzling how a single enzyme could exert control over the cyclization processes of such a wide range of substrates. Here, we used a library of ProcA3.3 precursor peptide variants and show that it is not the enzyme ProcM but rather its substrate sequences that determine the regioselectivity of lanthionine formation. We also demonstrate the utility of trapped ion mobility spectrometry-tandem mass spectrometry (TIMS-MS/MS) as a fast and convenient method to efficiently separate lanthipeptide constitutional isomers, particularly in cases where the isomers cannot be resolved by conventional liquid chromatography. Our data allowed identification of factors that are important for the cyclization outcome, but also showed that there are no easily identifiable predictive rules for all sequences. Our findings provide a platform for future deep learning approaches to allow such prediction of ring patterns of products of combinatorial biosynthesis.