Engineering of new-to-nature ribosomally synthesized and post-translationally modified peptide natural products

Abstract

Natural products have historically been important lead sources for drug development, particularly to combat infectious diseases. Increasingly, their structurally complex scaffolds are also envisioned as leads for applications for which they did not evolve, an approach aided by engineering of new-to-nature analogs. Ribosomally synthesized and post-translationally modified peptides (RiPPs) are promising candidates for bioengineering because they are genetically encoded and their biosynthetic enzymes display significant substrate tolerance. This review highlights recent advances in the discovery of highly unusual new reactions by genome mining and the application of engineering approaches to generate and screen novel RiPP variants. Furthermore, through the use of synthetic biology approaches, hybrid molecules with enhanced or completely new activities have been identified, which opens the door for future advancement of RiPPs as potential next-generation therapeutics.

Graphical Abstract

Introduction

Natural products (NPs) have played critical roles in the drug discovery and development process. A recent survey illustrates that more than 50% of all approved drugs from 1981 to 2019 are either NPs or derived thereof [1]. RiPPs are a large class of NPs that have received increasing attention in recent years. Their extensive post-translational modifications (PTMs) endow RiPPs with remarkable structural diversity, and an associated wide range of biological activities from antimicrobial to antiallodynic [2]. Enzymes that are involved in their biosynthesis have been studied extensively, which has enabled the discovery of novel mechanisms that have leveraged their potential in bioengineering.

RiPP biosynthesis features the translation of a precursor peptide, which in most cases comprises a C-terminal core peptide (CP) fused to an N-terminal leader region (Figure 1) [3,4]. The modification enzymes often recognize the leader peptide (LP) and install PTMs on the CP. The separation of substrate recognition in the LP and chemical modifications in the CP renders many RiPP biosynthetic enzymes highly tolerant of changes in the sequence of the CP. Following the PTMs, the LP is typically removed by a protease to reveal the final bioactive product. At present, more than 40 different Ripp classes have been reported that differ in the types of PTMs [4]. The rapid expansion of characterized RiPP classes has been aided by new RiPP-specific bioinformatic tools such as BAGEL [5], RODEO [6], RiPPER [7], and DeepRiPP [8] (for a comparison of these and other bioinformatics tools, see reference [4]). These tools have allowed discovery in either a class-specific manner using machine learning to find new examples of existing classes, or a class-agnostic manner to discover entirely new RiPP families [4]. This review provides an overview of recent progress (2017-2020) on the biosynthetic processes leading to this large group of natural products, with an emphasis on novel PTMs and bioengineering studies.

Figure 1.

General biosynthetic pathway of RiPPs that features a precursor peptide, modifying enzymes, protease, and transporter. The precursor peptide is composed of a leader peptide (LP), which contains enzyme recognition sites, and a core peptide (CP), which undergoes the PTMs that often include macrocyclization.

Biosynthesis of Lanthipeptides

Lanthipeptides are a large subfamily of RiPPs that contain intramolecular thioether linkages that are referred to as lanthionine (Lan) or methyllanthionine (MeLan) [9]. Lanthipeptides have a wide range of biological functions, including antimicrobial, antifungal, antiviral, and virulence activity. While the potent antimicrobial and low resistance properties of lanthipeptides such as nisin have been exploited in the food industry for decades [10], the current need for new antibiotics has catalyzed a flurry of recent studies on this class of peptides. Because lanthipeptides were the first class of RiPPs that were recognized to be made by PTMs [11], they have often served as a prototypical RiPP pathway for engineering purposes.

The biosynthesis of lanthipeptides involves dehydration of select Ser and Thr residues to 2,3-didehydroalanine (Dha) and (Z)-2,3-didehydrobutyrine (Dhb), respectively. Following dehydration, cysteine residues attack the dehydroamino acids through a Michael-type addition to form the Lan or MeLan thioether crosslinks in the final products (Figure 2a) [9]. In addition to these class-defining modifications, a range of other PTMs, such as methylation, hydroxylation, and epimerization, are found in individual lanthipeptides [12]. N- and C-terminal modification is common amongst RiPPs and is believed to protect the products from proteolytic degradation. One recently discovered example is the anti-HIV divamides that contain a trimethylated N-terminus introduced by the N-methyltransferase DivMT [13] (Figure 2b). Another interesting new subgroup of lipidated lanthipeptides is modified at both N- and C-termini [14,15]. In one member called microvionin, the N-terminus is modified by an N,N'-bismethylated guanidine fatty acid (Figure 2b) [14,16], thereby resembling lipopeptides of non-ribosomal origin. Genes encoding a putative 3-oxoacyl-acetyl carrier protein (ACP) synthase and an ACP were identified to form the fatty acid chain, indicating a natural hybrid fatty acid-RiPP pathway. An acyltransferase that links the fatty acid to the RiPP segment was recently identified and characterized, and offers exciting opportunities for engineering of hybrid compounds [15]. In addition to the lipidation of the N-terminus of microvionin, MicD catalyzes oxidative decarboxylation of the C-terminal Cys residue, which then forms a sulfur and carbon bridged crosslink called avionin (Figure 2a) [14]. Modifications of N- and C-termini are illustrated here for lanthipeptides, but are common amongst all RiPPs with a diverse collection of modification strategies [4].

Figure 2.

Lanthipeptides discussed in this review. The parts of the structures that are Ser/Thr- and Cys-derived are shown in red and blue, respectively, and other post-translationally modified residues are shown in green. a, Structure of the characteristic PTMs in lanthipeptides. Both the chemical structure and a shorthand notation are shown; the latter is used in panel b. b, Representative examples of lanthipeptides, including members with special terminal groups. Avi, avionin; Abu, 2-aminobutyric acid.

Biosynthesis of Pearlins

Peptide aminoacyl tRNA ligases (PEARLs) are a class of recently discovered enzymes that catalyze non-ribosomal peptide extensions using ribosomally-synthesized small peptides as scaffolds [17]. PEARLs share sequence similarity with LanB dehydratases involved in lanthipeptide biosynthesis that use glutamyl-tRNA to transfer the glutamyl group to the side chains of Ser/Thr residues [18]. Then, LanB enzymes eliminate the glutamate to generate Dha and Dhb. However, unlike LanBs that contain both glutamylation and elimination domains, PEARLs lack the elimination domains [17] (Figure 3a).

Figure 3.

Aminoacyl-tRNA dependent RiPP biosynthetic enzymes. a, Three different architectures of aminoacyl-tRNA dependent RiPP biosynthetic enzymes. For examples of lanthipeptides, thiopeptides, and pearlins see Figures 2 and 4. b, Biosynthetic gene cluster in P. syringae that encodes the PEARL TglB, and the biosynthetic pathway towards 3-thiaglutamate. c, TglB mechanism resulting in the addition of Cys to the C-terminus of the peptide TglA.

Among more than 600 PEARLs that were identified from a survey of bacterial genomes [17], the PEARLs that have been studied thus far are involved in the biosynthesis of natural products such as 3-thiaglutamate and the ammosamides (Figure 3b and 4a) [17,19]. TglB from the plant pathogen Pseudomonas syringae transfers a Cys from cysteinyl-tRNA to the C-terminus of a ribosomally-synthesized precursor peptide TglA (Figure 3b) [19]. Studies on the mechanism of TglB argue against a ping-pong mechanism in which the Cys is first transferred to the protein. Instead, the C-terminal carboxylate of the peptide substrate is phosphorylated in an ATP-dependent step followed by attack by the amino group of Cys-tRNA and finally hydrolysis of the tRNA (Figure 3c). Thus, this amide bond-forming process at the C-terminus of a peptide differs in several ways from ribosomal peptide extension. Most importantly PEARLs catalyze amide bond formation without an RNA template, and hence these enzymes must recognize both the substrate peptide and the aminoacyl-tRNA. In addition, the leaving group during amide bond formation is phosphate rather than the ribose at the terminus of tRNA. Finally, the hydrolysis of the tRNA in the last step of PEARL catalysis means that the activated ester of the aminoacyl-tRNA is never used. As such, this step resembles the peptidyl-tRNA hydrolase activity of release factors.

Figure 4.

Structures of select RiPPs discussed in this review. The enzymes that install the PTMs colored in red are discussed in the text or were used for the hybrid RiPPs in Table 1. a. Structures of the pearlin ammosamide A, the borosin omphalotin A, and the cyanobactin trunkamide. b. Examples of radical SAM enzymes catalyzing spliceotide (top) and cyclophane (bottom) formation. c. Structures of the sactipeptide subtilosin, the lasso peptide citrulassin A, the linear azole containing peptide goadsporin, and the thiopeptide thiocillin. For citrulassin, the C-terminal tail threads through the macrolactam in the active fold (not shown).

Following substrate cysteinylation, TglHI catalyzes an unusual β-carbon excision on Cys to produce peptide 1 (Figure 3b). Subsequently, the carboxy-S-adenosylmethionine (Cx-SAM) synthase TglE and a Cx-SAM–dependent methyltransferase TglF cap the thiol group of product 1 to form TglA-thiaGlu. The starting substrate TglA is then regenerated from TglA-thiaGlu by the protease TglG such that this scaffold peptide can be used in catalytic fashion [17]. The putative catalytic usage of the TglA substrate would relieve the burden of stoichiometric LP use in other RiPP pathways (Figure 1). 3-Thia glutamate was suggested to interfere with plant defense mechanisms that are activated by glutamate signaling pathways [17].

Several other PEARLs were identified in various bacterial genomes. AmmB is encoded in the ammosamide biosynthetic gene cluster in Streptomyces sp. CNR698 and appends a Trp from Trp-tRNA to the C-terminus of a ribosomally produced peptide by a mechanism like that shown in Figure 3c (Figure 4a) [17]. Thus, PEARLS are involved in the biosynthesis of small molecules with very different structures that have been called pearlins [4]. Given the two different PEARLs characterized to date that append unique amino acids (Cys and Trp), additional activities are anticipated to remain to be uncovered. The specificity of these enzymes for the peptide and aminoacyl-tRNA substrates is currently not yet known.

Biosynthesis of borosins, N-methylated RiPPs

Inspired by the excellent pharmacokinetic profile of cyclosporine, N-methylated cyclic peptides have served as candidates to improve key pharmacokinetic characteristics such as metabolic stability, membrane permeability, target specificity, and oral availability [20]. While methylation of peptide N-termini and side chains have been identified in several RiPPs [4], backbone N-methylation had been limited to nonribosomal peptides until the discovery of the biosynthetic pathway to the borosins. These peptides are ribosomally synthesized and then modified by a SAM-dependent N-methyltransferase [21,22]. The best-studied example is omphalotin A, a 12-mer or α-N-methylated head-to-tail cyclized peptide that is produced by the fungus Omphalotus olearius (Figure 4a). In an unusual architecture for RiPPs, the substrate peptide and the SAM-dependent methyltransferase are encoded by one gene, resulting in a single polypeptide (OphMA) with the N-terminus of the substrate fused to the C-terminus of the enzyme [21]. By co-expressing OphMA and the prolyl oligopeptidase OphP in a fungal heterologous system, the product was successfully methylated nine times [22].

Studies on the methyltransferase revealed N-to-C directionality of methylation [21], and crystallographic studies of the methyltransferase with its fused substrate revealed a complex catenane-like structure in which the homodimers interlock and methylate each other’s C termini [23,24]. More recently, a bioinformatic study revealed over 50 putative pathways in ascomycete and basidiomycete fungi and validated more than 10 autocatalytic borosin precursors, highlighting the potential of finding novel RiPP pathways in fungi [25].

Biosynthesis of spliceotides and cyclophanes

Natural ribosomal biosynthesis of peptides and proteins results exclusively in α-amino acid backbone topology. Recently, a study revealed a biosynthetic pathway that produces β-amino acid containing products [26]. PlpX, a radical S-adenosylmethionine (SAM) enzyme from Pleurocapsa sp. PCC 7319, catalyzes the formation of an α-ketoβ-amino acid containing product when coexpressed with its substrate peptide and an accessory protein PlpY. By incorporating isotopically labelled amino acids in the expression system, PlpX was shown to catalyze tyramine excision from the peptide backbone, followed by reconnection of the two peptide fragments to generate a β-amino acid containing peptide (Figure 4b) [26]. The product peptides were termed spliceotides. Bioinformatics studies identified genes encoding such enzymes in cyanobacteria, proteobacteria, and actinomycetes. Studies on the promiscuity of PlpX revealed that at least two α-ketoβ-amino moieties can be introduced in the CP with a minimal requirement of 11-residues [26], which opens the door for bioengineering opportunities to form diverse β-amino acid containing peptides from ribosomally synthesized precursors.

Another recently discovered recurring theme that involves radical SAM enzymes is the formation of carbon-carbon crosslinks between aromatic and aliphatic side chains of nearby residues. Recent examples include streptides (Lys-Trp crosslink) [27], ryptides (Arg-Tyr) [28], and triceptides (Trp/Phe-Asn/Lys/Gln/Arg/Asp/Ser crosslinks) [29] (e.g. Figure 4b) resulting in strained cyclophane products. In addition, cytochrome P450 enzymes have been implicated in crosslinks between two nearby aromatic side chains [30]. Notably, synthetic efforts towards one of these strained macrocyclic products revealed that two non-interconvertible stereoisomers are possible that are identical at all stereogenic centers, but only one of which is made naturally. Such isomerism was termed atypical atropisomerism [30], and given the many examples of overlapping macrocycles in RiPPs [4], it is likely that more examples will be revealed.

Generation of hybrid RiPPs

As illustrated in the preceding sections, the chemistry catalyzed by RiPP biosynthetic enzymes is remarkably diverse and unusual. Understanding of RiPP biosynthetic pathways has been leveraged to demonstrate the bioengineering potential to form new-to-nature hybrid peptides. Many studies have focused on engineering of the LP as it plays a critical role to guide modification enzymes to install PTMs on the core peptide [3]. From the very start of research on RiPP biosynthesis, scientists have been intrigued by the prospects of fusing LPs to designer peptides of interest that may then be substrates for the biosynthetic enzymes. Table 1 provides a historical overview of this strategy, which initially focused on chimera of natural LP and CP sequences from the same RiPP class, but which has evolved to ever more sophisticated combinations.

Table 1:

Chimeric substrate peptides generated by fusing leader peptides to non-cognate peptides to produce RiPPs or their variants. For structures of the compounds whose biosynthetic enzymes were used, see Figure 4. Not listed are the many examples of mutagenesis to obtain variants of natural RiPPs.

Leader peptide	Core peptide (NP)	Processing enzymes (structure generated)	Product(s)	Significance	Ref
subtilin LP	NisA (nisin Z)	nisin biosynthetic enzymes (lanthionine)	unmodified leader peptide of subtilin linked to nisin Z	first example of leader and core peptide chimera	[39]
subtilin LP	hybrid of NisA and SpaS (subtilin)	subtilin biosynthetic enzymes (lanthionine)	succinylated chimera of nisin and subtilin	first example to demonstrate critical recognition of the LP by the processing machinery	[40]
nisin LP	GdmA (gallidermin)	GdmD (AviCys), NisBC (lanthionine)	gallidermin	in vivo enzyme-substrate plug- and-play system	[41]
nisin LP	NisA (nisin, Fig. 2b)	LtnJ (D-Ala), NisBC (lanthionine)	nisin with D-Ala	in vivo enzyme-substrate plug- and-play system	[41]
nisin LP	ElxA (epilancin 15X, Fig. 2b)	nisin biosynthetic enzymes (lanthionine)	partially dehydrated epilancin 15X	accessing RiPPs using orthologous enzyme systems	[34]
ProcA3.2 LP	LctA (lacticin 481)	ProcM (lanthionine)	dehydrated and cyclized lacticin 481	accessing RiPPs using orthologous enzyme systems	[34]
HcaA LP NisA LP	hybrid of NisA and HcaA	HcaDF (thiazoline), NisBC (lanthionine)	thiazoline containing lanthipeptide	engineered macrocyclic RiPP hybrids	[33]
HcaA LP AlbA LP	AlbA (subtilosin, Fig. 4c)	HcaDF (thiazoline), AlbA (sactionine)	thiazoline containing sactipeptide	engineered macrocyclic RiPP hybrids	[33]
HcaA LP ProcA LP	ProcA3.3 (prochlorosin 3.3, Fig. 2b))	HcaDF (thiazoline), ProcM (lanthionine)	thiazoline containing lanthipeptide	engineered macrocyclic RiPP hybrids	[33]
HcaA LP ProcA LP	ProcA3.3 (prochlorosin 3.3, Fig. 2b)	HcaDF (thiazoline), ProcM (lanthionine), NpnJA (D-Ala)	D-alanine/thiazoline containing lanthipeptide	engineered macrocyclic RiPP hybrids	[33]
HcaA LP ProcA LP	ProcA3.3 (prochlorosin 3.3, Fig. 2b)	HcaDF (thiazoline), ProcM (lanthionine), MibD (decarboxylati	C-terminally decarboxylated, azoline-containing lanthipeptide	engineered macrocyclic RiPP hybrids	[33]
fusilassin LP	CitA D8E (citrulassin, Fig. 4c)	fusilassin biosynthetic enzymes (lasso fold)	citrulassin A D8E	accessing RiPPs using orthologous enzyme systems	[36]
nisin LP	PneA1 and PneA2 (pneumococcin)	nisin biosynthetic enzymes (lanthionine)	pneumococcins A1 and A2	accessing RiPPs using orthologous enzyme systems	[35]
cyanobactin LP	TruE variants (trunkamide, Fig 4a)	trunkamide biosynthetic enzymes (thiazoles, N-to-C cyclization)	cyanobactins with nonproteinogenic amino acids	heterologous expression system to incorporate ncAAs	[42]
nisin LP	NisA with C-terminal short peptides	nisin biosynthetic enzymes	nisin variants	improved activity against Gram-negative microorganisms	[43]
nisin LP	54 putative class I and II lanthipeptides	nisin biosynthetic enzymes	30 new lanthipeptides	accessing RiPPs using orthologous enzyme systems	[44]
nisin LP	randomized motifs from 11 lanthipeptides	nisin biosynthetic enzymes	>100 hybrid peptides, some with improved bioactivities	library generation and hybrid peptides that overcome nisin (self)resistance problems	[45]

A recent study constructed a library of variants of lanthipeptides by shuffling peptide modules derived from naturally-occurring lanthipeptides to develop new structures with improved antimicrobial activity and that bypass resistance problems [31]. The peptides were designed with the NisA LP from the nisin precursor attached to CPs that were combinatorially varied in several functional regions: a lipid II binding region, a hinge region, and a pore formation region (Figure 2b). The regions were selected from 12 natural lanthipeptides and different regions were shuffled by DNA synthesis to obtain a library of 6000 variants. The variants were modified by the nisin biosynthetic enzymes inside Lactococcus lactis, secreted using the natural export machinery, and activated via LP removal by the addition of external protease NisP. A high-throughput nano-Fleming inhibition assay was used to screen the bioactivity of the variants by compacting fluorescently labeled producer and sensor cells into beads [31,32]. The beads with a small number of sensor cells were identified as containing potential producers of antimicrobial hybrids. With this combinatorial approach, a number of hybrid variants were identified with potent activities against strains that are not sensitive to wild-type nisin.

Taking utilization of the unique LP-dependent biosynthetic logic of RiPP biosynthesis one step further, knowledge of specific recognition sequences within LPs has allowed the design of hybrid LPs that engage enzymes from different RiPP families. Thus, hybrid lanthipeptides containing thiazolines, as well as sactipeptides containing thiazolines, were generated inside Escherichia coli (Table 1) using biosynthetic enzymes involved in the biosynthesis of prochlorosin (Figure 2b), subtilosin, and azole containing peptides (e.g. Figure 4c) [33].

Chimeric substrates have also been used to access natural RiPPs using non-cognate biosynthetic machinery such as the production of the lanthipeptides lacticin 481 and pneumococcin with the prochlorosin [34] and nisin biosynthetic enzymes [35], respectively, the lasso peptide citrulassin with the fusilassin biosynthetic enzymes [36], and many examples of cyanobactin biosynthesis with non-cognate enzymes [37,38].

In vitro translation systems and incorporation of noncanonical amino acids (ncAAs)

Although significant progress has been made in RiPP engineering using in vivo expression systems, it is challenging to control the timing and quantity of PTM enzyme expression. In vitro translation technology can overcome these hurdles since no living cells are involved and all enzymes are purified externally. Thus, cell-free protein synthesis systems have the potential advantage of improving the homogeneity of the final product, preventing toxicity that a novel antibiotic can pose to the heterologous host, and simplifying the incorporation of ncAAs by genetic code reprogramming [46,47]. Combined with the use of engineered flexizymes (flexible tRNA acylation ribozymes), tRNA can be acylated with various donors without the requirement of structural similarity as posed by canonical amino acyl-tRNA synthetases (ARSs) [48]. N-Methylated, D-, β-, and γ-amino acids [48], as well as phenylselenocysteine [49] have been successfully incorporated into RiPPs by using cell-free protein biosynthesis for the substrate peptide (Figure 5). In these systems, unnecessary translation components such as ARSs and release factors can be excluded to simplify the scheme. In vitro translation systems have been successfully used to study the tolerance of RiPP enzymes involved in the biosynthesis of cyanobactins [50,51], the linear azole-containing peptide goadsporin [52] (Figure 4c), the thiopeptides thiocillin [49] (Figure 4c) and lactazole A [53], and the lanthipeptide nisin [54] (Figure 2b). Although these strategies are generally difficult to scale up (8-10 μM represents the practical upper limit [53]), the modification of non-native precursors has provided greatly expanded information about substrate tolerance and sometimes mechanism of RiPP enzymes and has allowed the synthesis of artificially designed RiPP derivatives [47].

Figure 5.

Summary of the flexible in vitro translation (FIT) system and its use in building nonstandard peptides and reconstituting biosynthetic pathways that result in new-to-nature RiPP analogs. Flexizymes charge tRNAs with ncAAs, in vitro translation incorporates the ncAAs into the peptide substrate, and the RiPP biosynthetic enzymes then install the PTMs in the final product.

Methods for production and screening of libraries of RiPP variants

Because of their genetically encoded precursors, RiPPs are well-suited for library generation. One recent study reported the generation of a library of 10⁶ non-natural lanthipeptides by heterologously expressing the lanthipeptide synthetase ProcM with peptides containing the cognate LP fused to a library of CP variants in E. coli [55]. Because of the highly substrate-tolerant nature of ProcM, the members of the peptide library were successfully dehydrated and cyclized. The resulting macrocyclic peptide library was screened to identify a member that prevented the interaction between the HIV p6 protein and the UEV domain of the human TSG101 protein, an interaction that is required for the HIV viral budding process. By coupling lanthipeptide generation to a bacterial reverse two-hybrid system, a peptide was identified that inhibited the interaction.

Surface display technologies facilitate the identification of peptides that bind to a protein or molecule of interest from a pool of random sequences. Two independent studies used phage display to screen for peptides with novel biological activities. The studies fused a lanthipeptide precursor library that contains natural leader sequences and randomized CPs to the pIII protein of M13 phage and co-expressed these peptides along with a lanthipeptide synthetase in E. coli. This approach led to the identification of cyclic peptides that bind to streptavidin and urokinase-type plasminogen activator [56], and the generation of variants of the lanthipeptide nisin that were screened for binding to its target lipid II [57]. mRNA display techniques are also well suited for adaptation to RiPP biosynthesis [58,59], and a lanthipeptide yeast display system was also recently established [57]. In the latter study, two lanthipeptides were identified that bind to αvβ3 integrin as potential ligands for tumor imaging. These studies provide blueprints that can be extended to other RiPP families. Indeed, it has already been established that fusing proteins to the C-terminus of lasso peptide precursors still allows their formation suggesting that the display of lasso peptides should be feasible [60].

Conclusions

RiPPs are genetically encoded NPs with high engineering potential. The recent explosion of genomic information as well as the development of new genome mining tools have accelerated the discovery of new RiPP classes [4]. In turn, these developments have led to exciting new biochemical transformations as highlighted in this review. Notably, most RiPPs are macrocyclic peptides, a class of compounds that is of high contemporary interest as a modality that in terms of size falls between typical small molecules and biologics [61]. Given their advantageous properties compared to linear peptides [62], macrocyclic peptides are heavily investigated for their ability to target protein-protein interactions [63]. In principle, RiPP biosynthetic pathways can greatly expand the types of mostly monocyclic macrocycles that can be accessed by synthetic chemistry and provide access to polycyclic compounds that may result in higher affinity or specificity. Because RiPP pathways are often exceptionally substrate tolerant, they enable production of novel peptides through bioengineering approaches such as the generation of hybrid RiPPs, incorporation of noncanonical amino acids, and surface display technologies for peptide library screening. Although the size of the RiPP family is rapidly expanding, the number of commercially approved RiPPs still remains limited. Future work will likely focus on developing hybrid molecules that have enhanced bioactivity and stability, and building new platforms to screen cyclic peptide libraries to identify lead compounds for next generation therapeutics.

Highlights.

• Discovery of novel post-translational modifications in natural product biosynthesis

• An enzyme that methylates the backbone amides of ribosomally synthesized peptides

• Use of biosynthetic enzymes to make macrocyclic peptide libraries

• Hybrid compounds generated by combining enzymes from different pathways

• Enzymatic generation of macrocyclic peptides displayed on phage, yeast, or mRNA