Type II Non-ribosomal Peptide Synthetase Proteins: Structure, Mechanism, and Protein-Protein Interactions

Abstract

Many medicinally-relevant compounds are derived from non-ribosomal peptide synthetase (NRPS) products. Type I NRPSs are organized into large modular complexes, while type II NRPS systems contain standalone or minimal domains that often encompass specialized tailoring enzymes that produce bioactive metabolites. Protein-protein interactions and communication between the type II biosynthetic machinery and various downstream pathways are critical for efficient metabolite production. Importantly, the architecture of type II NRPS proteins makes them ideal targets for combinatorial biosynthesis and metabolic engineering. Future investigations exploring the molecular basis or protein-protein recognition in type II NRPS pathways will guide these engineering efforts. In this review, we consolidate the broad range of NRPS systems containing type II proteins and focus on structural investigations, enzymatic mechanisms, and protein-protein interactions important to unraveling pathways that produce unique metabolites, including dehydrogenated prolines, substituted benzoic acids, substituted amino acids, and cyclopropanes.

Graphical Abstract

1. Introduction

Natural products derived from microorganisms have many diverse biological functions and properties. Evolutionarily, natural products are metabolites that confer a selective advantage to an organism in its environment, while others are agents of microbial warfare that promote invasion and pathogenesis. Clinically, the majority of marketed drugs from the past three decades are natural products or synthetic compounds inspired by natural products.^1,2 The biosynthesis of these compounds has been studied extensively due to the large spectrum of bioactivities exhibited in humans. Investigations into natural product biosynthesis enables the discovery of novel chemical transformations and also provides a scaffold for future metabolic pathway engineering to produce compounds with enhanced pharmacological properties.

Non-ribosomal peptide synthetases (NRPSs) are a large family of biosynthetic enzymes that generate medicinally-relevant natural products from amino acid precursors.^3,4 Unlike the peptides and proteins generated by the ribosome, NRPSs incorporate amino acids into their biosynthetic machinery for elongation into secondary peptide metabolites. The NRPSs are frequently categorized into type I and type II based on differences in the overall architecture of the synthetases (Fig. 1). Type I NRPSs are large modular complexes containing all the enzymes necessary to generate a peptide product in an assembly line fashion analogous to type I fatty acid synthases (FASs) and polyketide synthases (PKSs). Type II NRPS proteins are commonly standalone enzymes or didomains that coordinate to form unique amino acid derivatives (Fig. 1). Unlike type II FAS and PKS, the type II NRPS proteins are linear, non-iterative pathways that contain specialized tailoring enzymes and combine with other pathways to generate a final product. Tailoring enzymes are featured in both type I and type II NRPS systems; however, type II tailoring enzymes often provide further diversification uncommon in type I systems, such as dehydrogenation, cyclopropanation, or functional group migration.

Fig. 1.

Phylogenetic trees of peptidyl carrier proteins (PCPs) and adenylation (A) domains in type II NRPS proteins and selected type I NRPSs.

Before the establishment of the type II NRPS terminology, the standalone adenylation domain EntE from enterobactin biosynthesis was characterized as an AMP ligase that presumably generated an AMP intermediate for direct incorporation into downstream pathways.⁵ Almost a decade later, the corresponding EntE PCP partner and AMP-intermediate acceptor EntB was fully characterized.⁶ Soon after, the term “type II” NRPS was coined by Shen and coworkers in the characterization of the bleomycin type II PCP BlmI.⁷ In subsequent years, many natural product gene clusters were investigated with the improvement of DNA sequence techniques at the end of the 20^th century. Gene cluster findings led to the discovery of new, unique type II NRPS proteins (Fig. 2). Over the past two decades, many type II NRPS enzymes have been characterized and include the emergence of unique homologous systems. The unique tailoring enzymes and standalone architecture make the type II NRPS proteins ideal targets for combinatorial biosynthesis and metabolic pathway engineering to generate enhanced pharmaceutical compounds.

Fig. 2.

Selected gene clusters containing type II NRPS proteins. Genes associated with type II NRPS proteins are in color and type I NRPS genes are in grey. The clusters are sorted by dehydrogenated prolines (A), substituted benzoic acids (B), halogenated aliphatic amino acids (C), cyclopropanes (D), β-hydroxylated amino acids (E), and substituted aromatics (F).

Herein, we provide an extensive overview of the biochemical, structural, and mechanistic studies of type II NRPS enzymes. NRPSs have been the subject of several reviews over the past five years,^8–25 and only one review focuses on NRPS type II thioesterases.²⁶ The emergence of modern genome mining methods and tools has illuminated several additional standalone enzymes or didomains essential for producing unique chemical scaffolds, and elucidating protein-protein interactions in these systems is critical to advance combinatorial biosynthesis efforts in this area.

The peptidyl carrier protein (PCP) and the adenylation (A) domain establish the basis for type II NRPS systems described in the current review. We consider type II NRPS proteins as standalone PCPs, standalone A domains, and PCP-A di-domains that participate in biosynthesis with other carrier-protein dependent pathways, including type I NRPS, PKS, and FAS. Furthermore, several of the homologous families described below contain both standalone and di-domain systems, illustrating that they are all type II pathways. We have included a phylogenetic tree of PCPs and A domains to guide the reader (Fig. 1). The tree illustrates the similarities between type II NRPS proteins and differences from type I NRPSs. Based on the tailoring enzymes and the phylogenetic analysis, we have established six type II NRPS sub-families: dehydrogenated prolines; substituted benzoic acids; halogenated aliphatic amino acids; cyclopropanes; β-hydroxylated amino acids; and substituted aromatic amino acids. We have also included individual miscellaneous systems that may have undiscovered homologous pathways.

2. Dehydrogenated prolines

The dehydrogenated prolines consist of pyrrole and pyrroline moieties that are prepared via type II NRPS proteins (Fig. 3A). Pyrroles are found in many diverse compounds, including the prodiginines, pyoluteorin, chlorizidine A, and clorobiocin. In addition to the PCP and A domain, all type II pyrrole and pyrroline biosynthetic gene clusters consist of a dehydrogenase gene (Fig. 2A). The prodiginine compounds have been isolated from numerous species and have diverse biological properties.²⁷ The prodiginine derivative obatoclax was recently examined in stage II clinical trials as an anticancer therapeutic for hematologic tumors.²⁸ Pyoluteorin is an antifungal agent and the producing species Pseudomonas putida serves as a promising biopesticide during crop production.²⁹ The recently discovered marine Streptomyces sp. metabolite chlorizidine A exhibits notable anti-tumor activity targeting human α-enolase and glyceraldehyde-3-phosphate dehydrogenase within the glycolytic pathway.^30,31 The proposed chlorizidine A pathway suggests a unique interaction between NRPS and FAS machinery.³² The Streptomyces species’ metabolite clorobiocin and other aminocoumarin compounds inhibit bacterial DNA gyrases making the compounds highly effective antibiotic agents.³³ The Streptomyces sp. metabolite hormaomycin is a structurally complex compound that exhibits highly potent activity against Coryneform bacteria and influences interspecies differentiation and metabolism.^34,35 A pyrroline intermediate is generated by a type II NRPS system in the biosynthesis of anatoxin-a, a neurotoxin produced by cyanobacteria that contributes to water contamination and health issues.^36–38 Overall, the dehydrogenated proline natural products are structurally and functionally diverse and display a wide range of bioactivities.

Fig. 3.

Pyrrole natural products and biosynthesis. (A) Natural products containing a pyrrole derived from a type II NRPS. (B) Mechanistic proposal for pyrrole biosynthesis by PCP, adenylation (A) domain, and dehydrogenase (DG). The black bar and teal circle above the PCP denotes phosphopantetheine.

2.1. Proline activation

The first pyrrole type II genes to be identified were from the pyoluteorin, coumermycin, and undecylprodigiosin pathways (Fig. 2).^39–41 The core enzymes for pyrrole formation are the type II PCP, adenylation (A) domain, and FAD-dependent dehydrogenase. The continued discovery of homologous enzymes illustrates the prevalence of the family in nature.^42–48 In the pyoluteorin pathway, pyrrole formation is initiated by the A domain PltF, which activates ʟ-proline to prolyl-AMP and facilitates transfer to the pantetheine arm of PltL (Fig. 3B).⁴⁹ Interestingly, the homologous A domains (PltF/RedM) in pyoluteorin and undecylprodigiosin biosynthesis are specific for their native PCP, illustrating the importance of protein-protein recognition in this family.⁴⁹ In contrast, the prodigiosin PCP was found to be a promiscuous substrate for the three A domains from the pyoluteorin, prodigiosin, and undecylprodigiosin pathways.⁵⁰ Structural analysis of the PCP PigG and comparison to the PltL structure revealed a loop region important for recognition by the A domain and the region may influence specificity with the PCP partner.⁵⁰

2.2. FAD-dependent dehydrogenase

The proline dehydrogenases perform a four-electron oxidation on the tethered proline to generate pyrrole (Fig. 3B). The two step catalysis was confirmed in the homologous dehydrogenases from the biosynthesis of coumermycin and clorobiocin by electrospray ionization-fourier transform mass spectroscopy (ESI-FT MS) analysis.⁵¹ Recently, an analogous pyrroline type II system was discovered in the biosynthesis of anatoxin.^52,53 In contrast to the pyrrole forming dehydrogenase, the anatoxin-a dehydrogenase AnaB selectively catalyzes a two electron oxidation.⁵³ Bioinformatic analysis of AnaB indicates the enzyme belongs to the acyl-CoA dehydrogenase superfamily and, therefore, is proposed to catalyze an α/β dehydrogenation on proline and then facilitate an aza-allylic isomerization to generate (S)-1-pyrolline-5-carboxylic acid tethered to PCP.^54,55 Indeed, the crystal structure of AnaB (Fig. 4) revealed a acyl-CoA dehydrogenase fold, but interestingly, details of the active site suggests the protein is an oxidase utilizing molecular oxygen as the electron acceptor.⁵⁶ The use of oxygen as an electron acceptor is supported by the recent x-ray crystal structure PigA, a homologous proline dehydrogenase from the prodigiosin biosynthetic pathway.⁵⁷ The cocrystal structure of FAD with PigA reveals a homotetramer with FAD bound at the interface of a dimer. Furthermore, the active site displays a putative binding site for molecular oxygen, which is consistent with the supposed oxygen binding site, FAD binding, and tetramic structure of AnaB. The proposed mechanism is different from the proposed electron transport mechanism of dehydrogenases. Structural analysis of other dehydrogenases in this family will offer insights into the mechanism of proline oxidation.

Fig. 4.

Structure of the dehydrogenase AnaB (PDB ID: 4IRN) from anatoxin A biosynthesis. The enzyme exists as a homotetramer, but the homodimer is displayed here. The monomers together create the FAD binding sites. The FAD cofactors are depicted in black.

Recently, the type II NRPS proline dehydrogenase from tetrabromopyrrole biosynthesis, Bmp3, was cocrystallized with its cognate PCP Bmp1 and FAD.⁵⁸ Akin to the functional homotetrameric structure, FAD binding, and oxygen binding of PigA and AnaB, Bmp3 also forms a homotetramer with FAD bound at the interface of 2 monomers. The PCP Bmp1 was also bound at the interface of two monomers. The PCP-DH interface consists of hydrophobic contacts via Bmp1’s alpha helices II and III and by Bmp3 helices G and K. The proposed proline oxidative mechanism begins with the C-2 proton abstraction by Glu242, a C-3 hydride transfer to FAD, followed by FADH2 oxidation from molecular oxygen. Then, imine formation between N1 and C-2 occurs with subsequent C5 proton elimination by Glu242 to form the product of AnaB. Bmp3 then installs the second double bond upon a C4 proton elimination and a consecutive C2 proton elimination to create the final pyrrole. This mechanism is supported via mutagenesis studies of the active site, where upon mutation of Glu242, natural product formation is completely lost. These recent structural analysis of dehydrogenase structures have shed light on potential type II NRPS proline oxidative mechanisms that are likely not to be found in type I NRPS systems.

2.3. FADH₂, O₂-dependent halogenase

Many of the PCP-tethered pyrroles are further modified by an FAD-dependent halogenase that catalyzes chlorination or bromination at different positions on the ring.^{44,48,59–61} The pyoluteorin FADH₂ halogenase PltA was the first PCP dependent halogenase to be characterized enzymatically.⁶⁰ PltA catalyzes a two electron electrophilic aromatic substitution to form 5-chloropyrrolyl-PCP and then 4,5-chloropyrrolyl-PCP (Fig. 5C). The halogenases utilize FADH₂ and O₂ to generate hypochlorous acid, which is proposed to form a chloramine species with a proximal lysine residue for electrophilic pyrrole substitution (Fig. 5B).⁶² The recent crystal structure of PltA highlights K73, the proposed lysine residue involved in chlorine transfer (Fig. 5A).⁶³ Interestingly, the C-terminal region of the PltA structure obstructs the halogenase active site and does not provide enough space for the pyrrolyl-PltL to access the region (Fig. 5A). In addition, the pyrrole in the NMR structure of pyrrolyl-PltL is inaccessible to downstream enzymes.⁶⁴ These studies illustrate that recognition between protein partners must induce a conformational change, reiterating the importance of protein-protein interactions in this group. In addition, the halogenase HrmQ from the hormaomycin pathway was heterologously expressed in the clorobiocin producer Streptomyces roseochromogenes and, interestingly, chlorinated the naturally non-chlorinated pyrrole of clorobiocin.⁴⁸ The reaction presumably occurs while the pyrrole is PCP-tethered based on previous studies where PltA only chlorinates the pyrrole while attached to PltL.⁶⁰ The combinatorial biosynthetic study suggests that some of the FADH₂ halogenases are promiscuous towards other PCPs, but this has not been shown through in vitro studies.

Fig. 5.

Pyrrole chlorination in pyoluteorin pathway. (A) Structures of PltA (PDB ID: 5DBJ) and PltL (PDB ID: 2N5I). The FAD, conserved K73, and the C-terminal region are highlighted on PltA. The Ppant-pyrrole and residues that interact with the pyrrole are highlighted on PltL. (B) The proposed FAD mechanism for chlorination of the PltL-tethered pyrrole. (C) The proposed chlorination of PltL-tethered pyrrole.

In contrast to the mono- and dichlorinated pyrroles, a tetrabrominated pyrrole moiety was discovered in Pseudoalteromonas sp. PS5 as a precursor to pentabromopseudilin. ⁶⁵ To gain further insight into the PCP-bound pyrrole tetrabromination mechanism of bmp2, the crystal structure of FAD bound bmp2 was solved.⁶⁵ In comparison to other pyrrole halogenase structures, the putative pyrrole binding site lacked 3 strictly conserved amino acids. Mutagenesis of these residues completely abolished tetrabromination, suggesting their importance in modulating halogen additions amongst the FAD dependent halogenases.

2.4. Pyrrole and pyrroline transfer to downstream pathways

The final pyrrole PCP products transfer to a variety of pathways. Many of the pyrrole intermediates are transferred to type I PKS systems facilitated by the initial ketosynthase (KS).^{27,43,44,46,66} Alternatively in the clorobiocin pathway, in vivo mutant studies revealed the predicted acyltranferase CloN2 is important for transfer of the acid-pyrrole to the 3’-hydroxyl position of a ʟ-noviosyl sugar (Fig. 6A).⁶⁷ Further genetic analysis indicated that the PCP CloN1 and acyl transferase CloN7 are also necessary for acid-pyrrole transfer.⁶⁸ CloN2 is proposed to transfer the pyrrole from CloN5 to CloN1; however, in vitro studies have not confirmed the function because the protein is insoluble when heterologously expressed.⁶⁹ The homologous enzymes of CloN2, CloN1, and CloN7 are also present in the coumermycin A₁ gene cluster (Fig. 6B). In vitro analysis has confirmed that CouN7 transfers the pyrrole from CouN1 to the novobiocin scaffold descarboamoylnovobiocin to generate the pyrrole product.⁷⁰ In fact, CouN7 is able to transfer a panel of pyrrole analogs from CouN1 to descarboamoylnovobiocin generating a compound with improved antibacterial activity compared to the original scaffold.⁷¹ The homolog of CloN2/CouN2 has been identified in the nargenicin A₁ gene cluster, while a CloN1/CouN1 or CloN7/CouN7 homolog has not been discovered (Fig. 6B).⁴⁵ The promiscuity of CouN7/CloN7 for transferring products to downstream scaffolds is promising as a tool for future engineering efforts. In chlorizidine A biosynthesis, heterologous expression and mutant analysis suggests a unique bifurcated pathway, where the PCP-tethered pyrrole is transferred to both fatty acid and polyketide synthase, followed by elongation and rejoining of the intermediates.³² The pyrroline substrate in anatoxin-a biosynthesis is presumably transferred to the polyketide synthase AnaE by the ketosynthase domain. Based on genetic analysis, the pyrroline moiety is a cryptic intermediate, as the functional group allows for catalysis of the bicyclic anatoxin-a scaffold by the predicted cyclase Orf1.⁵² Currently, there is no structural information on the protein-protein interactions associated with transfer of dehydrogenated prolines from type II NRPS proteins to downstream pathways.

Fig. 6.

Pyrrole transfer to hydroxyl ring scaffolds. (A) Enzymes involved in pyrrole transfer from clorobiocin, coumermycin A1, and nargenicin A1. NgnN1 and NgnN7 have not been identified yet. (B) Structures of final products after pyrrole transfer and methyl modification.

3. Substituted benzoic acids

The substituted benzoic acids include 2,3-dihydroxybenzoic acid (DHB), salicylic acid, and 4-methyl-3-hydroxyanthranilic acid (MHA) (Fig. 7). The DHB moiety is found in many catecholate-type siderophores such as enterobactin, bacillibactin, anguibactin, acinetobactin, and vibriobactin. Siderophores chelate ferric iron from the environment and are important for cell signaling and oxidative stress response.^72,73 In particular, the DHB moieties in catecholate siderophores directly chelate the metal ion and therefore are essential components.⁷⁴ Gene fusion data and sequence analysis of the enterobactin pathway revealed four biosynthetic genes responsible for DHB production (entC, entE, entB, entA) (Fig. 2B).⁷⁵ The enterobactin genes are homologous to other siderophore genes. Yersiniabactin, derived from Yersinia enterocolitica, contains salicylic acid, a moiety also important for metal chelation.⁷⁶ The siderophore also protects pathogens from copper toxicity during urinary tract infections in mice and humans.⁷⁷ The antibiotic chlorothricin, from Streptomyces antibioticus, contains a salicylate derivative and inhibits the mevalonate pathway during microbial cholesterol biosynthesis.^78,79 MHA is present in the anti-tumor agent actinomycin D, which was originally isolated from Streptomyces chrysomallus. Actinomycin D is currently used as a treatment for a variety of cancers.^80–83 The two MHA moieties make up the core scaffold of actinomycin D and, therefore, are essential to the compound’s properties (Fig. 7A).

Fig. 7.

Substituted benzoic acid products and biosynthesis. (A) Natural products containing the substituted benzoic acids. (B) Biosynthesis of DHB. ICS, isochorismate synthase; ICL, isochorismate lyase; DHB DG, 2,3-dihydro-2,3-DHB dehydrogenase

3.1. Dihydroxybenzoic acid activation

Based on early enzymatic isolation experiments, the formation of DHB from chorismic acid in enterobactin biosynthesis is catalyzed by three enzymes.^84–87 Two decades later, in vitro studies revealed isochorismate synthase (EntC), isochorismate lyase (EntB), and DHB dehydrogenase (EntA) were the enzymes responsible for DHB production (Fig. 7B).^88–90 Before incorporation into downstream pathways, DHB is tethered to a type II PCP by the A domain EntE.⁵ Interestingly, the type II PCP is EntB, which was further characterized as a bifunctional enzyme that harbors both the isochorismate lyase (ICL) and type II PCP domains (Fig. 7B).⁶ The ICL domain is not necessary for the PCP domain interaction with either the PPtase EntD or EntE, as truncation of ICL did not diminish the catalytic efficiency of either of these interacting enzymes. To date, the effect of acylated PCP on the activity of ICL is unknown. The homologous enzymes of EntA, EntB, EntC, EntE have also been characterized in the biosynthesis of many siderophores, including bacillibactin, anguibactin, vibriobactin, and acinetobactin.^91–98

The interaction between PCP and A domains has been most extensively investigated in the siderophore pathways. The A domain EntE is capable of acylating the excised surfactin PCP SrfB1, but at a significantly reduced rate compared to the native PCP EntB, illustrating the specificity between A domain and PCP.⁹⁹ A directed evolution system was developed to mutate the vibriobactin type II PCP VibB for increased recognition by the enterobactin NRPS machinery in an EntB-knockout E. coli strain.¹⁰⁰ In vitro studies with the evolved mutant VibB, revealed that residues in the predicted loop 1 single-turn helix and helix III were important for aminoacylation activity by the adenylation domain EntE. The important interaction between the PCP and A domain were further confirmed when the structure of the enterobactin EntB (PCP) and EntE (A domain) complex was determined in an aminoacylation conformation via a mechanistic-based inhibitor (Fig. 8).^101,102 The structure reveals that loop 1 of the EntB interacts exclusively with the EntE C-terminal region, with several salt bridges and other ionic interactions present between the interface. The other important interaction is helix II of EntB with the hairpin turn (D476-G471) of the EntE C-terminal domain between two long β-strands, which is stabilized by hydrophobic and ionic interactions. The structure permitted mutational studies to increase aminoacylation activity of the non-homologous acinetobactin A domain BasE for the PCP EntB. The mutant BasE, containing mutated residues corresponding to R494, E500, and R506 of the EntE C-terminal region (Fig. 8), had increased aminoacylation activity towards the PCP EntB compared to the WT BasE.

Fig. 8.

Crystal structure of EntB PCP and EntE A domain (PDB ID: 3RG2) tightly bound by a mechanistic inihibitor. The dynamic C-terminal region of the A domain EntE is colored in dark yellow. The N-terminal region of the A domain EntE is colored in light yellow. The four helical bundle of EntB PCP is colored in salmon. The covalently attached pantetheine and adenosine vinylsulfonamide inhibitor are in black. EntB residues D570 and D557 interact with EntE residues R494, E500, and R506. A mutant BasE A domain, containing residues corresponding to R494, E500, and R506 of EntE, had increased aminoacylation activity towards the PCP EntB compared to the WT BasE.

Salicylate activation in yersiniabactin biosynthesis is analogous to DHB activation, except the aryl-PCP is part of the N-terminus of the type I NRPS HMWP2.^100,103 The aryl-PCP has been excised and utilized in combinatorial studies with DHB A and C domains to yield important information on protein-protein interactions of the type II siderophore enzymes.^100,104 Based on genetic analysis of chlorothricin biosynthesis, methyl-salicylate is activated by ChlB3 and loaded onto the PCP ChlB2.⁷⁸ The methyl-salicylate is further chlorinated and methoxylated, but the modifications are unknown to occur while tethered to ChlB2.

3.2. Methyl hydroxyanthranilic acid activation

In actinomycin D biosynthesis, MHA is derived from a tryptophan metabolite.¹⁰⁵ The A domain ACMS I activates MHA to MHA-AMP and transfers the substrate to the PCP AcmACP.^106,107 AcmACP is a standalone enzyme unlike the DHB PCP/ICL bifunctional enzymes, although the actinomycin D A domain and PCP have high sequence similarities to the siderophore A domains and PCPs (Fig. 1).^6,107

3.3. Dihydroxybenzoic acid transfer to type I and II NRPS proteins

The DHB/MHA moieties are transferred to different type I NRPS synthetases in the siderophore pathways, giving rise to structural variation. In enterobactin biosynthesis, the condensation domain of the type I NRPS EntF catalyzes amide bond formation between DHB-tethered EntB PCP and ʟ-serine-tethered EntF PCP.¹⁰⁸ The EntF crystal structure revealed significant conformational differences in the condensation domain compared to other type I NRPSs¹⁰⁹ and most likely adopts a unique conformation when interacting with EntB. In bacillibactin biosynthesis, the condensation domain in the first module of DhbF catalyzes amide bond formation between DhbB-tethered DHB and the DhbF PCP-tethered glycine.⁹²

The vibriobactin pathway has a DHB branch point, where the DHB-tethered VibB PCP has two interacting condensation domains, VibH and VibF (Fig. 9A). VibH is a standalone type II condensation domain that transfers the PCP-tethered DHB to one of the primary amines on norspermidine (NSPD).¹¹⁰ Combinatorial and mutant studies with the excised aryl-PCP of HMWP2 revealed that loop 1 and helix II of the aryl-PCP is important for VibH recognition and may be a general recognition surface for the enzyme.¹⁰⁴ The VibH crystal structure was also determined and provides detailed information on the NSPD binding site.¹¹¹ Mutant analysis on VibH revealed that residues W264 and D130 are important for binding to DHB-S-EntB and NSPD, as the residues are located near the Ppant channel opening and the lining of the solvent channel, respectively (Fig. 9B). Based on this structural study, the VibH condensation mechanism is likely distinct from the analogous mechanisms of acyltransferases (AT), TycB, and VibF. The other VibB partner VibF is a type I NRPS that contains two condensation/cyclization domains, Cy1 and Cy2. Mutational studies have shown that Cy2 condenses DHB (tethered to VibB) to threonine (tethered to VibF PCP) followed by Cy1 cyclization of the intermediate to an oxazoline ring.¹¹² Kinetic analysis revealed that VibF can transfer the dihydroxyphenyl-(DHP)-oxazoline-carboxylate onto the primary amine of NSPD, but VibF is five-fold more efficient at transfer to the primary amine of DHB-N₁-NSPD, indicating that VibF acts after VibH. Finally, VibF transfers another DHB-oxazoline to the secondary amine to generate vibriobactin (Fig. 9A).⁹⁶ An oxazoline ring is also present in acinetobactin and analysis of the gene cluster suggests the Cy1 and Cy2 domains of BasD are responsible for the ring formation.⁹⁷ Oxazoline biosynthesis is analogous to thiazoline biosynthesis in anguibactin, where the Cy2 and Cy1 homologs of AngN catalyze peptide bond formation between DHB (attached to AngB) and cysteine (attached to AngM), followed by cyclization of the intermediate to a thiazoline ring.¹¹³

Fig. 9.

The bifurcated transfer of dihydroxy benzoic acid (DHB) in vibriobactin biosynthesis. (A) DHB is transferred from VibB by both VibF and VibH. VibH transfers DHB to norspermidine (NSPD) and VibF Cy2 transfers DHB to the PCP-tethered threonine. After cyclization, the VibF oxazoline intermediate is transferred to the other primary amine and then the secondary amine. (B) The crystal structure of VibH (PDB ID: 1L5A). Mutational studies revealed that residues D130 and W264 are important for recognition of DHB-S-EntB and condensation of DHB and NSPD.

3.4. Methyl hydroxyanthranilic acid transfer to type I NRPS

In the actinomycin pathway, the condensation domain of ACMS II transfers MHA from AcmACP to threonine tethered to the PCP on module I of ACMS II.^114,115 The condensation domain of ACMS II does not contain a cyclization domain and therefore, the threonine is not converted into an oxazoline ring. The type I NRPS also has a high sequence similarity to the downstream siderophore type I NRPSs.⁹² It is important to note there are no studies that show DHB incorporation into downstream NRPS domains without the corresponding DHB PCP. The interaction between the DHB PCPs and downstream condensation domains likely control substrate assimilation. Structural studies on the type II DHB PCP interaction with the condensation domains will be important for combinatorial biosynthesis with alternative pathways.

4. Halogenated aliphatic amino acids

Many natural products contain halogens that contribute to their biological properties,^116,117 and several type II NRPS systems are responsible for chlorination of aliphatic amino acids (Fig. 10A). Barbamide is a marine toxin originally isolated from the Caribbean variety of Lyngbya majuscule.¹¹⁸ The ichthyotoxin contains a tri-chlorinated methyl group that originates from ʟ- leucine.¹¹⁹ DNA sequence analysis of the barbamide gene cluster revealed 12 open reading frames and four genes (barA, barB1, barB2, barD) responsible for production of (4S)-5,5,5-trichloroleucine (Fig. 2C). The Pseudomonas syringae isolated phytotoxin syringomycin has an amphipathic lipopeptide structure that allows for insertion into lipid bilayers of plant membrane to form pores that are freely permeable to cations. Syringomycin contains a terminal threonine that is mono-chlorinated at the methyl carbon. Several genes analogous to the barbamide and syringomycin type II clusters were discovered in the biosynthetic gene cluster of cytotrienin, an antitumor drug produced by Streptomyces sp. RK-95–74.¹²⁰

Fig. 10.

(A) Structures of natural products containing or associated with chloromethylated amino acids derived from type II NRPS proteins. (B) The activation and chlorination of l-leucine in the biosynthesis of barbamide.

4.1. Aliphatic amino acid activation

In barbamide biosynthesis, an ʟ-leucine is incorporated into the barbamide pathway by the A domain BarD, which activates ʟ-leucine to ʟ-leucyl-AMP and facilitates transfer of the substrate to the type II PCP BarA (Fig. 10B).¹²¹ BarD is also capable of loading (4S)-5,5,5-trichloroleucine onto BarA, but the downstream A domain from the type I NRPS BarE has significantly more activity towards the non-halogenated amino acid. The syringomycin PCP and A domain form the didomain SyrB1, in which the A domain two orders of magnitude higher activity towards ʟ-threonine adenylation compared to 4-Cl-ʟ-threonine indicating chlorination occurs after ʟ-threonine attachment to SyrB1.^122,123 The A domain CytC1 from cytotrienin biosynthesis most likely activates either ʟ-valine or ʟ−2-aminobutyric acid based on ATP-³²PP_i exchange assays and transfers the substrate to the pantetheine arm of the PCP CytC2.¹²⁴ Although, it is not understood how the amino acid is incorporated into cytotrienin.

4.2. Non-heme Fe (II), α-ketoglutarate halogenase

In all the methyl chlorinated type II systems, an α-ketoglutarate (KG), non-heme Fe (II) halogenase is responsible for halogenation of the PCP-tethered amino acid.^{121,123–125} The type II α-KG/Fe(II) halogenases require the PCP for amino acid chlorination, as the enzyme cannot chlorinate the non-tethered substrate.¹²³ The barbamide pathway contains two αKG/Fe(II) halogenases BarB1 and BarB2 that perform tri-chlorination of the pro-R methyl group of leucine while tethered to BarA (Fig. 10B).^125,126 In vitro assays revealed that BarB2 introduces the first two chlorine atoms, while BarB1 adds the final chlorine. The syringomycin halogenase SyrB2 performs a mono-chlorination on the methyl carbon of threonine bound to SyrB1, while the cytotrienin halogenase CytC3 performs di-chlorination on the methyl group of either ʟ-valine or ʟ−2-aminobutyric acid bound to CytC2.¹²⁴ The structures of the halogenases SyrB2 (syringomycin) and CytC3 (cytotrienin) have been determined and reveal detailed information on the chlorination mechanism (Fig. 11A).^127,128 The αKG/Fe(II) dependent halogenases contain a common cupin fold, which consists of antiparallel β-strands in a jelly roll motif.¹²⁹ The iron cofactor is centered in the jelly roll and coordinated by two histidine residues. The coordination resembles the 2-His-1 carboxylate (Asp/Glu) triad seen in αKG/Fe(II) hydroxylases,¹³⁰ but the carboxylate is replaced by an alanine in the halogenases providing space for a chlorine atom to interact directly with the coordinated iron.^127,128 Molecular oxygen and αKG also coordinate with Fe (II) and cooperate to generate the Fe (IV)-oxo species responsible for hydrogen abstraction of the amino acid substrate (Fig. 11B).¹³¹ The substrate radical is then proposed to abstract the Fe coordinated chlorine atom. Because of the similarities to Fe hydroxylase, the Fe halogenase SyrB2 has been analyzed for hydroxylase activity.^132,133 Hydroxylation activity is observed when the PCP-tethered threonine is replaced with ʟ−2-aminobutyric acid and ʟ-norvaline indicating the importance of the aliphatic carbon chain placement in the active site. Computational simulations on the SyrB2 active site revealed that deeper substrate position favors chlorination and shallower position favors hydroxylation.¹³³

Fig. 11.

Fe, α-KG dependent chlorination of amino acid methyl groups. (A) Crystal structures of the Fe, α-KG dependent halogenases SyrB2 (PDB ID: 2FCT) and CytC3 (PDB ID: 3GJB). (B) Proposed chlorination mechanism of PCP-tethered amino acid methyl group.

4.3. Transfer of chloromethyl products

The transfer of the chloromethyl amino acids to downstream pathways is intriguing and ambiguous. In barbamide biosynthesis, the chlorinated substrate is released from BarA by the type II thioesterase BarC.¹²¹ The transfer mechanism of the precursor to the downstream type I NRPS BarE is uncertain. However, it is proposed that the trichloroleucine is converted into α-ketotrichloroisocaproic acid and transferred to the BarE PCP by the BarE A domain (Fig. 12A). ATP/PPi exchange activity indicates that the BarE A domain is highly specific towards α-ketoisocaproic and α-ketotrichloroisocaproic acid.¹²¹ The generation of α-ketotrichloroisocaproic acid is proposed to be catalyzed by BarH (a predicted amidohydrolase) and/or BarJ (a predicted aminooxidase) (Fig. 12A),¹³⁴ although studies have not confirmed the function of the proteins and whether the transformation occurs on the PCP BarA. In syringomycin biosynthesis, the 4-Cl-ʟ-threonine is transferred from SyrB1 to the module 9 PCP of SyrE by the aminoacyltransferase SyrC (Fig. 12B).¹³⁵ The threonine intermediate is covalently attached to Cys224 of SyrC and transferred directly to the holo-PCP, as the terminal module lacks an A domain. SyrC is also capable of recognizing non-native PCPs and transferring alternative amino acids to the terminal module of SyrE.¹³⁵ The transferase may be useful in combinatorial biosynthesis due its unique transfer to the terminal module and promiscuity towards various analogs. In cytotrienin biosynthesis, the thioesterase CytC4 cleaves the chlorinated product from CytC2, but where the substrate is incorporated into cytotrienin is unclear. The cyclopropane ring in cytotrienin is a potential location for the intermediate, as the CytC enzymes are analogous to the type II enzymes involved in cryptic chlorination and cyclopropanation in the coronatine and kutzneride pathways.^136–138 However, isotope feeding experiments have shown that the cyclopropane ring of cytotrienin is derived from methionine.¹³⁹ Currently, no cyclopropane gene candidate has been identified in the Streptomyces species.

Fig. 12.

Transfer of chlorinated type II products in barbamide (A) and syringomycin E (B) biosynthesis. (A) The PCP-tethered trichloroleucine is proposed to be converted to α-ketotrichloroisocaproic by BarJ (predicted amidohydrolase), followed by release of the product via the thioesterase BarC and incorporation into the type I NRPS BarE. (B) The 4-Cl-ʟ-threonine is transferred to the terminal module of SyrE by the aminoacyltransferase SyrC.

5. Cyclopropanes

Cyclopropane-containing natural products derived from type II NRPS pathways include coronatine and the kutznerides (Fig. 13A). Coronatine is produced by Pseudomonas syringae pv. sp. as a phytotoxin and virulence factor that causes chlorosis, stunts root growth, and promotes ethylene production.^140–143 Recent work has highlighted the potential benefits of cornatine in promoting crop drought tolerance,^144,145 inducing production of interesting phytophenol and sterol metabolites,¹⁴⁶ and as an anticancer agent.¹⁴⁷ Coronatine contains coronamic acid, a cyclopropane amino acid biosynthesized by CmaA–E, a cluster of type II NRPS enzymes (Fig. 2D).¹⁴⁸ The kutznerides are antimicrobial and antifungal agents produced by Kutneria sp. 744.¹⁴⁹ Structurally, the kutznerides are cyclic hexadepsipeptides derived from several nonproteinogenic amino acids and an alpha hydroxy acid. Interestingly, the kutznerides contain both threo- and erythro- isomers of β-hydroxy glutamic acid, the incorporation of which will be discussed in sections 6.1 and 6.2. Sequence analysis of the kutzneride biosynthetic gene cluster revealed KtzA-D (Fig. 2D), which are responsible for cyclopropane formation.¹⁵⁰

Fig. 13.

(A) Structures of select cyclopropane-containing non-ribosomal peptides derived from type II NRPS. Cyclopropane moiety is highlighted in red. (B) Biosynthesis of coronamic acid and coronatine. A=adenylation domain; PCP=peptidyl carrier protein.

5.1. Aliphatic amino acid activation

During coronamic acid biosynthesis, cyclopropane formation commences with CmaA, an A-PCP didomain that activates l-allo-isoleucine to l-allo-isoleucyl-AMP and transfers the amino acid onto the phosphopantetheine arm of the PCP to form covalent adduct l-allo-isoleucyl-S-CmaD (Fig. 13B).¹⁴⁸ Adenylation and subsequent enzymatic transformations also occurs with l-valine, as evidenced from in vitro studies and the isolation of trace quantities of norcoronamic acid, an l-valine derivative of coronamic acid.¹⁵¹ During the biosynthesis of kutznerides, A domain KtzB activates l-isoleucine to l-isoleucyl-AMP and transfers the amino acid onto the phosphopantetheine arm of stand-alone PCP KtzC. KtzB also adenylates other aliphatic amino acids (l-valine>l-leucine>l-allo-isoleucine.¹⁵⁰

5.2. Transacylation from A-PCP didomain to stand-alone PCP

During coronamic acid biosynthesis, subsequent processing to the final cyclopropane amino acid by downstream enzymes requires transfer from an A-PCP didomain to a standalone PCP domain. An unusual acyltransferase (AT) CmaE catalyzes aminoacyl transfer from CmaA to CmaD through transient aminoacylation on an active site cysteine (Fig. 13B).¹⁵² CmaE is a 32-kDa protein homologous to α/β hydrolases; it is 16% identical to aminoacyltransferase SyrC discussed in section 4.3. CmaD is a stand-alone PCP domain that ferries the substrate to subsequent enzyme active sites for conversion to coronamic acid. Aminoacyl transfer to CmaD is essential for biosynthesis, as subsequent enzymes cannot process CmaA-loaded substrates.¹⁵² The underlying reason why downstream enzymes recognize CmaD-loaded substrates over CmaA-loaded substrates is unknown and further investigations are warranted, as protein-protein interactions are likely critical for product formation.

5.3. Cryptic chlorination

The penultimate step to cyclopropane formation involves a cryptic chlorination strategy during coronamic acid biosythesis. Prior to sequencing the coronatine biosynthetic gene cluster, feeding studies suggested that the methyl group of l-allo-isoleucine was the source of the methylene bridge and a high oxidation state intermediate was presumed to be involved.^153,154 Later, studies confirmed that CmaB, a non-heme Fe²⁺, O₂ and α-ketoglutarate-dependent halogenase catalyzes regiospecific chlorination of C_γ of l-allo-Isoleucine (Fig. 13B).¹⁵² CmaB is homologous to halogenases SyrB2 (57% identity) and BarB2 (40% identity) from the syringomycin and barbamide pathways, respectively.^119,155 It is likely that the mechanism proceeds through radicals and involves the formation of a highly oxidized Fe⁴⁺-oxo intermediate, as discussed earlier (Fig. 11B). Importantly, CmaB does not halogenate free amino acids and requires that substrates are covalently tethered to CmaD, again illustrating the importance of protein recognition. Analogously, KtzD (74% similarity to SyrB2) chlorinates the carrier-protein bound substrate at C_γ during kutzneride biosynthesis.¹⁵⁶ Elucidating the critical recognition elements between PCP and halogenase will be necessary for future combinatorial biosynthetic efforts.

5.4. Cyclopropane ring closure

The last step of cyclopropane formation involves a deprotonation, followed by intramolecular halide displacement on PCP-tethered substrates. During coronamic acid biosynthesis, CmaC, a Zn²⁺-dependent enzyme deprotonates at C_α to form a stabilized thioester enolate intermediate followed by intramolecular chloride displacement to form the cyclopropane ring of coronamic acid (Fig. 13B).¹³⁷ Cyclopropane is the sole product of this reaction, with no azetidine formation. Further evidence for the proposed mechanism is provided by deuterium exchange experiments in which deuterium traps that carbanion. CmaC cyclizes numerous γ-chloro-aminoacyl-S-CmaD substrates, including chloro-allo-isoleucyl, chloro-valyl, 4-chloro-2-amino-butryl, and 4,4-dichloro-2-amino-butyryl, while free γ-chloro-amino acids and CmaD-bound δ-halo and ε-halo substrates are not tolerated. In addition, γ-halo substrates bound to a non-cognate PCP was recognized by CmaC, although yields of the resulting cyclopropane were reduced substantially, indicating the importance of protein-protein interactions in type II NRPS systems and the potential to engineer these pathways to produce novel peptide-based precursors. Similarly, during kutzneride biosynthesis, KtzA, a redox-inactive flavoprotein, catalyzes a dehydrochlorination to close the ring.¹⁵⁶

5.5. Transfer and release enzymes

To date, the precise biochemical and molecular details of how cyclopropane amino acids are incorporated into the final natural product have not been fully described. The coronamic acid gene cluster contains a putative thiosterase cmaT that most likely releases the CmaD-tethered substrate.¹⁵⁷ In vivo analysis revealed that the gene cfl (coronofacate ligase) is responsible for ligation of coronamic and coronafacic acid, a polyketide precursor¹⁵⁸ Coronofacate ligase resembles other adenylate forming enzymes and, therefore, likely adenylates coronafacic acid facilitating nucleophilic attack by the coronamic acid amine.¹⁵⁹ The kutzneride biosynthetic pathway contains KtzF, a putative standalone thioesterase that releases 2-(1-methylcycloprolyl)-glycine (MecPGly), where it is incorporated into downstream type I NRPS biosynthetic machinery.¹⁵⁰

6. β-hydroxylated amino acids

Many β-hydroxylated amino acid-containing natural products are from type II NRPS pathways and they possess a broad range of bioactivities. These include the kutznerides, zorbamycin, antiprotealide (and related salinosporamide analogues), echinomycin, novobiocin, balhimycin, and nikkomycins (Fig. 14).

Fig. 14.

Structures of select β-hydroxy amino acid-containing and β-hydroxy amino acid-derived non-ribosomal peptides from type II NRPS. b-hydroxy amino acid moiety or derivative is highlighted in red.

Zorbamycin is a glycopeptide antitumor agent derived from Streptomyces flavoviridis ATCC 21892 and belongs to the larger family of bleomycin natural products that cleave DNA at specific sequences.^160,161 Unlike bleomycin, which contains threonine, zorbamycin contains β-hydroxy-l-valine. Sequence analysis of the zorbamycin biosynthetic gene cluster revealed 40 ORFs, approximately half of which contain similarities to bleomycin or tallysomicin. The zorbamycin gene cluster contains three didomain NRPS modules and eight stand-alone NRPS domains (Fig. 2e), several of which are homologous to bleomycin gene clusters.¹⁶⁰

Salinosporamide A was isolated from Salinospora tropica and was found to have potent anticancer activity against the 20S proteasome.¹⁶² Structurally, salinosporamide A is derived from acetate, chlorobutyrate, and an unusual β-hydroxylated l-3-cyclohex-2’-enylalanine (CHA) moiety derived from the shikimate pathway.^163,164

Echinomycin (or quinomycin A) is derived from Streptomyces griseovariabilis subsp. bandugensis. and is a member of the quinomycin family of antibiotics, which display potent anticancer activity.^165–167 Echinomycin contains a quinoxaline core that intercalates with DNA to inhibit replication and transcription. Elucidation of the echinomycin biosynthetic gene cluster revealed several type II NRPS proteins involved in quinoxaline biosynthesis, including an A-PCP didomain (Qui18), a hydroxylase (Qui15), a thioesterase (Qui14), and a tryptophan 2,3-dioxygenase (TDO) (Qui17).

The Streptomyces spheroides derived antibiotic novobiocin contains a β-hydroxylated 3-amino-4-hydroxy-coumarin moiety (Fig. 14).^168,169 Novobiocin inhibits DNA gyrase, thus, blocking DNA replication. Several type II NRPS enzymes involved in aminocoumarin formation and incorporation into novobiocin have been characterized, including an A-PCP didomain (NovH), a cytochrome P450 monooxygenase (NovI), an oxidase (NovJ/NovK), and a standalone A domain (NovL).^168–170

Balhimycin is a glycopeptide antibiotic produced by Amycolatopsis balhimycina. The heptapeptide backbone of balhimycin is identical to vancomycin and chloroeremomycin, however, the halogenation and glycosylation patterns differ. Wohlleben and colleagues elucidated the biosynthetic gene cluster of the balhimycin peptide backbone and identified three multi-modular type I NRPS enzymes (BpsA–C) and a terminal stand-alone type II A-PCP didomain (BpsD) (Fig. 2e).

The nikkomycins are peptide nucleoside antibiotics produced by Streptomyces tendae Tü901.¹⁷¹ Nikkomycins are fungicides, insectides, and pesticides that inhibit chitin synthase.¹⁷² In particular, Nikkomycins I, X, and C_x contain a 4-formyl-4-imidazolin-2-one (imidazolone) base derived from l-histidine. Walsh and coworkers characterized three type II NRPS enzymes involved in imidazolone biosynthesis, including an A-PCP didomain (NikP1), a heme hydroxylase (NikQ), and a thioesterse (NikP2).¹⁷¹

6.1. Amino acid activation

During the biosynthesis of β-hydroxy amino acid containing natural products, both proteinogenic and non-proteinogenic amino acids undergo activation either by A-PCP didomains or stand-alone A domains. For example, during zorbamycin biosynthesis, the A-PCP didomain ZbmVIIb catalyzes formation of l-valyl-S-zmbVIIb from l-valine (Fig. 15A). Similarly, during novobiocin and nikkomycin production, the A-PCP didomains NovH and NikP1 selectively activate and load l-tyrosine and l-histidine onto the phosphopantetheine arm of their PCPs, respectively. The A-PCP didomain SalB in the salinosporamide A pathway activates an unusual amino acid l-3-cyclohex-2’-enylalanine (CHA), although the specificity conferring code predicts that it would activate and load aromatic amino acids. Moore and coworkers altered the substrate tolerance of SalB using PCR mutagenesis to produce antiprotealide, a salinosporamide A analogue in which the cyclohexyl group was replaced with an isopropyl group, providing an outlet for the engineering salinosporamide analogues.

Fig. 15.

Proposed biosynthesis of zorbamycin. (A) ZbmVIIc-catalyzed β-hydroxylation on L-valyl-ZbmVIIb occurs before ZbmVIId-mediated transfer onto C-PCP didomain ZbmVIIa. (B) ZbmVIIc-catalyzed β-hydroxylation after ZbmVIId-mediated transfer onto standalone PCP domain Zbm-orf32/33. Following β-hydroxylation, ZbmVIId shuttles β-hydroxy-L-valine from Zbm-Orf32/33 to ZbmVIIa. In both models, thioesterase Zbm-orf35 releases β-hydroxy-L-valine from the biosynthetic machinery. A=adenylation domain; PCP=peptidyl carrier protein; AT=acyltransferase; C=condensation domain

Notably, the quinoxaline core of echinomycin originates using an A-PCP didomain Qui18 which coexpresses and copurifies as a heterotetramer with Qui5, an MbtH-like protein. Association of Qui5 is postulated to be critical for solubility and proper protein folding after translation. Qui18 selectively activates and loads l-tryptophan onto the phosphopantetheine arm of the PCP; no other l- or d- aromatic amino acids are loaded.

Activation of amino acids by standalone A domains is less frequent during biosynthesis of β-hydroxy amino acid containing non-ribosomal peptides. Interestingly, the incorporation of β-hydroxy-l-glutamic acid in kutzneride biosynthesis involves a type II A-domain that activates and transfers the amino acid onto the third PCP domain of the KtzH module.^150,173 The A domain that precedes this PCP is truncated and cannot activate amino acids. In balhimycin biosynthesis, the specificity conferring code and sequence comparisons to homologous enzymes predicts that the A-PCP didomain BpsD activates and loads l-tyrosine; however, this has not been confirmed experimentally due to challenges in expressing soluble protein.¹⁷⁴ Importantly, inactivation of BpsD abrogates balhimycin production, which was restored upon supplementation with β-hydroxytyrosine, confirming the involvement of BpsD in the synthesis.¹⁷⁴

6.2. Transacylation from A-PCP didomain to downstream PCP

Select β-hydroxy amino acid biosynthetic pathways may require the intervention of AT domains to shuttle amino-acyl substrates from A-PCP didomains to downstream PCP domains for further processing. The zorbamycin gene cluster contains two putative proteins that fulfil this role: ZbmVIId and Zbm-orf35. The zorbamycin gene cluster also includes additional C-PCP didomains (ZbmVIIa) and stand-alone PCP domains (ZbmI, Zbm-orf32, Zbm-orf33). Given the presence of a putative AT and additional PCPs, either ZbmVIId or Zbm-orf35 may shuttle aminoacyl groups between various PCPs (Fig. 15B) similar to coronamic acid biosynthesis (Fig. 12B). Sequence alignment of ZbmVIId with other NRPS ATs/TEs revealed an active site cysteine residue, suggesting that ZbmVIId shuttles cargo between PCP domains. In contrast, Zbm-orf35 contains an active site serine residue, suggesting the protein is a canonical thioesterase that cyclizes or releases fully elucidated peptides from the biosynthetic machinery. Further studies are needed to pinpoint the precise timing of shuttling events and downstream modification during zorbamycin biosynthesis.

6.3. Stereospecific hydroxylation

The key step in the biosynthesis of β-hydroxy-amino acids is β-carbon hydroxylation of PCP-bound substrates. The Wohlleben group discovered a P450 monooxygenase OxyD downstream of BpsD in the balhimycin biosynthetic gene cluster.¹⁷⁵ Inactivation of OxyD abrogates balhimycin production, which was restored upon supplementation with β-hydroxytyrosine, confirming the involvement of OxyD in the biosynthesis of β-hydroxy-l-tyrosine. Later, the Cryle group confirmed that OxyD binds l-tyrosyl-BpsD and l-phenylalanyl-BpsD with similar affinity, but does not bind free amino acids.¹⁷⁶ Importantly, a 2.1 Å crystal structure of OxyD revealed information about conserved structural elements and residues that form the catalytic site and the binding site for the PCP-tethered substrate and PCP, respectively (Fig. 16).¹⁷⁶ Overall, OxyD consists of 14 α-helices (αA′–αL ), 3 β-sheets (β1–β3), and a heme cofactor. The overall conformation of the catalytic site is influenced by electrostatic and hydrogen bonding interactions between the various α-helices. In the catalytic site, the heme cofactor is anchored through electrostatic interactions and a conserved glutamate/threonine pair in αI activates oxygen. Sequence alignment of OxyD with other P450 monooxygenases that recognize amino-acyl-PCPs as substrates (including ZbmVIIc, Nov I, and NikQ-discussed within this review) revealed common structural motifs that may be important for PCP recognition. These regions include the B-B₂ loop, helices αF, αG, and αI, and beta sheet β1 (Fig. 16). This information provides predictive tools to assign putative function to uncharacterized P450s based on sequence and offers a tractable entry for future combinatorial and metabolic bioengineering efforts.

Fig. 16.

Structure of the P450 β-hydroxylase OxyD (PDB ID: 3MGX) from balhimycin biosynthesis. The forward facing interface is the predicted binding region for the BpsD PCP. Red highlights are conserved P450 β-hydroxylase regions that may be important for BpsD PCP binding.

During salinosporamide biosynthesis, SalD, a cytochrome P450 oxidase hydroxylates the β position of CHA. SalD appears to have broad substrate tolerance, evidenced by feeding studies of non-proteinogenic amino acids 3-cyclohexylalanine and 3-cyclopentylalanine to afford salinosporamide X1 and salinosporamide X2 (Fig. 14).¹⁶³ Hydroxylase Qui15, from echinomycin biosynthesis, is a heme-dependent cytochrome P450 that hydroxlates Qui18-bound l-tryptophan at the β-position during quinomycin biosynthesis. Novobiocin (Fig. 14) biosynthesis contains cytochrome P450 monooxygenase NovI to stereospecifically hydroxylate NovH-bound l-tyrosine at the β-position via a 2-electron oxidation. In nikkomycin biosynthesis, the heme hydroxylase NikQ stereospecifically hydroxylates NikP1-bound l-histidine at the β-position, presumably via a 2-electron oxidation; free histidine is not hydroxylated by NikQ.¹⁷¹ In addition, apo NikP1 was observed to bind well to NikQ relative to NikP1-bound L-histidine, which suggests that the PCP surface, not the substrate identity, is the most important factor in PCP-partner protein binding.¹⁷⁷

Currently, there are two hypotheses regarding the hydroxylation of valine during zorbamycin biosynthesis (Fig. 15). One model suggests that hydroxylase ZbmVIIc stereospecifically hydroxylates ZbmVIIc-bound l-valine directly (Fig. 15A). Next, β-hydroxyl-L-valine is transferred onto C-PCP didomain ZbmVIIa via transient acylation of AT ZbmVIId. A second model proposes that l-valine is first offloaded from ZbmVIIb onto the phosphopantetheine arm of a downstream standalone PCP (Zbm-orf32 or Zbm-orf33) via transient acylation of AT ZbmVIId, prior to β-hydroxylation (Fig. 15B). This is followed by ZbmVIIc-mediated hydroxylation, and ZbmVIId-mediated transfer onto ZbmVIIa. The precise order of events are unknown and further studies are warranted to understand the biosynthetic route.

The hydroxylation of PCP-bound l-glutamic acid during kutzneride biosynthesis requires two O₂ and α-ketoglutarate dependent oxygenases: KtzO and KtzP, which afford the threo- and erythro- stereoisomers of β-hydroxy-l-glutamate, respectively.¹⁷³ No hydroxylation of PCP-bound d-glutamic acid or unbound amino acids occurs. However, KtzO/P catalyzes hydroxylation of l-glutamic acid bound to noncognate PCPs with similar catalytic efficiency.¹⁷³

6.4. Release or modifications of β-hydroxy amino acids

Thioesterases (TEs) directly off-load the β-hydroxy amino acid from the synthetic machinery, where they are directly incorporated into the natural product or further modified by additional enzymes. Several of these enzymes have not been characterized biochemically and their roles are deduced based on sequence comparisons to known TEs. For example, the biosyntheses of zorbamycin and quinomycin involve putative TEs Zbm-orf35 and Qui14, respectively, which presumably release PCP-bound β-hydroxy amino acids.^160,165 The Zbm-orf35 released β-hydroxy-l-valine is not modified further, while the Qui14 offloaded β-hydroxy-l-tryptophan undergoes additional modification. Free β-hydroxy-l-tryptophan is further modified by tryptophan 2,3-dioxygenase (TDO) Qui17, which catalyzes an oxidative ring cleavage of indole in tryptophan to afford N-formyl-β-hydroxykynurenine, confirmed by LC-MS experiments on free (2S, 3S)-β-hydroxy-tryptophan.¹⁶⁵ It might also be possible that the oxidative cleavage occurs on Qui18-bound β-hydroxy l-tryptophan, though this has not been determined experimentally.

The roles of several TEs that release PCP-bound β-hydroxy amino acids have been confirmed experimentally, including NiKP2 and perhydrolase Bhp during the biosynthesis of nikkomycins and balhimycin, respectively.^171,178 NikP2 also releases l-histidine bound NikP1, though there is a two-fold preference for the β-hydroxy substrate. Following cleavage, β-hydroxy-l-histidine is modified downstream to the imidazolone core found in nikkomycins. Similarly, during balhimycin biosynthesis, the released β-hydroxy-l-tyrosine is modified by halogenase Bha, which catalyzes meta-chlorination on the aromatic ring. The role of Bha is consistent with halogenase VhaA from vancomycin biosynthesis, which catalyzes aromatic chlorination on both β-hydroxy-l-tyrosines on the VpsC PCP hexapeptide-bound intermediate.¹⁷⁹

A few β-hydroxy-amino acids undergo additional modifications while bound to the PCP. During novobiocin biosynthesis,NovJ/NovK, an NADP⁺-dependent heterotetramer that catalyzes benzylic oxidation of NovH-bound β-hydroxy-l-tyrosine to β-keto-L-tyrosine via a 2-electron mechanism.¹⁷⁰ Subsequently, an unidentified monooxygenase installs a hydroxy substituent on the ortho position of the aromatic ring. This is followed by ring closure and concomitant release of the aminocoumarin core from the NovH A-PCP didomain.¹⁸⁰ Likewise, during the biosynthesis of salinosporamides, it has been proposed that β-hydroxy-CHA-S-SalB is condensed with a PKS-derived precursor to afford a linear SalB-bound intermediate that is cyclized to the final product.^164,181

7. Substituted aromatic β-amino acids

Several aromatic-substituted β-amino acids are derived from type II NRPS pathways and are present in C-1027, maduropeptin, and kedarcidin (Fig. 17A). All of these natural products are chromoprotein antitumor antibiotics that contain a conjugated enediyne core. The core undergoes Bergmann cyclization to a benzenoid diradical intermediate that abtracts protons from DNA to introduce double-stranded breaks and interstrand crosslinks.¹⁸²

Fig. 17.

Substituted aromatic products derived from type II NRPS proteins. (A) Structures of select aromatic-containing non-ribosomal peptides derived from type II NRPS. The aromatic moiety is highlighted in red. (B) Proposed mechanism for the transformation of L-tyrosine to β-tyrosine by aminomutase SgcC4; MIO=4-methylideneimidazole-5-one. The pymol views are of SgcC4 crystal structures with cinnamate epoxide and MIO ligands bound (mimics step 2 of mechanism; PDB ID: 2RJR), and with α,α-difluoro-β-tyrosine inhibitor and MIO ligands bound (mimics step 4 of mechanism; PDB ID: 2RJS). (D) Biosynthesis of (S)-3-chloro-4,5-dihydydroxy-β-phenylalanine from the C-1027 chromophore; A=adenylation domain; PCP=peptidyl carrier protein.

C-1027 was isolated from Streptomyces globisporous and the biosynthetic gene cluster was cloned and sequenced by Shen and coworkers.^182,183 The C-1027 chromophore gene cluster consists of 67 open reading frames that code for four convergent biosynthetic building blocks including the enediyne core encoded by a polyketide synthase (PKS), a deoxy aminosugar, a benzoxazolinate, and an (S)-3-chloro-4,5-dihydroxy-β-phenylalanine amino acid encoded by type II NRPS proteins.¹⁸² Substitutions at the 3 and 5 positions of the aromatic ring of β-phenylalanine are critical for bioactivity, as deschloro and deshydroxy congeners of C-1027 demonstrate decreased potency in interstrand crosslinked assays in vitro.^184,185 The biosynthesis of (S)-3-chloro-4,5-dihydroxy-β-phenylalanine involves five NRPS genes, including: SgcC4, SgcC1, SgcC2, SgcC3, and SgcC (Fig. 2).

Maduropeptin was isolated from Actinomadura madurae ATCC 39114 and later cloned and sequenced by Shen and colleagues (Fig. 17A).¹⁸⁶ The maduropeptin chromophore gene cluster consists of 42 open reading frames that code for four convergent biosynthetic building blocks including an enediyne core encoded by a PKS, a deoxy aminosugar, a 3,6-dimethylsalicyate, and an (S)-3-(2-chloro-3-hydroxy-4-methoxy-phenyl)-3-hydroxypropionic acid encoded by a type II NRPS.¹⁸⁶ The biosynthesis of (S)-3-(2-chloro-3-hydroxy-4-methoxy-phenyl)-3-hydroxypropionic acid is encoded by eight genes: mdpC, mdpC1–4, and mdpC6–8. Notably, mdpC4, mdpC1, mdpC2, mdpC3, and mdpC are highly similar (>50%) to genes involved in the biosynthesis of (S)-3-chloro-4,5-dihydroxy-β-phenylalanine from the C-1027 pathway.

Kedarcidin was isolated, cloned, and sequenced from Streptoallteichus sp. ATCC 53650 (Fig. 17A).¹⁸⁷ The kedarcidin chromophore gene cluster consists of 117 open reading frames that code for five convergent biosynthetic building blocks including an enediyne core, two deoxy sugars, a naphthonate moiety, and an unusual (R)-2-aza-3-chloro-β-tyrosine, encoded by type II NRPS proteins (Fig. 2F).¹⁸⁷ The biosynthesis of (R)-2-aza-3-chloro-β-tyrosine is encoded by six genes: kedY and kedY1–5. Notably, kedY4, kedY1, kedY2, kedY3, and kedY are highly similar (>40%) to homologues involved in biosynthesis of β-amino and β-hydroxy acids in the C-1027 and maduropeptin pathways, respectively.

7.1. Conversion of α-amino acids to β-amino acids

The enzymes SgcC4 and Mdp4C4 (82% similarity) are tyrosine aminomutases that convert l-tyrosine to (S)-β-tyrosine during C-1027 and maduropeptin biosynthesis, respectively.^{186,188–190} The mutases require a 4-methylideneimidazole-5-one (MIO) cofactor.¹⁸⁸ The proposed mechanism involves 1,4-conjugate addition of the amine into the cofactor and deprotonation by Y63 at the β- position to eliminate isolable p-hydroxycinnamate and an MIO amine adduct (Fig. 17B). The active site architecture excludes water and enables another 1,4-conjugate addition of the MIO amine to hydroxycinnamate, and subsequent elimination regenerates MIO and affords the β-amino acid. Based on sequence homology, KedY4 (82% similarity to SgcC4) is a putative aminomutase that converts 2-aza-l-tyrosine to 2-aza-(S)-β-tyrosine during kedarcidin biosynthesis.¹⁸⁷

7.2. β-amino acid activation

The A domains involved in C-1027, maduropeptin, and kedarcidin biosynthesis specifically adenylate β-amino acids and not α-amino acids. SgcC1 activates and loads both (R)-β-tyrosine and (S)-β-tyrosine onto PCP SgcC2 (Fig. 17C), with a slight preference for the S-enantiomer.^188,191 SgcC1 does not adenylate aromatic α-amino acids, and substitutions at the meta positions on the aromatic ring of the β-amino acid are modestly tolerated. In vitro studies have shown that MdpC1 (59% similarity to SgcC1) activates and loads (S)-β-tyrosine onto PCP MdpC2 (52% similarity to SgcC2) and, like SgcC1, other aromatic β-amino acids are modestly tolerated.¹⁸⁶ Based on genetic analysis, KedY1 (44% similarity to SgcC1) is believed to activate 2-aza-(S)-β-tyrosine and load PCP KedY2 (67% similarity to SgcC2).¹⁸⁷

The finding that SgcC1, MdpC1, and KedY1 preferentially adenylates a β-amino acid over an α-amino acid contradicts the amino acid-specificity conferring code of A domains developed by Stachelhaus.¹⁹¹ Sequence analysis between SgcC1 and α-amino acid specific A domains, revealed that SgcC1 contains a proline adjacent to the conserved aspartate that is normally involved in a key electrostatic interaction with the α-amine. The proline was conserved in other A domains that activate unusual amino acids that do not contain an amine at the alpha-position. Taken together, these findings suggested that the proline likely induces conformational protein changes that stabilize the electrostatic interactions between aspartate and the β-amine. To confirm the importance of this proline, a P→A mutant of SgcC1 was cloned and expressed and found to be ~140-fold less efficient for adenylation of (S)-β-tyrosine. Adenylation activity was completely abrogated for (R)-β-tyrosine and meta-substituted β-tyrosine. Substitutions on the aromatic ring of the β-amino acid occur downstream while tethered to the PCP.

7.3. FADH₂, O₂-dependent halogenation

Halogen aromatic substitution occurs first in the enediyne compounds. In C-1027 biosynthesis, the FADH₂ and O₂-dependent halogenase SgcC3 chlorinates the 3-position on the aromatic ring (Fig. 17C).¹⁹² There is a preference for chlorination over bromination, while fluorination and iodination do not occur. In comparison, halogenase MdpC3 (65% similarity to SgcC3) chlorinates the 2 position on the aromatic ring during maduropeptin biosynthesis.¹⁸⁶ Based on sequence homology, halogenase KedY3 (81% similarity to SgcC3) putatively installs a chlorine at the 3 position on the aromatic ring during KED biosynthesis. The presence of FADH₂, O₂, and a PCP-tethered substrate is a common paradigm with halogenases, as evidenced from the halogenases from pyrrole biosynthesis. The mechanism is likely similar to PltA-catalyzed halogenation of pyrrolyl-PltL discussed previously (section 2.2, Fig. 5).

7.4. FADH₂, O₂-dependent monooxygenase

After halogenation, SgcC, an FADH₂- and O₂-dependent monooxygenase, hydroxylates the 5-position of SgcC2-bound 3-chloro-(S)-β-tyrosine (Fig. 17C).¹⁹³ Early genetic studies revealed that inactivation of sgcC resulted in the production of 22-deshydroxy-C-1027.¹⁸² Interestingly, hydroxylation of 3-bromo- and 3-iodo-(S)-β-tyrosyl-S-SgcC2 occurs more readily than 3-chloro-, while hydroxylation of 3-fluoro- is disfavoured, likely due to the electron withdrawing nature of these substituents.¹⁹³ The 2.6 Å X-ray crystal structure of SgcC (Fig. 18) revealed that this enzyme contained an acyl-CoA dehydrogenase (ACAD) fold,¹⁹⁴ reminiscent of that observed in prolyl-ACP dehydrogenase AnaB (Fig. 4). This structure is one of only a handful of structurally elucidated ACAD-fold containing enzymes that recognizes carrier-protein bound substrates. Like most ACADs, SgcC is tetrametric and comparison to TtHpaB, a 4-hydroxyphenylacetate-3-monoxygenase from Thermus thermophilis HB8, reveals that the enzyme likely undergoes conformational changes to accommodate both the FAD cofactor and the β-tyrosinyl substrate. Moreover, docking analysis between SgcC and SgcC2 reveals that the interface is composed of extensive hydrogen bonds and electrostatic interactions that span subunits A/B and C/D (Fig. 18). Notably, the site of phosphopantetheinylation on SgcC2 (Ser31) is oriented towards the active site of SgcC. Additional studies aimed at elucidating the interface of SgcC-SgcC2 will be essential for future combinatorial engineering efforts in this area.

Fig. 18.

Structure of the monoxygenase SgcC (PDB ID: 4OO2) from C-1027 biosynthesis. The enzyme is a homotetramer. Residues highlighted red (A: R170, K220; B: G332, D335, R337) are proposed to interact with the PCP partner SgcC2.

The sequences of several enediyne gene clusters overlap with SgcC, including those from the maduropeptin and kedarcidin biosynthetic pathways. For instance, MdpC (81% similarity to SgcC), a putative monooxygenase, installs a hydroxide at the 3 position on the aromatic ring, though the activity has not been characterized in vitro.¹⁸⁶ KedY (70% similarity to SgcC) is also a putative monooxygenase in the kedarcidin gene cluster that is proposed to hydroxylate KedY2-tethered 2-aza-l-phenylalanine to form (R)-2-aza-3-chloro-β-tyrosine.¹⁸⁷ KedY likely installs a hydroxide at the 4-position of 2-aza-l-phenylalanine, although this enzyme has not been characterized biochemically.

7.5. Additional tailoring enzymes

During maduropeptin biosynthesis, additional modifications occur on the MdpC2-tethered substrates to generate the final β-hydroxy moiety. While many of these tailoring enzymes have not been characterized in vitro, their putative functions are assigned based on sequence homology. MdpC7 is a putative PLP-dependent transaminase that eliminates the β-amine to form β-keto-tyrosinyl-MdpC2. Later, MdpC8 stereospecifically reduces the β-ketone to form the penultimate β-hydroxy acid. Methyltransferase MdpC6 methylates the phenol at the para position. It is unknown if this sequence of events occurs before chlorination and oxidation of the aromatic ring. The different regiochemical outcomes in the oxidation and halogenation of the aromatic rings in C-1027 and maduropeptin may provide outlets for engineering these two pathways to develop new enediyne analogues.

7.6. Transfer and release enzymes

To complete C-1027 biosynthesis, SgcC5 is a type II condensation enzyme that catalyzes ester bond formation between SgcC2-bound β-amino tyrosine and the enediyne core, as evidenced by studies with simplified substrate analogues.¹⁹⁵ SgcC5 prefers 3S enantiomers of SgcC2-bound β-amino acid and R enantiomers of the enediyne analogue. SgcC5 is promiscuous and catalyzes condensation with numerous tyrosyl-SgcC2 analogues. Interestingly, SgcC5 also catalyzes amide bond formation.¹⁹⁵ Recently, the crystal structure of SgcC5 revealed that the active site is located at a conserved hydrophobic cavity, providing additional insights into the mechanism of bond formation.¹⁹⁶ To initiate condensation, SgcC5 utilizes a conserved active site histidine as a base to deprotonate the amino group of the acceptor substrate. Mutagenesis of the histidine complete abolishes SgcC5 activity with (S)-β-tyrosinyl-SgcC2. Docking of SgcC5 with an SgcC2 homology model revealed a putative binding interface with a conserved salt bridge that may impart partner protein specificity. Based on sequence homology to SgcC5, MdpC5 (58% similarity) and KedY5 (55% similarity) are predicted to catalyze amide bond formation during maduropeptin and kedarcidin biosynthesis, respectively.^186,187

8. Miscellaneous systems

8.1. Structural characterization of PCPs with unknown pathway function

As mentioned earlier, BlmI from bleomycin biosynthesis was the first named type II NRPS PCP (Fig. 19A).⁷ Unlike type I PCPs, holo-BlmI is aminoacylated in trans by an excised type I A domain from Streptomyces verticillis that was foreign to the bleomycin biosynthetic pathway. Since this seminal finding, several type II PCPs have been discovered and biochemically characterized. Our phylogenetic analysis reveals that several type II PCPs, including BlmI, clade together (Fig. 1), consistent with Shen’s analysis.¹⁹⁷ Fourteen years after initial biochemical studies, an X-ray crystal structure of BlmI was solved.¹⁹⁷ The structure revealed a four alpha-helical bundle in an A/H (apo/holo) conformational state, similar to most type I PCPs and type II acyl carrier proteins (ACPs).

Fig. 19.

Structures of type II PCPs and A domains with unknown biosynthetic roles. (A) BlmI (PDB ID: 4I4D) is located in the bleomycin gene cluster. (B) A3404 (PDB ID: 4HKG) is derived from Acinetobacter baumannii. (C) The A-PCP didomain protein PA1221 (PDB ID: 4DG9) is derived from Pseudomonas aeruginosa. The PCP and A domain are mechanistically crosslinked by an adenosine vinylsulfonamide inihibtor (carbons colored black). (D) The chemical structures of congocidine and leinamycin, with the key structural motifs installed by type II NRPS proteins highlighted in red.

Shortly after structural characterization of BlmI, the structure of A3404, a type II PCP from Acinetobacter baumannii further illuminated slight structural differences between type I and type II PCPs (Fig. 19B).¹⁹⁸ A3404 has a conserved XGLDSX signature motif that identifies the phosphopantetheinylation site. In comparison to most type I ACPs and PCPs, the loop regions between helices I-II and helices II-III of A3404 contain acidic residues that are critical for interactions with partner proteins. Type II PCPs A3404, BlmI, SgcC2, and MdpC2 contain between five and six acidic residues between helices II and III, whereas type I ACPs and PCPs contain 2 acidic residues. Consequently, the array of acidic residues after helix II in A3404 impacts the orientation of helix III, which is positioned approximately 90° to helices II and IV.

The crystal structure of the PA1221 was one of the first PCP•A didomain structures determined (Fig. 19C).¹⁹⁹ The PA1221 gene was discovered in a biosynthetic operon of Pseudomonas aeruginosa. Although the product of the gene cluster is unknown, the PA1221 structure provided valuable information on the recognition surface between the PCP and A domain. The interface consists of features similar to features of the dehydrogenated proline and enterobactin (Fig. 8) PCP•A domain interface.^50,101

The biochemical characterization and crystallization of PCPs BlmI and A3404 as bona fide type II NRPS proteins enable us to elaborate upon Shen’s original criteria for type II PCPs: 1. The amino acid sequence of type II PCPs is similar to type I PCPs, especially with regard to the conserved LGGCS phosphopantetheinylation motif located at the N- terminus of helix II; 2. Like type I PCPs, type II PCPs are phosphopantetheinylated at the N-terminus of helix II; 3. Type II PCPs contain large clusters of acidic residues between helices I-II and helices II-III, which probably facilitate interactions with partner proteins, though this is not yet fully established, and additional type II structures will shed further light on the differences between these systems.

8.2. Congocidine biosynthesis

Congocidine (or netropsin) is a dipyrroloamide isolated from Streptomyces ambofaciens and Streptomyces netropsis (Fig. 19D).²⁰⁰ Congocidine is a promising anticancer agent that selectively binds A-T base pairs in the minor groove of DNA. There are several type II NRPS proteins implicated in the biosynthesis, including PCP domain Cgc19, C domains Cgc2 and Cgc16, and A domain Cgc3* (or Cgc22). Detailed biochemical studies revealed that Cgc3* activates 4-acetamidopyrrole-2-carboxylic acid as an AMP intermediate. This is followed by transfer onto the phosphopantetheine arm of Cgc19, and deacetylation to 4-amino-2-pyrrolyl-S-Cgc19. Cgc16 catalyzes amide bond formation between 4-amino-2-pyrrolyl-S-Cgc19 and 4-guanidinoacetamido-2-pyrrolyl-S-Cgc19. The resulting Cgc19-tethered guanidinoacetyldipyrrolamide is released from the biosynthetic machinery via Cgc2-catalyzed condensation with guanidinoamine, and further N-methylated to the final product by an S-adenosyl-l-methionine (SAM)-dependent methyltransferase.

8.3. Leinamycin biosynthesis

Leinamycin is an antitumor antibiotic derived from an NRPS-PKS hybrid pathway from Streptomyces atropolivaceus S-140 (Fig. 19D).²⁰¹ Leinamcyin contains a 1,3-dioxo-1,2-dithiolane core that induces DNA cleavage by converting molecular oxygen into oxygen radicals.^202,203 The biosynthetic gene cluster contains two type II NRPS domains, including A domain LnmQ and PCP domain LnmP that initiate biosynthesis.^204,205 In vitro, LnmQ activates d-alanine as a d-alaninyl-AMP intermediate. Alternatively, glycine is the only other amino acid that is tolerated. Following activation, d-alanine is loaded onto the phosphopantetheine arm of LnmP. LnmQ and LnmP protein-protein interactions are critical, as evidenced by lack of aminoacyl transfer when each protein was incubated with an excised noncognate partner protein. Subsequently, the d-alaninyl-S-LmnP intermediate is transferred to downstream NRPS and PKS modules.

8.4. d-Alanylation of lipoteichoic acids

The cellular walls of Gram-positive bacteria are decorated with polymeric lipoteichoic acids. d-Alanylation of LTAs is critical for altering the negative charge of the membrane and ligand recognition and bacterial strains with defects in LTA biosynthesis are more susceptible to antibiotics. Interestingly, d-alanylation involves activation and transfer of D-alanine via a mechanism that is reminiscent of non-ribosomal peptide biosynthesis. In the cytoplasm, DltA is an adenylation enzyme that specifically activates d-alanine as a d-alanyl-AMP intermediate. Next, transacylation onto phosphopantetheinylated DltC occurs. Currently, there are two models that explain how d-alanine is translocated across the plasma membrane. One model posits that d-alanyl-S-DltC is tranported by DltB, a putative membrane transporter.²⁰⁶ Another study suggests a second transacylation of alanine onto bactoprenol pyrophosphate prior to DltB-mediated transport.²⁰⁶ Regardless of the mode of transport, further studies are warranted to understand how d-alanine is on-loaded onto LTA after translocation.

9. Outlook

The type II NRPS systems produce unique moieties in natural product biosynthesis, and their architecture offers potential for the development of engineered biosynthetic pathways. The systems are capable of performing unusual modifications (hydroxylations, halogenations, dehydrogenations, and cyclopropanations, etc) to various substrates. Furthermore, transfer location of the product into downstream pathways can vary from the starting unit (pyoluteorin pyrrole) to the terminal module (syringomycin chlorinated threonine). In addition, these type II NRPS systems commonly deliver its highly functionalized substrates to other carrier protein dependent pathways, which furthers the biosynthetic complexity of the natural product.

Although many of the enzymes have been extensively characterized individually, the knowledge of protein-protein interactions in these pathways can be significantly advanced. The atomic level detail of the interface created by the PCP and its partner proteins are required to effectively manipulate partner protein recognition and activity. Despite differing primary sequences, the conserved secondary structures of the PCP-partner protein complex may reveal that successful combinatorial biosynthesis is just a single point mutation away. Recently, various groups have begun to solve the structures of PCP-partner protein complexes to reveal the mechanism behind partner protein recognition. Elucidating the multiple interfaces will enhance our understanding of partner protein recognition and will contribute to the engineering of the unique and promising type II NRPS pathways.