Key words

1 Introduction

Polyketide natural products are a proven source of high-value bioactive small molecules [1]. These include compounds that have become mainstays for the treatment of human and animal diseases, and others that have found application as agrochemicals, flavors, fragrances, and nutraceuticals [2, 3]. Illustrative examples of clinically relevant polyketides include the statin family of cholesterol lowering agents, erythromycin A and related macrolide antibiotics, and the antiparasitic avermectins, which are used extensively in veterinary medicine (Fig. 1) [3–5]. The wealth of structural and functional diversity found within the polyketides is almost unsurpassed amongst natural products [6]. This, coupled to their exploitable bioactivities, has made both the compounds themselves, and the cellular machineries responsible for their biosynthesis, a source of intrigue and fascination for researchers worldwide.

Fig. 1
figure 1

Illustrative examples of polyketide natural products and their respective bioactivities

Full size image

Polyketides of bacterial origin represent a significant proportion of the natural product pool [1]. These compounds function to confer evolutionary fitness to the producing host, acting as defense agents or signaling molecules [7, 8]. Commonly, they are biosynthesized through the action of giant multifunctional megaenzymes, termed modular polyketide synthases (PKSs), which utilize sequential condensation chemistry to construct elaborate product scaffolds from simple carboxylic acid substrates [9, 10]. This biosynthetic logic permits the assembly of a suite of molecules that are in principle of almost unlimited chemical diversity. In addition, the highly modular architecture of these systems raises intriguing possibilities for their modification and manipulation through the addition, removal, or substitution of synthase modules or domains therein [11, 12]. This approach has been exploited to generate an even larger portfolio of useful molecules, the so-called unnatural natural products [13]. With the advent of synthetic biology and the increasing adoption of the concepts and practices of rational design, it is inevitable that synthase reengineering will become more reliable and expedient in the future [14, 15]. Here we outline the basic principles of modular PKS enzymology, highlighting recent developments in the characterization of these systems, and discuss future prospects for harnessing their biosynthetic potential.

2 Modular PKSs and the Biosynthetic Process

Modular PKSs are amongst the largest and most sophisticated enzymes known. In extreme cases their proteinogenic mass can exceed 5 MDa. The distinguishing structural feature of these systems is their assembly line-like architecture, which comprises a series of linearly arranged, multi-domain extension modules, housed in sequence within giant polypeptide chains [1, 16]. Biosynthesis upon these systems proceeds in step-wise fashion, through the processive transfer and extension of the nascent polyketide chain, as it progresses from one module to the next. Each module within the synthase complex incorporates a single carboxylic acid substrate, derived from a pool of simple precursors, into the pathway product (Fig. 2) [1, 6, 17, 18]. The first module of the PKS initiates biosynthesis through the selection and loading of a starter unit. Each subsequent module within the PKS catalyzes the thiotemplated addition of a defined extender unit to this initial priming substrate. In some cases, however, phenomena including module ‘skipping’ or iterative ‘stuttering’ have been reported [19, 20]. The structures and properties of polyketides biosynthesized via this route are dictated by the number, identity, and incorporation order of the starter and extender units from which they are assembled. These factors are defined by the sequence of the modules that constitute the PKS assembly line. For this reason modular PKSs can be considered to possess an inherent biosynthetic programming, where synthase nucleotide sequence is colinearly related to product chemistry [21, 22]. This relationship provides a powerful tool for the prediction and analysis of synthase function, requiring knowledge only of PKS gene sequence. In addition, given that the open reading frames (ORFs) that encode PKS assembly lines and their associated tailoring enzymes are commonly clustered within the genomes of biosynthetically competent microorganisms, PKS nucleotide sequences provide a useful analytical tool by which synthase components can be readily identified and functionally assigned [23, 24].

Fig. 2
figure 2

(a) Chemical structures of example PKS starter and extender units. (b) Individual steps involved in extender unit selection and chain incorporation. (c) Illustrative chain modifications catalyzed by module embedded PKS domains

Full size image

To perform the required set of chemical reactions necessary to extend a polyketide chain by a single acyl unit each PKS extension module must minimally house three essential domains. These are; an acyltransferase (AT) domain, responsible for the selection and loading of carboxylic acid extender units derived from their coenzyme-A (CoA) thioesters; an acyl carrier protein (ACP), possessing a phosphopantetheine arm, which acts as a site of covalent tethering for both the growing product chain and selected extender units; and a ketosynthase (KS) domain, which catalyzes the carbon–carbon bond forming Claisen condensation of the downstream product chain with the upstream ACP tethered extender unit (Fig. 2). This minimal KS-AT-ACP module architecture may be elaborated to include additional auxiliary domains that further modify the incorporated acyl unit. Such modifications may take place at the reactive α-position or the β-carbonyl. An example of the former is the introduction of an α-methyl substituent via a reaction catalyzed by a C-methyltransferase (CMeT), which exploits S-adenosyl methionine as a methyl source (Fig. 2). Alternatively the β-carbonyl of the incorporated acyl unit may be reduced in a reaction catalyzed by a ketoreductase domain (KR) to form a β-hydroxy substituent. This may in turn undergo dehydration, catalyzed by a dehydratase domain (DH), yielding an α/β- unsaturated thioester. Finally an enoyl reductase (ER) may act to reduce this species yet further yielding a product fully reduced at the β-position (Fig. 2). Both the substrate selectivity of the module embedded AT domain and the module’s compliment of auxiliary domains, offer complementary mechanisms for achieving diverse product chemistry [1, 6, 17, 18].

Upon reaching the final module of the PKS assembly line the immature polyketide product is liberated from the synthase via cleavage of its covalent tether. In the majority of modular PKSs this process is catalyzed by a thioesterase that forms either the C-terminal domain of the synthase, or functions as a stand-alone enzyme [25]. Frequently, following release from the PKS, the biosynthesized product is subjected to additional site-specific tailoring modifications catalyzed by free-standing enzymes. Tailoring modifications are in many cases implicitly required to establish the bioactivity of the synthase product. Examples include glycosylation, halogenation, methylation, and hydroxylation among others [1, 10, 26].

Modular PKS enzymology is readily distinguishable from that of other polyketide synthases. Iteratively acting type I PKSs (iPKSs) found in fungi are smaller but inherently more sophisticated systems, which use a single set of functional domains to accomplish the biosynthesis of their products, in a manner analogous to that of the type I fatty acid synthase [27–30]. Uniquely, the domains that comprise iPKSs often only catalyze their specific transformations during a subset of chain extension cycles, establishing a defined biosynthetic program that violates the colinearity of synthase sequence and product chemistry. Type I I PKSs are transiently assembled iteratively acting complexes composed of discrete proteins [31, 32]. Consequently type II systems are significantly smaller and less intricate than both modular and iteratively acting type I PKSs. Biosynthesis upon type II PKSs proceeds in two discernable phases, initiation and chain extension, which are catalyzed by the minimal type II PKS assembly, comprising two KS like condensing enzymes and an ACP. The minimal KS-ACP machinery may be further elaborated through the incorporation of a selection of additional domains including ketoreductases, cyclases, and aromatases. Type III PKSs, which are widely distributed amongst bacteria, fungi, and plants, are in essence free standing KS domains that act to catalyze the sequential condensation of acetate units to a CoA derived starter unit [33–35]. Chemical diversity in the products of type III systems arises due to the choice of starter unit, the number of chain extension steps catalyzed, the mechanism of cyclization, and product modification through the action of allied tailoring enzymes.

3 The DEBS Paradigm

Early attempts to delineate the enzymology of modular PKSs focused on the deoxyerythronolide B synthase (DEBS). This system is responsible for the biosynthesis of 6-deoxyerythronolide (6-dEB), the macrolide nucleus of the clinically relevant antibiotic erythromycin A [4]. Studies of DEBS resulted in the establishment of many of the central tenets of modular PKS enzymology and as such this system serves as a useful test case for illustrating the basic principles of PKS function [36].

DEBS is encoded for by a series of ORFs clustered on the genome of the soil-dwelling actinomycete Saccharopolyspora erythraea [37]. This cluster comprises the genes eryAI-III that encode the core biosynthetic machinery of DEBS (DEBS 1-3), along with a number of associated tailoring and regulatory proteins (Fig. 3) [36, 38]. DEBS was the first modular PKS gene cluster to be sequenced, thus providing initial evidence for the distinctive modular assembly line-like architecture of bacterial PKSs [21, 37]. This analysis also proved instrumental in establishing the colinear relationship between synthase nucleotide sequence and product chemistry.

Fig. 3
figure 3

Domain organization of the 6-deoxyerythronolide B synthase (DEBS) and the biosynthetic route to erythromycin A. Domains are colored based on their host module. Numbered arrows indicate the direction and order of product chain extension and transfer

Full size image

Each module within DEBS consists of a discrete set of domains that are responsible for the incorporation and reductive tailoring of a designated extender unit at a precise location within the growing product chain. Each domain is functionally specific and contributes to the incorporation of a single extender unit within its host module. Biosynthesis upon DEBS is initiated by a loading module, located at the N-terminus of DEBS 1, which consists of an AT, nominally selective for a propionate starter unit, and an ACP onto which this unit becomes tethered. It has been reported that this initial loading step is more promiscuous than was initially presumed and that in addition to propionate, non-native extender units can also be accepted by the starter module AT [39–41].

The remainder of DEBS comprises six extension modules distributed across DEBS 1-3 and a C-terminal TE domain that catalyzes the release and cyclization of the immature PKS product, the 14-membered 6-dEB macrolactone [42]. The modules within DEBS are numbered in the order in which they act during biosynthesis and each houses a single AT, ACP, and KS domain. Each of the extender module embedded AT domains of DEBS is selective for (2S)-methylmalonyl-CoA. In addition to their compliment of core domains, modules 1, 2, 5, and 6 also house active KRs. Consequently, extender units incorporated at these locations undergo reductive modification yielding a hydroxyl substituent at their β-position. The stereochemistries of the hydroxyl groups are dictated at the level of protein structure. The KR domain can attack the β-ketone from either side of the acyl chain, resulting in either a d- or l-configuration. The structure of the KR domain dictates the direction of substrate entry into the enzyme’s active site, which in turn determines the face of the keto-group that is presented for hydride transfer from NADPH [43]. Module 4 of DEBS has a full complement of reductive domains (KR, DH, and ER). Extender units incorporated at this location undergo complete β-keto reduction, yielding an α/β alkane. By contrast, DEBS module 3 houses no functional reductive domains resulting in retention of the β-keto group. Notably, however, this module does contain an oft unreported ‘broken’ KR that controls epimerization of the incorporated methylmalonyl extender unit, a function that was previously ascribed to this module’s KS domain [44, 45].

To generate the fully functional DEBS PKS there is a requirement to assemble DEBS 1-3 into a single complex. This is facilitated by protein–protein interaction interfaces at the N- and C-termini of the individual DEBS polypeptides, which serve to link these three proteins together to form the intact megaenzyme. DEBS docking domains comprise complimentary N- and C-terminal helical bundles of 30–50 and 80–130 residues respectively [46]. Complex formation between complimentary docking domains is implicitly required to permit transfer of the growing product chain between neighboring modules that are housed on different polypeptides.

Upon release from DEBS the cyclized 6-dEB macrolactone is subjected to post-PKS tailoring, catalyzed by a complement of stand-alone enzymes encoded within the ery gene cluster. 6-dEB is hydroxylated and glycosylated to form erythromycin D, which is then converted to erythromycin A through the action of the O-methyl transferase EryG and the hydroxylase EryK [47]. Interestingly these two transformations occur in a sequence independent fashion, forming the isolatable intermediates erythromycin B and erythromycin C respectively.

4 The Structural Enzymology of Modular PKSs

Significant insights into the enzymology of PKS chain extension and processing have been provided by structural studies of isolated synthase domains. Crystal structures have been reported for AT, ACP, KS, KR, DH, and ER domains amongst others, and NMR spectroscopy has been used extensively for the structural characterization of ACPs [38, 48–51]. These data have proven informative in establishing the roles and contributions of each domain during biosynthesis. Here we provide structural descriptions of the three core PKS domains (AT, ACP, and KS), but direct readers to references [49, 50] for more comprehensive descriptions of PKS domain structure.

Acyltransferases have been shown to possess a distinct two sub-domain architecture comprising a larger hydrolase like sub-domain fused to a smaller ferredoxin like sub-domain [52–55]. The AT active site sits at the interface of the two sub-domains, at the base of a solvent exposed channel, and houses an invariant His-Ser catalytic dyad. ATs are proposed to employ a ping-pong bi-bi mechanism that proceeds via an acyl-enzyme intermediate. The intermediate is stabilized during catalysis by an oxyanion hole formed by backbone amides from neighboring amino acids within the enzyme active site. Resolution of the acyl-enzyme intermediate occurs only in the presence of thiol nucleophiles rendering this intermediate sufficiently stable to permit its isolation and characterization in vitro [56]. The topology of the AT active site is in part dictated by amino acids that form defined substrate-selectivity motifs. Residues that occupy these positions play a role in dictating the specific acyl extender unit that the AT selects [57, 58]. Substitution of these motifs has been used to alter AT selectivity giving rise to new natural products, though the universality of this approach remains questionable [59–61]. Recently it has been proposed that extender unit selectivity is dictated more generally by a combination of structural features distributed throughout the enzyme fold [50].

ACPs are small negatively charged helical bundles that provide a site of anchorage for acyl intermediates during biosynthesis [51, 62]. The covalent attachment of intermediates to the ACP occurs via a post-translationally modified serine residue bearing a phosphopantetheine arm, which forms part of a conserved Asp-Ser-Leu motif. This motif is located at the N-terminus of helix 2 of the protein, which is considered the key portion of the ACP for mediating interactions with each of its respective binding partners [63–65]. In many instances ACPs exhibit a high degree of specificity for their cognate intra-module domains [63, 66]. Undoubtedly this property is dictated by a combination of ACP sequence and the identity of the substrate or intermediate to which the ACP is tethered. Although originally considered as somewhat as a passive component of the biosynthetic machinery, it is becoming increasingly clear that ACPs play a more active role. This includes for example shielding of tethered acyl units to allow their presentation at appropriate time points or locations within their host module [67–69].

Ketosynthase domain s are dimeric proteins with a conserved thiolase fold [52, 53, 70]. Structurally KSs are composed of two α-β-α-β-α protomers arranged in the form of a five-layered core, within which three layers of α-helices are separated by two layers of β-sheets. Although there is some structural divergence between KSs, they all exhibit a small number of universal features. These include retention of the overall fold described above, extensive and highly hydrophobic dimer interfaces, and a conserved active site cysteine that acts as a site for the covalent attachment of substrates and intermediates. Variations in KS structure are largely confined to the enzyme’s active site and associated regions, and have significant impact on substrate selectivity [52, 53, 71]. Examples of this include the identity and location of the key catalytic active site residues (excluding the universally conserved Cys) and the steric and electrostatic topology of the active site and its associated solvent exposed access channel. Such variations influence the ability of KSs to act upon intermediates of different chain length, saturation state and stereochemistry, and inadvertently provide a proofreading or gate-keeping function [66, 72, 73]. KS catalyzed Claisen condensation can be achieved using either a Cys-His-His or Cys-His-Asn catalytic triad, along with an intermediate stabilizing oxyanion hole.

Despite continued progress in the structural characterization of isolated PKS domains, it is only recently that structural techniques have been successfully applied to the study of intact PKS modules [74–77]. These analyses have afforded a step-change in the understanding of PKS enzymology and provided insight into the dynamic nature of chain extension and processing. Most significant amongst these studies has been the elucidation of the structure of module 5 from the pikromycin PKS (PikAIII) using cryo-electron microscopy (cryo-EM) [74, 75]. Uniquely, in this study, a number of structures were determined in a range of states that mimic all stages of the module’s catalytic cycle. In addition, the resolution range within which these structures were obtained (7–11 Å) allowed the unambiguous placement of high resolution crystal structures and NMR models of domains homologous to those of PikAIII, yielding a series of pseudo-atomic models describing the process of chain extension in its entirety. In an elegant series of experiments, PikAIII, which possesses a KS-AT-KR-ACP domain architecture, is shown to adopt a distinctive symmetrical arch-like structure comprising a single PikAIII homodimer. Dimerization occurs via the KS domain, which sits at the top of the arch. The AT and KR domains from each monomer form descending struts, which together with the capping KS dimer form a single central reaction chamber accessible by the active sites of each of the KS, AT and KR domains that constitute the module (Fig. 4). The relative positions of each domain deviate significantly from those proposed in structural models of intact PKS modules. The location of the PikAIII module’s two ACPs are also resolved and are shown, using targeted chemical modifications designed to mimic a range of reaction cycle intermediates, to occupy distinct locations within the chamber during catalysis (Fig. 4). These include; a holo-ACP form bearing a phosphopantetheine arm, where the ACPs are located adjacent to their respective AT or KR domains (state 1); a form in which the ACPs carry a pentaketide intermediate, locating them next to their respective AT’s active sites (state 2); a form in which the ACPs are acylated with methylmalonate, and within which they occupy positions below their cognate KSs, primed for active site entry (state 3); a form within which the ACPs carries a β-ketohexaketide, mimicking the system following chain extension, where the ACPs locate next to their respective KRs ready for ketoreduction (state 4); and finally a form mimicking the culmination of the chain extension processes, where the ACPs, carrying β-hydroxyhexaketide groups, are expelled from the PKS reaction chamber, facilitating chain transfer to the down-stream module (state 5). In each of the structures reported the two ACPs occupy equivalent positions on either side of the module dimer, suggesting that they operate in synchronous fashion. This is a likely consequence of space constraints within the reaction chamber and steric hindrance imposed by the downstream KS dimer. Analysis of the reported PikAIII cryo-EM structures also hints at dynamic motions in other domains within the PKS module and is a powerful illustration of the value of analyzing the structures of individual proteins in the context of their interacting partners.

Fig. 4
figure 4

(a) Solid rendering of the cryo-EM map of holo PikAIII (EMD-5647). (b) Cartoon representation of holo PikAIII showing the relative position of each domain within the module. (c) Conformational states of PikAIII observed by cryo-EM (state 1 EMD-5647, state 2 EMD-5663, state 3 EMD-5653, state 4 EMD-5664, state 5 EMD-5666). Each state equates to a defined step within the module’s catalytic cycle. The location of each PikAIII ACP is indicated by a yellow circle. In states 1 and 3 the ACPs occupy overlapping positions at the front and rear of the module

Full size image

5 Trans-AT Synthases: A New Paradigm in Modular PKS Enzymology

A significant recent development in the study of PKSs was the identification of a second class of modular synthases whose domain and module architectures diverge significantly from those of the canonical DEBS like systems [78–80]. This new family of modular PKSs, the trans-AT synthases, are notable for their highly mosaic structures that incorporate disparate biosynthetic features within a single megaenzyme complex [81]. Trans-AT PKSs have been shown to have an evolutionary lineage distinct from that of the cis-AT DEBS like systems [46]. Although trans-AT PKSs make use of the same step-wise sequential condensation chemistry employed by cis-AT synthases, they exploit a much broader repertoire of functional domains, make use of trans-acting elements to modify synthase intermediates, and exhibit module architectures that diverge from the classical KS-AT-ACP paradigm [46, 81, 82]. Here we describe two examples of biosynthetic peculiarities common to trans-AT PKSs, but direct readers to more comprehensive reviews of this area [81, 82].

The defining feature of trans-AT synthases is the absence of module embedded AT domains throughout the PKS . Substrate loading in these systems is instead provided in trans, by free-standing trans-acting ATs encoded within the synthase gene cluster [83]. Trans-acting ATs are found either as stand-alone enzymes, or as di- or tri-domain fusions proteins partnered with decarboxylases (DCs), ERs, and/or proof-reading acyl-hydrolases (AH; Fig. 5) [78, 84–86]. The role of trans-ATs in substrate loading has been demonstrated both in vivo and in vitro, and initial structural and functional characterization of these enzymes suggests that they possess the same general structure and catalytic mechanism as module embedded ATs, but are able to furnish multiple ACPs throughout the PKS complex with extender units [87–90]. The mechanism of trans-AT recruitment to each extension module remains to be established, however, it has been suggested that there may be distinct acyltransferase docking domains located within trans-AT PKS modules that facilitate this process [91, 92].

Fig 5
figure 5

(a) Domain organization within polypeptides that house trans-acting acyltransferases. (b) Scheme for the introduction of β-methyl branches in to polyketides via the HCS cassette route. Only the ACP domains of the modules that support the product chain are shown. ECH enoyl CoA hydratase

Full size image

Another distinctive biosynthetic feature common to trans-AT PKSs is their use of a multi-protein hydroxymethylglutaryl CoA synthase (HCS) enzyme cassette to catalyze the introduction of methyl groups at β-carbon positions within product incorporated acyl extender units [93]. To achieve this acetyl-ACP is condensed with the unreduced β-carbon of the nascent polyketide chain forming a β-hydroxy-β-carboxymethyl intermediate. This intermediate is then subjected to sequential dehydration and decarboxylation to form the β-methyl group (Fig. 5) [89, 94–97]. Structural studies of trans-AT PKS di-domain ACPs upon which β-branching takes place have identified a distinct amino acid signature that appears to target the branching machinery to these locations. This discovery raises the intriguing possibility of implementing β-branching chemistry at a range of locations throughout PKSs, by introducing appropriate amino acid signatures into ACPs that do not, in their native state, support branch formation.

6 Reengineering of PKS Assembly Lines

The highly modular architectures of PKSs make these systems attractive targets for reengineering as a mechanism for accessing unnatural natural products with novel or enhanced functionality. This approach gains credence from the notion that polyketide scaffolds have been evolutionarily selected for optimum performance within their producer’s environmental niche, and as such they may be of limited utility in wider contexts. The targeted modification of polyketides through manipulation of the cellular machineries responsible for their biosynthesis may therefore yield derivatives of the parent compound with, for example, improved clinical efficacy. The tractability of PKS reengineering was initially demonstrated in DEBS [98]. These investigations resulted in the establishment of a set of rules for the purposeful manipulation of modular synthases [98]. Despite these initial successes, the reengineering of modular PKSs has, however, proved to be much more challenging than was initially anticipated. Though with recent advances in the delineation of synthase enzymology, the development of more robust experimental tools for targeted high-throughput genetic manipulation and DNA synthesis, and improvements in analytical techniques and computation, PKS reengineering appears poised for a renascence.

A number of strategies have been proposed for the successful manipulation of PKSs. These focus on targeted modifications at the module, domain, or amino acid level, or involve the refactoring of precursor biosynthesis or post-PKS tailoring. Approaches targeted at the module level have focused on the deletion, insertion or substitution of intact modules within PKS complexes. This approach has been successfully applied to DEBS, where substitution of the loading module of DEBS 1 with that from the tylosin PKS resulted in a hybrid system selective solely for a propionate starter unit [99]. Similarly, replacement of the DEBS 1 starter module with that from the oleandomycin PKS yielded a hybrid system within which acetate was exclusively incorporated as the starter unit [99]. Although pioneered in DEBS, loading module substitution has been successfully implemented in other systems. For example, replacement of the loading module initiating avermectin biosynthesis, which specifically incorporates isobutyryl-CoA, with that of the cyclohexanecarboxylic (CHC) phoslactomycin PKS loading module from Streptomyces platensis, permitted the biosynthesis of the antiparasitic veterinary medicine doramectin [100]. To achieve this outcome it was also necessary to express the five proteins responsible for the biosynthesis of the CHC-CoA precursor in tandem with the hybrid PKS.

Targeted manipulation at the domain level requires less extensive interference with the PKS , however, the importance of intra-module protein–protein interactions should always be considered. The tolerance of DEBS to domain swapping has been probed extensively [101, 102]. Each extension module within this PKS contains an AT domain selective for (2S)-methylmalonyl-CoA. Substitution of these domains with ATs selective for malonyl, ethylmalonyl, or methoxymalonyl extender units has permitted the biosynthesis of regioselectively modified polyketides with the expected chemical composition [102–106]. Substitution of individual domains in DEBS modules 2, 5, and 6, with counterparts from the rapamycin PKS, possessing different substrate specificities and reductive capabilities, as well as the insertion of additional domains from the same system, has been used to produce an extensive range of 6-dEB analogs [102, 107, 108].

To minimize the deleterious effects of PKS reengineering, the targeted mutation of individual residues within modules or domains represents an attractive, less invasive approach. As the number of publically available genome sequences and protein structures increases exponentially, so increases the ability of researchers to make informed site-specific changes that confer or modulate protein function. This may include targeted changes that impact catalytic activity, substrate selectivity, cofactor binding, or stereoselectivity. Examples of the use of this approach in DEBS include deactivation of the enoyl reductase domain of module 4 permitting the biosynthesis of Δ6,7-anhydroerythromycin C [109], and targeted alteration of the substrate selectivities of extender module AT domains to allow malonyl-CoA and fluoromalonyl-CoA to be accepted as substrates [110]. This approach has been further informed by computational methods. For example the use of quantum mechanics/molecular mechanics (QM/MM) methods was instrumental in the design of a mutagenesis strategy to reengineer DEBS AT6 to accept a non-natural 2-propargylmalonyl extender unit [111]. The polyketide biosynthesized by this modified PKS possesses a synthetically functionalizable handle that can be exploited to generate an even greater number of useful derivatives.

In addition to reengineering strategies that focus explicitly on the PKS itself, complimentary approaches that target the allied starter and extender unit biosynthetic pathways, or focus on post synthase tailoring reactions have been investigated. The former have included the targeted mutation of acyl-CoA synthetases [112, 113], or the replacement of endogenous acyl-CoA pathways with those that generate alternative precursors [114]. For the latter, many successful approaches have involved the repurposing of glycosyltransferases to generate polyketide products with altered glycosylation patterns and consequently more favorable toxicities, solubilities, and bioavailabilities [115, 116].

7 Conclusions

In recent years significant progress has been made in the genetic, chemical, biochemical and structural characterization of modular PKSs and their constituent parts. Despite this, many questions still remain, and without doubt the study of modular synthases will remain a fertile area of research for many years to come. Studies of PKS structure and function are providing unexpected insights into the assembly, operation, and dynamics of these systems, and the frequency with which new PKS gene clusters are being identified and annotated continues to rise exponentially.

Notably, one area of modular PKS research that has progressed more slowly than was hoped is that of synthase reengineering. Methods developed to enable the rational redesign of modular PKSs and their products have proven nontrivial to implement successfully across multiple, often closely related systems, and consequently many approaches that showed initial promise have failed to deliver. That said, new fundamental insights into the enzymology of these systems will undoubtedly expedite the development of the necessary tools and technologies that are required for robust, broadly implementable synthase reengineering.