01/08/10

History of polyketide research

Polyketides as a research area dates back to 1893, where Collie and Myers isolated the first polyketide; orcinol (Collie & Myers 1893). Later the same year Collie proposed a rough mechanism for the synthesis of orcinol and related compounds, based solely on their structure. In this, he states that they most likely are synthesized by repetitive condensation or polymerization reactions (Collie 1893). Collie later showed that orcinol and related compounds did not fit into any of the previously described chemical classes, and therefore proposed that they should be placed in a new class called polyketides. Its members being characterized by the repetitive occurrence of a -CH₂-CO- “motif”, which he named ketide (Bentley & Bennett 1999).

For the following 45 years, only chemists explored the world of polyketides, and many new compounds were isolated and chemically characterized from both filamentous fungi and eubacteria (Streptomyces sp.). Research in the biosynthesis mechanisms was initiated in 1953, when Birch and Donovan suggested a new biosynthesis pathway for polyketides, which in many aspects resembled the mechanism suggested for fatty acid biosynthesis. This hypothesis became known as the polyacetate hypothesis and stated that: “Polyketides are formed by the head-to-tail linkage of acetate units, followed by a cyclization by an aldol reaction or by acylation to phenols” (Birch & Donovan 1953). In successive studies with the newly developed radionuclear labelling technique, it was possible to show that the ketide groups found in polyketides originated from acetate units derived from the primary metabolism of the producing organism. This theory has proven extremely successful in explaining the biochemical relationship between the different isolated derivates from wild types and deletion mutants, incapable of producing the original polyketide in question (Bentley & Bennett 1999).

With the implementation of modern recombinant DNA techniques (Ligase (1967), Restriction enzymes (1970), recombinant DNA (1972), sequencing (1983) and PCR (1984) (Kielberg et al. 2003)) in the mid 1980’s, it became possible to analyse the genetic basis for the production of polyketides. The first polyketide to be understood in genetic and molecular biological terms was the blue pigment actinorhodin produced by Streptomyces coelicolor. Early classical genetic analysis of six classes of act mutants had shown that they were closely linked, and by recombinant DNA technology it was possible to clone a single DNA piece that could compensate for all six known classes of mutants (Rudd & Hopwood 1979) (Hopwood 1997). The genes responsible for the production of a single polyketide (and other secondary metabolites) are typically organized in clusters of tightly linked genes (operons in bacteria and true gene clusters in fungi).

Polyketide synthases

Polyketide synthases (PKSs) are structurally and functionally related to fatty acid synthases (FAS’s), as both enzyme classes catalyzes the condensation of activated primary metabolites (acetyl-CoA and malonyl-CoA) to form b-ketoacetyl polymers linked to the enzyme by thioester bonds.

                     CO₂-CH₂-CO-S-CoA   +   CH₃-CO-S-PKS   => CH₃-CO-CH₂-CO-S-PKS   +   CoA-H + CO₂

In the fatty acid synthesis, this condensation is followed by b-ketoreduction, dehydration and enoyl reduction to yield the final fully reduced (saturated) fatty acid. In polyketide synthesis these reduction steps are partly or completely omitted in a controlled fashion, resulting in a highly diverse polyketide chain with respect to the occurrence of b-ketone, b-hydroxyl and alkyl groups (Fujii et al. 2001).

Polyketide synthases (PKSs) has typically been categorized based on their number of subunits (a single or multiple) and mode of synthesis (linear or iterative) (Table 2). Eleven different catalytic domains is generally recognized in PKSs (Table 3). The simplest functional PKS consists of a KS, an AT, an ACP and a TE domain. The domains responsible for the addition of a single ketide unit to the growing polyketide and the following modification is denoted a module (http://www.nii.res.in/nrps-pks.html) (Fujii et al. 2001).

The best characterized class of PKSs is the type I modular, but the functional information derived from these typically also apply to the other classes of PKSs.

Table 2          The differences between the three types of PKSs with respect to structure, synthesis mechanism, evolutionary relation and distribution (Watanabe & Ebizuka 2004).

Table 3          The different types of domains found in PKSs. The eleven different domains can be divided into three groups based on which part of the synthesis they  participates in (C = condensation reaction, R = reduction of b-ketone and M = other post-condensation modifications).

Type I (modular) PKSs

The final number of ketide units in polyketides synthesized by type I (modular) PKS equals the number of modules found in the PKS. This is a result of the linear synthesis mode of these PKSs, where the growing intermediate is passed along the PKS from module to module (Figure 3). The TE domain mediates the release of the final polyketide.

This type of PKSs is found in bacteria and is responsible for synthesis of clinical and economical important macrolide polyketides, such as the erythromycin A and rifamycin (http://www.bio.cam.ac.uk/~pflgroup/research.htm).

Figure 3        The module structure of the three Type I (modular) PKSs responsible for the synthesis of erythromycin, with the growing polyketide chain shown, the newest ketide group in each step is highlighted with red. The thick lines at the top denote the extent of the individual module, note that erythromycin is synthesized by three separate enzymes consisting of two modules each (After http://www.bio.cam.ac.uk/~pflgroup/research.htm). Abbreviations for the domains and their function can be found in Table 3. LD = loading module, M1-M6 modules and TE = Thioesterase domain.

Polyketide synthases and prediction of their products

Polyketide synthases are known from both eukaryotic and prokaryotic systems. This family of enzymes catalyze the fusion of short carbon chains into long polymers, via successive rounds of Claisen condensation reactions. However, while catalyzing similar reactions there are several different classes of PKSs, differing in their domain architecture and mode of synthesis.
Type I PKS are characterized by being multizymes (A single polypeptide chain housing multiple different active sites capable of catalyzing different reactions) posing all the necessary enzymatic domains for the formation of a polyketide. In type II PKS the required catalytic domains are located on individual proteins that interact to form a functional PKS enzyme complex. The type III PKS (chalcone synthase-like) differ from the two other types by not relying on acyl carrier protein domains (Meier and Burkart, 2009).
The type I PKS can furthermore be subdivided into a modular and an iterative. The modular type poses multiple copies of each type of active site, organized into modules that are responsible for the addition and modification of a single ketide unit. The starter unit is loaded into the enzyme in an N’terminal loading domain. The growing polyketide chain is then passed from module to module until it reaches the C’terminal end of the enzyme where it is released by a thioesterase domain. This means that it is possible to predict the final polyketide length and which types of modifications the individual ketides units will harbour, just by deciphering the order of catalytic sites found along the PKS and the number of modules (Meier and Burkart, 2009). The iterative type of PKS (iPKS) only pose a single copy of each catalytic domain, however these can be deployed repeatedly during synthesis of a single polyketide molecule, as described in the next two sections. Type I iPKSs are typically further subdivided based on which modifications they can introduce into the growing polyketide chain during synthesis. However the action of the core domains (also known as the minimal PKS) remain the same in all subclasses.

Minimal iPKS: Action of the core domains (non-reducing PKSs)

Polyketide biosynthesis in many aspects resembles the fatty acid synthesis, by utilizing the same active sites and reaction mechanisms. The synthesis can be divided into several steps, shown in figure 5. First the starter unit, in the form of an acetyl, is loaded into the b-ketosynthase domain (KS) of the enzyme, a process that is mediated by the acyl-carrier-protein domain (ACP) (step 1 in figure 5). The acetyl is delivered to the enzyme in the form of acetyl-CoA and bound to the enzyme via a thioester bond (Nelson and Cox, 2005). ACP domains in non-reducing iPKSs have been shown to be able to auto-malonylate (Hitchman et al., 1998). The ACP domain includes a long flexible prosthetic group (4’-phosphopatetheine) that functions as a “crane” that moves the substrates, intermediates and products between the different active sites found in the iPKS (Nelson and Cox, 2005).

The second substrate for the Claisen condensation reaction, the extender unit, is then loaded into the AT domain by the Malonyl-CoA domain (MAT) (step 2 in figure 5). The MAT is only found in the non-reducing iPKS where it facilitates the loading of malonyl between CoA and the acyl-transferase (AT) domain. The extender unit is typically a malonyl delivered to the enzyme by CoA. The KS domain catalyze the Claisen condensation reaction between the starter and extender units, driven by decarboxylation of the extender unit (Proctor et al., 1999) (Step 3 in figure 5). At this point two different options exist: 1) add another ketide unit or 2) release the polyketide chain from the enzyme. In option 1 the product is transferred back to the KS domain to prepare for a second iteration (step 9 in figure 5) and another extender unit is loaded into the enzyme. For option 2 to occur the polyketide have to have reached its predetermined length which is unique for each iPKS. The polyketide is transferred to the thioesterase domain (TE) that catalyzes its release from the enzyme (Hendrickson et al., 1999) (Step 10 in figure 5). The products of non-reducing PKSs typically undergo intra-chain aldol or Claisen reactions catalyzed by a Claisen-type cyclase domain (CLC), which is related to the TE, resulting in the formation of aromatic structures (Fujii et al., 2001). Recent results have proven the existence of a product template domain (PT) that is responsible for situating the polyketide chain correctly to ensure that only one product type is formed (Crawford et al., 2009).
Mammalian fatty acid synthases have been shown to function as homodimers (head-to-head arrangement), meaning that they pose two copies of each domain. Experiments with heterodimers, where one of the momomers have been mutated, suggest that the two monomers feed each other substrates. It is likely that a similar situation exist for the iPKS, but it has not been experimentally validated (Witkowski et al., 2004).
 

Action of modifying domains
The ketide units that are added to the growing polyketide chain by the core set of PKS domains (KS, AT and ACP) can be subjected to modifications catalyzed by modifying domains, if such are present in the enzyme. In reducing iPKSs the ketone group of a ketide unit can be reduced to various degrees catalysed by ketoreductase (KR), dehydratase (DH) and enoyl reductase (ER) domains (Kroken et al., 2003). The KR domain is responsible for reducing the ketone group to a hydroxyl group (Step 5 in figure 5), the DH domain further reduces the hydroxyl group to an enoyl group (Step 6 in figure 5), which in turn can be reduced to an alkyl group catalyzed by the ER domain (Step 7 in figure 5). Reducing iPKSs can also contain domains that add methyl (CmeT) or acetyl (CacT) groups to the reduced polyketide chains, resulting in branching of the backbone chain (Song et al., 2004).

Subdivision of type 1 iPKSs

Fungal iterative Polyketide synthases have traditionally been divided into three groups based on their modifying domains: non-reducing iPKS (NR iPKS), partial reducing iPKS (PR iPKS) and fully reducing or highly reducing iPKS (FR iPKS). The NR iPKS are characterized by only having the core set of iPKS domains (AT, KS, ACP and TE/CYC) as well as MAT and PT domains. The PR iPKS contain, in addition to the core set of domains, a KR domain and possibly also a DH resulting in products with hydroxyl and enoyl groups. The HR iPKS contains all the domains needed to reduce the b-ketone group to an alkyl group, meaning that they in addition to the core domains also have KR, DH and ER domains. However it is important to note that the presence of a modifying domain does not necessarily mean that it is used in every iteration of synthesis, a good example of this is found in the fusarin C biosynthetic pathway. This classification system reflects the evolutionary history of the iPKS as proven by the analysis of KS domains made by (Kroken et al., 2003). This is however a very rough classification that does not encompass the complex nature of iPKS biosynthetic systems, especially if one includes the different types of possible hybrid enzymes and multienzyme systems: such as systems where two iPKSs interact to form a common product (see zearalenone), or where an iPKS and a NRPS have been fused to form a single enzyme (see fusarin C), or systems where iPKS products are modified by NRPS like enzymes (see fumonisins).

 Animation of a minimal type I Iterative PKS in action (PowerPoint created by Rasmus J.N. Frandsen) (click to download)

References