Report

Cloning and Heterologous Expression of the Epothilone Gene Cluster

See allHide authors and affiliations

Science  28 Jan 2000:
Vol. 287, Issue 5453, pp. 640-642
DOI: 10.1126/science.287.5453.640

Abstract

The polyketide epothilone is a potential anticancer agent that stabilizes microtubules in a similar manner to Taxol. The gene cluster responsible for epothilone biosynthesis in the myxobacteriumSorangium cellulosum was cloned and completely sequenced. It encodes six multifunctional proteins composed of a loading module, one nonribosomal peptide synthetase module, eight polyketide synthase modules, and a P450 epoxidase that converts desoxyepothilone into epothilone. Concomitant expression of these genes in the actinomyceteStreptomyces coelicolor produced epothilones A and B.Streptomyces coelicolor is more amenable to strain improvement and grows about 10-fold as rapidly as the natural producer, so this heterologous expression system portends a plentiful supply of this important agent.

The epothilone polyketides (1) stabilize microtubules by means of the same mechanism of action as the anticancer agent Taxol (2). However, epothilones are advantageous in that they are effective against Taxol-resistant tumors and are sufficiently water soluble that they do not require deleterious solubilizing additives (3). For these reasons, epothilone is widely perceived as a potential successor to Taxol (4).

The paucity of epothilones currently obtainable represents a major impediment to clinical evaluation of this important agent. The epothilone producer Sorangium cellulosum yields only about 20 mg liter−1 of the polyketides and has a 16-hour doubling time that makes production in this organism economically impractical (1). Further, while epothilones A (1) and B (2) are the most abundant congeners (produced in a 2:1 ratio) in fermentation extracts, 12,13-desoxyepothilone B (4; epothilone D) has the highest therapeutic index but is produced in only trace amounts (4). Owing to the lack of a satisfactory fermentation process, the total synthesis of epothilones has been pursued as a source of material and in order to develop structure-activity relations (3, 5,6). The Danishefsky (3, 5) and Nicolaou (6) research groups have reported tour de force efforts for the complete synthesis of epothilone and numerous analogs. However, given the complexity of the over 20 synthetic-step processes, fermentation-based methods are likely to reign as preferred practical approaches for large-scale production of the epothilones.

Here we demonstrate the production of epothilones A and B in a “fermentation-friendly” heterologous host. To accomplish this, we cloned and sequenced the entire 56-kb epothilone gene cluster, which encodes a polyketide synthase (PKS), including a nonribosomal peptide synthetase module, and a cytochrome P450 epoxidase. Introduction of all the genes of the cluster into Streptomyces coelicolor CH999 led to the production of epothilones A and B. Heterologous production of the cytochrome P450 EpoK in Escherichia coli and an in vitro assay provided direct evidence that this enzyme catalyzes the conversion of desoxyepothilone (3 and 4) into epothilone (1 and 2) as the final step in epothilone biosynthesis.

Type I PKSs and nonribosomal peptide synthetases (NRPSs) are large multifunctional protein complexes organized in a modular fashion. Each PKS module activates and incorporates a two-carbon (ketide) unit building block into the polyketide backbone. The number and order of modules, and the types of ketide-modifying enzymes within each module, determine the structural variations of the resulting products. The epothilones show two interesting structural variations when compared to a prototypical polyketide such as 6-deoxyerythronolide B: a thiazole moiety and a geminal dimethyl group. A gene cluster that includes a NRPS module flanked by PKS modules, one of which contains an embedded methyl transferase, could produce such variations.

Using polymerase chain reaction (PCR)–generated hybridization probes (7), we isolated four overlapping cosmid clones from a genomic library of S. cellulosum strain SMP44. DNA sequence analysis revealed eight open reading frames (ORFs) that span over 56 kb (Fig. 1). They include epoA(encoding the 149-kD loading domain), epoB (158 kD, a NRPS module), epoC (193 kD, PKS module 2), epoD(765 kD, PKS modules 3 to 6), epoE (405 kD, PKS modules 7 and 8), epoF (257 kD, PKS module 9 plus a thioesterase domain), epoK (47 kD, a cytochrome P450), and an ORF immediately downstream of epoK that encodes a protein with three membrane-spanning regions (ORF1).

Figure 1

Modular organization of the epothilone polyketide synthase (PKS). Functional domains of each of the epothilone PKS modules are shown. Stepwise synthesis of epothilones begins at EpoA and ends with the cyclization by the TE domain in EpoF to yield either epothilones C and D or the hypothetical molecule containing the OH group at C-13. Abrevia- tions: KS, β-ketoacyl ACP synthase; KSy, β-ketoacyl ACP synthase containing a tyrosine substitition of the active-site cysteine; AT, acyltransferase; DH, dehydratase; ER, enoylreductase; KR, ketoreductase; MT methyltransferase; ACP, acyl carrier protein; TE, thioesterase; C, condensation; A, adenylation; PCP, peptidyl carrier protein.

The domain organization of the epothilone gene cluster is consistent with the structure of epothilone. The role of the enoylreductase (ER) domain within the loading module is unknown; it may be cryptic or it may play a role in the oxidation of the thiazoline to the thiazole. The only function absent is a dehydratase (DH) domain in module 4, which would generate a cis double bond between carbons 12 and 13. Dehydration could occur either in the next module (which possesses an active DH domain) by an atypical process, or by action of a post-PKS modifying enzyme. Another intriguing feature of the PKS is that the acyltransferase (AT) domain of module 4 accepts either malonyl or methylmalonyl extender units. This relaxed specificity is consistent with the PKS producing both epothilones A and B in the absence of an identifiable separate methyltransferase. A methyltransferase (MT) domain is integrated into module 8 between the DH and ketoreductase (KR) domains and is believed to methylate C-4 of the epothilones to generate the gem-dimethyl function. Similar MT domains have been observed in the PKSs for lovastatin, fumonisin, and yersiniabactin biosynthesis (8). Another notable feature of the epothilone polyketide megasynthase is the presence of an NRPS module flanked by two PKS modules. This NRPS module contains signature sequences for recognizing cysteine as well as a cyclization domain, which leads to the formation of the thiazole (9).

For heterologous expression of the epothilone gene cluster, and production of epothilone, we used the well-characterized actinomyceteS. coelicolor. In contrast to S. cellulosum,S. coelicolor is well understood genetically and genomically and has a doubling time of only 2 hours. Vector systems for the expression of PKS gene clusters in this organism have been described (10) and used to synthesize a variety of bacterial and fungal natural products (11). The large epothilone biosynthetic gene cluster was cloned into two compatible plasmids (12). The epoA, epoB, epoC, and epoD genes were cloned as an operon behind theactI promoter on a thiostrepton-resistant SCP2* derivative (13), whereas epoE, epoF,epoK, and ORF1 were fused as a second operon to theactI promoter on an apramycin-resistant pSET152 derivative (14). The plasmids were introduced into S. coelicolor CH999 (15), and transformants were grown on R2YE medium. The transformants produced epothilones A and B, as verified by high-performance liquid chromatography (HPLC), mass spectroscopy (MS) of the molecular ions, and, for epothilone A, mass fragmentation pattern (16). Recently, deletion of epoK and the downstream gene (ORF1) produced epothilones C and D (17). Initial yields of the epothilones in these studies were 50 to 100 μg liter−1. Given the high growth rate and pliability of S. coelicolor to genetic and conventional strain improvement, this system promises to evolve into the preferred producer of epothilone.

In addition, epoK gene was expressed in E. coli.EpoK was purified (18) and shown to have an ultraviolet (UV)–visible spectrum characteristic of a cytochrome P450 enzyme. The purified protein converted desoxyepothilone B (4) to epothilone B (2) (19), indicating that the epoxidation reaction is the last step in the biosynthetic pathway.

The production of epothilones A and B in S. coelicolor demonstrates that the polypeptides encoded by theepoA-F and epoK genes, and the small molecule precursors in the heterologous host are sufficient for epothilone biosynthesis. The availability of a heterologous expression system portends rapid advancement in several important areas. First, protein and metabolic engineering of the expression system are now possible that, together with conventional strain improvement approaches, will enhance productivity and increase availability of the epothilones. Compared to the poorly understood and slow-growing S. cellulosum, S. coelicolor offers major advantages, because it is readily amenable to genetic manipulation and replicates about 10-fold faster. Second, it should now be possible to construct an expression system for the currently most attractive clinical candidate, desoxyepothilone, as the sole fermentation product. This could be achieved by two relatively simple modifications that have precedent in the manipulation of PKS gene clusters (20): (i) substitution of the nonspecific AT of module 4 with a methylmalonyl-specific AT to prevent formation of epothilones A and C, and (ii) inactivation or omission of epoK to prevent conversion of desoxyepothilone to epothilone B. Finally, as demonstrated for erythromycin (21), the availability of cloned genes and a plasmid-borne expression system will allow facile manipulation of the epothilone PKS to produce potentially superior epothilone analogs.

  • * To whom correspondence should be addressed. E-mail: julien{at}kosan.com

REFERENCES AND NOTES

View Abstract

Stay Connected to Science

Navigate This Article