Polyketide synthase chemistry does not direct biosynthetic divergence between 9- and 10-membered

Geoff P. Horsmana,1, Yihua Chena,1, Jon S. Thorsona,b, and Ben Shena,b,c,2

aDivision of Pharmaceutical Sciences, School of Pharmacy, bNational Cooperative Drug Discovery Group, and cDepartment of Chemistry, University of Wisconsin, Madison, WI 53705

Edited by Jerrold Meinwald, Cornell University, Ithaca, NY, and approved May 17, 2010 (received for review March 16, 2010)

Enediynes are potent antitumor antibiotics that are classified as been sequenced, as exemplified by C-1027 (9), NCS (10), and 9- or 10-membered according to the size of the core (MDP) (11) for the 9-membered enediynes and structure. However, almost nothing is known about enediyne core CAL (12) and dynemicin (DYN) (13) for the 10-membered biosynthesis, and the determinants of 9- versus 10-membered en- enediynes, respectively (Fig. 1A). Common to all clusters is a five ediyne core biosynthetic divergence remain elusive. Previous work gene cassette (Fig. 1B, black) that encodes a unique enediyne- identified enediyne-specific polyketide synthases (PKSEs) that can specific iterative type I polyketide synthase (PKSE), a thioester- be phylogenetically distinguished as being involved in 9- versus ase (TE), and three additional genes of unknown function. The 10-membered enediyne biosynthesis, suggesting that biosynthetic PKSE generates an enediyne core precursor from iterative divergence might originate from differing PKSE chemistries. Recent condensation of malonyl-CoA (9); the high sequence identities in vitro studies have identified several compounds produced by the (41–68%, Table S1) and identical domain organization among PKSE and associated thioesterase (TE), but condition-dependent PKSEs (Fig. S1) suggest a common catalytic function and pro- product profiles make it difficult to ascertain a true catalytic differ- duct. These data have inspired speculation that PKSEs from both ence between 9- and 10-membered PKSE-TE systems. Here we re- classes generate an early common intermediate, which is directed

port that PKSE chemistry does not direct 9- versus 10-membered to divergent fates by the action of 9- or 10-membered-specific BIOCHEMISTRY enediyne core biosynthetic divergence as revealed by comparing accessory enzymes (Fig. 2A) (14). This hypothesis is consistent the products from three 9-membered and two 10-membered with the identification of 6–7 additional genes adjacent to the PKSE-TE systems under identical conditions using robust in vivo as- core cassette that are common only among the 9-membered gene says. Three independent experiments support a common catalytic clusters (Fig. 1B, gray). However, phylogenetic analysis of PKSEs function for 9- and 10-membered PKSEs by the production of a reveals that they group into two clades according to enediyne core i heptaene metabolite from: ( ) all five cognate PKSE-TE pairs in ring size. This ability to distinguish 9- versus 10-membered PKSEs Escherichia coli ii ;() the C-1027 and cognate PKSE- led to a predictive familial classification model for new enediyne- Streptomyces lividans iii TEs in K4-114; and ( ) selected native pro- encoding genes (14) and suggests that a possible catalytic differ- ducers of both 9- and 10-membered enediynes. Furthermore, PKSEs ence between the two PKSE families may be responsible for and TEs from different 9- and 10-membered enediyne biosynthetic generating the respective core structures. machineries are freely interchangeable, revealing that 9- versus Indeed, recent evidence for a functional distinction between 10-membered enediyne core biosynthetic divergence occurs 9- and 10-membered PKSEs is consistent with PKSE-directed beyond the PKSE-TE level. These findings establish a starting point biosynthetic divergence of the two enediyne families (Fig. 2B). for determining the origins of this biosynthetic divergence. For instance, we previously reported that NcsE and SgcE—the PKSEs involved in the biosynthesis of the 9-membered enediynes biosynthesis ∣ C-1027 ∣ thioesterase NCS and C-1027, respectively—produced heptaene (1) when co- expressed with their associated TEs, NcsE10 and SgcE10 (15, 16). nediyne natural products are some of the most potent antitu- In contrast, Liang and coworkers observed methyl hexaenone (2) Emor antibiotics discovered (1), and (NCS) as the major product of the in vitro combination of CalE8 and and calicheamicin (CAL) enjoy clinical success in Japan (2) CalE7, the PKSE-TE pair from the 10-membered enediyne and the United States (3), respectively. While their unique CAL (17). This apparently inefficient keto-reduction, or “ketor- mechanism of action (4, 5) and extraordinary biological activity eductase (KR) skipping,” by CalE8 in the final round of iterative generate interest among clinicians, the unusual molecular archi- chain extension was therefore proposed as the first point of tecture of the enediynes motivates elucidation of the biosynthetic biosynthetic divergence between 9- and 10-membered enediyne logic directing their assembly. Most structurally remarkable is cores (Fig. 2B). the enediyne core, which is composed of two acetylenic groups However, subsequent reports using in vitro methods to deci- conjugated by a double bond within either a 9- or 10-membered pher functional differences between 9- and 10-membered PKSEs ring (Fig. 1A). Enediyne biosynthesis has been studied most reveal more complex behavior. For example, Townsend and cow- extensively for the 9-membered enediyne C-1027, revealing a 1 2 convergent process in which several peripheral moieties are orkers detected , , and several truncated polyketides from in appended to the enediyne core (6). Indeed, recent elucidation of peripheral pathway biochemistry has enabled rational design Author contributions: G.P.H., Y.C., and B.S. designed research; G.P.H. and Y.C. performed of novel enediyne analogs possessing altered biological activities research; G.P.H., Y.C., and B.S. analyzed data; J.T. contributed new reagents/analytic tools; (7, 8). However, despite rapid progress toward understanding and G.P.H., Y.C., and B.S. wrote the paper. peripheral moiety construction, the biosynthesis of the enediyne The authors declare no conflict of interest. core structure remains a mystery. This article is a PNAS Direct Submission. The enediynes are classified according to the enediyne core 1G.P.H. and Y.C. contributed equally to this work. ring size as either 9- or 10-membered, and how these two classes 2To whom correspondence should be addressed. E-mail: [email protected]. biosynthetically diverge is an intriguing question. Several biosyn- This article contains supporting information online at www.pnas.org/lookup/suppl/ thetic gene clusters for both 9- and 10-membered enediynes have doi:10.1073/pnas.1003442107/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1003442107 PNAS Early Edition ∣ 1of5 Downloaded by guest on September 26, 2021 Fig. 1. Selected 9- and 10-membered enediynes whose biosynthetic gene clusters have been sequenced. (A) Chemical structures of the 9- (C-1027, NCS, and MDP) and 10-membered (CAL, DYN, and ESP) enediynes with the enediyne core structures shown in bold. (B) Regions of enediyne gene clusters encoding enediyne core-biosynthesis with genes common among both 9- and 10-membered enediynes shown in black and genes common only among the 9-membered enediynes shown in gray. The pksE genes for 9- and 10-membered systems are named E and E8, respectively, while TE genes are named E10 and E7, respectively.

vitro analysis of CalE8 and CalE7, indicating partial KR function We herein report that PKSE chemistry does not direct 9- versus in the final round of chain extension by CalE8 (dashed line in 10-membered enediyne core biosynthetic divergence (Fig. 2A) Fig. 2B). The variety of products was interpreted to be a conse- by comparing the products from three 9- (C-1027, NCS, MDP) quence of PKSE catalysis occurring in the absence of required and two 10-membered (CAL, DYN) PKSE-TE systems under accessory proteins and suggested that the apparent catalytic dif- identical conditions using robust in vivo assays. Three indepen- ference between 9- and 10-membered PKSEs may be an artifact dent experiments support a common catalytic function for 9- of assay conditions (18). Indeed, by optimizing in vitro reaction and 10-membered PKSEs by the production of 1 from: (i) all five conditions, Liang and coworkers variably observed 1, 2, and an cognate PKSE-TE pairs in E. coli;(ii) the SgcE-SgcE10 and unknown compound as major products from 10-membered CalE8-CalE7 cognate pairs in Streptomyces lividans K4-114; (CAL and DYN) and 9-membered (C-1027) PKSE-TE systems and (iii) selected native producers of both 9- and 10-membered (19). Although the assay could be modified to yield 1 as the major enediynes. Finally, as a first step toward investigating the possible product from each system, the condition-dependent product pro- role of protein–protein interactions in directing biosynthetic di- files of in vitro PKSE-TE systems caution against interpreting a vergence between 9- and 10-membered enediynes, we demon- common catalytic function for all PKSEs and motivate further strated no pathway-specific fidelity of the PKSE-TE interaction. study using more reproducible methods. These findings set the stage to further search for the origins of The difficulties with in vitro experimental systems for PKSE biosynthetic divergence between 9- and 10-membered enediynes. studies contrast sharply with our previously reported in vivo sys- tem, in which expression of pksE-TE from either the C-1027 or Results and Discussion NCS biosynthetic machineries in either Escherichia coli or Strep- The first step toward comparing 9- and 10-membered PKSEs was tomyces consistently provided 1 as the only major product (15). to prepare compatible E. coli constructs for coexpressing pksE and The apparent advantage of in vivo methods may reflect the dif- TE from each of the five sequenced enediyne gene clusters: C- ficulty of reconstituting the large multifunctional PKSE megaen- 1027 (sgcE-sgcE10), NCS (ncsE-ncsE10), MDP (mdpE-mdpE10), zyme in vitro in the absence of the mild and regenerative CAL (calE8-calE7), and DYN (dynE8-dynE7)(Table S2). Coex- conditions of the cell in vivo. We therefore elected to extend pression of each cognate pair in a 50 mL E. coli culture and HPLC the use of in vivo systems to compare 9- and 10-membered PKSEs analysis of the concentrated acetone extract revealed that 1 was with the intent of more faithfully reproducing the conditions pre- the only major product from each of the five PKSE-TE systems sented to the biosynthetic machinery in the native producers. (Fig. 3A,I–V). This compound was readily identified to be 1

2of5 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1003442107 Horsman et al. Downloaded by guest on September 26, 2021 BIOCHEMISTRY

Fig. 2. Two proposed origins of biosynthetic divergence between 9- and 10-membered enediynes. (A) Common PKSE chemistry based on current in vivo studies: All PKSEs generate a common intermediate, which is processed by respective pathway-specific enzymes to furnish different core structures. (B) Divergent PKSE chemistry based on previous in vitro studies: The phylogenetically related class of 9-membered PKSEs generate 1 via a β-hydroxy thioester, while the 10-membered PKSEs may stall at the final round of iterative keto-reduction to yield 2 as a major product.

by its distinctive absorption spectrum (Fig. S2) and expected mass 10-membered enediynes under enediyne-producing conditions. of 198.1 and confirmed to be identical to the previously charac- Although 1 is produced from pksE-TE expression in both E. coli terized authentic standard (15). Productions of 1 by either the and S. lividans, we could not definitively exclude the possibility 9- or 10-membered PKSE-TE systems under the conditions that production of 1 was an artifact of our simple two-enzyme examined are very similar as exemplified by the estimated yields system in a heterologous host. We therefore carefully examined of 1 at 1.3 Æ 0.5 mg∕L for CalE8-CalE7 and 2.8 Æ 1.0 mg∕L for the fermentation products of the native C-1027 producer S. glo- MdpE-MdpE10, respectively. Production was abolished when: bisporus C-1027. Strikingly, 1 is produced in S. globisporus at a (i) the TE was excluded (Fig. 3A, VI) or (ii) a ketosynthase-inac- high level (∼14 mg∕L) (Fig. 3C, I). Production of 1 was shown tivated SgcE-C211A enzyme (15) was used (SgcE KS°, Fig. 3A, to be sgcE-dependent—it was abolished in a ΔsgcE strain of VII), showing that production is both TE- and PKSE-dependent. S. globisporus C-1027, SB1005 (9) (Fig. 3C, II), and restored Taken together, the in vivo data demonstrates a common catalytic by complementation of SB1005 with a functional copy of the sgcE function for all PKSEs, implying that 9- versus 10-membered gene (Fig. 3C, III). Moreover, 1 was also produced independently enediyne pathway divergence is dictated by accessory enzymes, of C-1027, as demonstrated by its production in S. globisporus as previously proposed (14, 18). SB1010, a strain unable to produce C-1027 due to an inactivated Having successfully established a common chemistry for 9- and epoxide hydrolase gene sgcF (Fig. 3C, IV and Fig. S3) (22). 10-membered PKSE-TE systems in E. coli, we then showed that 1 Inspired by these findings, we examined another 9-membered en- is also produced by PKSE-TE expression in Streptomyces lividans ediyne producer Streptomyces carzinostaticus ATCC 15944 (NCS K4-114 (20), a host more closely related to the enediyne native producer) and two 10-membered enediyne native producers producers. Due to the condition-dependent product profile Micromonospora echinospora ssp. calichensis (CAL producer) observed using in vitro systems (17–19, 21), we reasoned that and Actinomadura verrucosospora ATCC 39334 (esperamicin the in vivo PKSE-TE chemistry may also be host-dependent, (ESP) producer); all strains tested produced 1 (Fig. 3C,V–VII) and we therefore sought to examine production in a host other under enediyne-producing conditions (Fig. S3). This fascinating than E. coli. Cognate PKSE-TE constructs for C-1027 (sgcE- observation further demonstrates that PKSEs from 9- and sgcE10) and CAL (calE8-calE7) were prepared (Table S2), and 10-membered systems utilize the same carbon chain extension expression of each construct in S. lividans K4-114 yielded 1 chemistry en route to different enediyne core structures. (Fig. 3B, I and II). This result further demonstrates a common Having established a common catalytic function for 9- and 10- chemistry catalyzed by both 9- and 10-membered PKSE-TE pairs membered PKSE classes, we reasoned that pathway divergence in a host possessing cellular conditions that are presumably more might involve protein–protein interactions between the PKSE representative of the native producer strains. and pathway-specific accessory enzymes. As a first step in this Finally, a common PKSE-TE chemistry was ultimately verified direction, we sought to probe the pathway-specific fidelity of by the production of 1 in selected native producers of both 9- and the PKSE-TE interaction, motivated in part by the observation

Horsman et al. PNAS Early Edition ∣ 3of5 Downloaded by guest on September 26, 2021 Fig. 3. HPLC chromatograms with UV detection at 370 nm showing production of 1 from selected: (A) recombinant E. coli strains coexpressing cognate pksE-TE pairs; (B) recombinant S. lividans K4-114 strains coexpressing cognate and mismatched pksE-TE pairs; (C) native enediyne producers (I) S. globisporus C-1027, (II) ΔsgcE mutant SB1005, (III) SB1011 (SB1005 complemented by expressing sgcE in trans), (IV) ΔsgcF mutant SB1010, (V) S. neocarzinostaticus ATCC 15944, (VI) M. echinospora ssp. calichensis, and (VII) A. verrucosospora ATCC 39334; and (D) recombinant E. coli strains coexpressing mismatched pksE-TE pairs.

that both PKSEs (14) and TEs appear to phylogenetically cluster lier observation that only the 9-membered pksE genes, but not according to enediyne structural class (Fig. S4). Although Liang calE8, could restore C-1027 or NCS production in pksE null and coworkers showed that SgcE10 could substitute for CalE7 to mutants (15). As an important first step toward elucidating path- similarly liberate 2 from CalE8 (21), the combinatorial flexibility way-directing interactions between PKSEs and accessory enzymes, afforded by our five PKSE and five TE constructs enables a more we demonstrated the production of 1 by all possible PKSE-TE thorough examination of pathway-specific effects by the conveni- combinations. This PKSE-TE promiscuity establishes that biosyn- ent coexpression of any PKSE-TE combination. Significantly, all thetic divergence between 9- and 10-membered enediyne cores five TEs were freely interchangeable with all five PKSEs to yield 1 requires more than just these two enzymes. Our findings now pro- from all 20 possible mismatched PKSE-TE combinations in vide a starting point from which to search for the origins of this E. coli (Fig. S2), as exemplified by representative chromatograms divergence. in Fig. 3D. We then verified this result in a different host, The present study raises additional questions and possibilities S. lividans K4-114 (20), to ensure that any pathway-specific effects with respect to the reproducible production of 1 from a wide vari- were not host-dependent. Specifically, we prepared constructs of ety of host systems. The high yields from the native producers sgcE-calE7 and calE8-sgcE10 (Table S2) for expression in S. livi- raises interesting questions regarding the biological relevance 1 dans K4-114 and observed the production of in both cases of this compound. For example, S. globisporus C-1027 produces (Fig. 3B, III and IV). The interchangeability of TEs between 9- ∼14 mg∕Lof1 and ∼1 mg∕L of C-1027 chromophore (9), which and 10-membered PKSE-TE systems clearly rules out the possibi- corresponds to successful C-1027 production in only about 1 of lity that pathway-specific divergence originates from the PKSE-TE every 70 enzyme turnovers. Finally, the apparent ubiquity of 1 in interaction and further supports a unified view of PKSE catalysis. all enediyne producers provides a convenient phenotypic screen In summary, we have presented definitive evidence that PKSE to identify new enediyne-producing organisms, complementing chemistry does not direct 9- versus 10-membered enediyne core the genotypic screenings we have reported previously (14, 23). divergence. We instead envision a universal PKSE-bound inter- mediate that is modified by pathway-specific accessory enzymes Methods en route to formation of the 9- or 10-membered enediyne core General. Oligonucleotide synthesis and DNA sequencing was performed by structures (Fig. 2A). Thus, the phylogenetic divergence between the University of Wisconsin Biotechnology Center. PCR was performed using 9- and 10-membered PKSEs may reflect enediyne core-specific Pfx polymerase (Invitrogen, Carlsbad, CA), and 3′ A-overhangs were added protein-protein interactions between the PKSE and correspond- with Taq polymerase from Invitrogen for 10 min at 72 °C prior to subcloning ing accessory enzymes. This hypothesis is consistent with our ear- into the pGEM-T Easy vector from Promega (Madison, WI). Mass spectrometry

4of5 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1003442107 Horsman et al. Downloaded by guest on September 26, 2021 was performed using electron impact ionization on a Waters (Micromass) CAL (25), and ESP (26) production, respectively. After 3 to 5 d fermentation, AutoSpec mass spectrometer (Milford, MA). Strains, plasmids, primers, the culture was centrifuged to collect the cell pellet. The supernatant was detailed procedures for making the pksE and TE expression constructs, se- adjusted to pH 4.0 using 1 M HCl and centrifuged again. The resulting pre- quence comparison and phylogenetic analysis of PKSEs and TEs, and HPLC cipitate was combined with the cell pellet and extracted with acetone. The analyses of 1 and bioassays for enediyne production are summarized in acetone extracts were subjected to HPLC as described below. Bioassays Tables S1, S2, and S3 and Figs. S1, S2, S3, and S4 and provided in the online against Micrococcus luteus ATCC 9431 to detect enediyne production were supporting information. performed as described previously (Fig. S3) (27).

Coexpression of pksE-TE Constructs in E. coli. Various combinations of pksE- Extraction and HPLC Analysis of Heptaene (1) from Recombinant Strains Coex- and TE-containing plasmids (Table S2) were cotransformed into E. coli pressing pksE-TE. Each 50 mL culture of the E. coli or S. lividans recombinant BL21(DE3) and plated on LB agar plates containing the appropriate combi- strains coexpressing pksE-TE was acidified to pH ∼ 3 and centrifuged to pellet nation of two antibiotics for selection. Single colonies were picked and the cells. The cell pellet was then resuspended in 20 mL acetone by vortexing. grown overnight at 37 °C in LB and then transferred to 50 mL LB cultures The acetone extract was centrifuged to pellet the cell debris, and the super- supplemented with the appropriate antibiotics. The cultures were grown natant was rotary evaporated to a volume of 1 mL, of which 100 μLwas at 37 °C until the optical density at 600 nm (OD600) reached ∼0.2, at which point the cultures were transferred to 18 °C incubators until they reached subjected to HPLC analysis. HPLC was carried out on a Varian HPLC system equipped with Prostar 210 pumps, a photodiode array detector, and an All- OD600 ∼ 0.4 and then supplemented with 0.1 mM IPTG to induce gene expres- μ 4 6 × 250 sion. The cultures were then incubated for an additional 2 d prior to harvest. tech Alltima C18 column (5 m, . mm, Grace Davison Discovery Sciences, Deerfield, IL) using mobile phases A (water, 0.1% trifluoroacetic Expression of pksE-TE Constructs in S. lividans K4-114. Various combinations of acid) and B (methanol, 0.1% trifluoroacetic acid). A gradient elution program pksE- and TE-containing plasmids (Table S2) were introduced into S. lividans was employed as follows: 60–100% B (0–15 min), 100% B (15–35 min), K4-114 by protoplast transformation according to standard protocols (24). 100–60% B (35–35.5 min), 60% B (35.5–40 min). The identity of 1 produced Single colonies of the resultant recombinant strains were used to inoculate from the recombinant strains or the native enediyne producers was estab- 5 mL cultures of MYME media supplemented with 10 μg∕mL thiostrepton lished by comparing its UV-Vis spectrum, mass spectrum, and HPLC retention and incubated at 30 °C for ∼5 days, and 1 mL of each starter culture was then time with a previously characterized authentic standard of 1 (15). used to inoculate a 50 mL culture, which was in turn incubated at 30 °C for ∼5–10 days. ACKNOWLEDGMENTS. We thank Dr. Y. Li, Institute of Medicinal Biotechnology, Chinese Academy of Medical Sciences (Beijing), for the wild-type S. globis- Production, Extraction, and HPLC Analysis of Heptaene (1) from Native Enediyne porus, and the Department of Chemistry, University of Wisconsin, Madison, Producers. S. globisporus strains were cultured in A9 medium using a for support in obtaining the MS data. This work was supported by the

two-step fermentation for C-1027 production as described previously (22). National Institutes of Health Grants CA78747 and CA113297 (to B.S.) and BIOCHEMISTRY S. carzinostaticus ATCC 15944, M. echinospora ssp. calichensis, and A. verru- CA84374 (to J.S.T.). G.P.H. is a Natural Sciences and Engineering Research cosospora ATCC 39334 were cultured as previously described for NCS (10), Council of Canada postdoctoral fellow.

1. Wang X-W, Xie H (1999) C-1027. Antineoplastic antibiotic. Drug Future 24:847–852. 14. Liu W, et al. (2003) Rapid PCR amplification of minimal enediyne polyketide synthase 2. Maeda H (2001) SMANCS and polymer-conjugated macromolecular drugs: Advantages cassettes leads to a predictive familial classification model. Proc Natl Acad Sci USA in cancer chemotherapy. Adv Drug Deliver Rev 46:169–185. 100:11959–11963. 3. Sievers EL, Linenberger M (2001) Mylotarg: Antibody-targeted chemotherapy comes 15. Zhang J, et al. (2008) A phosphopantetheinylating polyketide synthase producing a of age. Curr Opin Oncol 13:522–527. linear polyene to initiate enediyne antitumor antibiotic biosynthesis. Proc Natl Acad 4. Nicolaou KC, Dai W-M (1991) Chemistry and biology of the enediyne anticancer Sci USA 105:1460–1465. antibiotics. Angew Chem Int Ed Engl 30:1387–1416. 16. Horsman GP, Van Lanen SG, Shen B (2009) Iterative type I polyketide synthases for – 5. Galm U, et al. (2005) Antitumor antibiotics: , enediynes, and mitomycin. enediyne core biosynthesis. Methods Enzymol 459:97 112. Chem Rev 105:739–758. 17. Kong R, et al. (2008) Characterization of a carbonyl-conjugated polyene precursor in – 6. Van Lanen SG, Shen B (2008) Biosynthesis of enediyne antitumor antibiotics. Curr Top 10-membered enediyne biosynthesis. J Am Chem Soc 130:8142 8143. Med Chem 8:448–459. 18. Belecki K, Crawford JM, Townsend CA (2009) Production of octaketide polyenes by the 7. Kennedy DR, et al. (2007) Single chemical modifications of the C-1027 enediyne core, calicheamicin polyketide synthase CalE8: Implications for the biosynthesis of enediyne core structures. J Am Chem Soc 131:12564–12566. a radiomimetic antitumor drug, affect both drug potency and the role of ataxia- 19. Sun HH, et al. (2009) Products of the iterative polyketide synthases in 9-and 10-mem- telangiectasia mutated in cellular responses to DNA double-strand breaks. Cancer bered enediyne biosynthesis. Chem Commun 7399–7401. Res 67:773–781. 20. Ziermann R, Betlach MC (1999) Recombinant polyketide synthesis in Streptomyces: 8. Kennedy DR, Ju J, Shen B, Beerman TA (2007) Designer enediynes generate DNA Engineering of improved host strains. Biotechniques 26:106–110. breaks, interstrand cross-links, or both, with concomitant changes in the regulation 21. Kotaka M, et al. (2009) Structure and catalytic mechanism of the thioesterase CalE7 in of DNA damage responses. Proc Natl Acad Sci USA 104:17632–17637. enediyne biosynthesis. J Biol Chem 284:15739–15749. 9. Liu W, Christenson SD, Standage S, Shen B (2002) Biosynthesis of the enediyne 22. Lin SJ, Horsman GP, Chen YH, Li WL, Shen B (2009) Characterization of the SgcF antitumor antibiotic C-1027. Science 297:1170–1173. epoxide hydrolase supporting an (R)-vicinal diol intermediate for enediyne antitumor 10. Liu W, et al. (2005) The neocarzinostatin biosynthetic gene cluster from Streptomyces antibiotic C-1027 biosynthesis. J Am Chem Soc 131:16410–16417. carzinostaticus ATCC 15944 involving two iterative type I polyketide synthases. Chem 23. Zazopoulos E, et al. (2003) A genomics-guided approach for discovering and expres- – Biol 12:293 302. sing cryptic metabolic pathways. Nat Biotechnol 21:187–190. 11. Van Lanen SG, Oh TJ, Liu W, Wendt-Pienkowski E, Shen B (2007) Characterization of 24. Kieser T, Bibb MJ, Buttner MM, Chater KF, Hopwood DA (2000) Practical Streptomyces the maduropeptin biosynthetic gene cluster from Actinomadura madurae ATCC Genetics (The John Innes Foundation, Norwich, UK). 39144 supporting a unifying paradigm for enediyne biosynthesis. J Am Chem Soc 25. Maiese WM, et al. (1989) , a novel family of antitumor antibiotics: 129:13082–13094. Taxonomy, fermentation and biological properties. J Antibiot 42:558–563. 12. Ahlert J, et al. (2002) The calicheamicin gene cluster and its iterative type I enediyne 26. Lam KS, et al. (1993) Biosynthesis of esperamicin A1, an enediyne antitumor antibiotic. PKS. Science 297:1173–1176. J Am Chem Soc 115:12340–12345. 13. Gao Q, Thorson JS (2008) The biosynthetic genes encoding for the production of the 27. Liu W, Shen B (2000) Genes for production of the enediyne antitumor antibiotic dynemicin enediyne core in Micromonospora chersina ATCC53710. FEMS Microbiol C-1027 in Streptomyces globisporus are clustered with the cagA gene that encodes Lett 282:105–114. the C-1027 apoprotein. Antimicrob Agents Chemother 44:382–392.

Horsman et al. PNAS Early Edition ∣ 5of5 Downloaded by guest on September 26, 2021