Epoxomicin and Eponemycin Biosynthesis Involves gem- Dimethylation and an Acyl-CoA Dehydrogenase-Like Enzyme

Judith Zettler+,[a, b] Florian Zubeil+,[c] Andreas Kulik,[d] Stephanie Grond,[c] and Leonard Kaysser*[a, b]

Dedicated to the 60th birthday of Professor Lutz Heide

The a’,b’-epoxyketone moiety of proteasome inhibitors confers high binding specificity to the N-terminal threonine in catalytic proteasome b-subunits. We recently identified the epoxomicin and eponemycin biosynthetic gene clusters and have now conducted isotope-enriched precursor feeding studies and comprehensive gene deletion experiments to shed further light on their biosynthetic pathways. Leucine and two methyl groups from S-adenosylmethionine were readily incorporated into the epoxyketone warhead, suggesting decarboxylation of the thioester intermediate. Formation of the a’,b’-epoxyketone is likely mediated by conserved acyl-CoA dehydrogenase-like enzymes, as indicated by complete loss of epoxomicin and eponemycin production in the respective knockout mutants. Our results clarify crucial questions in the formation of epoxy- ketone compounds and lay the foundation for in vitro bio- chemical studies on the biosynthesis of this pharmaceutically important class of proteasome inhibitors.
Electrophilic functional groups are widespread in biological small molecules and comprise, for example, b-lactones and
-lactams, epoxides, carbamates, and a,b-unsaturated alde- hydes.[1] Often, these groups are essential for the biological ac- tivity of natural products targeting nucleophilic centers that are omnipresent in macromolecules of the living world. The a’,b’-epoxyketone moiety is especially intriguing, as it contains two adjacent electrophilic centers that confer specific binding
to the N-terminal catalytic threonine of proteasome subunits.[2]
First, the epoxyketone carbonyl is attacked by the b-hydroxyl of the threonine. Then, the a-amine attacks the epoxide ring to form a covalent morpholine adduct, leading to irreversible inhibition of proteolytic activity. Many important cellular pro- cesses are mediated by the proteasome, including apoptosis, cell growth, mitosis, and inflammatory mechanisms.[3] Unsur- prisingly, peptide epoxyketones have evolved as promising lead structures for the development of anticancer drugs.
Carfilzomib (Kyprolis, Onyx Pharmaceuticals; Figure 1) was the first epoxyketone proteasome inhibitor approved by the U.S. Food and Drug Administration (FDA) in 2012. The develop- ment of carfilzomib was greatly inspired by epoxomicin (Figure 1), a natural epoxyketone proteasome inhibitor pro- duced by Goodfellowiella coeruleoviolacea ATCC53904.[4] It binds to the b5-subunit of the proteasome, which exhibits chy- motrypsin-like activity, with high specificity.[5]
We recently isolated and cloned the gene clusters of epoxo- micin and eponemycin, another a’,b’-epoxyketone proteasome inhibitor, allowing the first insights into the biosynthesis of this group of molecules (Figure 1).[6] In 2015, Keller et al. reported on the macyranone (Figure 1) and Owen et al. on the clarepox- cin (Figure 1) and landepoxcin gene clusters, showing the archetypal character of the epoxomicin and eponemycin path- ways for epoxyketone formation.[7] A recent study on epone- mycin biosynthesis in Escherichia coli by Liu et al. will be dis- cussed in detail below.[8]
Both the epoxomicin (epx) and eponemycin (epn) gene clus- ters encode a linear non-ribosomal peptide synthetase/poly-

[a] J. Zettler,+ Prof. Dr. L. Kaysser
Pharmaceutical Biology, University of Ttibingen
Auf der Morgenstelle 8, 72076 Ttibingen (Germany) E-mail: [email protected]
[b] J. Zettler,+ Prof. Dr. L. Kaysser
German Centre for Infection Research (DZIF), partner site Ttibingen Auf der Morgenstelle 8, 72076 Ttibingen, (Germany)
[c] F. Zubeil,+ Prof. Dr. S. Grond
Institute of Organic Chemistry, University of Ttibingen Auf der Morgenstelle 18, 72076 Ttibingen (Germany)
[d] A. Kulik
Interfaculty Institute for Microbiology and Infection Medicine Ttibingen (IMIT)
Microbiology/Biotechnology, University of Ttibingen Auf der Morgenstelle 28, 72076 Ttibingen (Germany)
[+] These authors contributed equally to this work.
Supporting information for this article can be found under http://
ketide synthetase (NRPS/PKS) assembly line that most likely catalyzes the construction of an acylated peptide and the extension of the terminal leucine residue. Comparison of the pathways indicates the involvement of an acyl-CoA dehydro- genase (ACAD) and a cytochrome P450 (CYP) enzyme in the formation of the a’,b’-epoxyketone, as the respective genes are highly conserved.

Feeding of [methyl-13C]-l-methionine and analysis by NMR spectroscopy
Based on genetic information, we previously suggested two main possibilities regarding how the epoxyketone warhead could be built: 1) the PKS module incorporates C2 from acetate and C1 from S-adenosyl methionine (SAM), and the terminal
ChemBioChem 2016, 17, 792 – 798 792 ti 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim





















Figure 1. A) Chemical structures of proteasome inhibitors. B) Biosynthetic gene clusters of epoxomicin and eponemycin.

Figure 2. MS analysis of extracts from cultures of S. hygroscopicus

thioester or carboxylic acid is reduced by a 4-eti reduction; or 2) the PKS module incorporates C2 from acetate and C2 from SAM, followed by a spontaneous or enzyme-assisted decarbox- ylation.[6] Other possibilities such as the incorporation of a valine-derived dimethylmalonyl unit,[9] seem rather unlikely due to the presence of a methylation domain in the terminal PKS module but cannot be ruled out. In all cases, the epoxide might be installed by the conserved CYP.
In order to clarify this question, we performed feeding ex- periments with [methyl-13C]-l-methionine in the eponemycin and the epoxomicin producer wild-type strains, Streptomyces hygroscopicus ATCC53709 and G. coeruleoviolacea ATCC53904, respectively. The culture extracts were analyzed by HPLC-MS and compared to cultures grown without the isotope-enriched methionine supplement (Figure 2A, D–F). MS spectra obtained from extracts of the supplemented S. hygroscopicus cultures
ATCC53709 and G. coeruleoviolacea ATCC53904, supplemented with stable isotope-labeled precursors. Isotopic distribution and MS/MS spectrum of quasi-molecule ions A) eponemycin m/z 399.3 [M+H] + in unfed cultures,
B) eponemycin m/z 403.3 [M+H+4]+ after feeding with 13C5-l-valine, C) epo- nemycin m/z 401.3 [M+H+2]+ after feeding with 1,2-13C2-l-leucine, D) epo- nemycin m/z 401.3 [M+H+2]+ after feeding with [methyl-13C]- l-methionine, E) epoxomicin m/z 555.5 [M+H] + in unfed cultures, and F) epoxomicin m/z 558.3 [M+H+3] + after feeding with [methyl-13C]-l-methionine.
showed the m/z 399.3 [M+H] + main signal for eponemycin. The increasing signal intensity of the m/z 400.3 [M+H+1] + and m/z 401.3 [M+H+2] + isotope ions indicate the incorpora- tion of 13C-methyl from SAM into the eponemycin molecule. MS/MS spectra of eponemycin [M+H+2] + showed fragment ions m/z 307.1 and m/z 325.1, which were also found in the MS/MS spectra of eponemycin [M+H] + from non-supplement- ed cultures (Figure 2A and D). The presence of an m/z 188.0
(m/z 186.1+2) signal, however, strongly suggests the incorpo- ration of the 13C-methyl group into the epoxyketone warhead (Figure 2D and Figure S1 in the Supporting Information). To clarify the number and position of SAM-derived methyl groups in the molecule unambiguously, we isolated 1.5 mg of the iso- tope-labeled eponemycin for detailed high-resolution (HR) MS and nuclear magnetic resonance (NMR) studies.
The 13C NMR intensities of eponemycin from [methyl-13C]-l- methionine-supplemented fermentation showed a significant increase in the C-1 and C-8 signals (Figure S9). C-2 showed a very low signal pattern, which could be explained by 1J(C,C) coupling to C-1 and C-8. In order to quantify the 13C enrich- ment, specific 13C incorporation was calculated with unlabeled eponemycin isolated from cultures as a reference, according to an established method.[10] Specific 13C incorporation for C-1 (49.1, dppm = 49.6) and C-8 (17.5, dppm = 61.6), and absolute values below 0.6 for all other carbon signals, indicated exclu- sive 13C incorporation into the epoxyketone moiety.
Complete ratios for all carbon signals, specific incorporations calculated based on eponemycin and dihydroeponemycin, and 1H and 13C NMR data are given in the Supporting Information (Tables S1 and S2). In combination with the isotopic ratio, mea- sured by using HR-MS (Figures S2 and S3), the overall incorpo- ration was calculated as 22.4% unlabeled, 62.7% single, and 14.9% double 13C incorporation in C-1 and C-8 of the epoxyke- tone moiety.
Similar results were found in [methyl-13C]-l-methionine-sup- plemented cultures of G. coeruleoviolacea ATCC53904 through LC-MS analysis. Isotope ions m/z 556.5 [M+H+1] + , m/z 557.5 [M+H+2] + , and m/z 558.5 [M+H+3] + indicated the incorpora- tion of up to three methyl groups into an epoxomicin mole- cule (Figure 2E and F). MS/MS spectra of epoxomicin [M+H+3] + were inconsistent, but further LC-HR-MS/MS analy- sis displayed distinctive fragment ions with m/z 275.1886 and m/z 284.2058 that suggested the incorporation of two 13C- methyl groups into the epoxyketone warhead (Figure S4).

Feeding of [1,2-13C2]-l-leucine and [U-13C5]-l-valine
We previously speculated about an unusual biosynthetic mechanism to produce the (6-methyl)heptanoyl moiety of epo- nemycin, involving activation and C2-extension of leucine by the stand-alone adenylation-thiolation (MbtH-A-T) domain EpnJ and the FabH homologue EpnD. To explore this possibili- ty, we fed [1,2-13C2]-l-leucine to cultures of S. hygroscopicus ATCC53709. The MS spectra of extracts from the supplement- ed cultures showed a distinctive increase in the ratio of the m/z 401.3 [M+H+2] + signal intensity in comparison to MS spectra from non-supplemented cultures (Figure 2C). This strongly indicates the incorporation of one [1,2-13C2]-l-leucine molecule into eponemycin. MS/MS spectra of eponemycin [M+H+2] + displayed fragment ions with m/z 188.0 and 309.2, which suggests that the 13C2 isotopes from labeled leucine were present in the epoxyketone warhead but not in the fatty acid portion of the molecule. Typically, even-numbered iso- branched chain fatty acids are generated through successive rounds of C2 extension of valine-derived 2-methylpropanoyl-

CoA. To test this hypothesis, we supplemented cultures of S. hygroscopicus ATCC53709 with [U-13C5]-l-valine. LC-MS analy- sis revealed a strong MS signal with m/z 403.3 that corre- sponded to eponemycin [M+H+4] + (Figure 2B). The respec- tive MS/MS spectrum showed fragment ions with m/z 186.1
13C4]- intermediate into the fatty acid side chain.

Analysis of epn and epx in-frame gene deletion mutants
To obtain further information on the genetic background for eponemycin biosynthesis, we generated a set of knockout mu- tants for all genes in the postulated epn cluster, excluding only epnC (proteasome b-subunit), epnG (NRPS), and epnH (PKS). For this purpose the genes for a putative transcriptional regu- lator (epnA), a putative thioesterase (TE, epnB), a putative keto- synthase III (KS III, epnD), a putative acyl carrier protein (ACP, epnE), a putative ACAD (epnF), two putative CYPs (epnI and epnK), a putative partial NRPS module (MbtH-adenylation-thio- lation tridomain, epnJ), and a hypothetical protein (epnL) were individually replaced on fosmid epnLK01 with an apramycin resistance cassette by using the PCR targeting system.[11] In- frame deletions were obtained by the subsequent excision of the cassette by Flp-recombination. After introduction of the mutant fosmids into Streptomyces albus J1074, kanamycin-re- sistant mutants were cultivated and analyzed by LC-MS.
EpnA shows homology to LuxR-type regulators known from other secondary metabolite gene clusters.[12] The deletion of epnA resulted in the almost complete loss of the m/z 399.3 [M+H] + and m/z 401.3 [M+H] + signals at 13.5 and 13.3 min, respectively, in the base peak (BPC; Figure 3) and extracted ion











Figure 3. LC-MS analysis of epn mutant strains. Comparison of base peak chromatograms (BPCs) from culture extracts of S. albus J1074 wild-type and various derivatives containing the intact epn gene cluster or cluster versions with in-frame gene deletions. Mass peaks for eponemycin (EPN) and dihy- droeponemycin (DH-EPN) are highlighted.
chromatograms (EIC; Figure S5). The strong decrease in epone- mycin and dihydroeponemycin production suggests the role of a pathway-specific transcriptional activator for EpnA.
MS analysis of extracts from DepnB and DepnD mutant strains revealed the presence of eponemycin and dihydroepo- nemycin mass signals with m/z 399.3 [M+H] + and m/z 401.3 [M+H] + at 13.5 and 13.3 min, respectively (EIC, Figure S5). MS/
MS fragmentation confirmed the identity of the parent ions (data not shown). Notably, BPCs showed reduced production
of both molecules in both deletion mutants, yielding ti 1% of the amounts accumulated by the original heterologous pro- ducer (Figure 3). These results might indicate that EpnB (TE) and EpnD (KS III) are involved in, but not essential for, the for- mation of eponemycin and its dihydro derivative. In contrast, the DepnJ mutant completely abolished the production of both eponemycin and dihydroeponemycin (Figure 3 and Fig- ure S5), implying a crucial involvement of EpnJ (MbtH-A-T) in the construction of these two metabolites.
ACADs typically catalyze the dehydrogenation of acyl-CoA thioesters to the corresponding trans-2-enoyl-CoA by the re- duction of a flavin adenine dinucleotide (FAD) co-factor.[13] In- terestingly, conserved ACADs have been found in the epone- mycin (EpnF) and epoxomicin (EpxF) gene clusters, as well as in all other verified and postulated clusters for the biosynthesis of epoxyketone molecules.[6,7] This strongly suggests the in- volvement of these enzymes in the formation of the a’,b’-ep- oxyketone warhead. Indeed, in the extracts of the DepnF dele- tion mutant, the mass signals for eponemycin and dihydroepo- nemycin were entirely absent (Figures 3 and S5). Furthermore, we also analyzed the corresponding DepxF heterologous mutant strains containing the epoxomicin cluster, and again, could no longer detect the desired compound (m/z 555.7375 [M+H] + and 577.3572 [M+Na] + ; Figure S6). Our results there- fore show that the conserved ACADs EpnF and EpxF are essen- tial for the production of the epoxyketone proteasome inhibi- tors eponemycin and epoxomicin, respectively.
Another feature commonly found in epoxyketone gene clus- ters (e.g. for eponemycin, epoxomicin, clarepoxcin and lande- poxcin[7b] biosynthesis) are genes for CYP450 enzymes that share sequence similarities on the amino acid level of 36–88%. However, Keller et al. recently reported the gene cluster for the epoxyketone compound macyranone (Figure 1) from Cysto- bacter fuscus, which apparently lacks a CYP gene.[7a] To investi- gate the role of the CYP homologues EpnI and EpnK in the eponemycin pathway, we analyzed cell culture extracts of the respective mutant strains by LC-MS. In both mutants, produc- tion of eponemycin (m/z 399.3 [M+H] + ) and dihydroeponemy- cin (m/z 401.3 [M+H] + ) was completely abolished (Figure 3 and Figure S5). This indicates that the conserved CYPs play an important role in the eponemycin pathway. However, deletion of the epnI homologue epxC in the epoxomicin cluster resulted in an unchanged production of epoxomicin (Figure S6).

The results from our feeding studies with stable isotope-en- riched precursors showed the incorporation of l-leucine and

two methyl groups from SAM in the epoxyketone warhead. The dimethylation of polyketide intermediates by PKS C-meth- ylation domains (MTs) has been shown for other pathways.[14]
This finding implicates that one of the two carbons of the ace- tate unit introduced by the polyketide synthetase EpnH is lost in the downstream processing of the molecule. We would thus postulate that the formation of the epoxyketone warhead pro- ceeds through decarboxylation and subsequent installation of the epoxide (Figure 4A). During the preparation of our manu- script, Liu et al. published a study on the identification of the minimal gene cluster for the biosynthesis of epoxyketone pro- teasome inhibitors and came to the same conclusions.[8] They showed that in an E. coli heterologous system, only the genes for the NRPS/PKS machinery (epnG/epnH), the acyl CoA-dehy- drogenase (epnF), and the external addition of fatty acids are necessary to produce an epoxyketone compound. Feeding of isotope-enriched precursors in the same setup and MS analysis indicated that two methyl groups are incorporated and one carbon from acetate is lost by decarboxylation during epoxy- ketone biosynthesis. These findings are complementary to the results we obtained from our feeding studies in the epoxomi- cin and eponemycin native producer strains. We were able to show by NMR spectroscopy that the two methyl groups from SAM were retained as C-1 and C-8 in the epoxyketone moiety.
Moreover, we explored the role of the conserved acyl-CoA dehydrogenases EpnF/EpxF and CYP450s EpnI/EpxC by gene deletion experiments in Streptomyces heterologous expression systems. Indeed, analysis of both the DepnF and DepxF knock- out strains strongly indicates that the conserved ACADs are essentially involved in the formation of the epoxyketone war- head of proteasome inhibitors. This is in agreement with the results from Liu et al., who showed that in E. coli, only EpnF and the NRPS/PKS machinery are needed to form an epoxy- ketone.[8] Curiously, the deletion of both CYP genes epnI and epnK resulted in the abolishment of eponemycin production, whereas epoxomicin was still accumulated in the DepxC CYP mutant. Our results suggest that the conserved CYPs are not imperative for the transformation of the terminal carboxylic acid into the epoxyketone moiety but might instead partici- pate in common oxidative tailoring reactions (e.g., hydroxyl- ation; Figure 4A). This would be cohesive with the findings of Liu et al. and the fact that not all epoxyketone gene clusters encode for a CYP enzyme.[8] Notably, a detailed analysis of the DepnI and DepnK mutant strains revealed the accumulation of potential eponemycin congeners that might contain an intact epoxyketone warhead (Figure S7). Similar observations were made for the DepnB, DepnD, and DepnJ gene deletion mu- tants. LC-MS/MS analysis of culture extracts suggested the presence of putative derivatives with the a’,b’-epoxyketone but an altered fatty acid structure (Figure S7). Our feeding studies with [U-13C5]-l-valine indicate that the (6-methyl)hepta- noyl moiety is built from valine-derived 2-methylpropanoyl- CoA and acetate. One might thus speculate about a subsidiary function of EpnB, EpnD, or EpnJ in this pathway (Figure 4A).
Based on the evidence from our study and the study by Liu et al., we could potentially assume that the formation of the epoxyketone involves decarboxylation and an epoxidation re-















Figure 4. Biosynthetic model for the formation of epoxyketone proteasome inhibitors. A) Biosynthesis of eponemycin. B) Hypothetical EpnF mechanism.
action after the release of the enzyme-bound intermediate by the EpnH TE domain. Decarboxylation might occur spontane- ously, similar to what has been observed by Poust et al. for gem-dimethylated polyketide precursors,[14c] or might be enzyme-assisted. Given the nature of EpnF (and EpxF) as flavo- enzymes, it is tempting to speculate about a dehydrogena- tion/oxygenation reaction sequence that is driven by the pre- ceding decarboxylation event (Figure 4B). In this hypothetical model, turnover by EpnF would include two half-reactions, comparable to the group of FAD-dependent internal monooxy- genases that use their substrate as an electron donor instead of an external source, such as NAD(P)H.[15] First, EpnF would act as a dehydrogenase in analogy to common ACADs (reduc- tive half-reaction).[13] Then, the reduced flavin would react with molecular oxygen, allowing the transfer of an “OH+” equiva- lent to the electron-rich substrate and the generation of the desired epoxide. Elimination of water from flavin would com- plete the catalytic cycle (oxidative half-reaction). The work pre- sented herein sets the stage to further explore the role of the conserved ACADs and the validity of our biosynthetic model for a’,b’-epoxyketone formation by comprehensive in vitro bio- chemistry.
Experimental Section
Bacterial strains and general methods : Chemical, microbiological, and molecular biological agents were purchased from standard commercial sources. G. coeruleoviolacea ATCC53904, S. hygroscopi-
cus ATCC53709, S, albus J1046, and their respective derivatives were maintained and grown on either MS agar (2% soy flour, 2% mannitol, 2% agar; components purchased from Carl Roth) or TSB medium (Becton Dickinson, Heidelberg, Germany). E. coli strains were cultivated in LB medium (components purchased from Carl Roth), supplemented with appropriate antibiotics. DNA isolation and manipulations were carried out according to standard meth- ods for E. coli[16] and Streptomyces.[17]
Precursor feeding experiments : Stable isotope-labeled precursor incorporation experiments were carried out in 2ti50 mL cultures of S. hygroscopicus ATCC53709 (eponemycin producer) grown in soy- bean meal (3 g Lti 1), cottonseed (0.5 g Lti 1), glucose (3 g Lti 1), yeast extract (0.1 g Lti1), and CaCO3 (0.3 g Lti1, pH 7.0) or G. coeruleoviola- cea ATCC53904 (epoxomicin producer) grown in soluble starch (2 g Lti 1), soybean meal (1 g Lti 1), and CaCO3 (0.5 g Lti1, pH 7.0) at 28 8C. Isotope-labeled precursors were aseptically added to each culture in final concentrations of 5 mm. [1,2-13C2]-l-Leucine, [U- 13C5]-valine, and [methyl-13C]-l-methionine were purchased from Sigma–Aldrich. After six days, the cultures were extracted with EtOAc, and the extracts were analyzed with LC-MS.
Generation of gene deletion mutants in S. albus J1074 : Gene de- letion experiments were carried out in S. albus J1074, adapting a heterologous system for the production of eponemycins we es- tablished previously.[6] An apramycin resistance cassette [aac(3)IV]
was amplified from plasmid pIJ773 by PCR with primer pairs epnF_ F and epnF_R for the inactivation of epnF, epnI_F and epnI_R for the inactivation of epnI, epnK_F and epnK_R for the inactivation of epnK, epnA_F and epnA_R for the inactivation of epnA, epnB_F and epnB_R for the inactivation of epnB, epnD_F and epnD_R for the inactivation of epnD, epnE_F and epnE_R for the inactivation
of epnE, epnJ_F and epnJ_R for the inactivation of epnJ, epnL_F and epnL_R for the inactivation of epnL, epxC_F and epxC_R for the inactivation of epxC, and epxF_F and epxF_R for the inactiva- tion of epxF (for primer list, see Table S3). Genes epnA, epnB, epnD, epnE, epnF, epnI, epnJ, epnK, and epnL were replaced in E. coli BW25113/pKD46/epnLK01, and genes epxC and epxF were replaced in E. coli BW25113/pKD46/epxMS01 by using the PCR targeting system.[11] Fosmids epnLK01 and epxMS01 are pCC1FOS-based con- structs containing the complete epn gene cluster for the produc- tion of eponemycin, or the complete epx gene cluster for epoxomi- cin production, respectively. The vector backbone of merLK01 was modified by the introduction of FC31-components attP and int, an oriT, and the resistance marker neo, enabling conjugation, site-spe-
The resulting mutant fosmids were confirmed by restriction analy- sis. To avoid polar effects, the resistance cassette was removed in E. coli BT340, taking advantage of the flanking Flp/FRT recognition site.[19] Positive fosmids were screened for their apramycin sensitivi- ty and verified by restriction analysis and PCR using primer pairs Test epnA_F and Test epnA_R for DepnA, Test epnB_F and Test epnB_R for DepnB, Test epnD_F and Test epnD_R for DepnD, Test epnE_F and Test epnE_R for DepnE, EpnF Test_F and EpnF Test_R for DepnF, Test epnJ_F and Test epnJ_R for DepnJ, Test epnI_F and Test epnI_R for DepnI, Test epnK_F and Test epnK_R for DepnK, Test epnL_F and Test epnL_R for DepnL, Test epx_C_F and Test epx_C_R for DepxC, and Test epxF_F and Test epxF_R for DepxF, re- spectively (for primer list, see Table S3). Fosmids epnLK01, epnJZ01 (DepnF), epnJZ02 (DepnI), epnJZ03 (DepnK), epnJZ04 (DepnA), epnJZ05 (DepnB), epnJZ06 (DepnD), epnJZ07 (DepnE), epnJZ08 (DepnJ), epnJZ09 (DepnL), epxJZ01 (DepxC), and epxJZ02 (DepxF) were transferred into E. coli ET12567[20] and introduced into S. albus J1074 by triparental intergeneric conjugation with the help of E. coli ET12567/pUB307.[21] Kanamycin resistance clones were se- lected and designated as S. albus J1074/epnLK01 (1–3), S. albus J1074/epnJZ01 (1–3; DepnF), S. albus J1074/epnJZ02 (1–3; DepnI), S. albus J1074/epnJZ03 (1–3; DepnK), S. albus J1074/epnJZ04 (1–3; DepnA), S. albus J1074/epnJZ05 (1–3; DepnB), S. albus J1074/
epnJZ06 (1–3; DepnD), S. albus J1074/epnJZ07 (1–3; DepnE), S. albus J1074/epnJZ08 (1–3; DepnJ), S. albus J1074/epnJZ09 (1–3; DepnL), S. albus J1074/epxJZ01 (1–3; DepxC), and S. albus J1074/
epxJZ02 (1–3; DepxF).

Complementation of the DepnF and DepnI mutants: The genes epnF and epnI were amplified by PCR with the primer pairs epnA_ F_HindIII/epnA_R_SpeI and epnI_F_HindIII/epnI_R_SpeI, respective- ly (Table S3). The PCR products were cloned into the HindIII and SpeI restriction sites of the E. coli–Streptomyces shuttle vector pUWL-Apra-oriT (oriT, aac(3)IV (ApraR), bla (CarbR), PermE*), and the resulting plasmids, pJZ03 and pJZ04, were introduced into S. albus J1074/epnJZ01 and S. albus J1074/epnJZ02, respectively, by inter- generic conjugation. Both mutants restored the production of epo- nemycin (data not shown).

Heterologous production of epoxyketones : A detailed procedure for the production and analysis of epoxyketone proteasome inhibi- tors from heterologous S. albus strains was described previously.[6]
In brief, spores of S. albus J1074 (wild-type or fosmid-containing) were inoculated in 10 mL tryptic soy broth (TSB). After 1 day of in- cubation at 30 8C, the cultures were transferred (2% final concen- tration) to R5 medium [50 mL; 103 g Lti1 sucrose, 0.25 g Lti1 K2SO4, 10.12 g Lti1 MgCl2·6H2O, 10 g Lti1 glucose, 0.1 g Lti1 casaminoacids,
5g Lti1 yeast extract, 5.73 g Lti1 TES (N-tris(hydroxymethyl)methyl-2- aminoethanesulfonic acid), 80 mg Lti 1 ZnCl2, 400 mg Lti 1 FeCl3·6H2O, 20 mg Lti1 CuCl2·2H2O, 20 mg Lti 1 MnCl2·4H2O, 20 mg Lti 1

Na2B4O7·10H2O, 20 mg Lti 1 (NH4)6Mo7O24·4H2O, 50 mg Lti1 KH2PO4,
3 g Lti1 l-proline, 2.94 g Lti1 CaCl2, and 280 mg Lti1 NaOH]. After
6days of incubation at 28 8C, eponemycin derivatives were isolated from the cultures by EtOAc extraction. The residual culture extracts were dissolved in MeCN and stored at ti 20 8C until LC/MS analysis. Analysis of epoxyketone derivatives : For routine analysis, sample aliquots (5 mL) were injected onto a Nucleosil 100, C18, 3 mm HPLC column (100ti2 mm i.d. fitted with a precolumn 10ti2 mm, Dr. Maisch GmbH, Ammerbuch, Germany) coupled to a mass spec- trometer with an electrospray ionization (ESI) interface (LC/MSD Ultra Trap System XCT 6330; Agilent 1200 series; Agilent Technolo- gies). Chromatography was carried out at a flow rate of 0.4 mLminti1 with a linear gradient from 30% to 100% solvent B over 20 min (solvent A: water with 0.1% formic acid; solvent B: acetonitrile with 0.06% (v/v) formic acid). For MS analysis, ESI (posi- tive and negative ionization) was performed in Ultra Scan mode with a capillary voltage of 3.5 kV and drying gas temperature of 350 8C. To obtain HR-MS data, samples were applied to a Dionex Ultimate 3000 HPLC system (Thermo Fisher Scientific), coupled to a maXis 4G ESI-QTOF mass spectrometer (Bruker Daltonics). The following solvent composition was used to separate the analytes: 0.1% formic acid in water as solvent A and 0.06% formic acid in MeOH as solvent B, with a gradient from 10 to 100% B over 20 min, followed by 100% B for 5 min at a flow rate of 0.3 mLminti1. The separation was carried out on a Nucleoshell C18 column, 2.7 mm, 150ti2 mm (Macherey–Nagel). The ESI source was operated at a nebulizer pressure of 2.0 bar, and dry gas was set to 8.0 Lminti1 at 200 8C. MS/MS spectra were recorded in auto MS/MS mode with collision energy stepping enabled. The scan rates for full scan and MS/MS spectra were set to 1 Hz and 7 Hz, respective- ly. Sodium formate was used as internal calibrant in each analysis. Molecular formulae were calculated from monoisotopic masses by using the SmartFormula function of DataAnalysis (Bruker Dalton- ics).
Purification of eponemycin : Eponemycin was purified from 1.5 L of culture. During purification, the eponemycin content of the chromatographic fractions was detected by HR-MS. Evaporation of the EtOAc extract gave an oily crude mixture, which was eluted by flash chromatography (Chromabond Flash RS 40 SiOH column, Ma- cherey–Nagel) in a stepwise manner at a flow rate of 40 mLminti 1 with CHCl3/MeOH (100:0 for 7 min, then 95:5) as a mobile phase. Eponemycin-containing fractions were pooled (469 mg) and ap- plied to a Sephadex LH-20 (75ti2.6 cm, MeOH, 0.5 mLminti1, GE Healthcare) column. Eponemycin-containing fractions were again pooled (46 mg) and further purified by preparative HPLC (Kromasil C18, 7 mm, 250ti20 mm, H2O/MeCN [65:35], 16 mLminti 1, Dr. Maisch GmbH). UV detection was performed at 210 nm. Eponemy- cin eluted at 24.4 min and was evaporated to dryness in vacuo.
NMR analysis of eponemycin : 1H and 13C NMR spectra were re- corded on a Bruker Avance III HD 400 spectrometer (Bruker Bio- spin) operating at a proton frequency of 400.2 MHz with CDCl3 as the solvent and internal standard. All spectra were recorded at room temperature. Unequivocal assignments for all proton and 13C signals were made for eponemycin, as well as for commercial dihy- droeponemycin. Initial assignments of 1H and 13C spectra were done by using literature data, which were then confirmed by HSQC and, if required, by HMBC spectra. A tabulation of the com- plete assignments of eponemycin and dihydroeponemycin as well
13C, and HSQC spectra of eponemycin isolated from an l- [methyl-13C]methionine-supplemented culture and 13C spectra of dihydroeponemycin and unlabeled eponemycin are given in the Supporting Information (Table S1, Figures S8–S12). Prior to calcula-
tion of the specific incorporation, 13C spectra of all compounds were scaled to the same intensity by using C-2’’ as an unlabeled reference signal. All calculated incorporation rates are given in Table S2.


The authors thank Profs. Lutz Heide and Bradley S. Moore for their support and helpful scientific discussions, as well as Markus Kramer and Paul Schuler for their assistance in recording NMR spectra. This work was supported by grants from the German Re- search Foundation (DFG) to L.K. (KA 3071/4-1) and A.K.(SFB 766).

Keywords: acyl-CoA dehydrogenase · biosynthesis ·
epoxomicin · epoxyketone warhead · proteasome inhibitor

[1]M. Gersch, J. Kreuzer, S. A. Sieber, Nat. Prod. Rep. 2012, 29, 659–682.
[2]M. Groll, K. B. Kim, N. Kairies, R. Huber, C. M. Crews, J. Am. Chem. Soc. 2000, 122, 1237–1238.
[3]S. Frankland-Searby, S. R. Bhaumik, Biochim. Biophys. Acta 1825, 64–76.
[4]M. Hanada, K. Sugawara, K. Kaneta, S. Toda, Y. Nishiyama, K. Tomita, H. Yamamoto, M. Konishi, T. Oki, J. Antibiot. 1992, 45, 1746–1752.
[5]K. B. Kim, J. Myung, N. Sin, C. M. Crews, Bioorg. Med. Chem. Lett. 1999, 9, 3335–3340.
[6]M. Schorn, J. Zettler, J. P. Noel, P. C. Dorrestein, B. S. Moore, L. Kaysser, ACS Chem. Biol. 2014, 9, 301–309.
[7]a) L. Keller, A. Plaza, C. Dubiella, M. Groll, M. Kaiser, R. Muller, J. Am. Chem. Soc. 2015, 137, 8121–8130; b) J. G. Owen, Z. Charlop-Powers, A. G. Smith, M. A. Ternei, P. Y. Calle, B. V. Reddy, D. Montiel, S. F. Brady, Proc. Natl. Acad. Sci. USA 2015, 112, 4221–4226.
[8]J. Liu, X. Zhu, W. Zhang, ChemBioChem 2015, 16, 2585–2589.

[9]K. Herold, Z. Xu, F. A. Gollmick, U. Grafe, C. Hertweck, Org. Biomol. Chem. 2004, 2, 2411–2414.
[10]a) K. Biemann, Mass Spectrometry: Organic Chemical Applications, McGraw-Hill, New York, 1962 ; b) A. I. Scott, C. A. Townsend, K. Okada, M. Kajiwara, R. J. Cushley, P. J. Whitman, J. Am. Chem. Soc. 1974, 96, 8069–8080.
[11]B. Gust, G. L. Challis, K. Fowler, T. Kieser, K. F. Chater, Proc. Natl. Acad. Sci. USA 2003, 100, 1541–1546.
[12]S. Chen, X. Huang, X. Zhou, L. Bai, J. He, K. J. Jeong, S. Y. Lee, Z. Deng, Chem. Biol. 2003, 10, 1065–1076.
[13]S. Ghisla, C. Thorpe, Eur. J. Biochem. 2004, 271, 494–508.
[14]a) K. Ishida, K. Fritzsche, C. Hertweck, J. Am. Chem. Soc. 2007, 129, 12648–12649; b) A. Schenk, Z. Xu, C. Pfeiffer, C. Steinbeck, C. Hertweck, Angew. Chem. Int. Ed. 2007, 46, 7035–7038; Angew. Chem. 2007, 119, 7165–7168; c) S. Poust, R. M. Phelan, K. Deng, L. Katz, C. J. Petzold, J. D. Keasling, Angew. Chem. Int. Ed. 2015, 54, 2370–2373; Angew. Chem. 2015, 127, 2400–2403.
[15]M. M. Huijbers, S. Montersino, A. H. Westphal, D. Tischler, W. J. van Ber- kel, Arch. Biochem. Biophys. 2014, 544, 2–17.
[16]J. Sambrook, D. W. Russell, Molecular Cloning, A Laboratory Manual, Cold Spring Harbor Laboratory Press, New York, 2001.
[17]T. Kieser, M. Bibb, M. Buttner, K. Chater, D. Hopwood, Practical Strepto- myces Genetics, The John Innes Foundation, Norwich, 2000.
[18]L. Kaysser, P. Bernhardt, S. J. Nam, S. Loesgen, J. G. Ruby, P. Skewes-Cox, P. R. Jensen, W. Fenical, B. S. Moore, J. Am. Chem. Soc. 2012, 134, 11988–11991.
[19]P. P. Cherepanov, W. Wackernagel, Gene 1995, 158, 9–14.
[20]D. J. MacNeil, J. Bacteriol. 1988, 170, 5607–5612.
[21]F. Flett, V. Mersinias, C. P. Smith, FEMS Microbiol. Lett. 1997, 155, 223– 229.

Manuscript received: October 23, 2015 Accepted article published: January 20, 2016 Final article published: February 25, 2016