Combining total synthesis and genetic engineering to probe dihydropyran formation in ambruticin biosynthesis

The ambruticins are a family of potent antifungal polyketide derived natural products isolated from the myxobacterium Sorangium cellulosum. Their unusual structures include a trisubstituted cyclopropyl group and two oxygen heterocycles, a tetrahydropyran (THP) and dihydropyran (DHP). Herein we report a flexible modular approach for the total synthesis of ambruticins which is used to prepare ambruticins F and S as well as in the first total synthesis of 20,21-dihydroambruticin F. The flexible strategy unites 3 fragments via Julia–Kocienski olefinations and provides important standards for investigation of dihydropyran formation in ambruticin biosynthesis. Cultures of wild-type S. cellulosum So ce10 produce mainly ambruticin S and the VS series of metabolites. An efficient electroporation method enabled gene knockout experiments which revealed that the ΔambP-S mutant of S. cellulosum accumulated the bisTHP polyketide 20,21-dihydroambruticin F. In contrast, the ΔambN-S mutant gave ambruticin F with the 20,21-alkene as the major metabolite confirming that AmbP and AmbO (a Rieske enzyme and flavin-dependent monooxygenase respectively) are implicated in 20,21-alkene formation. The results of feeding studies to a Sorangium strain containing only ambP and ambO are in accord with formation of the 20,21-alkene occurring prior to generation of the C3 to C7 dihydroxylated tetrahydropyran in ambruticin biosynthesis.


Introduction
The ambruticins (e.g.ambruticins F and S) are a family of polyketide derived natural products, isolated from the myxobacterium Sorangium (Polyangium) cellulosum (Scheme 1). 1,2hey exhibit potent antifungal activity against a range of fungal pathogens including Coccidioides immitis and Blastomyces dermatitidis. 3,4Their antimycotic activity originates from interaction with the high-osmolarity glycerol (HOG) protein kinase signalling pathway. 5][9][10] Global warming has exacerbated this by fuelling the worldwide spread of fungi. 11,12Research into lead compounds such as the ambruticins has been stimulated by the urgent need to develop antimycotics with novel modes of action to address antifungal resistance. 13he ambruticins share several key structural features including a tetrahydropyran (THP) and dihydropyran (DHP) with a hydrocarbon linker incorporating an unusual trisubstituted cyclopropyl ring and 4 olenic bonds.The structures of the major metabolites from S. cellulosum differ at C-5 and include an alcohol (ambruticin S) or amines such as ambruticins VS-5 and VS-4 (Scheme 1A).Ambruticin F (2), the C-5 epimer of ambruticin S (4), has been isolated as a minor metabolite, however, spectral data were not recorded. 14][17][18][19][20][21] In 2006, Reeves reported analysis of the ambruticin biosynthetic gene cluster (BGC) and proposed several unusual features of the modular biosynthetic pathway based on the results from isotopic labelling studies alongside gene knockout experiments. 22In the tailoring phase of the polyketide pathway, the THP ring is proposed to be formed by the AmbJ catalysed 6,7epoxidation of ambruticin J (1), followed by selective intramolecular attack on the oxirane ring to afford ambruticin F (2) (Scheme 1A). 22,24AmbQ and AmbN are implicated in oxidation of 2 to 5-ketoambruticin (3), and subsequent reduction to ambruticin S (4).In a parallel pathway reductive amination of ketone 3 (by AmbR) was proposed to give ambruticin VS-5 (5), which undergoes N-methylation (by AmbS) to deliver ambruticin VS-4 (6).
The biosynthetic origin of the DHP ring of the ambruticins is unusual.Hahn and co-workers revealed that the oxygen heterocycle is initially formed from linear polyketide 7 by AmbDH3 catalysed dehydration and subsequent intramolecular oxa-Michael addition (IMOMA) giving THP 8 (Scheme 1B). 23mbDH3 has broad substrate specicity as demonstrated in its use in the chemoenzymatic synthesis of the diarylheptanoid centrolobine. 25The results of gene knockout experiments led Reeves to propose that dehydrogenation of the THP to a DHP occurs later in the biosynthetic pathway, catalysed by the Rieske oxygenase AmbP and the NAD(P)/FAD-dependent oxidoreductase AmbO. 22LC-MS analysis of culture extracts of mutants of S. cellulosum where ambP or ambO were disrupted indicated that a series of 20,21-dihydroambruticins was produced at similar relative levels to each of the corresponding ambruticins (with the 20,21-alkene) found in the wild-type strain. 22Hence selective desaturation may occur at various points in the biosynthetic pathway.However, no NMR spectral data for the bisTHP products were reported.Herein we describe the total syntheses of both 20,21-dihydroambruticin F (9) and ambruticin F (2) alongside the results of gene knockout experiments, giving the rst denitive experimental evidence for desaturation catalysed by AmbP and AmbO in ambruticin biosynthesis.

Results and discussion
Total syntheses of 20,21-dihydroambruticin F, ambruticin F and ambruticin S As 20,21-dihydroambruticin F ( 9) is novel and no literature NMR data are available for ambruticin F (2), we required authentic samples of both to conrm the structures of accumulated products in our proposed gene knockout experiments. 14First the total synthesis of 20,21-dihydroambruticin F was investigated based on a exible modular approach whereby three fragments 11, 12 and 13 were to be united by two Julia-Kocienski olenations (Scheme 2). 26This synthetic strategy would then be readily adapted for the synthesis of ambruticin F by using the DHP containing sulfone 14 rather than the saturated oxygen heterocycle 11 as a coupling partner.
For the C9-C13 fragment (13), reduction of the known aldehyde 20 (ref.20) with DIBAL-H followed by a Mitsunobu reaction and oxidation with mCPBA provided the phenyl tetrazole sulfone 13 (Scheme 3B). 17Oxidation of the intermediate sulde was also achieved using molybdate and H 2 O 2 in ethanol giving similar yields of sulfone 13, but the reaction proved less amenable to scale up. 32Julia-Kocienski reaction of aldehyde 12 with sulfone 13 afforded the desired E-olen 22 in 70% yield.LiHMDS in a mixture of DMF and HMPA was required to achieve the required trans-selectivity in the reaction, which was conducted successfully on >7.5 mmol scale. 17Selective deprotection of the primary alcohol with TBAF and acetic acid in THF followed by oxidation with Dess-Martin periodinane (DMP) furnished aldehyde 10 in 73% yield over the two steps. 33he nal building block required for the total synthesis of 20,21-dihydroambruticin F (9) was tetrahydropyran 11.Prins cyclisations have been widely used in natural product synthesis and we investigated the use of a SnCl 4 mediated Prins cyclisation for the stereoselective construction of the THP ring. 34,35eaction of homoallylic alcohol 25 with 2-tert-butyldimethylsilyloxypropanal 26, gave THP 28 with all four substituents in the required equatorial position but in a disappointing 27% yield due to in situ deprotection of the silyl ether (Scheme 4).Using the more robust TBDPS protecting group gave 29 in 41% yield.Radical dechlorination of 29 followed by deprotection of the silyl ether and subsequent oxidation of the resultant alcohol gave ketone 32 in 80% yield over the 3 steps from 29. 35 An alternative approach to ketone 32 began with the known DHP 34 prepared in 7 steps and 20% overall yield from 33. 20,36 Following oxidation of 34, hydrogenation of 35 with Crabtree's catalyst afforded the desired THP fragment 32 in 89% yield as a single diastereomer (Scheme 4). 37The observed selectivity was rationalised by hydrogenation being directed to the top face of the alkene by the ketone.
Ketone 32 was coupled with known phosphorus diamide 36 under conditions developed by Hanessian in the total synthesis of ambruticin S (Scheme 5). 19The olenation provided a 15 : 1 mixture of alkene isomers which, aer silyl deprotection, were separated to afford the desired E-alkene 38 in 39% yield over the two steps.The same modular strategy was used for the synthesis of ambruticin F (2).In this case the known sulfone 14 (ref.20) was coupled to aldehyde 10 to afford alkene 41 in 38% yield as a 10 : 1 mixture of inseparable E/Z isomers (Scheme 6).Again, it was found that use of NaHMDS in a mixture of THF and HMPA in the coupling reaction gave an improved yield (80%) but the selectivity was 3 : 1 in favour of the E-alkene.Deprotection of bissilyl ether 41 with TBAF in THF followed by ester hydrolysis completed the rst total synthesis of ambruticin F (2).As NMR data of the natural product 2 were not available for comparison with the synthetic material, we aimed to convert 2 to a fully characterised metabolite.
6][17][18][19] To correlate our synthetic ambruticin F with ambruticin S, a biomimetic oxidation/reduction sequence was undertaken (Scheme 6). 22eaction of dihydroxy ester 42 with Fétizon's reagent (Ag 2 CO 3 on Celite) led to selective oxidation of the 5-axial alcohol to ketone 43 in 54% yield, whereas with DMP in DCM the ketone was isolated in only 33% yield. 14,38Reduction of ketone 43 with NaBH 4 in MeOH gave a separable mixture of 1,2-syn-diol (ambruticin F ethyl ester 42) in 35% yield and 1,2-anti-diol (ambruticin S ethyl ester) in 47% yield.Finally, hydrolysis of the ethyl ester of ambruticin S completed the total synthesis of ambruticin S (4), and all spectral data were consistent with those reported for the natural product as well as products from previous total syntheses (ESI, Fig. S5 and S6 †). 17,19,38gineering the ambruticin biosynthetic pathway We found that cultures of the wild-type strain of S. cellulosum So ce10 produce predominantly ambruticin S and the VS series of compounds, and ambruticin F could not be detected by LC-MS in culture extracts (Fig. 1A).Hence, to investigate production of both ambruticin F (2) and 20,21-dihydroambruticin F (9), it was necessary to engineer the biosynthetic pathway in S. cellulosum So ce10.
Strains of the genus Sorangium are, however, very slow growing, notoriously difficult to ferment and challenging to genetically engineer.Gene disruption of So ce10 has only been successfully achieved by Reeves and co-workers in 2006 and since then no further genetic engineering of this strain has been reported. 22In their study, target genes were disrupted through conjugation of a plasmid that could insert into the middle of the gene by a single recombination event.Mutants generated by this method could potentially undergo a second homologous recombination event and revert to the wild-type.To circumvent the challenges of conjugation and lability of mutants obtained, we developed an efficient electroporation method for transferring DNA into So ce10 cells and introduced a plasmid that contains an antibiotic selection marker anked by upstream and downstream fragments of the target gene.This superior approach led to far more stable mutant strains generated via a double crossover recombination event.
Previous studies had shown that disrupting ambQ from the ambruticin biosynthetic gene cluster led to accumulation of ambruticin F but along with ambruticin S and the VS series of compounds found in the wild-type strain, potentially due to the function of AmbQ being complemented by AmbN. 22This led us to propose that disruption of both ambQ and ambN, as well as other downstream genes, would afford ambruticin F as the major metabolite.Therefore, the DambN-S mutant strain of S. cellulosum So ce10 was constructed by replacing the continuous region of ambN, ambQ, ambR and ambS genes with a hygromycin selection marker (Fig. 2A).Fermentation of the DambN-S mutant gave ambruticin F as the major product which was isolated and characterised (Fig. 2B).The NMR data of the puried natural ambruticin F matched perfectly with those of the synthetic sample conrming the structure of the natural product (ESI, Fig. S3 and S4 †).
Reeves proposed that 20,21-dehydrogenation of THPs to DHPs in the ambruticins is catalysed by the two-component enzyme system including the Rieske oxygenase AmbP and the NAD(P)/FAD-dependent oxidoreductase AmbO and indicated that this intriguing selective desaturation may occur at various points in the biosynthetic pathway. 22To investigate whether these enzymes could indeed be implicated in the conversion of 20,21-dihydroambruticin F to ambruticin F, cultures of S. cellulosum So ce10 were grown in which the continuous region of ambP, ambO, ambN, ambQ, ambR and ambS genes (designated as the DambP-S mutant) were disrupted.Analysis of the crude extract by HPLC (Fig. 2C) revealed one major product which was isolated and NMR analysis and comparison with synthetic material conrmed that 20,21-dihydroambruticin F (9) had accumulated.Similar to the DambP-S mutant in which the ambP and ambO genes were deleted together, deletions of ambP and ambO individually from the DambN-S mutant, designated as DambN-S/DambP and DambO-S respectively, also yielded 9 as the major metabolite (ESI, Fig. S1 and S2 †).These studies fully support the important role of AmbP and AmbO as the key enzymes involved in desaturation of tetrahydropyrans leading to 20,21-alkenes in ambruticin biosynthesis.
In a parallel experiment, deletions of the jerP and jerO genes were carried out in the closely related jerangolid biosynthetic pathway in S. cellulosum So ce307 (Fig. 3A).Both JerP and JerO share 78% sequence identity to AmbP and AmbO. 22Compared with the production of jerangolid A (44) by wild-type S. cellulosum So ce307, both the DjerP and DjerO mutants produced jerangolid H (45) lacking the 13,14-double bond as the major compound (Fig. 3B).
With trisubstituted tetrahydropyrans 20,21-dihydroambruticin F (9) and jerangolid H (45) in hand, we next turned to biotransformation assays with AmbP and AmbO.Despite various attempts, no clear turnover was observed with either puried enzymes of AmbP and AmbO or their expressing E. coli whole cells.We therefore further engineered the DambN-S mutant strain and deleted all other ambruticin biosynthetic genes except ambP and ambO, thus generating a Sorangium strain that only contains ambP and ambO at the amb BGC region (designated as So ce10-AmbPO).Interestingly, feeding jerangolid H to this Sorangium strain showed conversion to jerangolid A (Fig. 4A), whereas 20,21-dihydroambruticin F (9) was recovered unchanged in an analogous feeding experiment with So ce10-AmbPO (Fig. 4B).These results are inconsistent with the original proposal that the desaturation step may occur late in the biosynthetic pathway. 22Bearing in mind that ambruticin J (1) with the 20,21-alkene has been isolated from the DambJ mutant, 22 desaturation is now proposed to happen at an earlier stage of the biosynthetic pathway, most likely before the western THP ring is formed.Note: During the review process of this manuscript, Guth et al. reported reconstitution of the activity of JerP and JerO by extensive optimisation of the whole-cell biotransformation conditions in E. coli, similar conversion of jerangolid H to jerangolid A was observed. 39

Conclusions
In summary, we have completed the rst total synthesis of 20,21-dihydroambruticin F (9), whereby three fragments were combined by two Julia-Kocienski olenations.The two THP   rings were constructed on multi-gram scales by a biomimetic epoxidation-cyclisation strategy for the C1-C8 fragment and a Prins cyclisation for the second ring.The modular approach was readily adapted to allow completion of the rst total synthesis of ambruticin F (2) which in turn was converted to ambruticin S (4).These synthetic studies provided key standards to enable investigation of dihydropyran formation in ambruticin biosynthesis.Cultures of wild-type S. cellulosum So ce10 produced mainly ambruticin S and the VS series of metabolites.A method was developed to carry out gene knockout experiments reliably which revealed that the DambP-S mutant of S. cellulosum accumulated the novel bisTHP polyketide 20,21-dihydroambruticin F whilst the DambN-S mutant gave ambruticin F with the 20,21-alkene as the major metabolite.Similar results were obtained when ambP and ambO were deleted individually in these strains.These studies conrm that AmbP and AmbO (a Rieske enzyme and avin-dependent monooxygenase respectively) are implicated in 20,21-alkene formation.Furthermore, the results of feeding studies with Sorangium strains containing only ambP and ambO are in accord with desaturation occurring prior to formation of the C3-C7 tetrahydropyran in ambruticin biosynthesis, laying the foundation for future investigations into the timing and mechanism of this intriguing biotransformation.
Scheme 5. Deprotonation of sulfone 11 with 1.2 equivalents of KHMDS in DME, followed by treatment with 1.2 equivalents of aldehyde 10 gave an inseparable mixture of alkene isomers (E/Z = 10 : 1) in 40% yield.The use of NaHMDS in a mixture of THF and HMPA resulted in an improved yield of 78% for the required coupling but reduced the stereoselectivity (E/Z = 3 : 1).A screen of further solvents and bases failed to improve either the yield or selectivity.Silyl deprotection of 39 with TBAF followed by hydrolysis of the ethyl ester gave the target, 20,21dihydroambruticin F(9).The same modular strategy was used for the synthesis of ambruticin F (2).In this case the known sulfone 14 (ref.20)was coupled to aldehyde 10 to afford alkene 41 in 38% yield as a 10 : 1 mixture of inseparable E/Z isomers (Scheme 6).Again, it was found that use of NaHMDS in a mixture of THF and HMPA in the coupling reaction gave an improved yield (80%) but the selectivity was 3 : 1 in favour of the E-alkene.Deprotection of bissilyl ether 41 with TBAF in THF followed by ester hydrolysis completed the rst total synthesis of ambruticin F (2).As NMR

Fig. 2 (
Fig. 2 (A) Organisation of the ambruticin biosynthetic gene cluster where genes disrupted in this study are highlighted in red.(B) HPLC traces of the DambN-S mutant of S. cellulosum So ce10 compared with synthetic standard of ambruticin F (2). (C) HPLC traces of the DambP-S mutant of S. cellulosum So ce10 compared with synthetic standard of 20,21-dihydroambruticin F (9).