Letter Cite This: Org. Lett. XXXX, XXX, XXX−XXX
pubs.acs.org/OrgLett
Solid-Phase-Based Total Synthesis and Stereochemical Assignment of the Cryptic Natural Product Aurantizolicin Ansgar Oberheide,† Sebastian Pflanze,‡ Pierre Stallforth,‡ and Hans-Dieter Arndt*,† †
Friedrich-Schiller-Universität, Institut für Organische und Makromolekulare Chemie, Humboldtstr. 10, D-07743 Jena, Germany Leibniz Institute for Natural Product Research and Infection Biology, Hans Knöll Institute (HKI), Chemistry of Microbial Communication, Beutenbergstr. 11A, D-07745 Jena, Germany
‡
Downloaded via NAGOYA UNIV on January 12, 2019 at 03:15:07 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
S Supporting Information *
ABSTRACT: The total synthesis and stereochemical assignment of the polyazole cyclopeptide aurantizolicin was achieved by connecting the solution synthesis of building blocks with solid-phase peptide synthesis. Macrothiolactonization and an aza-Wittig reaction provided the natural product macrocycle in high yield as well as key stereoisomers. NMR comparison as well as isolation of the natural product from the producer organism Streptomyces aurantiacus confirmed the presence and sequence of one L-Ile and one D-allo-Ile residue in aurantizolicin.
M
arine organisms produce a broad panel of structurally diverse metabolites that often display unique chemical properties as well as distinct biological activities.1 Especially natural products containing five-membered heterocycles such as imidazoles, oxazoles, or thiazoles receive interest with respect to their isolation, biosynthesis, bioactivity, and synthetic accessibility.2 Beyond the large class of linear azol(in)e-containing peptides (LAPs),3 several macrocyclic peptides 1,4 2,5 3,6,7 4,8 and 59,10 have been isolated and structurally characterized that feature five catenated azoles bridged by a lipophilic tripeptide unit (Figure 1). While being produced by different marine actinobacteria, all the octapeptide-derived macrocycles curacozole (2),5 urukthapelstatin A (3),6,7 mechercharstatin (4),8 and YM-216391 (5)9,10 inhibit the cell growth of different human cancer cell lines with high potency. Neither the molecular mechanism of action of these natural products nor the environmental cues triggering their production have been clarified to date. Closely related to compounds 1−5 is the smaller marthiapeptide A11 (6) that shows antibacterial activity against Gram-positive bacteria but is considerably less cytotoxic. The biosynthesis of YM-216391 (5) was shown to feature posttranslational modifications of a ribosomally synthesized octaprepeptide.12 Based on the analysis of bacterial genomes and biosynthesis genes, the structures of aurantizolicin4 (1) and curacozole5 (2) were predicted. These natural products were isolated in small amounts and structurally assigned. However, no YmG epimerase homolog had been identified,12 leaving amino acid stereochemistry unclear. Heterologous overexpression of 1 succeeded recently by using promoter exchange.13 Marfey’s analysis suggested the presence of one Dallo-Ile residue in the product such obtained. We have therefore designed a synthesis of aurantizolicin (1) that allows © XXXX American Chemical Society
Figure 1. Polyazole cyclopeptides 1−5 of the YM-216391 family and the related but smaller marthiapeptide A (6).
for the rapid modification of all α-amino acid residues (Scheme 1). Ideally, the central thiazole ring could be simplified via an intramolecular aza-Wittig reaction to the less strained14 azidothiolactone 7 that should be easily available from the ω-mercapto carboxylic acid 8. This highly modified octapeptide might be assembled by three consecutive amide couplings to identify the biazoles 9 and 11 as well as two Ile derivatives Received: December 10, 2018
A
DOI: 10.1021/acs.orglett.8b03940 Org. Lett. XXXX, XXX, XXX−XXX
Letter
Organic Letters
oxazole 17 in a two-step procedure including oxazoline formation by using diethylaminosulfur trifluoride (DAST)16 and oxidation with DBU/BrCCl3.17 Methyl ester 17 was cleanly deprotected with lithium hydroxide to give carboxylic acid 18. In order to prepare the building block for the macrocyclization chemistry, the N-Boc-thiazolidine was deprotected using TFA and the free thiol was reprotected with TrCl in TFA. Then, a ZnSO4-catalyzed diazo transfer to the free amine gave access to the advanced azide building block 9 in 66% yield over three steps. Crucial for the success of the diazo transfer was the use of ZnSO4 instead of ZnCl2.18 This sequence features a late-stage incorporation of the reactive azide functionality which was found to be labile during oxazole formation as well as to be prone to racemization and βelimination under basic conditions.14 Furthermore, the use of toxic tin reagents for the saponification of the methyl ester is avoided.19 For the synthesis of bioxazole 11, suitably protected glycine 19 and serine 20 were coupled to dipeptide 21 (90% yield, Scheme 3) and cyclodehydrated with DAST. Oxidation with
Scheme 1. Synthesis Planning for Aurantizolicin (1)
Scheme 3. Synthesis of Bioxazole 11a
10 as essential building blocks. By using solid-phase peptide synthesis (SPPS) for chain assembly, rapid and high yielding construction of the target molecule should be achieved. The synthesis of biazole 9 began with transforming Lcysteine into the fully protected thiocarboxamide 13,15 which was then combined with methyl bromopyruvate in a Hantzschtype synthesis to afford thiazole 14 in 86% yield (Scheme 2). Saponifying methyl ester 14 using LiOH provided carboxylic acid 15 (97% yield). Subsequent amide coupling with racemic β-hydroxy phenylalanine methylester hydrochloride yielded thiazolopeptide 16 as a diastereomeric mixture in 84% yield. The β-hydroxy amide 16 was then smoothly transferred into Scheme 2. Synthesis of Biazole 9a
a
EDCl = 1-ethyl-3-[3-(dimethylamino)propyl]carbodiimide hydrochloride, HOBt = 1-hydroxybenzotriazole.
DBU/BrCCl3 afforded oxazole 22 with 69% yield over both steps. The methyl ester was cleaved (LiOH), and the corresponding acid 23 was coupled with serine methyl ester hydrochloride to give β-hydroxy amide 24 in 79% yield. Bioxazole 25 was obtained in 65% yield by using DAST and DBU/BrCCl3.20 The synthesis of building block 11 was completed by saponification of the methyl ester (LiOH) and exchange of the Boc for an Fmoc protecting group for SPPS. Carboxylic acid 11 was obtained in 65% yield over three steps. With both biazole building blocks 9 and 11 in hand, the pseudo-octapeptide 8 was assembled on solid support (Scheme 4). 2-Chlorotrityl resin was selected to allow retaining the STr group upon peptide release (HFIP). Bioxazole 11 was attached to the resin at 0.60 mmol/g which resulted in a loading of 0.37 mmol/g. Excess bioxazole 11 from the resin loading was partially recovered (72% of theoretical yield). Cleavage of the terminal Fmoc groups was performed with 20% piperidine in DMF. The coupling of the two α-amino acids (Fmoc-Ile and
a
DAST = diethylaminosulfur trifluoride, DBU = 1,8-diazabicyclo[5.4.0]undec-7-ene, HBTU = O-(benzotriazole-1-yl)-1,1,3,3-tetramethyluronium, pyr = pyridine, TFAA = trifluoroacetic anhydride, Tr = trityl. B
DOI: 10.1021/acs.orglett.8b03940 Org. Lett. XXXX, XXX, XXX−XXX
Letter
Organic Letters
Scheme 4. Assembly of the Cyclization Precursor 8 by SPPS and Completion of the Total Synthesis of Aurantizolicin (1)a
a The diastereomeric cyclopeptides 26 and 27 were analogously synthesized. HFIP = 1,1,1,3,3,3-hexafluoro-2-propanol, 2,6-lutid. = 2,6dimethylpyridine, PyBOP = (benzotriazol-1-yl-oxy)tripyrrolidinophosphonium hexafluorophosphate, TFA = trifluoroacetic acid.
Fmoc-D-allo-Ile) succeeded under regular conditions for SPPS.21 A slight excess of the advanced biazole 9 (1.5 equiv) was found to be optimal for the final coupling. HFIP-induced release from resin delivered octapeptide 8 in 70% yield. Benefits for increasing amounts of 9 were minute, whereas reduced stoichiometry compromised yields (1.1 equiv: 56%). Acid-mediated detritylation of peptide 8 afforded the ωmercapto carboxylic acid as the cyclization precursor. The crucial macrocyclization proceeded smoothly.14 Slow addition of the cyclization precursor to PyBOP in moderate dilution delivered macrothiolactone 7 in 80% yield. The PPh3-mediated aza-Wittig reaction of azidothiolactone 7 formed the corresponding thiazoline, which was directly oxidized with DBU/BrCCl3 to provide aurantizolicin (1, 48% yield). The comparison of the reported 1H chemical shifts of the natural product to our synthetic sample revealed a very high similarity (average Δδ = 13.3 ppb). Compared to linear peptides for which extensive NMR studies have been performed to distinguish L-Ile and D-allo-Ile,22 changes of the configuration of amino acids embedded in a macrocycle or protein can lead to strongly altered conformations and thus individual key signatures in NMR spectra.23 To confirm the stereochemical assignment and further test our synthesis design, we synthesized the diastereomeric cyclopeptides 26 and 27 within 7 days in 22% and 17% yield respectively, starting from loaded resin (see Supporting Information (SI) for details). For these isomers either both isoleucine residues feature the L-configuration (cyclopeptide 26) or the order of LIle and D-allo-Ile is inverted (cyclopeptide 27).24 Analysis of the 1H NMR spectra of the three synthetic stereoisomers (Figure 2) revealed large deviations in the 1H chemical shifts of stereoisomers 26 and 27 when compared to reported values (average Δδ = 108 and 47.5 ppb, respectively). As expected, the configuration of the tripeptide does not strongly influence the 1H chemical shifts of the polyazole backbone. However, marked differences between the stereoisomers 26 and 27 and the naturally occurring aurantizolicin were observed for the CHα and NH signals of the tripeptide,
Figure 2. Deviations of 1H chemical shifts (ppm) between the synthetic cyclopeptides 1, 26, and 27 compared to reported values in DMSO-d6 (500 and 600 MHz, respectively). Positions 1−16 represent the tripeptide section, and positions 17−37, the polyazole backbone.
whereas the data of synthetic cyclopeptide 1 perfectly match those reported (see SI). Furthermore, 13C NMR comparison led to similar conclusions. The ample amount of material available from synthesis also allowed detecting and assigning all carbon resonances (see SI for details). Taken together, our data confirm the tripeptide sequence of the naturally occurring aurantizolicin (1) is composed of Gly-L-Ile-D-allo-Ile. For the structurally very similar curacozole (2), analogous results were obtained. The spectra of synthetic cyclopeptide 1 and of curacozole (2) coincide strikingly. The crucial CHα signal chemical shifts as well as coupling constants match up very well (see SI). Based on these data, we conclude that the configuration of the recently reported curacozole (2) is identical to that of aurantizolicin (1) and other known members of this class of natural products. In order to unequivocally confirm that synthetic aurantizolicin is identical to the natural product, we isolated it from the C
DOI: 10.1021/acs.orglett.8b03940 Org. Lett. XXXX, XXX, XXX−XXX
Letter
Organic Letters original producer S. aurantiacus.4,13 Extensive screening efforts allowed us to identify culture conditions that led to the expression of the aurantizolicin biosynthesis genes in sufficient amounts for HPLC- and LC-MS-based comparison. Mass selected LC-MS experiments as well as comparison of HPLC profiles showed that synthetic 1 and natural aurantizolicin match perfectly, whereas the diastereoisomers 26 and 27 displayed different retention times (see SI). Beyond confirmation of structural identity, our experiments show that expression of the aurantizolicin biosynthesis genes is tightly regulated. This finding prompts the question for cues necessary to trigger the production of this cytotoxic metabolite of S. aurantiacus in Nature. In summary, we completed the first total synthesis of aurantizolicin (1) and of two stereoisomers 26 and 27 by designing a rapid assembly of the cyclization precursor peptides on solid support, followed by an optimized, highyielding macrocyclization procedure. NMR data and the comparison with isolated materials from the producer strain S. aurantiacus confirm that synthetic cyclopeptide 1 is identical to the naturally occurring aurantizolicin. This efficient new synthesis format provides sufficient material for in-depth biological studies and will equally facilitate investigations of non-natural analogs. The stable synthetic L-Ile epimer suggests that epimerization does not occur spontaneously.25 Future research must show how the D-allo-Ile residue is formed during biosynthesis.
■
(2) (a) Jin, Z. Nat. Prod. Rep. 2013, 30, 869−915. (b) Jin, Z. Nat. Prod. Rep. 2016, 33, 1268−1317. (3) Arnison, P. G.; Bibb, M. J.; Bierbaum, G.; van der Donk, W. A.; et al. Nat. Prod. Rep. 2013, 30, 108−160. (4) Skinnider, M. A.; Johnston, C. W.; Edgar, R. E.; Dejong, C. A.; Merwin, N. J.; Rees, P. N.; Magarvey, N. A. Proc. Natl. Acad. Sci. U. S. A. 2016, 113, E6343−E6351. (5) Kaweewan, I.; Komaki, H.; Hemmi, H.; Hoshino, K.; Hosaka, T.; Isokawa, G.; Oyoshi, T.; Kodani, S. J. Antibiot. 2019, 72, 1−7. (6) Matsuo, Y.; Kanoh, K.; Imagawa, H.; Adachi, K.; Nishizawa, M.; Shizuri, Y. J. Antibiot. 2007, 60, 256−260. (7) Matsuo, Y.; Kanoh, K.; Yamori, T.; Kasai, H.; Katsuta, A.; Adachi, K.; Shin-ya, K.; Shizuri, Y. J. Antibiot. 2007, 60, 251−255. (8) Kanoh, K.; Matsuo, Y.; Adachi, K.; Imagawa, H.; Nishizawa, M.; Shizuri, Y. J. Antibiot. 2005, 58, 289−292. (9) Sohda, K.-y.; Hiramoto, M.; Suzumura, K.-i.; Takebayashi, Y.; Suzuki, K.-i.; Tanaka, A. J. Antibiot. 2005, 58, 32−36. (10) Sohda, K.-y.; Nagai, K.; Yamori, T.; Suzuki, K.-i.; Tanaka, A. J. Antibiot. 2005, 58, 27−31. (11) Zhou, X.; Huang, H.; Chen, Y.; Tan, J.; Song, Y.; Zou, J.; Tian, X.; Hua, Y.; Ju, J. J. Nat. Prod. 2012, 75, 2251−2255. (12) Jian, X.-H.; Pan, H.-X.; Ning, T.-T.; Shi, Y.-Y.; Chen, Y.-S.; Li, Y.; Zeng, Y.-W.; Xu, J.; Tang, G.-L. ACS Chem. Biol. 2012, 7, 646− 651. (13) Pei, Z.-F.; Yang, M.-J.; Li, L.; Jian, X.-H.; Yin, Y.; Li, D.; Pan, H.-X.; Lu, Y.; Jiang, W.; Tang, G.-L. Org. Biomol. Chem. 2018, 16, 9373−9376. (14) Schwenk, S.; Ronco, C.; Oberheide, A.; Arndt, H.-D. Eur. J. Org. Chem. 2016, 2016, 4795−4799. (15) LaMattina, J. W.; Wang, B.; Badding, E. D.; Gadsby, L. K.; Grove, T. L.; Booker, S. J. J. Am. Chem. Soc. 2017, 139, 17438−17445. (16) Burrell, G.; Evans, J. M.; Jones, G. E.; Stemp, G. Tetrahedron Lett. 1990, 31, 3649−3652. (17) Phillips, A. J.; Uto, Y.; Wipf, P.; Reno, M. J.; Williams, D. R. Org. Lett. 2000, 2, 1165−1168. (18) Loos, P.; Ronco, C.; Riedrich, M.; Arndt, H.-D. Eur. J. Org. Chem. 2013, 2013, 3290−3315. (19) Nicolaou, K. C.; Estrada, A. A.; Zak, M.; Lee, S. H.; Safina, B. S. Angew. Chem., Int. Ed. 2005, 44, 1378−1382. (20) Double cyclodehydration reaction of the tripeptide NBoc-GlySer-Ser-OMe by using DAST smoothly delivered the corresponding bioxazoline. However, the subsequent oxidation reaction turned out to be problematic, giving the desired bioxazole 26 in low yield only (20%). (21) (a) El-Faham, A.; Albericio, F. Chem. Rev. 2011, 111, 6557− 6602. (b) Kates, S. A.; Albericio, F. Solid-phase synthesis: A practical guide; CRC Press: New York, 2000. (22) Anderson, Z. J.; Hobson, C.; Needley, R.; Song, L.; Perryman, M. S.; Kerby, P.; Fox, D. J. Org. Biomol. Chem. 2017, 15, 9372−9378. (23) (a) Kikuchi, H.; Hoshikawa, T.; Fujimura, S.; Sakata, N.; Kurata, S.; Katou, Y.; Oshima, Y. J. Nat. Prod. 2015, 78, 1949−1956. (b) Dang, B.; Shen, R.; Kubota, T.; Mandal, K.; Bezanilla, F.; Roux, B.; Kent, S. B. H. Angew. Chem., Int. Ed. 2017, 56, 3324−3328. (24) The presence of D-Ile or two D-allo-Ile residues was found unlikely considering the biosynthesis logic of these natural products. This analysis was substantiated in parallel to our work by Pei et al.; see ref 13. (25) Spontaneous changes of configuration have been observed for ribosomally expressed azole-containing cyclopeptides; e.g., see ref 14 and: Schoof, S.; Arndt, H.-D. Chem. Commun. 2009, 7113.
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.orglett.8b03940. General methods; synthetic procedures and physical data; NMR spectra (PDF)
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. ORCID
Hans-Dieter Arndt: 0000-0002-0792-1422 Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS A.O. and S.P. gratefully acknowledge PhD fellowships from the Carl-Zeiss-Stiftung and the IRLS for Microbial and Biomolecular Interactions, respectively. P.S. and H.D.A. were supported by the DFG (SFB1127 ChemBioSys). P.S. is thankful to the Leibniz Association for funding. The authors thank Dr. Johannes Arp (HKI) for initial culturing experiments and the NMR platform Jena for excellent support. This work benefitted in part from equipment grants of the TMWWDG (43-5572-321-12040-12) and the DFG (INST 275/331-1 FUGG).
■
REFERENCES
(1) (a) Blunt, J. W.; Copp, B. R.; Keyzers, R. A.; Munro, M. H. G.; Prinsep, M. R. Nat. Prod. Rep. 2017, 34, 235−294. (b) Blunt, J. W.; Carroll, A. R.; Copp, B. R.; Davis, R. A.; Keyzers, R. A.; Prinsep, M. R. Nat. Prod. Rep. 2018, 35, 8−53. D
DOI: 10.1021/acs.orglett.8b03940 Org. Lett. XXXX, XXX, XXX−XXX
