Total Syntheses of Cystobactamids and Structural Confirmation of Cystobactamid 919-2

: The cystobactamids are a family of antibacterial natural products with unprecedented chemical scaffolds that are active against both Gram-positive and Gram-negative pathogens. Herein, we describe the first total synthesis of cystobactamid 919-2 from three fragments. Our convergent synthesis enabled both the confirmation of the correct structure and the determination of the absolute configuration of cystobactamid 919-2.

2 also efficiently inhibits the growth of several Gram-negative pathogens,s uch as E. coli and A. baumannii. These compounds have been identified as inhibitors of bacterial type IIa topoisomerases,and the potencyof2 is comparable to or even exceeds that of ciprofloxacin, af luoroquinolone antibiotic currently used in the clinic, thus offering exciting new alternatives for type II topoisomerase inhibitors. [5] More recently,cystobactamid 919-2 was reisolated by Kim and co-workers from cultures of myxobacteria Corallococcus coralloides M23, together with coralmycins A( 5)a nd B. [6] Comparison of spectroscopic data of 2 with coralmycin A revealed ambiguities regarding the NOE correlations and optical rotation data (see below), which led to ar evision of the relative stereochemistry of the b-methoxyasparagine moiety in cystobactamid 919-2 (4). With an additional hydroxy group on the benzoic acid unit, coralmycin Ai s even more potent against Gram-negative pathogens than cystobactamid 919-2 ( Figure 1).
Theisolation of cystobactamids from the natural producer is difficult, and yields are very low.F or further development, compound supply must be sustained by an optimized biotechnological approach or scalable chemical synthesis. Both cystobactamids and coralmycins are nonribosomal peptides that consist of six unnatural amino acids,i ncluding one b-methoxyasparagine and five aromatic amino acids. [4] The para-aminobenzoic acid (PABA) chain scaffold is unique, and has not been found before in natural products. [7] Furthermore,b oth the nitro group [8] and the isopropyl ether of the 3-hydroxy-4-aminobenzoic acid moieties [9] are rare amongst natural products.C hallenges associated with the synthesis of cystobactamids include the efficient synthesis of the monomeric PA BA derivatives and their low reactivity in amide bond formation. Incorporation of the b-methoxyasparagine unit was also found to be nontrivial. Another unforeseen problem was the difficulty in separating the diastereomers at al ate stage owing to their poor solubility and tendency to aggregate.
Thea ctivated ester fragment 6 was prepared by esterification of the known acid 9 with pentafluorophenyl trifluoroacetate (CF 3 CO 2 C 6 F 5 ). [10] Thec yclic anhydride fragment 7 was prepared from readily available l-diethyl tartrate (10). Compound 10 was transformed into the azido alcohol 11 in two steps following al iterature procedure. [11] Methylation of the alcohol under almost neutral conditions (Ag 2 O/MeI) afforded the methyl ether product. Attempted hydrolysis of the ester under acidic or basic conditions led to partial epimerization. Alternatively,t he azide group was first reduced by hydrogenation to the primary amine.A cidic hydrolysis of the diester afforded the corresponding dicarboxylic acid without epimerization. Finally,t reatment of the acid with trifluoroacetic anhydride (TFAA) afforded the cyclic anhydride fragment 7 with concomitant protection of the amine as its trifluoroacetamide (Scheme 2). [12] Regioselective alkylation of 2,3-dihydroxybenzaldehyde (12)w ith 2-PrBr at the 3-OH position [13] and subsequent acetylation of the 2-OH group afforded the corresponding protected aldehyde (Scheme 3). Regioselective nitration [14] followed by oxidation of the aldehyde with Oxone gave the carboxylic acid 13. [15] Thestructure of 13 was confirmed by   X-ray crystallographic analysis of as ingle crystal (see the Supporting Information). Thes econd fragment 15 was prepared from 3-hydroxy-4-nitrobenzoic acid (14). Esterification of 14 with t-BuOH, [16] alkylation of the phenol with 2-PrI, and reduction of the nitro group afforded aniline 15. Direct coupling of acid 13 and aniline 15 under standard amide coupling conditions worked poorly.T herefore,acid 13 was first transformed into the corresponding acid chloride, and further reaction with aniline 15 under Schotten-Baumann conditions afforded the amide product in good yield. Reduction of the nitro group by hydrogenation gave aniline 16.Amidation of 16 with 4-nitrobenzoyl chloride and another hydrogenation afforded the trimeric aniline fragment 8. Removal of both the acetyl and the tert-butyl group afforded cystobactamid 507 (3). Thes pectroscopic data of synthetic 3 match those reported in the literature (Scheme 3). [4,17] With ample amounts of fragments 6-8 in hand, we investigated the fragment coupling (Scheme 4). Reaction of aniline 8 with cyclic anhydride 7 in toluene cleanly afforded a2 .8:1 regioisomeric mixture of the ring-opened carboxylic acid products. [18] Thec arboxylic acids from the ring-opening reaction were esterified with trimethyl orthoacetate to form the corresponding methyl esters 17 and 17' '. [19] Them ajor product 17' ' was separated from its regioisomer by flash column chromatography and was isolated in 69 %yield. It was identified to be the b-isomer of the ring opening, as confirmed by X-ray analysis of as ingle crystal (Scheme 4a nd the Supporting Information). [20] Apparently,t he additional methoxy group changed the regioselectivity,w hich usually favors the a-isomer in similar reactions of aspartic anhydride with anilines. [12a, 18] Remarkably,a minolysis of either methyl ester 17 or 17' ' using aqueous NH 3 in toluene afforded the same major product, primary amide 18,with concomitant removal of both the acetyl and trifluoroacetyl groups.W etherefore concluded that the reaction occurred via aspartimide intermediate 19, ak nown complication in peptide synthesis. [21] A b-selective ring opening of 19 with NH 3 then afforded 18,the structure of which was elucidated by 1 H- 13 Ca nd 1 H-15 NH MBC analysis of its acetate derivative 20 (see the Supporting Information, Figures S9-S13). Thechoice of organic cosolvent is critical for this reaction. Ther eaction became homogeneous in THF or CH 3 OH, and significant amounts of side products were formed, while the reaction became heterogeneous in toluene, and the product precipitated from the solution. Only as mall amount of side products was observed under these basic conditions.S eparation of 18 from side products by flash column chromatography on silica gel was found to be difficult owing to their similar retention factors (R f ), poor solubility, and aggregation behavior on the column. [22] It was found that trituration of the solid product mixture with hexanes/ethyl acetate (1:1) afforded analytically pure amine 18.Coupling of amine 18 with carboxylic acid 9 under amide coupling conditions (EDC,H OBT) did not afford ac lean product owing to the influence of the free phenolic hydroxy group. However,c oupling of 18 with the third fragment 6 occurred readily in DMF to give compound 21,a nd subsequent removal of the tert-butyl group with TFA [23] afforded acid 4.
Thespectroscopic data ( 1 Hand 13 CNMR;Figures S1-S3) of synthetic 4 fully match those of the natural cystobactamid 919-2 in both CD 3 OD and [D 6 ]DMSO. [4,6] TheN MR data of the b-methoxyasparagine moiety are highly characteristic and listed in Table 1.
Forf urther confirmation, the NMR spectra ( 1 Ha nd 2D) of amixed sample of synthetic and natural cystobactamid 919-2w ere also measured, and they were analyzed by HPLC coinjection. Only one set of peaks was observed ( Figures S4-S6), thus confirming the identity of the natural product. Synthetic cystobactamid 919-2 (4)a nd the natural product exhibited very similar antibacterial activities against several bacterial strains (Table 2).
TheC Ds pectrum also matched the reported one but surprisingly,t he specific rotation [+ 688 8 (c = 0.047, CH 3 OH); + 468 8 (c = 0.10, CH 2 Cl 2 /CH 3 OH 1:1); + 248 8 (c = 0.33, THF)] is of the opposite sign to the literature values [À14.38 8 (c = 0.057, CH 3 OH); [4] À14.48 8 (c = 0.08, CH 3 OH) [6] ], which originally led the Müller group,i nc ombination with NOE data, to assign the stereochemical configuration as shown for 1 and 2.K im and co-workers also observed discrepancies regarding the optical rotation data and concluded that the "determination of the absolute stereochemistry by optical rotation was difficult for cystobactamids". Even though cystobactamid 919-2 and coralmycin Ahave very similar spectroscopic data ( 1 H, 13 CNMR and CD) and proposed structures (see 4 and 5), opposite specific rotations were determined. [6] Ac areful investigation of our measurement showed that owing to the low solubility of cystobactamid 919-2 (4), trace amounts of insoluble sample greatly influenced the optical rotation. A consistent positive value was obtained only after filtration of the solution. In addition, the advanced intermediates in our synthesis also have positive specific rotation values. Therefore,w ep ropose that the absolute configuration (2S,3R)o fc ystobactamid 919-2 (4)i sc orrect as assigned by Kim and co-workers,but that the specific rotation value should be revised. TheC Ds pectrum and specific rotation (after filtration) of an authentic sample confirmed our proposal ( Figure S7). Indeed, configurational analysis of cystobactamids is not trivial and requires thorough examination. Consequently,t otal synthesis was crucial to remove any possible doubts. In summary,w eh ave completed total syntheses of cystobactamid 507 (3)a nd cystobactamid 919-2 (4). Our synthesis unequivocally established both the correct structure and absolute configuration of cystobactamid 919-2. Thec onvergent nature of our synthesis and scalable syntheses of each fragment will also enable the production of compounds for further biological studies.S yntheses of other isomeric natural products in this family and studies of their structure-activity relationships are in progress.F urther "azologization" of these natural products for photopharmacological studies will also be investigated. [24]

Conflict of interest
Theauthors declare no conflict of interest.