Keywords

2.1 Introduction

Continuous emergence of drug-resistant bacteria is increasingly posing a serious threat to human health. According to a report by the UK government’s O’Neill Commission, the number of deaths attributable to antimicrobial resistance (AMR) is expected to rise from current 700,000 (low estimate) to 10 million by 2050 unless appropriate countermeasures to AMR are taken [1]. Iterative chemical modifications of existing antibacterial agents have been continued to temporarily restore their activities against the bacteria that have acquired resistance, but such efforts to produce new-generation antibacterial agents have regrettably resulted in the appearance of more resistant and intractable strains of bacteria. To overcome such situations, natural product chemists have been seeking new antibiotic scaffolds with broad-spectrum antibacterial activity, novel modes of action, and low or no cross-resistance to existing antibacterial drugs [2, 3]. Amycolamicin described in this chapter is expected to become a promising bridgehead to tackle the antibiotic crisis.

Amycolamicin is an antibiotic isolated in 2009 from the culture of the soil actinomycete Amycolatopsis sp. MK575-fF5 by Igarashi and coworkers at BIKAKEN (Japan) [4]. Just after the discovery of amycolamicin, Singh et al. at Merck (USA) identified an antibacterial substance produced by the soil bacterium Kibdelosporangium sp. MA7385 and gave it the name of kibdelomycin [5]. Kibdelomycin had a surprisingly similar chemical structure and biological properties as amycolamicin, but from the distinct difference in their NMR spectra, the two natural products were considered to be different and probably diastereomeric to each other for a period of time. After some twists and turns in their structural determination [4,5,6,7,8], the structure of amycolamicin was finally assigned as 1 (Fig. 2.1) by the BIKAKEN group in 2012 through extensive spectroscopy combined with X-ray crystallographic analysis of its degradation product and some other synthetic and analytical methods [9] and that of kibdelomycin was unambiguously determined by the Merck group in 2014 to be the same as amycolamicin (1) on the basis of the cocrystal structures of kibdelomycin with its target proteins [10]. However, the question of why the NMR spectra of amycolamicin and kibdelomycin were different remained to be solved; this mystery was later settled by Li and coworkers’ synthetic study as described in Sect. 2.2.1.

Fig. 2.1
A structure has a 17-C ring system. C1 is bonded to methylene group. C4 is dashed bonded to O, which further bonded to a chair conformer of a 6-C ring that bonded to a 3-C chain, in which C3 is bonded to a 5-C ring. C5 is dashed bonded to H. C11 is bonded to isopropyl group. C12 is bonded to a 6-C ring.

Structure of amycolamicin (also known as kibdelomycin)

Full size image

This secondary metabolite produced by the actinomycetes exhibits potent antibacterial activity against an array of Gram-positive drug-resistant bacterial including methicillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant enterococci (VRE) as well as against some Gram-negative bacteria such as drug-resistant strains of Haemophilus influenzae [9]. It is also reported that amycolamicin is a strong antibiotic against two important human pathogens, Acinetobacter baumannii and Clostridium difficile [8, 11]. Amycolamicin selectively inhibits bacterial DNA synthesis through binding to bacterial type II DNA topoisomerases (DNA gyrase GyrB subunit and topoisomerase IV ParE subunit) in a unique multipoint U-shaped binding mode without affecting human topoisomerase IIα and with no apparent toxicity in mice [9,10,11]. Additionally, it does not show cross-resistance to various known DNA gyrase inhibitors such as novobiocin, coumermycin A1, and ciprofloxacin [11].

These pharmacological properties of amycolamicin (1) as a promising lead for an innovatively novel class of antibacterial agents and its unprecedented hybrid molecular architecture composed of two novel sugars [amykitanose (A) and amycolose (D)], a tetramic acid (B), a trans-decalin (C), and a dichloropyrrole carboxylic acid (E) prompted synthetic studies on this natural product, which recently culminated in the first total synthesis of 1 by Li and coworkers in 2021 [12]. Shortly after the first synthesis, two total syntheses of 1 by us [13] and by the Baran group [14] were successively disclosed in 2022. In this chapter, we first outline Li’s and Baran’s total synthesis of 1 in brief (Sect. 2.2), and then describe our total synthesis in detail (Sect. 2.3). Additionally, syntheses of the A, A/B, C, and D/E units of 1 implemented by other groups as well as by us are also presented shortly (Sect. 2.4).

2.2 Total Synthesis of Amycolamicin by the Li and Baran Groups

2.2.1 First Total Synthesis of Amycolamicin by the Li Group

The total synthesis of amycolamicin (1) by Li and coworkers is outlined in Scheme 2.1, where the carbon numbering follows that in ref 9. They first prepared N-acyl amycolose derivative 5 (D/E unit), trans-decalinoyl cyanide 6 (C unit), and N-amykitanosyl tetramic acid 7 (A/B unit) from 2, 3, and l-rhamnose 4, respectively, via key reactions written in Scheme 2.1. β-Selective glycosylation of 6 with 5 under modified Yu’s gold(I)-catalyzed N-glycosylation conditions [15] followed by C-acylation of 7 with the resulting glycoside afforded 1a (triethyl amine salt of amycolamicin), whose NMR spectra were identical to those reported by the Merck group for kibdelomycin except for redundant triethylamine signals, along with a small amount of amycolamicin. Acidic treatment of 1a gave 1, the NMR spectra of which matched those reported by the BIKAKEN group for amycolamicin. Based on these results, Li et al. revealed that kibdelomycin was a salt form of amycolamicin, which was the reason why they displayed distinct NMR spectra [12].

Scheme 2.1
A mechanism of compound 1. Compounds 2, 3, and 4 yield compounds 5, 6, and 7 through 12 steps, 9 steps, and 12 steps reactions, respectively. Compounds 5, 6, and 7 yield compound 1 a via 2-step reaction. Compound 1 a reacts reversibly in the presence of H C l slash M e O H to yield compound 1.

Outline of Li’s total synthesis of 1

Full size image

2.2.2 Total Synthesis by the Baran Group

In the total synthesis of 1 by Baran and coworkers summarized in Scheme 2.2, three key intermediates 11 (D/E unit), 12 (C unit), and 13 (A unit with l-valine residue) were prepared from furan derivative 8, Weinreb amide 9, and l-fucose 10, respectively. The trans-decalin system of 12 was constructed by an exclusively diastereoselective intramolecular Diels–Alder (IMDA) reaction. Glycosylation of 12 with 11 followed by one-carbon elongation of the resulting glycoside with S, Sʹ-dimethyl dithiocarbonate provided a β-keto thioester. The thioester intermediate was condensed with the N-glycoside 13 and the resulting β-keto amide was transformed into 1b (1ʺ-epi-amycolamicin) via the Dieckmann cyclization to install the tetramic acid ring (B unit). Treatment of 1b with aqueous formic acid gave a 4:3 equilibrium mixture of 1b and 1, the HPLC separation of which provided 1b and 1 in isolated yields of 46% and 32%, respectively. They revealed that 1b had nearly identical antibacterial activity as amycolamicin (1) and that some truncated analogs of 1 exhibited little to no activity [14].

Scheme 2.2
A mechanism of compound 1. Compounds 8, 9, and 10 yield compounds 11, 12, and 13 through 9 steps, 9 steps, and 10 steps reactions, respectively. Compounds 11, 12, and 13 yield compound 1 b via 4-step reaction. Compound 1 a reacts reversibly in the presence of H C O 2 H to yield compound 1.

Outline of Baran’s total synthesis of 1

Full size image

2.3 Total Synthesis of Amycolamicin by Our Group

2.3.1 Retrosynthetic Analysis of Amycolamicin

Our synthetic plan for amycolamicin (1) is depicted retrosynthetically in Scheme 2.3. Amycolamicin (1) would be obtained from methyl N-(β-ketoacyl)-N-glycosyl-l-valinate 14 via the Dieckmann condensation to construct the tetramic acid ring (B unit) and regioselective ring opening of the cyclic carbonate moiety with ammonia (or its appropriately protected derivative) leading eventually to the 3ʺ-acetoxy-4ʺ-carbamoyloxy portion of 1. The cyclic carbonate in 14 was expected to play not only as a protecting group of the 3ʺ, 4ʺ-diol system but also as a masked β-hydroxy carbamate structure. The N-glycosyl amide 14 was then dissected into thioester 15 and methyl N-glycosyl-l-valinate 16 with the intention of combining them by Ley’s N-acylation protocol. We envisaged that 16 might possibly be converted, regardless of its anomeric nature, into the desired α-anomer 14 in a stereoconvergent manner based on the following considerations: (1) the α-anomer 14 with the nitrogen-containing substituent at C1ʺ on the convex side of the cis-fused bicyclic ring system would be more stable than its β-anomer bearing the substituent on the concave side; and (2) the anomers of 14 would be inherently interconvertible under acidic conditions due to their N, O-acetalic nature. The N-glycoside 16 would readily be obtainable from l-fucose 10 via stereochemical inversion at the C2ʺ-position. The thioester 15, on the other hand, would be prepared by β-selective glycosylation of trans-decalinol 18 with pyranose 17 followed by the attachment of an S-tert-butyl thioacetate unit. To create the tetrasubstituted C3ʹ-stereogenic center in 17, we planned the diasteroselective addition of β-alkoxy vinyllithium reagent 19 to α,β-bisalkoxy ketone 20, which in turn would be derived through oxidation of the double bond of 21, whose enantiomer had previously been reported in the literature. For the preparation of 18, it would be appropriate to utilize the IMDA reaction of tetraenal 22, which was traced back to 2,3-dibromopropene 23 with the use of the Heck coupling and CBS reduction in mind.

Scheme 2.3
A retrosynthetic mechanism. Compound 1 undergoes Dieckmann cyclization + regioselective carbamoylation to yield compound 14, which undergoes steroconvergent N acylation to form compound 15. Compound 15 reacts with 16 to form 10. Compound 15 reacts to yield compound 17, followed by 20, 21, 22, and 23.

Retrosynthetic analysis of amycolamicin (1)

Full size image

2.3.2 Preparation of Cyclic Carbonate-Protected N-Glycosyl-l-Valine Methyl Ester 16

2.3.2.1 Initial Approach to 16

Methyl glycoside 25 (Scheme 2.4) was chosen as the key intermediate for the preparation of 16 since the glycoside had previously been synthesized from l-fucose 10 in 6 steps by Igarashi et al. in their structure determination studies on amycolamicin [9]. According to the literature, the pyranose 10 was converted into the corresponding methyl glycoside, which was then protected as its acetonide to give α-glycoside 24 in 59% yield after chromatographic purification (route A). The chemical yield of this two-step sequence was improved to 71% by using Amberlite IR-120(H) instead of HCl as the acid catalyst as well as by modifying the conditions for acetonide formation (route B) [16]. The stereochemical inversion at C2ʺ was performed by oxidation of 24 with PDC followed by reduction of the resulting ketone with LiAlH4 to give 2ʺ-epi-24 in 44% yield over two steps (route A). Utilization of DIBAL instead of LiAlH4 increased the two-step yield to 68% (route B). Methyl etherification of 2ʺ-epi-24 and subsequent acidic hydrolysis of the acetonide moiety delivered the known glycoside 25. Treatment of the diol 25 with 1,1ʹ-carbonyldiimidazole (CDI) and imidazole afforded cyclic carbonate 26 and hydrolysis of its glycosidic bond with TiBr4 in CH2Cl2/EtOAc [17] provided carbonate-protected bicyclic pyranose 27 as an anomeric mixture (α/β = 16:1). The conditions using TiBr4 for the hydrolysis of 26 was adopted based on our previous study on the hydrolysis of another methyl glycoside (see Sect. 2.4.4). Finally, N-glycosylation of methyl l-valinate with 27 under acidic conditions provided 16 as a 1:1.1 α/β anomeric mixture. The overall yield of 16 from l-fucose 10 via route B was 26% in 9 steps [18].

Scheme 2.4
A synthesis of methyl N glycosyl l valinate 16. Compound 10 reacts with catalyst A to form compound 24, which further reacts with the same catalyst to form 2 prime epi 24. The compound 2 prime epi 24 further reacts to form compound 25, followed by compounds 26, 27, and 16 via 9 steps with 26% yield.

Initial approach to methyl N-glycosyl-l-valinate 16

Full size image

2.3.2.2 Improved Approach to 16

The above-described first approach to the N-glycoside 16 from l-fucose 10 required a considerably lengthy nine-step sequence due to the use of two different protecting groups (acetonide and cyclic carbonate) for the C3ʺ/C4ʺ vicinal diol moiety, which inevitably necessitated a roundabout deprotection/reprotection manipulation, causing a modest overall yield of 26%. In our second approach to 16 shown in Scheme 2.5, only the cyclic carbonate group was utilized for the protection of the diol unit and, in addition, two one-pot processes were incorporated to improve the efficiency of the synthetic pathway.

Scheme 2.5
A synthesis route for methyl N glycosyl l valinate 16. Compound 10 reacts with catalyst A to form compound 26, which further reacts with C D I to form compound 29. The compound 29 further reacts to form compound 30, followed by compounds 31 and 16 via 5 steps with a 38% overall yield.

Improved approach to methyl N-glycosyl-l-valinate 16

Full size image

The new route commenced with direct β-selective glycosidation of the unprotected pyranose 10 with phenol in water using DMC (2-chloro-1,3-dimethylimidazolinium chloride) as a selective activator of the anomeric hydroxy group [19]. After considerable examination of reaction conditions, in which the amounts of reagents (DMC, 6 or 10 equiv; Et3N, 26 or 46 equiv), solvent (water or water/MeCN), and reaction temperature (0 or –10 °C) were varied, the best outcome was achieved when the reaction was conducted using 10 equiv of DMC and 46 equiv of Et3N in water at –10 °C, giving rise to phenyl β-glycoside 28 in 81% isolated yield (crude β/α ratio = 12:1). Use of 6 equiv of DMC resulted in the recovery of considerable amounts of 10 and the reactions in the mixed solvent (water/MeCM = 5:1 or 1:1) or at the higher temperature (0 °C) decreased the anomeric selectivity. Treatment of 28 with CDI and imidazole in CH2Cl2 gave a 11:1 mixture of desired 3ʺ,4ʺ-carbonate 29 and its regioisomer 29ʹ, the former of which was isolated chromatographically in 88% yield. In this cyclic carbonate formation, the use of triphosgene and Et3N (or pyridine) in CH2Cl2 at 0 °C drastically reduced the 29/29ʹ ratio to 2.5:1–1.5:1 and implementation of the reaction in MeCN (triphosgene, Et3N, 0 °C) reversed the ratio to 1:1.5. Stereochemical inversion at C2ʺ of 29 was performed in one pot by oxidation of 29 with IBX followed by reduction of the resulting ketone intermediate with NaBH4, delivering 30 in 82% yield (dr > 99:1). The oxidation step in this one-pot operation was problematic; exposure of 29 to various conditions including those using AZADOL/NaClO, AZADOL/PhI(OAc)2, Dess–Martin periodinane (DMP), SO3·Py, DMSO/Ac2O, PDC, sodium 2-iodobenzenesulfonate/Oxone [20], and so on resulted in the recovery of 29 or the formation of complex mixtures. The alcohol 30 was then O-methylated with MeI/Ag2O in MeCN to give 31, which was subjected to a one-pot process comprising acid hydrolysis of the glycosidic linkage and in-situ N-glycosylation with methyl l-valinate to furnish 16. This new route from 10 to 16 with no use of acetonide protection brought an improved overall yield of 38% in only five operational steps [18].

2.3.3 Preparation of TBS-Protected N-Acyl Amycolose 17

2.3.3.1 Initial Approach to 17

The key step for the preparation of 17 is the nucleophilic addition of the vinyllithium reagent generated from β-alkoxy vinyl bromide 38 to α-siloxy-β-alkoxy ketone 20 to provide tertiary alcohol 39 in a diastereoselective manner (Scheme 2.6, 20 → 39). We first prepared 21 starting from PMB-protected methyl (R)-lactate 32 by slight modification of the reaction conditions previously reported for obtaining ent-21 from ent-32 [21]. The lactate 32 was one-carbon homologated with (dimethoxyphosphoryl)methyllithium to give 33, which was then subjected to the Horner–Wadsworth–Emmons (HWE) reaction with acetaldehyde under Masamune–Roush conditions to provide alkoxy enone 34. Chelation-controlled reduction of 34 with Zn(BH4)2 afforded the allylic alcohol 21 (77% isolated yield, crude dr = 11:1), the enantiomeric excess of which was determined to be > 99:1 by 1H NMR analysis of its (R)- and (S)-MTPA esters. Contrary to our expectation based on literature precedents [22, 23], the Sharpless asymmetric epoxidation of 21 using (–)-diisopropyl tartrate (DIPT) was sluggish and delivered a mixture of desired epoxy alcohol 35 (erythro isomer) and its threo isomer in a modest diastereoselectivity of 4.8:1 probably due to an undesirable effect of the PMB-oxy substituent at the C5ʹ chiral center. The epoxy alcohol 35 was protected as TBS ether 36, which was then subjected to epoxide ring opening with sodium azide for its conversion into azido alcohol 37 using the following additives and solvents (at 100 °C, pressure bottle): (1) NH4Cl/EtOH–H2O (3:1, 18 h), (2) NH4Cl/MeO(CH2)2OH–H2O (8:1, 17 h), (3) NH4Cl/DMSO–H2O (8:1, 17 h), (4) PPTS/MeO(CH2)2OH–H2O (3:1, 10 d), (5) PPTS/1,4-dioxane–H2O (3:1, 10 d), and (6) LiClO4/MeCN (9 d) [24]. All of these conditions, however, gave mixtures of 37 and its regioisomer 37ʹ in low selectivities of 1.1:1–2.5:1, providing 37 in unsatisfactory yields of up to 37% after chromatographic purification. Use of Me3N·HCl as the additive, however, considerably improved the 37/37ʹ ratio to 4.9:1, affording 37 in an acceptable isolated yield of 57%, although the reaction required 7 days at 100 °C to go to completion. The alcohol 37 was oxidized with DMP to give ketone 20, to which the lithium anion prepared by treatment of PMB-oxy-substituted Z-vinyl bromide 38 [25, 26] with t-butyllithium was added. The resulting addition product 39 was obtained as a single diastereomer in a good yield of 84%. This outcome dovetailed nicely with the results obtained in a systematic study by Evans et al., in which anti-substituted α-(TBS-oxy)-β-(PMB-oxy)aldehyde A was converted into B with dr ≥ 99:1 upon exposure to some lithium enolates [27]. The azido alcohol 39 was transformed into 42 by the Staudinger reduction (n-Bu3P, MeOH) followed by condensation of the resulting amine 40 with known pyrrole carboxylic acid 41 [28] in 96% yield over two steps. The Z-geometry of 42 was assigned based on the NOE between the two olefinic protons as well as from their coupling constant (7.2 Hz) close to those for Z-enol ethers [29]. It is worth mentioning that the reduction step did not proceed to completion even after 3 days when Ph3P was used instead of n-Bu3P. The bis-PMB ether 42 was treated with DDQ for the purpose of obtaining 17 by oxidative removal of the two PMB groups in a simultaneous manner, but the reaction stopped after only the C5ʹ-PMB group was deprotected, affording diol 42ʹ. Use of CAN was also fruitless, providing a complex mixture. Fortunately, exposure of 42 to TFA in CH2Cl2 gave a successful outcome, furnishing 17 in 83% yield via 43; the cyclic intermediate 43 could be isolated by quenching the reaction before completion. The overall yield of 17 from 32 was 14% through 11 steps [26]. Additionally, 17 was converted into N-acyl amycolose 17ʹ (α/β = ca 1:1) by treatment with TBAF and also into its methyl α- and β-glycosides (44α and 44β, respectively) by exposure to TMSCl/MeOH [30]. All of these three compounds (17ʹ, 44α, and 44β) are known as cytotoxic degradation products of amycolamicin [7], and the structure of 44β was previously established unambiguously by X-ray crystallographic analysis [9].

Scheme 2.6
A synthesis route for T B S protected N acyl amycolose 17. Compound 32 reacts with the catalyst to form compound 33, which further reacts with M e C H O, L i C i, and T H F to form compound 34. The compound 34 further reacts to form compounds 21, 35, 36, 37, 20, 39, 40, 42, 43, 44, 17, and 17 prime.

Initial approach to TBS-protected N-acyl amycolose 17

Full size image

2.3.3.2 Improved Approach to 17

Our initial approach to 17 described above left problems in two processes: (1) modest diastereoselectivity (dr = 4.8:1) and yield (56%) in the Sharpless asymmetric epoxidation (21 → 35); and (2) insufficient regioselectivity (4.9:1) and yield (57%) as well as the very long reaction time (7 days at 100 °C) in the epoxide ring opening with NaN3 (36 → 37). To circumvent these issues, we modified the first approach as shown in Scheme 2.7. Protection of the allylic alcohol 21 followed by the Sharpless asymmetric dihydroxylation of the resulting TBS ether 45 using AD-mix-β provided diol 46 (crude dr = 98.5:1) in an excellent yield of 94%. Regioselective mono-tosylation of 46 was first attempted by its treatment with TsCl (3 equiv) and Et3N (5 equiv) in CH2Cl2 at 0 °C to room temperature. The reaction was, however, very sluggish and required 19 h to go to completion, during which the product 46ʹ gradually cyclized into tetrahydrofuran derivative 47, yielding a mixture of 46ʹ and 47 in a ratio of ca. 1:1. Upon use of pyridine instead of Et3N, the reaction was much slower and not completed even after 48 h of stirring at room temperature, delivering a mixture of 46, 46ʹ, and 47. Exposure of 46 to TsCl/n-Bu2SnO/Et3N in CH2Cl2 (rt, 12 h) [31] or to TsCl/Ag2O/KI in CH2Cl2 (4 d) [32] was also unsuccessful, resulting in almost exclusive formation of 47 or in the recovery of 46, respectively. Furthermore, silica gel column chromatographic purification also induced the cyclization of 46ʹ to 47. Fortunately, the troubles were overcome by conducting the mono-tosylation by Tanabe’s method using Me3N·HCl as the additive [33] and by performing the next Dess–Martin oxidation of the resulting intermediate 46ʹ as a one-pot operation. The reaction under Tanabe’s conditions was completed within 40 min at 0 °C (TLC monitoring) and subsequent direct addition of DMP to the reaction mixture furnished α-tosyloxy ketone 48 in a good yield of 81%. The tosylate 48 was found to be well-suited for the SN2 substitution with NaN3, providing the α-azido ketone 20 nearly quantitatively, which was transformed into 17 by the same four-step sequence as depicted in Scheme 2.6. These modifications significantly improved the overall yield of 17 from 32 from 14 to 32% without changing the number of steps [13].

Scheme 2.7
A synthesis route for T B S protected N acyl amycolose 17. Compound 32 reacts with the catalyst to form compound 45, which further reacts to form compounds 46 and 46 prime. The compounds 46 and 46 prime further react to form compound 48, which further reacts to form compound 20 and compound 17.

Improved approach to TBS-protected N-acyl amycolose 17

Full size image

2.3.4 Preparation of Trans-Decalin Aldehyde 18

The preparation of the trans-decalin aldehyde 18 utilizing the IMDA reaction as the key step is shown in Scheme 2.8. Alkylation of the dianion of dimethyl (2-oxopropyl)phosphonate with allylic bromide 23 gave 49, which was then subjected to the HWE olefination with E-crotonaldehyde under Masamune–Roush conditions to provide 50 as a 19:1 E/Z mixture in 51% yield over two steps; the moderate yield (51%) is ascribable to the formation of unidentified byproducts in the alkylation step, which proceeded in ca. 57% yield. The Heck reaction of the vinyl bromide 50 with acrolein diethyl acetal 51 leading to 52 needed an examination of reaction conditions [34, 35]. When the reaction was conducted using Pd(OAc)2 and K2CO3 in the presence of n-Bu4NBr and (o-tolyl)3P (DMF, 80 °C, 3 h), a mixture of 52 and undesired cyclopentenone derivative 53 (see the bottom of Scheme 2.8) was obtained in a ratio of ca. 1:2.2 (yield not determined), the latter of which would probably be formed through an intramolecular Heck reaction. Exposure of 50 and 51 to Pd(OAc)2/K2CO3/n-Bu4NOAc in the absence of the phosphine ligand (DMF, rt, 24 h) suppressed the formation of 53 completely, but the yield of 52 decreased to 18%. After some other experimentations, we found that the reaction performed without using any phase transfer catalyst and phosphine ligand (DMF, 40 °C, 72 h) successfully furnished 52 in a much better yield of 76% with no formation of 53. As a matter of fact, we first attempted the Heck reaction of the phosphonate 49 with the protected acrolein 51 to obtain 54, which would probably be convertible into 52 by the HWE reaction with E-crotonaldehyde. The Heck reaction between 49 and 51, however, gave a complex mixture, which prompted us to reverse the order of the two processes as described above. The CBS reduction of the ketone 52 proceeded uneventfully to give rise to 55 [36], the absolute configuration (R) and the enantiomeric excess (96%) of which were determined by the modified Mosher analysis. Since the ionic IMDA reaction of 55 in the presence of various Lewis acids (LiClO4, MgBr2, I2, InCl3, Sc(OTf)3, BF3·OEt2, Et2AlCl, etc.) to hopefully construct acetal-protected trans-decalin 56 resulted only in partial deprotection of the acetal group or in the formation of complex mixtures [37, 38], we decided to conduct the cycloaddition of the corresponding aldehyde 57, which was prepared by acidic hydrolysis of 55. To our delight, the IMDA reaction of 57 in CH2Cl2 (–20 to 0 °C) in the presence of Et2AlCl (2 equiv) furnished the desired cycloadduct 18 in 71% isolated yield over two steps with excellent diastereoselectivity [18/(18ʹ + 18ʺ + 18‴) = 96.4:3.6 (see Fig. 2.2)]. The highly preferential formation of the endo-equatorial product 18 means that this Et2AlCl-promoted IMDA reaction proceeded nearly exclusively via the endo-equatorial transition state depicted in Fig. 2.2. We are considering that one of the two equiv of Et2AlCl used should coordinate with the carbonyl oxygen and the remaining one equiv would react with the unprotected hydroxy group to form an O–Al bond [39]. The formation of the O–Al bond [probably RO–AlClEt or RO(H)–AlClEt2] is presumed to have directed the reaction to go through the endo-equatorial transition state [13]. The importance of the state of protection of the hydroxy group in the stereochemical course of this cycloaddition is apparent from the results of the following comparative experiments: (1) exposure of the MOM-protected congener of 57 (MOM-57) to Et2AlCl (1 equiv) gave a 1:2.8 mixture of MOM-18 (endo-equatorial product) and MOM-18ʹ (endo-axial product), modestly favoring the endo-axial product; and (2) treatment of the TBS-protected derivative of 57 (TBS-57) with Et2AlCl (1 equiv) delivered endo-axial product TBS-18ʹ highly preferentially (TBS-18/TBS-18ʹ = 1:18). These results obtained for the MOM- and TBS-protected substrates were consistent with those of extensive studies on dialkylaluminum chloride-promoted IMDA reactions of protected trienals 58 by Marshall et al. They revealed that the IMDA reactions of the MOM- or alkyl-protected trienals 58 gave the corresponding endo-equatorial and endo-axial cycloadducts almost non-selectively, while those of their TBS-protected congeners afforded endo-axial products in a highly selective manner [40, 41]. In addition, it should be noted that this cycloaddition could be realized only by using Et2AlCl among Lewis acids tested; the utilization of other acids (EtAlCl2, LiClO4, Me3Al, or BF3·OEt2) resulted in the formation of complex mixtures or in the recovery of the starting material 57. The directing effect of a protecting group-free hydroxy group on the stereochemical course of a dialkylaluminum chloride-promoted IMDA reaction was also observed in the preparation of 12 by Baran et al. (see Scheme 2.2) [14].

Scheme 2.8
A mechanism of trans-decalin aldehyde 18. Compound 23 reacts to yield compound 49, which further reacts to form compound 50, followed by 52, 55, and 57. Compound 57 reacts with A t 2 A l C l to form compound 18. 5 structures of compounds 53, 54, 56, 18 prime, and 58 are given below the reaction.

Preparation of trans-decalin aldehyde 18

Full size image
Fig. 2.2
A pathway of tetraenal 57. Compound 57 reacts to yield endo equatorial T S, endo axial T S, exo equatorial T S, and exo axial T S, which further react to form compound 18 endo equatorial, compound prime endo axial, compound double prime exo equatorial, and 18 triple prime exo axial.

Stereochemical course of the Et2AlCl-promoted IMDA reaction of tetraenal 57

Full size image

2.3.5 Completion of the Total Synthesis of 1 Through Coupling of the Three Segments 16, 17, and 18.

Toward the completion of the total synthesis of amycolamicin (1), we first addressed the preparation of the thioester 15 via the β-selective glycosylation of the trans-decalinol 18 with the N-acyl amycolose 17 (Scheme 2.9). With the intention of performing the glycosylation by the Schmidt protocol [42], the amycolose derivative 17 was exposed to trichloroacetonitrile and DBU in CH2Cl2 for preparing the corresponding acetimidate derivative. To our surprise, however, the product obtained in 86% yield was not the acetimidate, but instead bicyclic N,O-acetal 59 formed by nucleophilic attack of the amide nitrogen at C7ʹ to the activated anomeric carbon. To probe its possibility as a glycosyl donor, 59 and 18 were allowed to react in CH2Cl2 in the presence of MS 4 Å and various acid catalysts (BF3·OEt2, TiCl4, Cu(OTf)2, TBSOTf, PPTS, TsOH, and TfOH). Although the use of TiCl4 brought about the formation of a complex mixture, the desired glycosylation product 60 was obtained in varying yields by using the other catalysts, among which TfOH (1 equiv) gave the best result, delivering predominantly the β-anomer 60 in 67% isolated yield along with a small amount of its α-anomer (60/1ʹα-60 = 4.3:1). The stereochemistry of 60 was assigned based on the NOE between the 1ʹ-H and 5ʹ-H as well as the large J1ʹH,2ʹH and J4ʹH,5ʹH values. Recently, we achieved one-pot conversion of 17 into 60 in 64% yield via 59, which could be successfully prepared in situ by intramolecular dehydration of 17 mediated by DMC [19]. The aldehyde 60 was two-carbon elongated by its aldol reaction with S-tert-butyl thioacetate at –78 °C to afford aldol 61. In this reaction, raising the reaction temperature from –78 °C to room temperature caused partial dehydration of the aldol adduct, giving rise to a considerable amount of an α,β-unsaturated thioester. The Dess–Martin oxidation of 61 gave the β-keto thioester 15 (1.8:1 mixture of keto and enol forms), which set the stage for the pivotal step in our total synthesis of amycolamicin (1), i.e., the stereoconvergent N-acylation of the N-glycoside 16 (α-anomer/β-anomer = 1:1.1) with the thioester 15.

Scheme 2.9
A synthetic route of thioester 15. Compound 17 reacts with C l 3 C C N, D B U, and C H 2 C l 2 to yield compound 59, which further reacts in the presence of compound 18 to yield compound 60. The compound 60 reacts with M e C O S t B u, L i H M D S, and T H F to yield compounds 61 and 15.

Preparation of thioester 15

Full size image

Our expectation that the N-acylation of the N-glycoside 16 incorporating a cis-fused bicyclic carbonate system might possibly take place in a stereoconvergent manner is based on the following observations made by Sawa et al. during their structural determination of 1 (Fig. 2.3) [9]: (1) acidic methanolysis of amycolamicin (1, α-anomer) selectively cleaved the glycosidic bond between the amycolose and trans-decalin moieties to provide N-glycoside C as a 1:1 α/β anomeric mixture in 77% yield; and (2) acetonidation of the 3ʺ,4ʺ-vicinal diol portion of D (a degradation product of 1, α-anomer/β-anomer = 1.3:1) with 2,2-dimethoxypropane under acidic conditions gave acetonide E with the α/β ratio significantly increased to 11:1. We considered that the first observation by Sawa et al. indicated the presence of equilibrium between the two anomers of C under the acidic conditions and the second one would suggest that the cis-fused bicyclic nature of the sugar moiety in E might have driven the equilibrium toward the sterically less hindered α-anomer with the bulky substituent at C1ʺ on the convex side of the bicyclic ring system.

Fig. 2.3
A reaction of anomerization. Amycolamicin 1 reacts with H C l and M e O H at 50 degrees Celsius for 1 hour to yield compound C. Compound D reacts in the presence of M e 2 C O M e 2, T s O H, and D M F to yield compound E.

Partial anomerizations observed by Sawa et al.

Full size image

Beyond our expectation, the N-acylation of the anomeric mixture 16 (α/β = 1:1.1) having a cis-fused bicyclic carbonate system with the thioester 15 under modified Ley’s conditions (AgTFA, 2,6-di-tert-butylpyridine, MS 5 Å, THF, 0 °C, 45 min) [43] proceeded in an exclusively stereoconvergent manner to provide the α-anomer 14 (J1ʺH,2ʺH = 9.0 Hz) as a single anomer in 72% yield presumably via anomerization of the β-anomer of 14 to the thermodynamically more stable α-anomer 14 (Scheme 2.10). As to the stereoconvergency of this N-acylation reaction, however, there might be another possibility that the anomerization of the N-glycoside 16 preceded the N-acylation reaction, since we observed in NMR monitoring experiments that a 1:12 α/β mixture of 16 (obtained during its SiO2 chromatographic purification), on exposure to AgTFA in THF-d8 at 0 °C, changed quickly to a 1:1 α/β mixture of 16 after 20 min and reached an equilibrium (α/β = ca. 1.8:1) within 45 min regardless of the presence or absence of 2,6-di-tert-butylpyridine (see the bottom of Scheme 2.10). The α-anomer of 16 might be N-acylated quickly because its nitrogen substituent is situated on the convex side of the bicyclic ring system, while the β-anomer with the substituent on the concave side would resist the N-acylation due to severe steric hindrance and therefore might be N-acylated after swift anomerization to its α-anomer, possibly providing the α-anomer 14 preferentially. Regardless of which pathway is more plausible, the stereoconvergent N-acylation strategy described here could be a suitable option for the diastereoselective synthesis of analogous N-acyl N-glycosides.

Scheme 2.10
A pathway of amycolamicin 1. Compound 15 reacts in the presence of compound 16, which further reacts to form compound 14, followed by compounds 62 and 64. Compound 64 reacts in the presence of D D Q to yield compounds 65, 66, and 1.

Completion of the total synthesis of amycolamicin (1)

Full size image

The Dieckmann condensation of the β-keto amide 14 provided 62 with a tetramic acid ring installed, the carbonate ring of which was opened with 2,4-dimethoxybenzylamine 63 in one pot to give desired β-hydroxy carbamate 64 in 61% isolated yield from 14 along with 21% yield of its regioisomer possessing a carbamate group at C3ʺ, favoring the desired isomer 64 [44]. The stereochemistry of 64 was established based on the ROE correlations and coupling constant shown in Scheme 2.10. The N-arylmethyl group in 64 was removed by DDQ oxidation to afford β-hydroxy carbamate 65, which was acetylated in the presence of Li2CO3 to afford β-acetoxy carbamate 66 in 56% yield over two steps. The addition of Li2CO3 to the reaction mixture was essential for the successful outcome; without the salt, acetylation at the tetramic acid moiety also took place concomitantly. We also attempted direct ammonolysis of 62 into 65 by using ammonia instead of the amine 63, but the ammonolysis followed by acetylation of the resulting product gave 66ʹ (undesired regioisomer) predominantly (66/66ʹ = 1:10). Finally, removal of the TBS protecting group in 66 with TASF successfully finished the total synthesis of amycolamicin (1), the overall yield of which was 4.3% via a longest linear sequence of 19 steps from the PMB-protected methyl (R)-lactate 32.

2.4 Synthesis of Partial Structures of Amycolamicin

2.4.1 Synthesis of N-Acyl Amycolose 17ʹ by the Schobert Group

Schobert et al. reported the synthesis of N-acyl amycolose 17ʹ, which was employed as an intermediate in Li’s total synthesis of 1, in 12 steps from benzyl α-d-mannoside 67 (Scheme 2.11) [45]. The mannoside 67 was converted into 68 by a three-step sequence involving the Klemer–Rodemeyer fragmentation [46] to obtain a 2ʹ-deoxy-3ʹ-oxo intermediate. The vinyl group in 68 was utilized as a foothold to install the nitrogen functionality via epoxidation, and 6ʹ-deoxygenation was achieved by the Dang protocol [47].

Scheme 2.11
A mechanism of N acyl amycolose 17 prime. Compound 67 undergoes the Klemer Rodemeyer fragmentation Grignard reaction to yield compound 68 via a 3-step reaction, which further reacts to yield compounds 17 prime and 5 via a 9-step reaction.

Synthesis of N-acyl amycolose 17ʹ by Schobert et al.

Full size image

2.4.2 Synthesis of Trans-Decalinoyl Cyanide 6 by the Altmann and Schobert Groups

Altmann et al. synthesized Li’s trans-decalin intermediate 6 in 19 steps from chiral lactone 69 via the Me2AlCl-promoted IMDA reaction of 70a using Davies’ SuperQuat chiral auxiliary to prepare trans-decalin derivative 71a [Scheme 2.12(a)] [48, 49]. Schobert et al., on the other hand, performed an analogous IMDA reaction without using Lewis acids (70b → 71b) [Scheme 2.12(b)]. Their synthesis of 6 was performed in 17 steps from ethyl 4-iodobutanoate 72 [45].

Scheme 2.12
2 possible pathways of trans decalinoyl cyanide 6. a. Compound 69 reacts to yield compound 70 a via a 10-step reaction, which further reacts to form compounds 71 a and 6 via an 8-step reaction. b. Compound 72 reacts to form compound 70 b, followed by compounds 71 b and 6 via a 9-step reaction.

Synthesis of trans-decalinoyl cyanide 6 by Altmann et al. (a) and Schobert et al. (b)

Full size image

2.4.3 Synthesis of N-amykitanosyl Tetramic Acid 7 by Schobert Et Al. And by Us

The Schobert group also reported a formal synthesis of Li’s N-amykitanosyl tetramic acid intermediate 7 in 16 steps from l-rhamnose 4 [Scheme 2.13a] [45]. The 6-deoxypyranose 4 was converted into glycosyl o-hexynylbenzoate 73 via stereochemical inversion at C4ʺ by an oxidation/reduction sequence. The benzoate 73 had been previously transformed into 7 by Li et al. via α-selective N-glycosylation of tetramic acid derivative 74 with 73 using Yu’s gold-catalyzed N-glycosylation protocol [12, 15]. We also performed a nine-step synthesis of 7 from l-fucose 10 [Scheme 2.13(b)]. Benzyl N-glycosyl-l-valinate 75 (α/β = 1.1:1) prepared in line with the procedures depicted in Scheme 2.5 was exposed to the Bestmann’s ylide [50] to afford α-N-glycosyl tetramic acid derivative 76 as a single anomer. In this case also, the reaction took place in a stereoconvergent manner thanks probably to the cis-bicyclic nature of 75 (cf. Scheme 2.10, 15 + 16 → 14).

Scheme 2.13
2 mechanisms of N amykitanosyl tetramic acid 7. a. Compound 4 reacts to yield compound 73 via 12-step reaction, which further reacts to form compound 7 via 4-step reaction. b. Compound 10 yields compound 75 via 5-step reaction, which further reacts to form compound 76, followed by compound 7.

Synthesis of N-amykitanosyl tetramic acid 7 by Schobert et al. (a) and Kuwahara et al. (b)

Full size image

2.4.4 Synthesis of Amykitanose 79 by Us

Since amykitanose 79 itself located at the rightmost end of 1 had not been synthesized, we implemented its synthesis for future biological studies [16] (Scheme 2.14). Cyclic orthoester 77 prepared from l-fucose 10 in five steps was subjected to hydrolysis with TsOH·H2O in CHCl3 to preferentially afford 78 as a thermodynamically more stable product along with a small amount of its regioisomer (3ʺ-hydroxy-4ʺ-acetoxy derivative) (78/regioisomer = 11.5:1). In this hydrolysis, the use of CHCl3 as the solvent was essential to achieve the high regioselectivity; the reactions in THF, MeCN, AcOH/H2O, EtOAc, and toluene instead of CHCl3 resulted in low selectivity of 1.6:1, 1.6:1, 1.9:1, 3.6:1, and 3.9:1, respectively. The methyl glycoside 78 was converted into 79 by carbamoylation with trichloroacetyl isocyanate followed by hydrolysis using TiBr4 [17]. The use of aqueous TsOH or TFA for the hydrolysis mainly brought about the removal of the acetyl group.

Scheme 2.14
A mechanism of amykitanose 79. Compound 10 reacts to yield compound 77 via 5-step reaction, which further reacts with T s O H, H 2 O, and C H C l 3 to yield compound 78. The compound 78 further reacts with C l 3 C C O N C O, C H 2 C l 2, T i B r 4, and C H 2 C l 2 slash E t O A c to yield compound 79.

Synthesis of amykitanose 79 by Kuwahara et al.

Full size image

2.5 Conclusion

The total synthesis of amycolamicin 1 was achieved by combining three segments: cyclic carbonate-protected methyl N-glycosyl-l-valinate 16, TBS-protected N-acyl amycolose 17, and hydroxy trans-decalin aldehyde 18. The key steps for the preparation of 16, 17, and 18 were stereochemical inversion of a l-fucose derivative by a one-pot oxidation/reduction sequence (29 → 30), exclusively diastereoselective addition of a vinyllithium reagent to an α-siloxy-β-alkoxy ketone (20 → 39), and Et2AlCl-promoted highly diastereoselective IMDA reaction of an unprotected hydroxy tetraenal (57 → 18), respectively. The assembly of the three segments was conducted as follows: (1) β-selective glycosylation of a trans-decalinol with an N,O-acetalic glycosyl donor derived from 17 (18 + 59 → 60); (2) two-carbon elongation of the resulting glycoside to form a β-keto thioester (60 → 15); and (3) exclusively stereoconvergent N-acylation of 16 with the thioester 15 to afford an N-acyl α-N-glycoside as a single anomer (15 + 16 → 14). The total synthesis was completed by four additional steps involving the Dieckmann condensation to construct a tetramic acid ring (14 → 62) and regioselective ring opening of a cyclic carbonate with an (arylmethyl)amine (62 → 64) leading eventually to the β-acetoxy carbamate moiety of amycolamicin. The overall yield of our total synthesis of 1 was 4.3% via a longest linear sequence of 19 steps from a known PMB-protected methyl (R)-lactate (32).