Enantioselective Transfer Hydrogenative Cycloaddition Unlocks the Total Synthesis of SF2446 B3: Aglycone of the Arenimycin and SF2446 Type II Polyketide Antibiotics

Abstract

The first total synthesis and structure validation of an arenimycin/SF2446 type II polyketide is described, as represented by de novo construction of SF2446 B3, the aglycone shared by this family of type II polyketides. Ruthenium-catalyzed α-ketol-benzocyclobutenone [4+2] cycloaddition, which occurs via successive stereoablation-stereoregeneration, affects a double dynamic kinetic asymmetric transformation (DYKAT) wherein two racemic starting materials combine to form the congested angucycline bay region with control of regio-, diastereo- and enantioselectivity. This work represents the first application of transfer hydrogenative cycloaddition and enantioselective intermolecular metal-catalyzed C–C bond activation in target-oriented synthesis.

Graphical Abstract

Introduction

Arenimycin A is a secondary metabolite of the marine actinomycete Salinispora arenicola (strain CNR-647) that was initially obtained from the Grand Bahama Island mangrove tunicate Ecteinascidia turbinate,^1a and is coproduced with arenimycin B from Salinispora arenicola (strain CNB-527).^1b,2 Arenimycins C and D were isolated through “mining” and heterologous expression of soil metagenomes.^1c The closely related bacterial metabolites SF2446 A1, A2, B1–B3 were isolated from the soil actinomycete Streptomyces SF2446.³ Type II polyketides of the angucycline family,⁴ arenimycins A-D and SF2446 A1, A2, B1–B3 possess a highly uncommon benzo[a]naphthacene quinone ring system,⁵ and are most notably differentiated by their N-glycoside moieties (Figure 1).⁶ Arenimycins A-D display potent activity against rifampin- and methicillin-resistant Staphylococcus aureus (MRSA).¹ Despite their promising antibacterial activity, the exceedingly low natural abundance of the arenimycins and SF2446 congeners has prohibited elucidation of their mechanism of action, yet their isolation from both marine and terrestrial microbes suggest the fundamental importance of their pharmacophore. The structural assignment of the arenimycins and SF2446 family type II polyketides has not been corroborated by total synthesis and efforts toward their de novo construction remain absent in the literature.

Figure 1.

Arenimycins A-D and SF2446 A1, A2, B1–B3 and α-ketol-benzocyclobutenone [4+2] cycloaddition.

Our laboratory has developed a suite of catalytic enantioselective C–C bond formations that convert lower alcohols to higher alcohols.⁷ The redox-economy and site-selectivity of these hydrogen auto-transfer processes has allowed diverse type I polyketide natural products to be prepared in significantly fewer steps than previously possible.⁸ In more recent work, this reactivity mode was used to inform parallel protocols for convergent type II polyketide construction, resulting in the development of several unique ruthenium-catalyzed [4+2] cycloadditions.^9,10 In particular, a powerful ruthenium-catalyzed [4+2] cycloaddition of 1,2-diols, ketols or diones with benzocyclobutenones was developed,^9d,e which provides entry to type II polyketides bearing bridgehead diol motifs, which are often challenging to install.¹¹ Furthermore, these cycloadditions are capable of forming angucycline ring systems bearing highly congested “bay regions” with control of regio-, diastereo- and enantioselectivity.^9g In the first example of enantioselective intermolecular metal-catalyzed C–C bond activation in natural product total synthesis,¹² we herewith report the total synthesis and structure validation of SF2446 B3, the aglycone common to the arenimycins and SF2446 family type II polyketides, via catalytic asymmetric α-ketol-benzocyclobutenone [4+2] cycloaddition.

Results and Discussion

Given the nuanced requirements of the α-ketol and arylcyclobutenone partners, their selection in connection with the present synthesis of SF2446 B3 is herewith described in detail (Table 1). In prior work on the development of the catalytic asymmetric α-ketol-benzocyclobutenone [4+2] cycloaddition,^9g enantioselectivity was not only dependent on the chiral ligand (DM-SEGPHOS), but was also influenced by structural features of both the benzocyclobutenone and the tetralone-derived ketol. As illustrated (Table 1, entries 1–4),^9g an essentially racemic reaction could be rendered highly enantioselective upon successive introduction of alkoxy substituents flanking the ketone moieties of the benzocyclobutenone (7a to 7b, Table 1, entries 1 and 2) and tetralone-derived α-ketol (12a to 12c, Table 1, entry 4). In the optimal case, benzocyclobutenone 7b and the α-ketol 12c, which incorporates an ortho-benzyl ethyl moiety, were converted to cycloadduct 13bc in 96% yield as 97:3 ratio of enantiomers (Table 1, entry 4). Based on these data, it was posited that the structurally related benzyl ether 12d required to access arenimycin/SF2446 core would be a competent partner for cycloaddition with benzocyclobutenone 7b. To our surprise it was not (Table 1, entry 5). As the corresponding methyl ether-containing α-ketol 12e participates in cycloaddition with benzocyclobutenone 7b (Table 1, entry 6), it became apparent the C3 benzyl ether is not well tolerated in the presence of the adjacent carbomethoxy group. The methyl ether of α-ketol 12e, however, is not an ideal protecting group for the C3 phenol of the arenimycin/SF2446 aglycone, so the related 1,3-dioxinone-containing α-ketol 12f was explored and, to our delight, could be converted to cycloadduct 13bf in good yield with high levels of regio- and stereocontrol (Table 1, entry 7).

Table 1.

Ruthenium-catalyzed [4+2] cycloaddition of benzocyclobutenones 7a-h with α-ketols 12a-g.^a

^aYields of material isolated by silica gel chromatography. Enantioselectivities were determined by chiral stationary-phase HPLC analysis. Diastereoselectivities were determined by ¹H NMR analysis of crude reaction mixtures.

^bDiol (200 mol%).

^cRu₃(CO)₁₂ (10 mol%), (R)-DM-SEGPHOS (30 mol%), 140 °C, 48 hr.

^dRu₃(CO)₁₂ (3.3 mol%), (R)-DM-SEGPHOS (10 mol%), 140 °C. ND = Not determined. See Supporting Information for further details.

Having identified α-ketol 12f as a competent partner for ruthenium-catalyzed [4+2] cycloaddition, we turned our attention to the evaluation of potential cyclobutenone partners. The naphthoquinone-containing cyclobutenones 7c and 7d would be ideal as they incorporate native structural features of SF2446 B3. However, cycloadditions of 7c and 7d were not successful due to competing catechol formation from the transient 1,2-dione derived from α-ketol 12e (which was used due to its availability at the time) (Table 1, entries 8 and 9). Speculating that the quinone moiety of 7c and 7d may be incompatible with intervening ruthenium hydrides, the cycloaddition of the corresponding hydroquinone methyl ether 7e was attempted. However, under a variety of conditions, including temperatures as high as 200 °C, hydroquinone 7e failed to react and could be recovered in high yield from the reaction mixture (Table 1, entry 10). These data suggested that the greater loss of resonance stabilization energy may preclude cycloreversion of naphthocyclobutenone 7e or may shorten the lifetime of the ketene methide that is reversibly formed to a prohibitive extent; an hypothesis corroborated by the successful cycloaddition of the angular naphthocyclobutenone 7f with cyclohexane diol 12g (Table 1, entry 11)¹³ and DFT calculations (Figure 2). For this reason, the tetrahydronaphthocyclobutenones 7g and 7h were evaluated. The cycloadditions of 7g with α-ketols 12c-12e again revealed that the C3 benzyl ether is not well tolerated in the presence of the C1 carbomethoxy group (Table 1, entries 12–14). Finally, as aromatization of the oxa-bicycle moiety of cycloadducts derived from 7g was problematic, the tetralone-derived cyclobutenone 7h was subjected to cycloaddition with α-ketol 12f. The cycloadduct 13hf was formed in 74% yield with high levels of regio-, diastereo-, and enantioselectivity (Table 1, entry 15).

Figure 2.

Cycloreversion and loss of resonance stabilization energy vis-à-vis ruthenaindanone formation.^a

^aSee Supporting Information for further details.

The synthesis of benzocyclobutenone 7h begins with the palladium-catalyzed hydroxylation¹⁴ of 6-methoxy-α-tetralone 1 followed by hydrobromic acid mediated demethylation of the phenolic ether (Scheme 1).¹⁵ Chemoselective iodination¹⁶ of the resulting resorcinol 2 provides the iodide 3, which upon regioselective triflation¹⁷ delivers sulfonate 4. Successive exposure of 4 to sodium borohydride and dibromomethane provides the aryne precursor 5.¹⁸ Finally, [2+2] cycloaddition of the aryne derived from 5 with the silyl ketene acetal 6 delivers the benzocyclobutenone 7h in a total of 8 steps (longest linear sequence, LLS) from 6-methoxy-α-tetralone 1.¹⁹

Scheme 1.

Synthesis of tetrahydronaphthocyclobutenone 7h via aryne-mediated [2+2] cycloaddition.^a

^aYields of material isolated by silica gel chromatography. See Supporting Information for further details.

The preparation of α-ketol 12 begins with condensation of dimethyl oxoglutarate and acetylacetone to provide (after O-benzylation) the isophthalate 9 (Scheme 2).²⁰ Staunton-Weinreb-type annulation of 9 using methyl acrylate followed by concomitant acid-mediated decarboxylation and debenzylation provides the 8-hydroxy-α-tetralone 10.²¹ Conversion of the 8-hydroxy-α-tetralone 10 to the 1,3-dioxinone 11²² followed by Rubottom-type oxidation²³ delivered α-ketol 12f in a total of 8 steps (LLS) from dimethyl oxoglutarate and acetylacetone.

Scheme 2.

Synthesis of α-ketol 12f.^a

^aYields of material isolated by silica gel chromatography. See Supporting Information for further details.

The conversion of benzocyclobutenone 7h and α-ketol 12f to the penultimate synthetic intermediate en route to SF2446 B3, quinone 16, is highlighted in Scheme 3. Ruthenium-catalyzed cycloaddition^9f of tetralone-derived cyclobutenone 7h with α-ketol 12f convergently assembles 13hf, which incorporates the complete carbon skeleton of the aglycone, including the congested angucycline bay region, with high levels of regio-, diastereo-, and enantioselectivity. Selective mono-methylation of the bridgehead 3°,3°-diol at C5,C18 by way of the tin ketal occurred exclusively at the less hindered C18-hydroxyl group.²⁴ Exposure of the methyl ether to hydrochloric acid in organic media resulted in concomitant hydrolysis of the 1,3-dioxane at C13,C15 and elimination of the C13 hydroxyl to form the C12,C13 olefin 14. In a 3-step sequence, the C6 TIPS ether of 14 was removed, the C12,C13 olefin was converted to the bromohydrin and the C6 and C13 hydroxyls were simultaneously oxidized to form the triketone 15. Dehydrobromination mediated by DBU provided the highly insoluble phenol, which was directly subjected to copper-catalyzed oxidation to deliver the brick-red quinone 16.²⁵

Scheme 3.

Conversion of benzocyclobutenone 7h and α-ketol 12f to the penultimate synthetic intermediate, quinone 16.^a

^aYields of material isolated by silica gel chromatography. See Supporting Information for further details.

Quinone 16 is a potential gateway to diverse arenimycins and SF2446 family type II polyketides. Direct aminoglycosylation via addition of β−1-amino-l-rhamnose 18 to quinone 16 followed by aerobic oxidation of the resulting hydroquinone was envisioned. While related oxidative amine additions to symmetric naphthoquinones have been reported,²⁶ to our knowledge, the formation of N-glycosidic linkages in this manner has not been described. Potential selectivity issues associated with this strategy involved anomerization of 1-aminorhamnose and oxidative aminoglycosylation at C11 versus C12. A model system to probe the issue of regioselective oxidative aminoglycosylation was found in juglone 17. Exposure of juglone 17 to β−1-amino-l-rhamnose 18 in the presence of a copper catalyst in a reaction vessel under an atmosphere of air gave no reaction (Scheme 4). Under microwave conditions at higher loadings of catalyst (20 mol%), the N-glycoside formed but underwent cleavage in situ to form mixtures of 2- and 3-amino juglone. Gratifyingly, at lower loadings of the copper catalyst (5 mol%), glycosidic cleavage could be suppressed and the regioisomeric C2 and C3 N-glycosides 19a and 19b could be isolated in 85% yield as a 3:1 mixture. Despite the low reaction temperatures, microwave irradiation was required to promote conversion of juglone to adducts 19a and 19b. Single crystal X-ray diffraction analysis confirmed the structural assignment of 19a as the β-rhamnoside. Having established a favorable regiochemical bias in the model system of juglone 17, the optimized conditions for direct regioselective oxidative aminoglycosylation were applied to quinone 16. However, attempted methanolysis of the 1,3-dioxinone moiety under a variety of conditions resulted in only partial cleavage of the 1,3-dioxinone moiety to form the surprisingly stable formaldehyde hemiacetal.²⁷ As 1,3-dioxinone methanolysis occurred smoothly on α-ketol 12f, it appeared that the hindered bay region of adducts derived from 16 impeded deprotection. However, under more forcing conditions, quinone 16 could be reacted with 18 to deliver the aglycone SF2446 B3 in 21% yield over the 2-step sequence as a single regioisomer. The NMR spectral data and optical rotation of synthetic SF2446 B3 matched that of the naturally occurring material, thus validating its structural assignment.

Scheme 4.

Development of direct regioselective oxidative aminoglycosylation and total synthesis of SF2446-B3.^a

^aYields of material isolated by silica gel chromatography. See Supporting Information for further details.

Conclusions

In summary, we report the first total synthesis and structure validation of an arenimycin/SF2446 type II polyketide, SF2446 B3; the aglycone shared by this family of type II polyketides. Whereas classical convergent type II polyketide total syntheses typically rely on fragment union via Hauser-Kraus-type annulations²⁸ or aryne mediated cycloadditions,²⁹ we demonstrate the utility of a unique enantioselective ruthenium-catalyzed cycloaddition for convergent type II polyketide construction: the [4+2] cycloaddition of α-ketols and benzocyclobutenones. This cycloaddition occurs via successive stereoablation-stereoregeneration, enabling a double dynamic kinetic asymmetric transformation (DYKAT) in which two racemic starting materials combine to construct the congested angucycline bay region with control of regio-, diastereo- and enantioselectivity. The synthetic challenges posed by the arenimycin/SF2446 type II polyketides also led us to develop of a novel method for the direct copper-catalyzed aerobic oxidative aminoglycosylation of quinones that occurs with high levels of regiocontrol. These studies and prior work from our laboratory contribute to a growing lexicon of catalytic methods for polyketide construction beyond premetalated reagents.