A Short Asymmetric Route to the Bromophycolide A and D Skeleton

Abstract

An asymmetric synthesis of the Bromophycolide D ring system has been achieved in 7 steps from a known geranylgeranylated benzoate, via bromonium-promoted transannular cyclization of a macrocyclic intermediate.

In 2005, Kubanek and coworkers isolated bromophycolides A and B (Figure 1) from the Fijian red algae Callophycus Serratus.^1a Although instantly appealing to the organic chemist's eye, these novel structures are the first macrocyclic halogenated terpene-benzoate structures that have been isolated. Additionally, they and subsequently-discovered bromophycolides C-Q show interesting biological activity, including growth inhibition of resistant bacterial strains (MRSA and VREF), anti-tumor and anti-malarial activity.¹

Figure 1. Bromphycolides A (1, Δ^6,19), B (2) and D (3, Δ^19,23).

No report of synthetic approaches to these molecules has yet appeared, perhaps because their multiple halogenated stereocenters make the bromophycolides intimidating targets. However, the stereochemical pattern of bromination suggested to us that a concise strategy might be feasible (Scheme 1). As shown in structure 4, not all stereocenters in 3 are derived from brominations on the same alkene face. This leads to the biosynthetic question of whether 1) multiple brominating enzymes are responsible for conferring opposite selectivity for the various positions, or 2) a single enzyme brominates multiple positions, but is capable of brominating either face when the substrate is heavily biased. We imagined that if macrocycle such as 5 were the biosynthetic precursor to 3, alkene positions 1,2 and 3 might be well enough differentiated to undergo regio- and diastereo-selective reactions with a bromonium source even outside the context of an enzyme pocket.² Although 5 is a 19-membered macrocycle, numerous elements limit its conformational flexibility. Three E-configured double bonds define arrays of four coplanar carbon atoms with locked 180° torsion angles and the benzoate contributes another coplanar array of five atoms. Additionally, the large t-butyl-like substituent should prefer a pseudo-equatorial orientation. Examination of models suggested that intermediate 5 should undergo regio- and diastereoselective transannular cyclization between alkenes 1 and 2 (TS 6) much more readily than between alkenes 2 and 3 (TS 6′) to accommodate a chair transition state and all of the above-mentioned constraints. If successful, this strategy would provide a concise route to the desired ring system 7 (and not 7′).

Scheme 1. Stereochemical Analysis and Synthetic Proposal.

We thus set about to synthesize transannular cyclization substrate 5 (Scheme 2). Known geranylgeranylbenzoate 8 was prepared in 3 steps from commercially available starting materials.³ The first asymmetric center was conveniently installed by Sharpless asymmetric dihydroxylation. To address the regiochemical problem of differentiating the “terminal” trisubstituted alkene from the other three, we employed ligand 9, developed specifically for this purpose by Corey.⁴ By running the reaction to ∼70% conversion, we were able to isolate the desired regioisomer 10 in 49% yield (71% brsm) and 92 % ee.⁵

Scheme 2. Preparation of Substrate 4.

The R- alcohol of 10 was inverted in the process of installing the tertiary bromide. Thus, mesylation of the secondary alcohol of 10 and ring closure with K₂CO₃ cleanly afforded epoxide 11. Saponification at the optimal temperature of 65 °C cleanly afforded acid 12 without harming the epoxide. There was precedent suggesting that MgBr₂·Et₂O would open the epoxide of 12 regio-selectively to give tertiary bromide 13.⁶ However, we anticipated competing polyene cyclizations upon lewis acid activation of the epoxide.⁷ Indeed, addition of a full equivalent of Bu₄NBr was necessary to supress polyene cyclizations – presumably bromide in high concentrations can compete with intramolecular alkene attack on the activated epoxide. Under these conditions, we could obtain bromohydrin 13 in good yield and regioselectivity.

Despite the hindered nature of the secondary neopentyl alcohol of 13 and the observed base-sensitivity of its vicinal bromohydrin,⁸ Shiina macrolactonization⁹ proceeded in high yield.¹⁰ The resulting macrocycle 5 crystallized during storage at -20 °C, and X-ray crystal structure determination confirmed its S- absolute configuration.¹¹

With facile access to hundreds of milligrams of 5, we investigated bromonium-initiated transannular cyclization (Scheme 3). After significant experimentation,¹² we found that 1.1 equivalents of Snyder's recently reported reagent 14 (bromodiethylsulfonium bromopentachloroantimonate, BDSB)¹³ in 1 M LiClO₄/Et₂O¹⁴ afforded a 19% combined isolated yield of the desired products 7 and 15. NMR and LC analysis of the crude product showed it to be a ∼9:3:2:1 mixture of 16:7:15:17. The major product, 16, however, invariably decomposed during normal and reverse-phase chromatography. Because we suspected 16 was an allylic bromide, we selectively solvolyzed it by treatment of the crude product mixture with methanol. From the solvolyzed mixture we isolated and identified five compounds: 7 and 15, as well as moderately stable allylic bromide 17 and 32% of 18a/b, the two major new compounds produced during methanolysis. Since we could not characterize 16 as a pure compound, its structure is tentatively assigned based on its known conversion to 18a/b. The reaction of 5 with 14 evidently proceeds primarily through bromonium intermediate 19 (Scheme 3). Likely due to geometry, attack on this bromonium by the alkene (black arrow) is not fast enough to compete with loss of a proton from either of two positions (red or blue arrow), giving rise to 16 and 17 The superiority of 14 compared with brominating reagents such as NBS is likely due to the absence of any basic leaving group that could accelerate processes leading to 16 and 17.¹⁵ In the bromonium polyene cyclization literature, ¹⁶ low yields are typical, likely due to the prevalence of deprotonation, although 14 has been shown to improve yields in unconstrained systems.¹³

Scheme 3. Transannular Cyclization of 4.

Careful NMR analysis confirmed our assignment of structure 7. It was easily distinguished from regioisomeric structure 7′ (see Scheme 1) based on HMBC correlations.¹⁷ It was more difficult to rule out the possibility of diastereomeric structure 7″, in which the configuration of the cyclohexane ring is reversed relative to the ester stereocenter (Scheme 4). MMFF-based Monte-Carlo conformational searches of 7 and 7″ found the minimum-energy conformations shown in Scheme 4. The depicted structure and conformation of 7 is tentatively assigned based on chemical shift, coupling constant and 800 MHz COSY, TOCSY, HSQC, HMQC and ROESY data. One allylic methylene proton resonates unusually upfield at 0.39 δ, indicating its proximity to the aromatic pi cloud. Importantly, the NOE features highlighted in red would be difficult to observe in 7″. We thus conclude that, to the extent cyclization occurs, it does so regio- and diastereoselectively,¹⁸ according to our predicted model.

Scheme 4. Conformation and NMR Characteristics of 7.

Conversion of the trisubstituted alkene of 7 to a bromohydrin with the correct stereochemistry would give protected bromophycolide D (3). Interestingly, the conformation of 7 is such that the trisubstituted alkene points its methyl “down”, exposing the Re face to solvent. Bromonium formation on this face would give intermediate 20, and subsequent anti attack by water would give the requisite bromohydrin in 3. However, the back side of this brominium is apparently too blocked for water to attack, as treatment of 7 with bromonium sources in aqueous/organic solvent mixtures afforded primarily products resulting from deprotonation of 20.¹⁹

In conclusion, we have developed a very concise asymmetric approach to bromophycolide skeleta, in only seven steps from known geranylgeranyl benzoate 8. Our approach demonstrates a remarkable degree of regio- and diastereocontrol in the differentiation of three nearly identical alkenes within a macrocycle. Advances toward completion of the synthesis will be reported in due course.