Synthetic Studies Toward the Bryostatins: A Substrate-Controlled Approach to the A-Ring

Abstract

The synthesis of a C₁-C₁₃ A-ring subunit of bryostatin 1 is detailed. The key features of the approach include the convergent fragment assembly with a highly stereoselective construction of the C₇-C₈ bond indicated above.

Pettit and co-workers reported in 1982 the isolation and structural identification of bryostatin 1 (1) from the bryozoan Bugula neritina¹. Seventeen structurally related congeners have since been isolated and identified, and the family remains of significant interest to the biological, medical, and synthetic communities.² Bryostatin 1 exhibits an impressive array of biological properties including anticancer activity, synergistic anticancer activity with established therapeutic agents such as vincristine,³ and activity against Alzheimer's disease.⁴ Bryostatin 1 is known to bind to PKCα with nanomolar affinity, but this elicits different biological responses than those associated with binding by the tumor-promoting phorbol esters.⁵ The reasons for these differences and the mechanisms by which bryostatin 1 affects the aforementioned areas of therapeutic interest remain unclear.

In 1990, Masamune and co-workers disclosed the first total synthesis of bryostatin 7.⁶ More recently, both the Evans⁷ and Yamamura⁸ groups have reported bryostatin total syntheses. In addition, the promising biological profile of bryostatin 1, coupled with its scarcity from natural sources, have encouraged a number of other synthetic efforts in this area.⁹ Wender and co-workers have also reported on the synthesis and biological studies of several analogues of bryostatin 1.¹⁰

The synthetic strategy chosen for implementation is detailed in Figure 1. Methodology developed in these laboratories for the construction of 2,6-disubstituted-4-methylene tetrahydropyrans^{11, 9g, 9h} was envisioned to join an A-ring hydroxy allylsilane 2 and C-ring enal 3 with concomitant formation of the B-ring. The A-ring containing the necessary pendant allylsilane could be derived from methyl ester 4 through application of the Bunnelle reaction.¹² Unraveling 4 to linear synthon 5 reveals the C₅ oxygenation to be 1,3-anti to both of the flanking protected hydroxyl groups, which suggested that this hydroxyl stereocenter could be exploited in the installation of both the C₃ and C₇ stereocenters via 1,3-asymmetric induction. Accordingly, a PMB ether was deemed to be an appropriate protecting group for this hydroxyl on the basis of its ability to participate in chelation-controlled processes,¹³ and its ease of removal under very specific conditions. This disconnection would, however, require addition of a nucleophile such as 6 to set the C₇ stereocenter and install the gem-dimethyl moiety. Little precedent exists for such a transformation. In the context of bryostatin synthesis, with the exception of one report,¹⁴ most A-ring approaches commence with material that already contains the gem-dimethyl group. Finally, a Mukaiyama aldol reaction was envisioned for stereoselective introduction of the C₃ stereocenter and the required masked carboxylic acid functionality at C₁.

Figure 1.

Retrosynthetic analysis.

The synthesis of allylstannane 6 commenced with a catalytic asymmetric allylation (CAA)¹⁵ reaction on commercially available α,β-unsaturated aldehyde 9 to afford the desired homoallylic alcohol in exceptional yield and enantioselectivity (Scheme 1). This particular allylation deserves comment. Complete consumption of the aldehyde was observed after just 12 h; typically, the CAA process requires approximately 72 h to reach completion. The rapidity of this reaction is likely to be associated with the electron withdrawing unsaturated ester moiety; i.e., the substrate is a vinylogous glyoxalate. This seemingly superfluous unsaturation was deemed necessary due to previous observations made by Brown and co-workers¹⁶ in which a boron-mediated allylation into the saturated aldehyde corresponding to 9 was accompanied by significant lactonization of the product.

Scheme 1.

Synthesis of Tetrasubstituted Allylstannane 6

After considerable experimentation, conjugate reduction of the α,β-unsaturated ester was accomplished efficiently by application of Semmelhack's Cu(I)/Red-Al^® protocol.¹⁷ Introduction of the gem-dimethyl moiety was accomplished by condensation of ester 10 with acetone to provide tertiary alcohol 11. Elimination of the hydroxyl group by treatment with SOCl₂/pyridine yielded the terminal olefin which subsequently underwent base-mediated olefin migration to afford 12.¹⁸ Full reduction of the ester proceeded without difficulty to afford the corresponding alcohol, setting the stage for installation of the stannyl group. A one-pot mesylation/Bu₃SnLi displacement¹⁹ ensued to provide allylstannane 6. The 37% overall yield for this 8 step sequence provided expedient access to this stannane.

Attention was next turned to the synthesis of the C₁-C₇ subunit. This first required the generation of Mukaiyama aldol substrate 16 (Scheme 2). A CAA reaction was relied upon once again to provide homoallylic alcohol 14 in 90% yield and 93% ee. Conversion of the alcohol to PMB ether 15 was accomplished by reaction with p-methoxybenzyl trichloroacetimidate and catalytic CSA. It is worth noting that unavoidable silyl migration was observed under the typical KH/PMBBr conditions. Additionally, the use of CSA was found to offer superior results than those obtained with other commonly employed acids such as TfOH, PPTS, and BF₃·OEt₂. Deprotection of the primary TBDPS ether with TBAF and subsequent Parikh-Doering oxidation gave the requisite aldehyde 16 in 92% yield over the two reactions.

Scheme 2.

Synthesis of Aldehyde 16

A variety of Lewis acids and conditions were screened in the Mukaiyama aldol reaction (Table 1). The use of MgBr₂·OEt₂ afforded moderate diastereoselectivities for this transformation (entries 1 and 2). A slight improvement in aldehyde facial selectivity was observed when the monodentate Lewis acid BF₃·OEt₂ was employed (entry 3). While it is noteworthy that the aldehyde proved stable to exposure to TiCl₃(Oi-Pr), no enhancement in selectivity resulted from the use of this Lewis acid (entry 4). A dramatic improvement in selectivity was realized when TiCl₂(Oi-Pr)₂ was used in place of TiCl₃(Oi-Pr), but the conversion was modest as 40% of aldehyde 16 was recovered (entry 5). However, a further increase in aldehyde facial selectivity and a significant improvement in conversion was observed when the number of equivalents of the mixed titanium Lewis acid was increased (entry 6). Optimization of this reaction revealed that the use of 2.5 equivalents of the TiCl₂(Oi-Pr)₂ afforded a 95% yield of aldol adduct 17, as a 41:1 mixture of diastereomers, as ascertained by HPLC analysis (entry 7).²⁰ The suspected relative stereochemical relationship between the C₃ and C₅ stereocenters was confirmed by application of Rychnovsky's acetonide NMR method.²¹

Table 1.

Mukaiyama Aldol Survey

entry	Lewis acid	equiv	temp (°C)	solvent	dr
1	MgBr2·OEt2	2.0	−20	CH2Cl2	4 : 1
2	MgBr2·OEt2	2.0	−78 to −20	CH2Cl2	4.5 : 1
3	BF3·OEt2	1.1	−78	CH2Cl2	5 : 1
4	TiCl3(OiPr)	1.0	−78	PhCH3	5 : 1
5	TiCl2(OiPr)2	1.0	−78	PhCH3	32 : 1a
6	TiCl2(OiPr)2	2.0	−78	PhCH3	37 : 1b
7	TiCl2(OiPr)2	2.5	−78	PhCH3	41 : 1c

^a40% of 16 recovered.

^b10% of 16 recovered.

^c95% of 17 isolated.

Preliminary coupling studies suggested that the C₃ hydroxyl protecting group might remotely influence the facial bias in the stannane addition to the aldehyde. Thus, two differentially protected aldehydes were synthesized in preparation for the coupling studies (Scheme 3). Secondary alcohol 17 was protected as the TBS ether by treatment with TBSOTf and lutidine. Ozonolytic cleavage of the terminal olefin provided aldehyde 18 in an 82% yield over the two steps. The analogous C₃ TBDPS protected aldehyde was synthesized by silylation with TBDPSCl followed by OsO₄/NMO dihydroxylation and cleavage of the resulting diol by Pb(OAc)₄.²² Aldehyde 19 was accessed in 91% yield from alcohol precursor 17.

Scheme 3.

Synthesis of Two Aldehyde Coupling Partners

With both aldehyde and stannane coupling partners in hand, attention was directed towards the critical coupling reaction. Surprisingly, both MgBr₂·OEt₂ and TiCl₂(Oi-Pr)₂ failed to promote this addition. It was reasoned that aldehyde 18 may need increased activation in order for the addition of the relatively hindered stannane 6 to occur. Dimethylaluminum chloride appeared to be a suitable Lewis acid candidate to explore as it is recognized for its exceptional chelating ability.²³

In the event, subjection of aldehyde 18 to Me₂AlCl (2.5 equiv) in CH₂Cl₂ gave the desired addition product 20 in good yield but with modest 3.5:1 diastereoselectivity (Table 2, entry 1).²⁴ An increase in the amount of Lewis acid used to 5.0 equivalents improved the level of stereoselectivity observed (entry 2). Reactions using the TBDPS ether containing aldehyde 19 confirmed our suspicions that the C₃ protecting group might influence the stereochemical outcome; under identical conditions to those employed in entry 1, the desired adduct was obtained in comparable yield but with 7:1 diastereoselectivity (entry 3). A dramatic improvement in diastereoselectivity was realized by a change of solvent (entry 4). When the same reaction was performed in toluene a single addition product was obtained in 79% yield. As entry 5 demonstrates, when the reaction was carried out using 5.0 equiv of Me₂AlCl in toluene, an 88% yield of 21 resulted under these optimized conditions. The expected stereochemistry was corroborated via NOE data obtained after closure of the A-ring (Scheme 4).

Table 2.

Optimization of the Coupling Reaction

entry	R	equiva	solvent	yield (%)b	dr
1	TBS	2.5	CH2Cl2	83	3.5 : 1
2	TBS	5.0	CH2Cl2	73	7 : 1
3	TBDPS	2.5	CH2Cl2	78	7 : 1
4	TBDPS	2.5	PhCH3	79	single isomerc
5	TBDPS	5.0	PhCH3	88	single isomerc

^aRefers to Lewis acid.

^bAll reactions were performed with 1.3 equiv of stannane, yield based on aldehyde.

^cBy ¹H and ¹³C NMR analysis.

Scheme 4.

Completion of the A-ring Synthesis

The only remaining tasks were formation of the A-ring and elaboration of the terminal olefin to the methyl ester. Toward this end, acylation of the newly formed hydroxyl group was accomplished by exposure to Ac₂O and DMAP. It was anticipated that oxidative cleavage could be carried out simultaneously on both the C₉ and terminal olefins in order to minimize the number of chemical operations required to reach the A-ring target. Unfortunately, no conditions were found which would effect this transformation. Independent oxidative operations on the olefins were thus carried out as follows. Dihydroxylation of the terminal olefin followed by NaIO₄ cleavage of the resulting diol was accomplished in good yield to provide aldehyde 22. No product resulting from the dihydroxylation of the C₉ olefin was detected, undoubtedly as a consequence of the considerable steric demand imposed the proximal gem-dimethyl group. Pinnick oxidation²⁵ of the aldehyde and methylation of the resulting carboxylic acid with (trimethylsilyl) diazomethane provided methyl ester 5 in 91% yield over the two steps. With the methyl ester in hand, oxidative cleavage of the PMB group was accomplished in 91% yield by reaction with DDQ. Finally, ozonolysis of the remaining olefin afforded a mixture of ketol and open-chain keto-alcohol. This equilibrating mixture underwent conversion to the cyclic methyl ketal with concomitant deprotection of the TBS ether under acidic methanolic conditions to give A-ring target 4 in 55% yield from alkene 5.

In conclusion, A-ring subunit 4 was accessed from aldehyde 13 in 17 linear steps in 15% overall yield. This connective fragment assembly approach provides an efficient means for introduction of both the gem-dimethyl group and the C₇ stereocenter in a highly stereoselective manner. Efforts to utilize this approach in a total synthesis program are in progress.