A Straightforward Synthesis of Polyketides via Ester Dienolate Matteson Homologation

Abstract Application of ester dienolates as nucleophiles in Matteson homologations allows for the stereoselective synthesis of highly substituted α,β‐unsaturated δ‐hydroxy carboxyl acids, structural motifs widespread found in polyketide natural products. The protocol is rather flexible and permits the introduction of substituents and functionalities also at those positions which are not accessible by the commonly used aldol reaction. Therefore, this ester dienolate Matteson approach is an interesting alternative to the “classical” polyketide syntheses.

The most popular approaches take advantage of asymmetric aldol reactions, in all their variations, to generate directly the polyketide substitution pattern. [10] An alternative approach uses asymmetric allylations/crotylations in combination with oxidative double bond cleavage to generate the same structural motifs. [11] No question, thesep rotocols are straightforward for the syntheses of classical polyketides, but are less suited for posttranslational modified natural products, having O-or Cfunctionalitiesa t" unusual" positions.
Our group is also involved in the synthesis of biological active naturalp roducts, focusing on cyclic peptides and peptide-polyketide hybrids. [12] To become independent of the aldol-motif we developed recently as ynthesis of lagunamide based on aM atteson homologation. [13] This stereoselective prolongation of chiral boronic esters was introduced by Donald Matteson already 40 years ago. [14] Key step of this protocol is the highly stereoselective formationo faa-chloro boronic ester A (Scheme 1) which can be subjected to nucleophilic substitution under S N 2-conditions with aw ide range of nucleophiles, [15] such as Grignard reagents, alkoxides or certain enolates. [16,17] This allows for the stepwise stereoselective incorporationo fs ubstituents and functionalitiesi nto ag rowing carbonc hain. For al ong time most synthetic applicationso f this elegant method were reported by Mattesonh imself, [18] but in the last years it found several nice applications in natural product [19] or drug syntheses. [20] Aggarwal developed av ersion wheret he stereochemical outcome of each prolongation step can be controlled by using sparteinasac hiral ligand. [1 5b, 21] We appliedt he classical Mattesona pproachi nt he synthesis of the polyketide fragment of lagunamide Af rom ester B, [13] wherea ll four stereogenic centers werec ontrolled by the chiral diol in the boronic ester.K ey intermediate was the prolonged a-chloro boronic ester C whichwas oxidized to the cor-   responding aldehyde and subjectedt oaH orner-Wadsworth-Emmons olefination [22] to the desired a-methylated a,b-unsaturated ester D.U nfortunately,t his last step was rather slow,r equiring 4days for completenessa nd providing a4:1 (E/Z)-mixture of the a,b-unsaturated ester. Therefore, we tried to shorten the synthetic sequence significantly with the option to get also access to other stereoisomers for SAR studies. The idea was to stop the Matteson homologation sequence after generation of the first three stereogenic centers and to introduce the "non chiral" part (red) of the polyketide in one step, including the a,ß-unsaturated ester moiety.I nt his case the OH-group is not introduced via S N 2r eaction but by oxidation of the boronic ester and shouldt herefore be obtained with the oppositec onfiguration as before, what makest his approach an interesting complement to the previousp rotocol. As nucleophiles we wanted to use deprotonated butenoic esters, vinylogous enolates whichh aveb een used previously in aldol additions.
In comparison to normale ster enolates, reactions of the vinylogeous enolates are more difficult to control, because besides a-a nd g-p roducts also (E/Z)-isomers of the resulting double bond can be obtained and it is sometimes difficult to separatet he regio-and stereoisomers. While alkylations preferentiallyo ccur at the a-position, [23] the outcome of allylations dependso nt he counter-ion of the dienolate. [24] Li-Dienolates also give rise to a-products, the g-products can be obtained preferentially with Cu-dienolates. Kinetically controlled aldol re-actions yield mainly the a-product, [25] while the g-products are obtained under thermodynamical conditions. [26,27] Mukaiyama aldol reactions using vinylogeouss ilylketenacetals as nucleophiles also give ah igherr atio of g-substitution product, [28] while this approachw as used frequently in polyketide syntheses, [29] for example, in as ynthesis of lagunamide Aa nd related compounds. [30] To the best of our knowledge dienolates have never been applied in Mattesonh omologations so far,a nd only two publicationsd escribe the use of simple ester enolates. [16] To figure out if the dienolates can be used in Matteson homologations at all we investigated the reactiono fc rotylesters with phenyl ethyl boronic ester 1 (Table 1). Firste xperiments were carriedo ut under "typical Mattesonc onditions" using CHCl 2 Li as an ucleophile.C rotonic esters cannot be deprotonated with LDA, because 1,4-addition of the amide is at oo fast process, [29a] but this side reaction can be suppressed by addition of HMPA. [23] To avoid the usage of this nasty reagent we decidedt ou se DMPU (N,N'-dimethylpropylene urea) instead. [31] But no desired homologation product could be observed, only amixture of undefinedproducts.Weassumed, that the reactivity of the a-chloro boronic ester might not be high enough and decided to switch to the corresponding a-bromo derivative. These brominated esters are more reactive, but showa lso ah igher tendency for epimerization, causing products with lower stereoselectivity. [32] Therefore, it is recommended to use such esters withoutp urification and storaget oa void decomposition and epimerization. With a-bromo boronic ester 2,t he desired product could be obtainedi na cceptable yield, althougha samixture of the linear (E)-configured g-product (3a) and the branched a-product (3a')( as diastereomericm ixture) ( Table 1, entry 1). After oxidation, the desired d-hydroxylated unsaturated ester could be obtained in pure form. Ethyl crotonate gave almost the same result as the tert-butylester (entry 2). No reaction was observedw ith silylketenacetals, even in the presence of Lewis acids. Therefore, we decided to investigate the influence of the counter ion, whichh as an influence in aldol reactions. [24] But in our case the addition of copper salts had no influence on the a/g-selectivity,o nly the yield Scheme1.Matteson homologation and application.  3a nd 4). In the presence of magnesium salts no complete conversion could be observed (entry 5) and in all cases more or less 1:1m ixtures of the regioisomersw ere obtained. Therefore, we switched to the tiglic esters 4a-c (R', R'' = H) which give rise to the substitution pattern we were looking for.T oo ur big surprise, with these ester enolates the linear gproducts 5a-c were formed almost exclusively (Table 2, entries 1-3). The yields were almost the same as in the previous example, only in case of the methyl ester 5c the yield wasa little lower (entry 3). In all examples only the formation of the (E)-isomer was observed. To investigate the scope andl imitations of this protocol also substituted tiglic esters 4d-f were evaluated. Introduction of an additional methyl group on the double bond (4d)w as well accepted and the tetra substituted double bond in 5d was obtained in good (E/Z)-ratio of 94:6 (entry 4). Addition of another methyl group at the g-position (4e)h ad no influenceo nt he yield but on the regioselectivity (entries5 and 6). The g/a-selectivity of 5e was significantly lower,c ompared to the examples without substituent on the g-position, but was still in as ynthetically useful range. While the a-productw as an inseparable mixture of diastereomers and (E/Z)-isomers, the g-productw as obtained with good diastereoselectivity,e specially in case of the ethyl ester 5e (entry 5). The product mixturesw ere analyzed after oxidation, the corresponding alcohols could be separated by flash chromatography.I ts hould be mentioned that the yields given in tables 1and 2are isolated yields after chromatography.
All boronic esters 5 were oxidized to the corresponding unsaturated d-hydroxyesters. Depending on the oxidation protocol different products become available. Using "typical" oxidation conditions methyle ster 5c was also hydrolyzed to the free carboxylic acid 6c (Scheme 2). In contrast, under the same conditions tert-butyl esters (5a)a re not affected. Using Na 2 CO 3 as am ilder base also the ethyl ester of 5b remains untouched. In the last two cases the hydroxy esters 6a and 6b could not be separated from the chiral diol used as auxiliary in the boronic ester.But this problem can be solvedbya dding commercially availablem ethyl boronic acidf orming the chiral methyl boronic ester B (Scheme 1), which can be used again in Matteson homologations. In principle, the originalb oronic esters used in the homologations equence can be "recovered", what makest his protocol rather economic.A ll alcohols 6 were obtained with an ee > 90 %c learly indicating that no significant epimerization in the a-bromoboronic ester occurred.
Finally we reacted tiglic ester 4b with as eries of different boronic esters (see Supporting Information) to illustrate the scope and applicability of this protocol (Figure 2). Linear as well as branched and functionalized boronic esters can be used giving yields in the range of 61-78 %, independent of the substitution pattern and the chiral auxiliary used. Only one set of signals are observed in the NMR spectra of the purified products 7,i ndicating ah igh diastereoselectivity.
As ap roof of conceptw ea lso synthesized ap rotected polyketide fragment of epi-lagunamideAstartingf rom known boronic ester 8.H omologation of 8 with CHBr 2 Li followed by reaction with the dienolate of ethyl tiglate provided 9 in high yield, whichw as then oxidized to the desired polyketide fragment 10.M oreover,w er ecovered the chiral auxiliary DICHED after oxidation (Scheme3).
In conclusion, we could show that ester dienolates are good nucleophiles for Matteson homologations. Especially enolates of tiglic esters give excellent regioand stereoselectivities, and in combinationw ith as ubsequent oxidation of the prolonged