Iterative assembly-line synthesis of polypropionates with full stereocontrol

The polypropionate motif is ubiquitous, being characteristic of the most important family of natural products for human health, the polyketides. Numerous strategies have been devised to construct these molecules with high stereocontrol, but certain stereoisomers remain challenging to prepare. We now describe the development of an iterative assembly line strategy for the construction of polypropionates. An assembly line strategy for the synthesis of deoxypolypropionates has already been described. However, the introduction of carbinol units required the development of new building blocks and new reaction conditions. This has been achieved by the use of enantioenriched lithiated α-chlorosilanes [1-((2'-lithiochloromethyldimethylsilyl)-methyl)-2-(methoxymethyl)-pyrrolidine], thus enabling the programmed synthesis of polypropionates in a fully stereocontrolled manner, including the stereochemically challenging anti-anti isomers. The versatility of the approach is exemplified in its extension to the synthesis of 1,3-related polyols. The methodology now allows access to a much wider family of polyketide natural products with stereochemistry being dialled in at will.


INTRODUCTION
Polypropionates are common structural motifs in what is arguably the most important family of natural products, the polyketides. [1][2][3] They have been extensively mined, investigated and exploited as chemotherapeutic agents for the promotion of human health. [4][5][6] Polyketide-derived drugs include antibiotics, cancer chemotherapeutics, immunosuppressants, cholesterol-lowering agents, and antifungals. [7][8][9][10] Their structural and stereochemical complexity, coupled with their important biological activity, have made them attractive targets for over half a century. This intense interest has resulted in the development of important types of methodology for the stereocontrolled synthesis of polypropionates [11][12][13] including Evans' aldol, 14 and more recently, Krische's catalytic crotylation reactions. [15][16] Indeed, the synthesis of highly complex polyketide natural products using Evans' methodology has been one of the major achievements in organic synthesis of the 20 th century. However, this methodology often involved numerous redox processes, which can now be avoided using Krische's 21 st century catalytic crotylations, providing a step change in efficiency. Powerful though these methods are, and despite extensive investigations into the asymmetric synthesis of acyclic molecules, certain structural motifs, for example, the anti-anti isomers, remain difficult to make with high diastereoselectivity. [17][18] In recent years, we have developed iterative synthetic strategies that allow the synthesis of acyclic molecules containing multiple contiguous stereogenic centers through reagent-controlled homologation of boronic esters. This process enabled the conversion of a simple boronic ester into a molecule bearing 10 contiguous methyl substituents with full stereocontrol in an effectively "one-pot" process, without purification of intermediates. 19 Different stereoisomers could be obtained simply by modifying the sequence of chiral reagents added. Moreover, the power of this approach was demonstrated in two short syntheses of polydeoxypropionate natural products, (+)-kalkitoxin and (+)-hydroxyphthioceranic acid (Figure 1a). 20 The core of these complex molecules was constructed by iterative homologations of boronic esters by using chiral lithiated benzoate esters (S/R)-1 and chloromethyllithium 2 21 as key building blocks. As with Krische's and also Negishi's ZACA methodology, [22][23] no redox processes or functional-group interconversions were required between iterations, enhancing the efficiency of the process.
However, polydeoxypropionates represent only a small subset of the vast family of polyketide natural products, the majority containing polypropionates, that is, carbon chains with alternating methyl-and hydroxyl groups. Thus, in order to extend our iterative methodology to the much larger class of polyketide natural products, [24][25][26] new carbenoid building blocks bearing oxygen functionality were required. However, even if a carbenoid could be identified bearing a protected oxygen, boronic esters possessing -ethers present problems with further homologation because the boronate complex is prone to undergo β-elimination rather than the desired 1,2-migration ( Figure 2a). [27][28] To avoid this competing reaction, we considered masking the oxygen functionality as a silyl group. The stereoretentive oxidative cleavage of silicon-carbon bonds, by using reactions conditions developed by Tamao and Fleming, and variants thereof, is well described. 29 Using this approach, we now describe our success in developing an iterative assembly-line synthesis strategy to encompass the synthesis of the much larger class of polyketide natural products, the polypropionates (Figure 1b). By alternating the addition of a novel class of chiral α-silylchloromethyllithium reagents (S/R)-3 with lithiated benzoate esters (S/R)-1 multiple times, simple boronic esters have been transformed into carbon chains bearing alternating hydroxyl and methyl groups. Furthermore, because each iterative homologation shows very high levels of reagent control, and is blind to pre-existing stereogenic centers present in the boronic ester, this methodology enables stereochemistry to be dialled in at will with essentially full control.

RESULTS AND DISCUSSION
We began our studies by identifying a suitable organosilyl reagent for the key stereocontrolled lithiation-borylation reactions. However, like a phenyl group, the silyl group renders an adjacent carbanion configurationally unstable even at low temperature thereby making control of stereochemistry especially challenging. [30][31][32] Indeed, using a silyl-substituted lithiated carbamate generated by the deprotonation of TMSCH2OCb (Cb = N,N-diisopropylcarbamoyl) with s-BuLi in the presence of (−)-sparteine in a lithiation-borylation reaction, the homologated boronic ester was obtained, but in racemic form. 33 Moreover, Blakemore and co-workers recently reported that α-silylmethyllithium carbenoid 4, generated in the presence chiral isopropyl-substituted bis(oxazoline) (BOX) ligand, reacted with phenethyl pinacol boronic ester 6a to give α-silylalkyl boronic ester 7 in 69% yield but with only moderate enantioselectivity (57% ee, Figure 2b). 34 In contrast, related reactions of benzylic organolithiums in the presence of the same chiral bis(oxazoline) ligand have been shown to give high enantioselectivity, [35][36] showing that phenyland silyl-stabilized carbanions behave quite differently. We therefore considered an alternative approach (Figure 2c) based on the work of Chan and co-workers. They reported that lithiated benzylsilane 10 bearing a tethered chiral (methoxymethyl)-pyrrolidinomethyl moiety could be trapped with alkyl halides in good yield and high diastereoselectivity (Figure 2d). 37 In detailed mechanistic studies, Strohmann showed that the alkylation of lithiated benzylsilanes 10 occurs with inversion of configuration. [38][39] We therefore re-designed the α-silyl benzoate ester, as used by Blakemore, to incorporate the chiral (methoxymethyl)pyrrolidinomethyl side-arm. Subjecting the new organosilyl reagent 5 to the lithiation-borylation reaction with a primary alkylboronic ester smoothly provided the desired product 8a in high yield (80%) with high diastereoselectivity ( Figure 2c). Although the reaction worked well with a simple primary boronic ester, we observed no conversion with the more hindered i-propylboronic ester 6b under the same reaction conditions. We therefore explored the use of additives to promote the reaction 40 (see Assuming that lithiated α-silylmethyl benzoate ester 5 was generated with the same sense of diastereoselectivity as that confirmed for Chan's αbenzylsilane 10, its trapping with the boronic ester must have taken place with retention of configuration, thus contrasting with the invertive trapping of organolithium 10 with MeI. 38 We believe that the more pronounced tetrahedral nature of organolithium 5 together with its potential to complex with the oxygen atoms of the boronic ester, thereby directing the reagent to the same face as the lithium atom, accounts for the origin of the retention of configuration observed. 43 In order to promote the 1,2-migration of the intermediate boronate complex 9 and avoid the use of Mg(ClO4)2 we considered replacing the benzoate ester group (OTIB group) with a better leaving group (Cl − ). Indeed, one of the benefits of using the tethered chiral auxiliary approach is that small groups with high leaving-group ability (such as Cl − ) can be incorporated, the directing and stabilising attributes of the benzoate ester group, which are required for sparteine-mediated generation of lithium carbenoids, now being excess to requirements. Thus, the ammonium salt of α-chloromethyl silane 12, was prepared and deprotonated using 2.0 equiv. of s-BuLi in Et2O at −78 °C, followed by treatment with phenylethyl boronic ester 6a to give the homologated αsilylalkyl boronic ester (8a, Table 1) in excellent yield (92%) and very high d.r. In the case of ipropylboronic ester 6b, 1,2-metallate rearrangement occurred while warming to room temperature without the need for additional Lewis acids, thus providing the corresponding product (8b, Table 1) in high yield with complete stereocontrol. As α-chloromethyl silane 12 was clearly the superior reagent, compared to the corresponding benzoate, we tested its scope with a range of boronic esters (Table 1). In general, the homologations of enantioenriched boronic esters with lithiated α-chloromethyl silane 3 proceeded smoothly to provide the corresponding products 8 in high yields with excellent levels of diastereoselectivity. Despite, the limitations in functional-group tolerance that are normally associated with reactions involving organolithiums, for example, the incompatibility of electrophilic carbonyl groups, OH and NH groups, and carbonbased acids (terminal alkynes and carbonyl groups with α-hydrogen atoms), a range of functional groups were tolerated, including alkenes 8c, protected alcohols 8d, tert-butyl esters 8j, and azides 8k. The tolerance of sterically hindered esters and azides, which can react with organolithiums, is attributed to the rapid trapping of the organolithium with the desired boronic ester functional group, as confirmed by the aforementioned ReactIR studies. Additionally, heterocycle-containing substrates, for example, thiophene-, pyridine-and Boc-protected pyrrolidine-containing boronic esters, could be homologated to give the corresponding products, 8f-h, respectively, in good yield and high levels of diastereoselectivity. To explore possible matched and mis-matched effects, the two enantiomers of the α-chloromethyl silane (S,S/R,R)-3 were reacted with three different chiral boronic esters. In all cases the products were obtained in good yield (Table 1, 8i-k) and in only one case was a slight mis-matched effect observed (Table   1, syn-8h vs. anti-8h). These results show that reagent control dominates over substrate control, an essential feature to its broader synthetic utility. The products contain rich functionality as the C-B bond in the resulting geminal borosilanes 8 can be selectively functionalized over C-Si bonds through a wide array of stereospecific transformations 44 , including Zweifel olefination, 41 arylation, 45 and alkynylation 46 giving access to a valuable class of chiral organosilanes.
Having demonstrated the scope of the reaction with different boronic esters, we then turned our attention to developing a protocol for iterative homologation. The building blocks used as the key repeat units were 1) chiral lithiated α-chloromethyl silanes (S,S)-3 and (R,R)-3, in which the silyl moiety is a surrogate for an oxygen atom and 2) chiral lithiated benzoate esters (S)-1 and (R)-1, which were readily available in high e.r. 19 from the corresponding stannanes, which were in turn synthesized using Hoppe-Beak sparteine-mediated lithiation. [47][48][49] We initially targeted the challenging all-anti stereotetrad in our studies.
We began by subjecting the α-silylalkyl boronic ester ent-8a to initial homologation with (R)-1, a reaction that worked well, giving the homologation product 13 in good yield and high d.r. ( Figure   3). However, a subsequent homologation with (R,R)-3 failed. We believe that coordination of the pyrrolidine nitrogen atom to the boron atom of the boronic ester attenuates its reactivity, preventing boronate-complex formation. Indeed, many signals in the 1 H-, 11 B-, and 13 C-NMR spectra of boronic ester 13 were broad indicating nitrogen-boron coordination. We therefore needed to remove the amino group to allow subsequent homologation to take place and considered using photoredox catalysis. Amino silanes have been used as precursors for α-amino radicals, generated through photoredox catalysis, but the focus has always been on the fate of the carbon-centred radical, not the silyl moiety. [50][51] We reasoned that upon oxidation of the amino group, attack of the silyl group by MeOH would give the α-amino radical 22 and the desired methoxysilane (14, Figure 4). We therefore tested the reaction of amino silane ent-8a in the Having demonstrated a highly effective assembly-line synthesis protocol, we sought to target specific diastereoisomers. As illustrated in Table 2 Finally, silane 28 was oxidized to the corresponding 1,3-related tetrols 29 in 56% yield with >95:5 d.r.

CONCLUSIONS
In conclusion, we have demonstrated that the lithiated α-silyl carbenoid 3, which can be generated with high diastereocontrol by incorporating a chelating side-arm, can be successfully employed in the reagent-controlled homologation of boronic esters to provide α-silylalkyl boronates with very high stereocontrol. Coupling this new building block with our established building blocks (lithiated benzoate esters) enables iterative homologation of boronic esters for the rapid and diastereoselective syntheses of polypropionates. This method is highly versatile as different diastereomers can be targeted simply by adjusting the sequence and the configuration of the reagents that are added. For example, the all-anti polypropionate stereotetrad, which remains challenging to synthesize with high selectivity by current aldol or crotylation strategies, could be easily prepared in good yield with excellent stereoselectivity by using our iterative assembly-line synthesis methodology. In addition, the scope of our approach was also extended to the synthesis of 1,3-related polyacetates by using chiral lithiated α-chloromethyl silanes and α-chloromethyllithium as key building blocks. This iterative methodology opens the door for its application, not just to a much wider family of polyketide natural products, but also to polyketideinspired unnatural products using a broader set of building blocks.