trans-Hydrogenation: Application to a Concise and Scalable Synthesis of Brefeldin A

The important biochemical probe molecule brefeldin A (1) has served as an inspirational target in the past, but none of the many routes has actually delivered more than just a few milligrams of product, where documented. The approach described herein is clearly more efficient; it hinges upon the first implementation of ruthenium-catalyzed trans-hydrogenation in natural products total synthesis. Because this unorthodox reaction is selective for the triple bond and does not touch the transannular alkene or the lactone site of the cycloalkyne, it outperforms the classical Birch-type reduction that could not be applied at such a late stage. Other key steps en route to 1 comprise an iron-catalyzed reductive formation of a non-terminal alkyne, an asymmetric propiolate carbonyl addition mediated by a bulky amino alcohol, and a macrocyclization by ring-closing alkyne metathesis catalyzed by a molybdenum alkylidyne.

constituents into the endoplasmatic reticulum. This massive but reversible morphological change is caused by binding of 1 to ap rotein complex consisting of ac atalytic guanine exchange factor (GEF) and the small Gprotein adenosine ribosylation factor 1( ARF1), which exerts key regulatory functions for vesicle budding and transport. [3,4] Tw oi ndependent crystal structures showed that 1 inserts in aw edgelike manner at the interface of these proteins and thereby brings the GDP/GTP exchange critical for the proper functioning of the ARF1 GTPase to ahalt. [5] Equally rich is the synthetic record of brefeldin. More than 40 different strategies in pursuit of 1 or its less potent sibling 2 have been described over the past four decades. [6][7][8][9][10][11][12] Although many original solutions were found, several recurring themes can also be noticed in this impressiveb ody of work. Them ost obvious one is the enduring dominance of macrolactonization for the formation of the 13-membered ring. Only af ew macrocyclizations through CÀCb ond formation have been pursued with varying success, [8] with ring-closing olefin metathesis (RCM) [13] at the D 10, 11 bond being the only catalytic method applied to date. [9] Since the current state of the art does not allow E-selectivity to be imposed on RCM by catalyst control, [14] the observed isomer ratios were case dependent and typically unfavorable.
Other groups chose to set the embedded E-olefins more concisely,f or instance through the trans-reduction of an appropriate alkyne precursor.W ith one exception, where atwo-step protocol of trans-hydrosilylation/proto-desilylation was pursued to form the enoate motif of 1, [10] they all resorted to the use of alkali metals in liquid ammonia. [11] Because of the harsh conditions,t his methodology necessitates considerable oxidation state and protecting group management en route to the final product and therefore needs to be carefully timed. We felt that the procedure for direct alkyne trans-hydrogenation recently disclosed by our group provides al arger window of opportunity and should qualify for applications to polyfunctional compounds where Birch reduction has no bearing. [15,16] Since this emerging methodology is as yet hardly understood and has never been applied to natural product chemistry,alate-stage implementation into aroute to 1 might help in scouting the strategic assets of this method, as well as any possible pitfalls.
Finally,aliterature survey showed that the amounts of brefeldin A(1)formed de novo in the numerous campaigns of the last four decades were minute and mostly in the singledigit milligram range,where documented. [6,17,18] Although 1 is accessible by fermentation, [19] this status quo is deemed inadequate by todays standards in the field of target-oriented synthesis. [20] Therefore we felt encouraged to pursue this prominent target once again, hoping that an ew route based on alkyne trans-hydrogenation would lead to am ore satisfactory solution.
The meso-diester 3 served as ac onvenient point of departure and was desymmetrized on al arge scale through as emi-hydrolysis catalyzed by pig liver esterase (Scheme 1). [21,22] Ther esulting mono-acid 4,w hich is also commercially available,w as converted into lactone 5 (ee = 96 %) prior to oxidative cleavage of the double bond. [22] An intramolecular Claisen condensation/decarboxylation sequence transformed the tricarbonyl compound 6 into product 7 in one operation. [11c] Although the yield was somewhat scale-dependent, multigram amounts of ketone 7 were procured upon slight modification of the literature procedure.I ts annulated bicyclics keleton renders the catalytic hydrogenation of the carbonyl group over platinum on charcoal rigorously stereoselective; [23] this step was basically quantitative provided that the medium was supplemented with NaOAc to avoid elimination of the nascent hydroxyl group.After TBS protection, lactone 9 was converted into the methyl-capped alkyne 11 by an iron-catalyzed reductive alkylation recently developed in our laboratory. [24] To this end, 9 was reacted with PPh 3 /CCl 4 and the resulting dichloroolefin 10 treated with MeLi in the presence of catalytic amounts of [Fe(acac) 3 ]a nd ortho-phenylenediamine to furnish the desired product 11 in 55 %y ield on a3gram scale (single largest batch). Not unexpectedly,i ntermediate 10 is sensitive and should be used without delay. [24] Moreover, inadvertent cleavage of the TBS-ether led to spontaneous addition of the alcohol across the activated dichloroalkene bond with formation of the stable tricyclic cage structure 12 ( Figure 1). [23] Upon consideration of these peculiarities,how-ever,t he iron-catalyzed alkyne formation nicely secured as ubstantial material supply.
Next, compound 11 was oxidized and the resulting aldehyde isomerized to the thermodynamically more stable trans-configured product 13 on exposure to K 2 CO 3 in MeOH (Scheme 2). Thef ollowing addition of the readily prepared propiolate 16 required careful optimization. Of the various procedures investigated, the method developed by Kojima and co-workers was the most practical in that only 1.5 equivalents of 16 were needed to reach full conversion of 13. [25] For high diastereoselectivity however,t he steering aminoalcohol 24 described in the literature had to be supplanted by the more bulky analogue 25,w hich furnished the desired adduct 17 with ad.r. of > 95:5. Subsequent reduction with Red-Al at low temperature [26] followed by TBS protection of the resulting allylic alcohol gave diyne 18 in readiness for macrocyclization.
This transformation was accomplished with the aid of the molybdenum alkylidyne complex 26, [27] which is arguably the most active and selective catalyst for alkyne metathesis known to date. [28] Although this catalyst is,apriori, operative at ambient temperature,t he formation of 19 required gentle heating, which is thought to reflect the strained nature of the incipient cycloalkyne. [29,30] With this proviso,t he macrocyclization proceeded well on a1.25 gscale (single largest batch).
The 1 HNMR spectrum of 19 (CDCl 3 )i sdistinguished by ar emarkable downfield shift of the enoate proton H3 (d = 7.28 ppm), which is attributed to deshielding by the anisotropy cone of the acetylene unit. If this is the case,t he compound must adopt arigid conformation that holds the two p-systems in close transannular proximity.T he structure of the derived diol 20 in the solid state confirmed this interpretation ( Figure 2): [23] it shows H3 to be inward oriented, directed towards the triple bond, and positioned slightly below the plane of the macrocyclic scaffold;t ight contacts with C10/C11 indicate significant transannular strain.
With an appreciable amount of cycloalkyne 19 in hand, the stage was set for the crucial trans-hydrogenation. Our model studies had identified [Cp*Ru(cod)Cl] as ag ood catalyst for this unorthodox transformation. [15] In fact, this complex furnished E-21 with excellent selectivity (E:Z >

Angewandte
Chemie 95:5) but resulted in substantial overreduction ( 40 %). Although this outcome remains unexplained at this point, we have reason to believe that the strained nature of this particular substrate and the presence of asecond coordination site for the active ruthenium center in close transannular proximity to the triple bond render the reduction of 19 particularly challenging. [31] Gratifyingly though, overreduction became av ery minor issue (< 5%)w hen [Cp*Ru-(MeCN) 3 ]PF 6 was used instead, which furnished E-21 in stereochemically pure form (E:Z > 99:1). Thee qually reducible enoate moiety was not touched to any noticeable extent nor was the lactone cleaved by the Lewis-acidic catalyst species generated in situ;n either functional group would subsist under Birch conditions. [32] However,s ome isomerization of the newly formed disubstituted double bond in 21 into at hermodynamically more favorable trisubstituted position at the ring junction (22)o rw ithin the five-membered ring (23)could not be suppressed. Although the presence of these isomers rendered product isolation more demanding,geometrically and positionally pure 21 was secured in appreciable 56 %yield when the reaction was performed on a > 1g scale.S tandard deprotection then furnished brefeldin A( 1 ) as ac olorless crystalline material. Its integrity and identity were confirmed by spectroscopic means as well as X-ray diffraction. [23] Then ew route to brefeldin A( 1 )o utlined above is no more than par with the shortest previous syntheses of this target in terms of step count. [6] However,itisdeemed competitive and arguably highly practical and therefore constitutes ap otentially relevant entry to the debate about synthetic efficiency in general. Most notably,a ll critical steps are under rigorous catalyst control. Likewise,the great share of catalysis was instrumental for the ready adaptation to the (multi)gram scale;t hus,s ubstantially more material was prepared than in any of the numerous campaigns described in the literature (for which the throughput has been properly documented). To this end, it was essential that catalytic ring-closing alkyne metathesis once more proved itself ac onvincing alternative to the previously executed macrocyclization reactions,b et hey based on CÀCb ond formation or traditional lactonization. Finally,the first late-stage implementation of ad irect alkyne trans-hydrogenation illustrates another recent advance in catalysis that allows chemoselectivity problems,f or which the established stoichiometric repertoire has no adequate answer, to be solved. At the same time,however,the present case also shows that abetter understanding of this still enigmatic process is necessary to avoid issues with possible alkene isomerization and overreduction. Work along these lines is ongoing in our laboratory.