Collective Total Synthesis of Casbane Diterpenes: One Strategy, Multiple Targets

Abstract Of the more than 100 casbane diterpenes known to date, only the eponymous parent hydrocarbon casbene itself has ever been targeted by chemical synthesis. Outlined herein is a conceptually new approach that brings not a single but a variety of casbane derivatives into reach, especially the more highly oxygenated and arguably more relevant members of this family. The key design elements are a catalyst‐controlled intramolecular cyclopropanation with or without subsequent equilibration, chain extension of the resulting stereoisomeric cyclopropane building blocks by chemoselective hydroboration/cross‐coupling, and the efficient closure of the strained macrobicyclic framework by ring‐closing alkyne metathesis. A hydroxy‐directed catalytic trans‐hydrostannation allows for late‐stage diversity. These virtues are manifested in the concise total syntheses of depressin, yuexiandajisu A, and ent‐pekinenin C. The last compound turned out to be identical to euphorhylonal A, the structure of which had clearly been misassigned.


Introduction
Thebicyclo[12.1.0]pentadecane framework of the casbane diterpenes represents the formal progenitor of the more complex polycyclic skeletons of the lathyrane,j atrophane, tigliane,d aphnane and ingenane families. [1,2] Despite their role as acentral node within this large biosynthetic network, casbane derivatives themselves are rare:the flowering plants of the genus Euphorbia and the biologically totally unrelated soft corals of the genus Sinularia are by far the major sources. From this small number of producing organisms,h owever, more than 100 different structural variants have been isolated to date,which feature abewildering diversity of oxygenation patterns and stereochemical permutations along the hydrocarbon rim.
Thef ew examples shown in Figure 1a re representative. Most naturally occurring casbane diterpenes comprise a cisdisubstituted cyclopropane unit, which appears in both antipodal formats (compare the depressin derivatives 2, 3 [3] with the "pseudoenantiomeric" compounds 6, [4] 7 [5] ). The same is true for the trans-cyclopropane series,asillustrated by 9 and 10,which are isomeric even though they derive from the same soft coral. [3] It is of note,h owever,t hat the absolute configuration of several casbanes was either not addressed by the isolation teams or has only been inferred by plausibility arguments,e specially in the older literature.A nother common structural variation concerns the oxygenation pattern of the "northern" sector, in that the C5 methylene group is frequently oxidized to the alcohol or ketone level, with or without concomitant oxygenation of the surrounding positions.O nce again, the stereochemistry is evidently not uniform as the comparison of 12 [6] and 13 [7] suggests,w hich supposedly have opposite configurations at C5 (in other cases such as 4, [8] the configuration of this center is simply unknown). Moreover,t he parent C18 methyl substituent can be transformed to the alcohol (11), [6] aldehyde (4, [8] 12, [6] 13 [7] )o rc arboxylic acid level (e.g.y uexiandajisu A, which is either 14 or ent-14). [9] Additional diversity is gained by isomerization (8, [10] 13 [7] ), (partial) saturation (9) [3] and/or epoxidation (5) [11] of the trisubstituted alkenes and/or by additional oxygenation of the "lower" sector of the constituent framework. [12] Thef act that casbane diterpenes were evolved by sessile marine sponges and, in parallel, by higher plants is fascinating and suggestive at the same time.I nt his context, it is worth remembering that Euphorbia plants (the "spurge" family) are commonly considered toxic;s ome of them play important roles in traditional Chinese medicine ("Lang Du") for the treatment of various ailments. [13] In line with this notion, pekinenin C( 12)w as shown to cause cell cycle arrest in the G0/G1 phase and apoptotic cell death via the mitochondrial as well as the death-receptor pathways. [14] Equally noteworthy is the role of certain casbanes as phytoalexins:w hile the amount of 1 in seedlings of the castor bean (Rhicinus communis)i sm inute, [15] exposure to certain fungi massively upregulates its biosynthesis as part of the plantsd efense mechanism. [16,17] Along the same lines, ent-10-oxodepressin (7)a saprototype member of the "enantiomeric" series was recognized as phytoalexin of rice. [5] This report also marks arare case of acasbane derivative isolated from ahigher plant not belonging to the Euphorbia genus;i nv iew of the importance of rice for the sustention of humankind, the discovery of the potential fungicidal properties of 7 is arguably relevant and warrants further scrutiny. Thei ncreasing awareness of the possible biological roles of the casbane diterpenes is not yet reflected in advanced synthetic studies.Sofar,the eponymous hydrocarbon 1 is the only member that has been targeted in the past:f ive independent chemical syntheses and an enzymatic approach were published prior to the turn of the millennium; [18][19][20][21][22][23] all activity seems to have ceased since then. Although adetailed analysis of this prior art is beyond the scope of this Research Article,i ti sn oteworthy that all approaches-except for the enzymatic route [23] -struggled with the macrocyclization step, which was low yielding in most cases.Whatismore,either the ring closure itself or the necessary functional group manipulations in its aftermath invariably led to isomer mixtures that required tedious separation. Any new foray must address these issues;m ost importantly,p ositional and/or configurational double bond isomerism should be strictly avoided. Since the known chemical syntheses gave racemic 1 unless the "chiral pool" was tapped (cis-chrysanthemic acid, (+ +)-2carene), [24] it is also appropriate and timely to envisage ac atalytic enantioselective route.I nr ecognition of the fact that some of the oxygenated casbane derivatives are of larger biological significance than the parent compound itself,ahigh level of synthetic flexibility is also deemed necessary;f rom the conceptual viewpoint, any (late-stage) diversity as integral part of the synthesis blueprint would mark anotable strategic advance. [25][26][27]

Results and Discussion
We conjectured that the trans-hydrometalation of internal alkynes,which has recently become afocal point of research in our laboratory, [28] provides valuable opportunities in this context (Scheme 1). Specifically,t reatment of propargyl alcohols with R 3 EÀH( E = Si, Ge,S n) in the presence of [Cp*RuCl] 4 as catalyst follows astereochemically unorthodox trans-addition mode and delivers the R 3 EÀ group with high fidelity to the proximal Catom of the former triple bond. This outcome is mechanistically well understood;i ne ssence,i t originates from the cooperativity between the unprotected ÀOH group of the substrate and the polarized [RuÀCl] unit of the catalyst. [29] When applied to ac ycloalkyne of type F,t he resulting alkenylstannane (germane,s ilane) G should allow either am ethyl, ah ydroxymethyl, af ormyl, or ac arboxyl group to be stitched to the C4 position of the casbane framework where such variation is common (see above). Since the directing C5ÀOH group can either be kept as such or be oxidized to the corresponding ketone (or even be removed), all basic oxygenation patterns in the "northern" sector should be accessible from as ingle platform. Cycloalkyne F,i nt urn, might be forged by ring closing alkyne metathesis (RCAM): [30][31][32] since the best available catalysts activate triple bonds exclusively while leaving all sorts of double bonds untouched, the integrity of the trisubstituted alkenes punctuating the carbon skeleton of F seems guaranteed. Thesame,however, cannot be taken for granted for the cyclopropyl group:any alkyne metathesis catalyst is aSchrock alkylidyne,w hich is nucleophilic at carbon. [33] This inherent polarization might entail (reversible) ring opening and/or decomposition, especially if forcing conditions are needed (see the Insert in Scheme 1): this is likely to be the case in the projected application given the strain of the resulting 14membered ring F that incorporates two E-configured alkenes as well as the linear alkyne;t he annelated cyclopropyl ring rigidifies the skeleton even further. Therefore,anapplication to the casbane series will also clarify an as yet unknown aspect of alkyne metathesis chemistry in general. Further analysis reveals that the building block C can be used twice en route to F,w hich reduces the synthetic exertion and helps render the route practical. Thec yclopropyl fragment B can be traced back to lactone A,w hich is available by rhodium catalyzed intramolecular cyclopropanation: [34,35] since this step proceeds under catalyst control and the elaboration of A into B can be performed with or without epimerization of aderived aldehyde intermediate,a ll possible isomers are accessible. Theb lueprint shown in Scheme 1h ence combines flexibility at the outset with late-stage diversity and should therefore bring ac onsiderable number of casbane derivatives into reach.
In the forward sense,d iazoester 15 was prepared by aliterature route from prenyl alcohol and diketene followed by diazo transfer (Scheme 2);i ts treatment with catalytic Rh 2 (S-MEPY) 4 furnished multigram amounts of 16 with high optical purity (93 % ee). [34,35] After reduction of the lactone with Dibal-H in CH 2 Cl 2 at low temperature,t he resulting crude lactol was subjected to Wittig olefination to give alcohol 17, [36] which was then oxidized to the corresponding aldehyde 18.Itisatthis point that the synthesis route bifurcates for the first time:w hen 18 was subjected to regular Corey/Fuchs alkynylation [37] and C-methylation under optimized conditions, [38] the cis-configured enyne 19 was obtained. Stirring of as olution of 18 in MeOH with K 2 CO 3 ,h owever,c auses equilibration with formation of the thermodynamically more stable trans isomer and hence opens access to the epimeric trans-configured building block 20 (dr ! 9:1). [39] Thes omewhat modest yields of 19 and 20 solely reflect the high volatility of these hydrocarbons.T he enantiomeric compounds are obviously equally well accessible. [40] Fragment 24 to be used twice during the synthesis was readily obtained on treatment of the lithium salt of 3-pentyn-1-ol (21)with the cyanocuprate derived from PhMe 2 SiLi and CuCN (Scheme 3). [41] Alternatively,t his very same product can be made from 2,3-dihydrofuran 22 by metalation/silylation followed by nickel catalyzed opening of the resulting compound 23 with MeMgBr. [42] Whereas the first method works well on a 3gscale and takes as ingle step,t he twostep procedure is more practical when it comes to making much larger quantities.Oxidation of 24 to the corresponding aldehyde followed by addition of propynylmagnesium bromide gave 26.A lthough it would be fairly straightforward to set this propargyl alcohol center in isomerically pure form, [43] no effort in this direction was undertaken at this stage:inview of the uncertainty concerning the configurational assignment of the C5ÀOH group in various casbane derivatives,access to both isomers was probably necessary anyway.C ompound 26 was then O-silylated prior to iododesilylation on reaction with NIS in CH 2 Cl 2 at À20 8 8Ci nt he presence of hexafluoroisopropanol (HFIP) and 2,6-lutidine following ap rocedure developed by Zakarian and co-workers. [44] Under these conditions,t he yield of alkenyl iodide 27 was high and no sign of double bond isomerization was observed within the limits of detection (E:Z ! 95:5, 1 HNMR). This product was then subjected to ap alladium catalyzed Negishi coupling [45] with the organozinc reagent derived from iodide 25, [46] which also derives from 24.S omewhat unexpectedly,h owever,t he iododesilylation of the resulting product 28 did not proceed well under the conditions that had worked so smoothly for the shorter building block 26; [44] several alternative methods were also tried but equally found low-yielding,e rratic, or unselective. [47] Since (inseparable) isomer mixtures need to be strictly avoided for the sake of practicality,considerable optimization was necessary at this stage.P art of the problem is the poor solubility of 28 in the polar medium comprised of HFIP as (co)solvent. Therefore we considered the cleavage of the TBDPS ether prior to iododesilylation, even though this order of the events bears risk and has hardly any precedent: [48] upon addition of an [I + ]s ource to the resulting polyunsaturated substrate 29,t he unprotected ÀOH group might interfere by attacking any transient iodonium ion and thus entail iodoetherification. This was indeed observed when the reaction was carried out with NIS in HFIP:the cyclic ether 30 was formed as am ixture of two diastereomers.A ddition of HOAc, [49] however, allowed the problem to be solved and the desired product 31 to be obtained in well reproducible 70 % yield with impeccable integrity of the double-bond geometry (E:Z ! 95:5, 1 HNMR). Although the generality of this finding needs yet to be explored in detail, it suggests that these modified Zakarian conditions might further increase the scope of iododesilylation and render the reaction applicable to otherwise "protecting group free" synthetic ventures.
Thec oupling of the building blocks commenced with an exquisitely chemoselective hydroboration of the terminal alkene in 19 in the presence of the internal alkyne,which was achieved on treatment with 9-H-9-BBN dimer in THF;t he choice of solvent is critical (Scheme 4). [50,51] Ther esulting alkylborane 32 was then subjected to aS uzuki coupling [52] with alkenyl iodide 31 using Ba(OH) 2 ·8 H 2 Oa sp romotor; [53,54] other bases gave inferior results.S ince the latest generations of alkyne metathesis catalysts endowed with (chelating) silanolate ligands work even in the presence of unprotected alcohols despite the inherent nucleophilicity/ basicity of the operative Schrock-type alkylidyne unit, [55,56] diyne 33 thus formed could be directly subjected to RCAM to give the cycloalkynes 34 and 35.The fact that the reaction had to be carried out in refluxing toluene is tentatively ascribed to the high ring strain of these compounds;t his aspect and the rather forcing conditions notwithstanding, the cyclopropyl unit survived. At this stage,t he two diastereomeric alcohols were easily separated by flash chromatography and the structure of 34 in the solid state was confirmed (Figure 2). It shows that the cis configuration of the three-membered ring is intact:neither cleavage nor scrambling had interfered during the metathesis of the adjacent triple bond.
With 34 and 35 in hand and the configuration of the C5À OH group rigorously established, the project reached the phase of late-stage diversification. To this end, 35 was subjected to ruthenium catalyzed trans-hydrostannation to give product 36 as the only detectable isomer in good yield. [29,57] Af ormal Stille cross-coupling with MeI under conditions previously developed in our laboratory allowed  the yet missing methyl substituent to be introduced, [58,59] before the resulting product was oxidized to the corresponding ketone 2.The NMR spectra as well as the specific rotation of synthetic 2 are in excellent agreement with those of depressin reported in the literature. [3] We hence confirm the constitution and absolute configuration assigned to this natural product derived from the soft coral Sinularia depressa collected off the shore of Hainan Province,C hina. [60] Next, stannane 36 was transformed into the corresponding aldehyde 4 upon treatment with MeLi (2 equiv) followed by quenching of the resulting organolithium species with DMF. [61] Compound 4 could represent euphorhylonal A, an atural product extracted from Euphorbia hylonoma;t he spectra of 4,h owever,w ere not matching.S ince the configuration of euphorhylonal AatC5had not been established by the isolation team and is therefore an obvious point of concern, [8] thee pimeric cycloalkyne 34 was subjected to an analogous sequence of trans-hydrostannation/formylation. Yet, the recorded spectra of the resulting aldehyde 5-epi-4 were not in accord with the literature either.I tw as hence clear that euphorhylonal Adoes not contain a cis-configured cyclopropane as proposed in the literature; [8] unfortunately, the published data do not provide any hint as to which of the two positions on the three-membered ring might have been misassigned and needs to be inverted. [8] Since we had planned aforay into the trans series anyway in order to showcase the flexibility of the synthesis blueprint, pekinenin C( 12) [6] isolated from the roots of Euphorbia pekinensis was the obvious target because it differs from nominal eyphorhylonal A( 4)s olely in the configuration of one center on the cyclopropane ring. This compound exhibits appreciable cytotoxicityr ooted in its ability to induce apoptotic cell death. [14] Thes ynthesis follows the established route in that the appropriate enyne 20 (see Scheme 2) was subjected to chemoselective hydroboration followed by crosscoupling of the resulting alkylborane with 31 to give diyne 39 in readiness for RCAM (Scheme 5). Using the same catalyst generated in situ from complex 37 and ligand 38, [55] the macrocyclization proceeded within minutes at 70 8 8C; the comparison with the cyclization of diyne 33,w hich had required reflux temperature,s uggests that the trans-config-ured cyclopropane imposes less strain on the transition state and the resulting bicyclicscaffold of 40 and 41.That the threemembered ring remained intact during the alkyne metathesis reaction echoes the results discussed above.O nce again, the diastereomeric alcohols could be readily separated at this stage;t he configuration of the C5ÀOH was determined by Mosher ester analysis (see the SI). Each isomer was then subjected to trans-hydrostannation and subsequent formylation to give isomers ent-12 and 44,r espectively. [62] Gratifyingly,t he spectral data as well as the specific rotation of synthetic ent-12 correspond well to those of euphorhylonal A; [8] the NMR spectrum is also identical with that of pekinenin C, whereas the sign of the [a] D is opposite. [6] We hence conclude that euphorhylonal Aa nd pekinenin Ca re not diastereomeric but almost certainly enantiomeric to each other. [63] This finding shows that phylogenetically closely related Euphorbia plants can produce antipodal secondary metabolites.

Conclusion
In summary,w eh ave developed as hort, efficient, selective,m odular, and inherently flexible approach to the casbane family of diterpenes,a si llustrated by the total synthesis of depressin, yuexiandajisu A, and ent-pekinenin C; the latter proved identical with natural euphorhylonal A which had obviously been misassigned by the isolation team. Thec hosen route also clarifies ap reviously unknown aspect of alkyne metathesis in that it proves that acyclopropane ring conjugated to the triple bond to be activated and catalytically transformed remains intact even under forcing conditions. Finally,t he present study further illustrates the virtues of directed trans-hydrometalation, areaction manifold that had been largely unknown until recently but proves versatile and enabling,n ot least as an entry point into av ariety of stereodefined trisubstituted alkenes. [28, 59,60a,64] Further studies along these lines,including approaches to even more complex target compounds,are underway and will be reported in due course.